diff options
| author | Hiroya Fujinami <make.just.on@gmail.com> | 2025-07-31 13:08:54 +0900 |
|---|---|---|
| committer | nagachika <nagachika@ruby-lang.org> | 2025-11-02 14:05:12 +0900 |
| commit | e02c892ade90ee401daf26292fbb99d32af7f619 (patch) | |
| tree | 86b15624967c622cde3621f9aa2321054ae46268 | |
| parent | 30f661951bb19f69cf55388c76ecaf7b0be41079 (diff) | |
Fix indents in Onigmo files to use spaces instead of tabs (#14047) [no ci]
| -rw-r--r-- | regcomp.c | 2272 | ||||
| -rw-r--r-- | regenc.c | 84 | ||||
| -rw-r--r-- | regenc.h | 8 | ||||
| -rw-r--r-- | regerror.c | 84 | ||||
| -rw-r--r-- | regexec.c | 2652 | ||||
| -rw-r--r-- | regparse.c | 2558 | ||||
| -rw-r--r-- | regparse.h | 4 |
7 files changed, 3834 insertions, 3828 deletions
@@ -640,17 +640,17 @@ compile_cclass_node(CClassNode* cc, regex_t* reg) else { if (ONIGENC_MBC_MINLEN(reg->enc) > 1 || bitset_is_empty(cc->bs)) { if (IS_NCCLASS_NOT(cc)) - add_opcode(reg, OP_CCLASS_MB_NOT); + add_opcode(reg, OP_CCLASS_MB_NOT); else - add_opcode(reg, OP_CCLASS_MB); + add_opcode(reg, OP_CCLASS_MB); r = add_multi_byte_cclass(cc->mbuf, reg); } else { if (IS_NCCLASS_NOT(cc)) - add_opcode(reg, OP_CCLASS_MIX_NOT); + add_opcode(reg, OP_CCLASS_MIX_NOT); else - add_opcode(reg, OP_CCLASS_MIX); + add_opcode(reg, OP_CCLASS_MIX); r = add_bitset(reg, cc->bs); if (r) return r; @@ -762,9 +762,9 @@ compile_length_quantifier_node(QtfrNode* qn, regex_t* reg) if (NTYPE(qn->target) == NT_CANY) { if (qn->greedy && infinite) { if (IS_NOT_NULL(qn->next_head_exact) && !CKN_ON) - return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower + cklen; + return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower + cklen; else - return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower + cklen; + return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower + cklen; } } @@ -776,17 +776,17 @@ compile_length_quantifier_node(QtfrNode* qn, regex_t* reg) if (infinite && qn->lower <= 1) { if (qn->greedy) { if (qn->lower == 1) - len = SIZE_OP_JUMP; + len = SIZE_OP_JUMP; else - len = 0; + len = 0; len += SIZE_OP_PUSH + cklen + mod_tlen + SIZE_OP_JUMP; } else { if (qn->lower == 0) - len = SIZE_OP_JUMP; + len = SIZE_OP_JUMP; else - len = 0; + len = 0; len += mod_tlen + SIZE_OP_PUSH + cklen; } @@ -800,10 +800,10 @@ compile_length_quantifier_node(QtfrNode* qn, regex_t* reg) else if (qn->upper == 1 && qn->greedy) { if (qn->lower == 0) { if (CKN_ON) { - len = SIZE_OP_STATE_CHECK_PUSH + tlen; + len = SIZE_OP_STATE_CHECK_PUSH + tlen; } else { - len = SIZE_OP_PUSH + tlen; + len = SIZE_OP_PUSH + tlen; } } else { @@ -841,31 +841,31 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) if (r) return r; if (IS_NOT_NULL(qn->next_head_exact) && !CKN_ON) { if (IS_MULTILINE(reg->options)) - r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT); + r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT); else - r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT); + r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT); if (r) return r; if (CKN_ON) { - r = add_state_check_num(reg, ckn); - if (r) return r; + r = add_state_check_num(reg, ckn); + if (r) return r; } return add_bytes(reg, NSTR(qn->next_head_exact)->s, 1); } else { if (IS_MULTILINE(reg->options)) { - r = add_opcode(reg, (CKN_ON ? - OP_STATE_CHECK_ANYCHAR_ML_STAR - : OP_ANYCHAR_ML_STAR)); + r = add_opcode(reg, (CKN_ON ? + OP_STATE_CHECK_ANYCHAR_ML_STAR + : OP_ANYCHAR_ML_STAR)); } else { - r = add_opcode(reg, (CKN_ON ? - OP_STATE_CHECK_ANYCHAR_STAR - : OP_ANYCHAR_STAR)); + r = add_opcode(reg, (CKN_ON ? + OP_STATE_CHECK_ANYCHAR_STAR + : OP_ANYCHAR_STAR)); } if (r) return r; if (CKN_ON) - r = add_state_check_num(reg, ckn); + r = add_state_check_num(reg, ckn); return r; } @@ -879,45 +879,45 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) if (infinite && qn->lower <= 1) { if (qn->greedy) { if (qn->lower == 1) { - r = add_opcode_rel_addr(reg, OP_JUMP, - (CKN_ON ? SIZE_OP_STATE_CHECK_PUSH : SIZE_OP_PUSH)); - if (r) return r; + r = add_opcode_rel_addr(reg, OP_JUMP, + (CKN_ON ? SIZE_OP_STATE_CHECK_PUSH : SIZE_OP_PUSH)); + if (r) return r; } if (CKN_ON) { - r = add_opcode(reg, OP_STATE_CHECK_PUSH); - if (r) return r; - r = add_state_check_num(reg, ckn); - if (r) return r; - r = add_rel_addr(reg, mod_tlen + SIZE_OP_JUMP); + r = add_opcode(reg, OP_STATE_CHECK_PUSH); + if (r) return r; + r = add_state_check_num(reg, ckn); + if (r) return r; + r = add_rel_addr(reg, mod_tlen + SIZE_OP_JUMP); } else { - r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP); + r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP); } if (r) return r; r = compile_tree_empty_check(qn->target, reg, empty_info); if (r) return r; r = add_opcode_rel_addr(reg, OP_JUMP, - -(mod_tlen + (int )SIZE_OP_JUMP - + (int )(CKN_ON ? SIZE_OP_STATE_CHECK_PUSH : SIZE_OP_PUSH))); + -(mod_tlen + (int )SIZE_OP_JUMP + + (int )(CKN_ON ? SIZE_OP_STATE_CHECK_PUSH : SIZE_OP_PUSH))); } else { if (qn->lower == 0) { - r = add_opcode_rel_addr(reg, OP_JUMP, mod_tlen); - if (r) return r; + r = add_opcode_rel_addr(reg, OP_JUMP, mod_tlen); + if (r) return r; } r = compile_tree_empty_check(qn->target, reg, empty_info); if (r) return r; if (CKN_ON) { - r = add_opcode(reg, OP_STATE_CHECK_PUSH_OR_JUMP); - if (r) return r; - r = add_state_check_num(reg, ckn); - if (r) return r; - r = add_rel_addr(reg, - -(mod_tlen + (int )SIZE_OP_STATE_CHECK_PUSH_OR_JUMP)); + r = add_opcode(reg, OP_STATE_CHECK_PUSH_OR_JUMP); + if (r) return r; + r = add_state_check_num(reg, ckn); + if (r) return r; + r = add_rel_addr(reg, + -(mod_tlen + (int )SIZE_OP_STATE_CHECK_PUSH_OR_JUMP)); } else - r = add_opcode_rel_addr(reg, OP_PUSH, -(mod_tlen + (int )SIZE_OP_PUSH)); + r = add_opcode_rel_addr(reg, OP_PUSH, -(mod_tlen + (int )SIZE_OP_PUSH)); } } else if (qn->upper == 0) { @@ -932,14 +932,14 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) else if (qn->upper == 1 && qn->greedy) { if (qn->lower == 0) { if (CKN_ON) { - r = add_opcode(reg, OP_STATE_CHECK_PUSH); - if (r) return r; - r = add_state_check_num(reg, ckn); - if (r) return r; - r = add_rel_addr(reg, tlen); + r = add_opcode(reg, OP_STATE_CHECK_PUSH); + if (r) return r; + r = add_state_check_num(reg, ckn); + if (r) return r; + r = add_rel_addr(reg, tlen); } else { - r = add_opcode_rel_addr(reg, OP_PUSH, tlen); + r = add_opcode_rel_addr(reg, OP_PUSH, tlen); } if (r) return r; } @@ -991,9 +991,9 @@ compile_length_quantifier_node(QtfrNode* qn, regex_t* reg) if (NTYPE(qn->target) == NT_CANY) { if (qn->greedy && infinite) { if (IS_NOT_NULL(qn->next_head_exact)) - return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower; + return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower; else - return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower; + return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower; } } @@ -1014,13 +1014,13 @@ compile_length_quantifier_node(QtfrNode* qn, regex_t* reg) if (qn->greedy) { #ifdef USE_OP_PUSH_OR_JUMP_EXACT if (IS_NOT_NULL(qn->head_exact)) - len += SIZE_OP_PUSH_OR_JUMP_EXACT1 + mod_tlen + SIZE_OP_JUMP; + len += SIZE_OP_PUSH_OR_JUMP_EXACT1 + mod_tlen + SIZE_OP_JUMP; else #endif if (IS_NOT_NULL(qn->next_head_exact)) - len += SIZE_OP_PUSH_IF_PEEK_NEXT + mod_tlen + SIZE_OP_JUMP; + len += SIZE_OP_PUSH_IF_PEEK_NEXT + mod_tlen + SIZE_OP_JUMP; else - len += SIZE_OP_PUSH + mod_tlen + SIZE_OP_JUMP; + len += SIZE_OP_PUSH + mod_tlen + SIZE_OP_JUMP; } else len += SIZE_OP_JUMP + mod_tlen + SIZE_OP_PUSH; @@ -1060,17 +1060,17 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) if (r) return r; if (IS_NOT_NULL(qn->next_head_exact)) { if (IS_MULTILINE(reg->options)) - r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT); + r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT); else - r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT); + r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT); if (r) return r; return add_bytes(reg, NSTR(qn->next_head_exact)->s, 1); } else { if (IS_MULTILINE(reg->options)) - return add_opcode(reg, OP_ANYCHAR_ML_STAR); + return add_opcode(reg, OP_ANYCHAR_ML_STAR); else - return add_opcode(reg, OP_ANYCHAR_STAR); + return add_opcode(reg, OP_ANYCHAR_STAR); } } @@ -1084,17 +1084,17 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) if (qn->lower == 1 && tlen > QUANTIFIER_EXPAND_LIMIT_SIZE) { if (qn->greedy) { #ifdef USE_OP_PUSH_OR_JUMP_EXACT - if (IS_NOT_NULL(qn->head_exact)) - r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_OR_JUMP_EXACT1); - else + if (IS_NOT_NULL(qn->head_exact)) + r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_OR_JUMP_EXACT1); + else #endif - if (IS_NOT_NULL(qn->next_head_exact)) - r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_IF_PEEK_NEXT); - else - r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH); + if (IS_NOT_NULL(qn->next_head_exact)) + r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_IF_PEEK_NEXT); + else + r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH); } else { - r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_JUMP); + r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_JUMP); } if (r) return r; } @@ -1106,34 +1106,34 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) if (qn->greedy) { #ifdef USE_OP_PUSH_OR_JUMP_EXACT if (IS_NOT_NULL(qn->head_exact)) { - r = add_opcode_rel_addr(reg, OP_PUSH_OR_JUMP_EXACT1, - mod_tlen + SIZE_OP_JUMP); - if (r) return r; - add_bytes(reg, NSTR(qn->head_exact)->s, 1); - r = compile_tree_empty_check(qn->target, reg, empty_info); - if (r) return r; - r = add_opcode_rel_addr(reg, OP_JUMP, - -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_OR_JUMP_EXACT1)); + r = add_opcode_rel_addr(reg, OP_PUSH_OR_JUMP_EXACT1, + mod_tlen + SIZE_OP_JUMP); + if (r) return r; + add_bytes(reg, NSTR(qn->head_exact)->s, 1); + r = compile_tree_empty_check(qn->target, reg, empty_info); + if (r) return r; + r = add_opcode_rel_addr(reg, OP_JUMP, + -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_OR_JUMP_EXACT1)); } else #endif if (IS_NOT_NULL(qn->next_head_exact)) { - r = add_opcode_rel_addr(reg, OP_PUSH_IF_PEEK_NEXT, - mod_tlen + SIZE_OP_JUMP); - if (r) return r; - add_bytes(reg, NSTR(qn->next_head_exact)->s, 1); - r = compile_tree_empty_check(qn->target, reg, empty_info); - if (r) return r; - r = add_opcode_rel_addr(reg, OP_JUMP, + r = add_opcode_rel_addr(reg, OP_PUSH_IF_PEEK_NEXT, + mod_tlen + SIZE_OP_JUMP); + if (r) return r; + add_bytes(reg, NSTR(qn->next_head_exact)->s, 1); + r = compile_tree_empty_check(qn->target, reg, empty_info); + if (r) return r; + r = add_opcode_rel_addr(reg, OP_JUMP, -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_IF_PEEK_NEXT)); } else { - r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP); - if (r) return r; - r = compile_tree_empty_check(qn->target, reg, empty_info); - if (r) return r; - r = add_opcode_rel_addr(reg, OP_JUMP, - -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH)); + r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP); + if (r) return r; + r = compile_tree_empty_check(qn->target, reg, empty_info); + if (r) return r; + r = add_opcode_rel_addr(reg, OP_JUMP, + -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH)); } } else { @@ -1159,7 +1159,7 @@ compile_quantifier_node(QtfrNode* qn, regex_t* reg) for (i = 0; i < n; i++) { r = add_opcode_rel_addr(reg, OP_PUSH, - (n - i) * tlen + (n - i - 1) * SIZE_OP_PUSH); + (n - i) * tlen + (n - i - 1) * SIZE_OP_PUSH); if (r) return r; r = compile_tree(qn->target, reg); if (r) return r; @@ -1246,29 +1246,29 @@ compile_length_enclose_node(EncloseNode* node, regex_t* reg) #ifdef USE_SUBEXP_CALL if (IS_ENCLOSE_CALLED(node)) { len = SIZE_OP_MEMORY_START_PUSH + tlen - + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN; + + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN; if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)) - len += (IS_ENCLOSE_RECURSION(node) - ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH); + len += (IS_ENCLOSE_RECURSION(node) + ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH); else - len += (IS_ENCLOSE_RECURSION(node) - ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END); + len += (IS_ENCLOSE_RECURSION(node) + ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END); } else if (IS_ENCLOSE_RECURSION(node)) { len = SIZE_OP_MEMORY_START_PUSH; len += tlen + (BIT_STATUS_AT(reg->bt_mem_end, node->regnum) - ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_REC); + ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_REC); } else #endif { if (BIT_STATUS_AT(reg->bt_mem_start, node->regnum)) - len = SIZE_OP_MEMORY_START_PUSH; + len = SIZE_OP_MEMORY_START_PUSH; else - len = SIZE_OP_MEMORY_START; + len = SIZE_OP_MEMORY_START; len += tlen + (BIT_STATUS_AT(reg->bt_mem_end, node->regnum) - ? SIZE_OP_MEMORY_END_PUSH : SIZE_OP_MEMORY_END); + ? SIZE_OP_MEMORY_END_PUSH : SIZE_OP_MEMORY_END); } break; @@ -1283,7 +1283,7 @@ compile_length_enclose_node(EncloseNode* node, regex_t* reg) if (tlen < 0) return tlen; len = tlen * qn->lower - + SIZE_OP_PUSH + tlen + SIZE_OP_POP + SIZE_OP_JUMP; + + SIZE_OP_PUSH + tlen + SIZE_OP_POP + SIZE_OP_JUMP; } else { #endif @@ -1348,11 +1348,11 @@ compile_enclose_node(EncloseNode* node, regex_t* reg) len = compile_length_tree(node->target, reg); len += (SIZE_OP_MEMORY_START_PUSH + SIZE_OP_RETURN); if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)) - len += (IS_ENCLOSE_RECURSION(node) - ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH); + len += (IS_ENCLOSE_RECURSION(node) + ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH); else - len += (IS_ENCLOSE_RECURSION(node) - ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END); + len += (IS_ENCLOSE_RECURSION(node) + ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END); r = add_opcode_rel_addr(reg, OP_JUMP, len); if (r) return r; @@ -1370,11 +1370,11 @@ compile_enclose_node(EncloseNode* node, regex_t* reg) #ifdef USE_SUBEXP_CALL if (IS_ENCLOSE_CALLED(node)) { if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)) - r = add_opcode(reg, (IS_ENCLOSE_RECURSION(node) - ? OP_MEMORY_END_PUSH_REC : OP_MEMORY_END_PUSH)); + r = add_opcode(reg, (IS_ENCLOSE_RECURSION(node) + ? OP_MEMORY_END_PUSH_REC : OP_MEMORY_END_PUSH)); else - r = add_opcode(reg, (IS_ENCLOSE_RECURSION(node) - ? OP_MEMORY_END_REC : OP_MEMORY_END)); + r = add_opcode(reg, (IS_ENCLOSE_RECURSION(node) + ? OP_MEMORY_END_REC : OP_MEMORY_END)); if (r) return r; r = add_mem_num(reg, node->regnum); @@ -1383,9 +1383,9 @@ compile_enclose_node(EncloseNode* node, regex_t* reg) } else if (IS_ENCLOSE_RECURSION(node)) { if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)) - r = add_opcode(reg, OP_MEMORY_END_PUSH_REC); + r = add_opcode(reg, OP_MEMORY_END_PUSH_REC); else - r = add_opcode(reg, OP_MEMORY_END_REC); + r = add_opcode(reg, OP_MEMORY_END_REC); if (r) return r; r = add_mem_num(reg, node->regnum); } @@ -1393,9 +1393,9 @@ compile_enclose_node(EncloseNode* node, regex_t* reg) #endif { if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)) - r = add_opcode(reg, OP_MEMORY_END_PUSH); + r = add_opcode(reg, OP_MEMORY_END_PUSH); else - r = add_opcode(reg, OP_MEMORY_END); + r = add_opcode(reg, OP_MEMORY_END); if (r) return r; r = add_mem_num(reg, node->regnum); } @@ -1421,7 +1421,7 @@ compile_enclose_node(EncloseNode* node, regex_t* reg) r = add_opcode(reg, OP_POP); if (r) return r; r = add_opcode_rel_addr(reg, OP_JUMP, - -((int )SIZE_OP_PUSH + len + (int )SIZE_OP_POP + (int )SIZE_OP_JUMP)); + -((int )SIZE_OP_PUSH + len + (int )SIZE_OP_POP + (int )SIZE_OP_JUMP)); } else { #endif @@ -1579,11 +1579,11 @@ compile_anchor_node(AnchorNode* node, regex_t* reg) r = add_opcode(reg, OP_LOOK_BEHIND); if (r) return r; if (node->char_len < 0) { - r = get_char_length_tree(node->target, reg, &n); - if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; + r = get_char_length_tree(node->target, reg, &n); + if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; } else - n = node->char_len; + n = node->char_len; r = add_length(reg, n); if (r) return r; r = compile_tree(node->target, reg); @@ -1595,14 +1595,14 @@ compile_anchor_node(AnchorNode* node, regex_t* reg) int n; len = compile_length_tree(node->target, reg); r = add_opcode_rel_addr(reg, OP_PUSH_LOOK_BEHIND_NOT, - len + SIZE_OP_FAIL_LOOK_BEHIND_NOT); + len + SIZE_OP_FAIL_LOOK_BEHIND_NOT); if (r) return r; if (node->char_len < 0) { - r = get_char_length_tree(node->target, reg, &n); - if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; + r = get_char_length_tree(node->target, reg, &n); + if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; } else - n = node->char_len; + n = node->char_len; r = add_length(reg, n); if (r) return r; r = compile_tree(node->target, reg); @@ -1641,10 +1641,10 @@ compile_length_tree(Node* node, regex_t* reg) int n = 0; len = 0; do { - r = compile_length_tree(NCAR(node), reg); - if (r < 0) return r; - len += r; - n++; + r = compile_length_tree(NCAR(node), reg); + if (r < 0) return r; + len += r; + n++; } while (IS_NOT_NULL(node = NCDR(node))); r = len; r += (SIZE_OP_PUSH + SIZE_OP_JUMP) * (n - 1); @@ -1673,17 +1673,17 @@ compile_length_tree(Node* node, regex_t* reg) #ifdef USE_BACKREF_WITH_LEVEL if (IS_BACKREF_NEST_LEVEL(br)) { - r = SIZE_OPCODE + SIZE_OPTION + SIZE_LENGTH + + r = SIZE_OPCODE + SIZE_OPTION + SIZE_LENGTH + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num); } else #endif if (br->back_num == 1) { - r = ((!IS_IGNORECASE(reg->options) && br->back_static[0] <= 2) - ? SIZE_OPCODE : (SIZE_OPCODE + SIZE_MEMNUM)); + r = ((!IS_IGNORECASE(reg->options) && br->back_static[0] <= 2) + ? SIZE_OPCODE : (SIZE_OPCODE + SIZE_MEMNUM)); } else { - r = SIZE_OPCODE + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num); + r = SIZE_OPCODE + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num); } } break; @@ -1732,26 +1732,26 @@ compile_tree(Node* node, regex_t* reg) Node* x = node; len = 0; do { - len += compile_length_tree(NCAR(x), reg); - if (NCDR(x) != NULL) { - len += SIZE_OP_PUSH + SIZE_OP_JUMP; - } + len += compile_length_tree(NCAR(x), reg); + if (NCDR(x) != NULL) { + len += SIZE_OP_PUSH + SIZE_OP_JUMP; + } } while (IS_NOT_NULL(x = NCDR(x))); pos = reg->used + len; /* goal position */ do { - len = compile_length_tree(NCAR(node), reg); - if (IS_NOT_NULL(NCDR(node))) { - r = add_opcode_rel_addr(reg, OP_PUSH, len + SIZE_OP_JUMP); - if (r) break; - } - r = compile_tree(NCAR(node), reg); - if (r) break; - if (IS_NOT_NULL(NCDR(node))) { - len = pos - (reg->used + SIZE_OP_JUMP); - r = add_opcode_rel_addr(reg, OP_JUMP, len); - if (r) break; - } + len = compile_length_tree(NCAR(node), reg); + if (IS_NOT_NULL(NCDR(node))) { + r = add_opcode_rel_addr(reg, OP_PUSH, len + SIZE_OP_JUMP); + if (r) break; + } + r = compile_tree(NCAR(node), reg); + if (r) break; + if (IS_NOT_NULL(NCDR(node))) { + len = pos - (reg->used + SIZE_OP_JUMP); + r = add_opcode_rel_addr(reg, OP_JUMP, len); + if (r) break; + } } while (IS_NOT_NULL(node = NCDR(node))); } break; @@ -1773,18 +1773,18 @@ compile_tree(Node* node, regex_t* reg) switch (NCTYPE(node)->ctype) { case ONIGENC_CTYPE_WORD: - if (NCTYPE(node)->ascii_range != 0) { - if (NCTYPE(node)->not != 0) op = OP_NOT_ASCII_WORD; - else op = OP_ASCII_WORD; - } - else { - if (NCTYPE(node)->not != 0) op = OP_NOT_WORD; - else op = OP_WORD; - } - break; + if (NCTYPE(node)->ascii_range != 0) { + if (NCTYPE(node)->not != 0) op = OP_NOT_ASCII_WORD; + else op = OP_ASCII_WORD; + } + else { + if (NCTYPE(node)->not != 0) op = OP_NOT_WORD; + else op = OP_WORD; + } + break; default: - return ONIGERR_TYPE_BUG; - break; + return ONIGERR_TYPE_BUG; + break; } r = add_opcode(reg, op); } @@ -1803,58 +1803,58 @@ compile_tree(Node* node, regex_t* reg) #ifdef USE_BACKREF_WITH_LEVEL if (IS_BACKREF_NEST_LEVEL(br)) { - r = add_opcode(reg, OP_BACKREF_WITH_LEVEL); - if (r) return r; - r = add_option(reg, (reg->options & ONIG_OPTION_IGNORECASE)); - if (r) return r; - r = add_length(reg, br->nest_level); - if (r) return r; - - goto add_bacref_mems; + r = add_opcode(reg, OP_BACKREF_WITH_LEVEL); + if (r) return r; + r = add_option(reg, (reg->options & ONIG_OPTION_IGNORECASE)); + if (r) return r; + r = add_length(reg, br->nest_level); + if (r) return r; + + goto add_bacref_mems; } else #endif if (br->back_num == 1) { - n = br->back_static[0]; - if (IS_IGNORECASE(reg->options)) { - r = add_opcode(reg, OP_BACKREFN_IC); - if (r) return r; - r = add_mem_num(reg, n); - } - else { - switch (n) { - case 1: r = add_opcode(reg, OP_BACKREF1); break; - case 2: r = add_opcode(reg, OP_BACKREF2); break; - default: - r = add_opcode(reg, OP_BACKREFN); - if (r) return r; - r = add_mem_num(reg, n); - break; - } - } + n = br->back_static[0]; + if (IS_IGNORECASE(reg->options)) { + r = add_opcode(reg, OP_BACKREFN_IC); + if (r) return r; + r = add_mem_num(reg, n); + } + else { + switch (n) { + case 1: r = add_opcode(reg, OP_BACKREF1); break; + case 2: r = add_opcode(reg, OP_BACKREF2); break; + default: + r = add_opcode(reg, OP_BACKREFN); + if (r) return r; + r = add_mem_num(reg, n); + break; + } + } } else { - int i; - int* p; + int i; + int* p; - if (IS_IGNORECASE(reg->options)) { - r = add_opcode(reg, OP_BACKREF_MULTI_IC); - } - else { - r = add_opcode(reg, OP_BACKREF_MULTI); - } - if (r) return r; + if (IS_IGNORECASE(reg->options)) { + r = add_opcode(reg, OP_BACKREF_MULTI_IC); + } + else { + r = add_opcode(reg, OP_BACKREF_MULTI); + } + if (r) return r; #ifdef USE_BACKREF_WITH_LEVEL add_bacref_mems: #endif - r = add_length(reg, br->back_num); - if (r) return r; - p = BACKREFS_P(br); - for (i = br->back_num - 1; i >= 0; i--) { - r = add_mem_num(reg, p[i]); - if (r) return r; - } + r = add_length(reg, br->back_num); + if (r) return r; + p = BACKREFS_P(br); + for (i = br->back_num - 1; i >= 0; i--) { + r = add_mem_num(reg, p[i]); + if (r) return r; + } } } break; @@ -1909,7 +1909,7 @@ noname_disable_map(Node** plink, GroupNumRemap* map, int* counter) Node* old = *ptarget; r = noname_disable_map(ptarget, map, counter); if (*ptarget != old && NTYPE(*ptarget) == NT_QTFR) { - onig_reduce_nested_quantifier(node, *ptarget); + onig_reduce_nested_quantifier(node, *ptarget); } } break; @@ -1918,18 +1918,18 @@ noname_disable_map(Node** plink, GroupNumRemap* map, int* counter) { EncloseNode* en = NENCLOSE(node); if (en->type == ENCLOSE_MEMORY) { - if (IS_ENCLOSE_NAMED_GROUP(en)) { - (*counter)++; - map[en->regnum].new_val = *counter; - en->regnum = *counter; - } - else if (en->regnum != 0) { - *plink = en->target; - en->target = NULL_NODE; - onig_node_free(node); - r = noname_disable_map(plink, map, counter); - break; - } + if (IS_ENCLOSE_NAMED_GROUP(en)) { + (*counter)++; + map[en->regnum].new_val = *counter; + en->regnum = *counter; + } + else if (en->regnum != 0) { + *plink = en->target; + en->target = NULL_NODE; + onig_node_free(node); + r = noname_disable_map(plink, map, counter); + break; + } } r = noname_disable_map(&(en->target), map, counter); } @@ -1995,8 +1995,8 @@ renumber_by_map(Node* node, GroupNumRemap* map, const int num_mem) { EncloseNode* en = NENCLOSE(node); if (en->type == ENCLOSE_CONDITION) { - if (en->regnum > num_mem) return ONIGERR_INVALID_BACKREF; - en->regnum = map[en->regnum].new_val; + if (en->regnum > num_mem) return ONIGERR_INVALID_BACKREF; + en->regnum = map[en->regnum].new_val; } r = renumber_by_map(en->target, map, num_mem); } @@ -2127,8 +2127,8 @@ quantifiers_memory_node_info(Node* node) { int v; do { - v = quantifiers_memory_node_info(NCAR(node)); - if (v > r) r = v; + v = quantifiers_memory_node_info(NCAR(node)); + if (v > r) r = v; } while (v >= 0 && IS_NOT_NULL(node = NCDR(node))); } break; @@ -2147,7 +2147,7 @@ quantifiers_memory_node_info(Node* node) { QtfrNode* qn = NQTFR(node); if (qn->upper != 0) { - r = quantifiers_memory_node_info(qn->target); + r = quantifiers_memory_node_info(qn->target); } } break; @@ -2157,17 +2157,17 @@ quantifiers_memory_node_info(Node* node) EncloseNode* en = NENCLOSE(node); switch (en->type) { case ENCLOSE_MEMORY: - return NQ_TARGET_IS_EMPTY_MEM; - break; + return NQ_TARGET_IS_EMPTY_MEM; + break; case ENCLOSE_OPTION: case ENCLOSE_STOP_BACKTRACK: case ENCLOSE_CONDITION: case ENCLOSE_ABSENT: - r = quantifiers_memory_node_info(en->target); - break; + r = quantifiers_memory_node_info(en->target); + break; default: - break; + break; } } break; @@ -2207,10 +2207,10 @@ get_min_match_length(Node* node, OnigDistance *min, ScanEnv* env) r = get_min_match_length(nodes[backs[0]], min, env); if (r != 0) break; for (i = 1; i < br->back_num; i++) { - if (backs[i] > env->num_mem) return ONIGERR_INVALID_BACKREF; - r = get_min_match_length(nodes[backs[i]], &tmin, env); - if (r != 0) break; - if (*min > tmin) *min = tmin; + if (backs[i] > env->num_mem) return ONIGERR_INVALID_BACKREF; + r = get_min_match_length(nodes[backs[i]], &tmin, env); + if (r != 0) break; + if (*min > tmin) *min = tmin; } } break; @@ -2220,7 +2220,7 @@ get_min_match_length(Node* node, OnigDistance *min, ScanEnv* env) if (IS_CALL_RECURSION(NCALL(node))) { EncloseNode* en = NENCLOSE(NCALL(node)->target); if (IS_ENCLOSE_MIN_FIXED(en)) - *min = en->min_len; + *min = en->min_len; } else r = get_min_match_length(NCALL(node)->target, min, env); @@ -2239,11 +2239,11 @@ get_min_match_length(Node* node, OnigDistance *min, ScanEnv* env) Node *x, *y; y = node; do { - x = NCAR(y); - r = get_min_match_length(x, &tmin, env); - if (r != 0) break; - if (y == node) *min = tmin; - else if (*min > tmin) *min = tmin; + x = NCAR(y); + r = get_min_match_length(x, &tmin, env); + if (r != 0) break; + if (y == node) *min = tmin; + else if (*min > tmin) *min = tmin; } while (r == 0 && IS_NOT_NULL(y = NCDR(y))); } break; @@ -2269,9 +2269,9 @@ get_min_match_length(Node* node, OnigDistance *min, ScanEnv* env) QtfrNode* qn = NQTFR(node); if (qn->lower > 0) { - r = get_min_match_length(qn->target, min, env); - if (r == 0) - *min = distance_multiply(*min, qn->lower); + r = get_min_match_length(qn->target, min, env); + if (r == 0) + *min = distance_multiply(*min, qn->lower); } } break; @@ -2284,28 +2284,28 @@ get_min_match_length(Node* node, OnigDistance *min, ScanEnv* env) if (IS_ENCLOSE_MIN_FIXED(en)) *min = en->min_len; else { - if (IS_ENCLOSE_MARK1(NENCLOSE(node))) - *min = 0; /* recursive */ - else { - SET_ENCLOSE_STATUS(node, NST_MARK1); - r = get_min_match_length(en->target, min, env); - CLEAR_ENCLOSE_STATUS(node, NST_MARK1); - if (r == 0) { - en->min_len = *min; - SET_ENCLOSE_STATUS(node, NST_MIN_FIXED); - } - } + if (IS_ENCLOSE_MARK1(NENCLOSE(node))) + *min = 0; /* recursive */ + else { + SET_ENCLOSE_STATUS(node, NST_MARK1); + r = get_min_match_length(en->target, min, env); + CLEAR_ENCLOSE_STATUS(node, NST_MARK1); + if (r == 0) { + en->min_len = *min; + SET_ENCLOSE_STATUS(node, NST_MIN_FIXED); + } + } } break; case ENCLOSE_OPTION: case ENCLOSE_STOP_BACKTRACK: case ENCLOSE_CONDITION: - r = get_min_match_length(en->target, min, env); - break; + r = get_min_match_length(en->target, min, env); + break; case ENCLOSE_ABSENT: - break; + break; } } break; @@ -2330,7 +2330,7 @@ get_max_match_length(Node* node, OnigDistance *max, ScanEnv* env) do { r = get_max_match_length(NCAR(node), &tmax, env); if (r == 0) - *max = distance_add(*max, tmax); + *max = distance_add(*max, tmax); } while (r == 0 && IS_NOT_NULL(node = NCDR(node))); break; @@ -2364,15 +2364,15 @@ get_max_match_length(Node* node, OnigDistance *max, ScanEnv* env) Node** nodes = SCANENV_MEM_NODES(env); BRefNode* br = NBREF(node); if (br->state & NST_RECURSION) { - *max = ONIG_INFINITE_DISTANCE; - break; + *max = ONIG_INFINITE_DISTANCE; + break; } backs = BACKREFS_P(br); for (i = 0; i < br->back_num; i++) { - if (backs[i] > env->num_mem) return ONIGERR_INVALID_BACKREF; - r = get_max_match_length(nodes[backs[i]], &tmax, env); - if (r != 0) break; - if (*max < tmax) *max = tmax; + if (backs[i] > env->num_mem) return ONIGERR_INVALID_BACKREF; + r = get_max_match_length(nodes[backs[i]], &tmax, env); + if (r != 0) break; + if (*max < tmax) *max = tmax; } } break; @@ -2391,13 +2391,13 @@ get_max_match_length(Node* node, OnigDistance *max, ScanEnv* env) QtfrNode* qn = NQTFR(node); if (qn->upper != 0) { - r = get_max_match_length(qn->target, max, env); - if (r == 0 && *max != 0) { - if (! IS_REPEAT_INFINITE(qn->upper)) - *max = distance_multiply(*max, qn->upper); - else - *max = ONIG_INFINITE_DISTANCE; - } + r = get_max_match_length(qn->target, max, env); + if (r == 0 && *max != 0) { + if (! IS_REPEAT_INFINITE(qn->upper)) + *max = distance_multiply(*max, qn->upper); + else + *max = ONIG_INFINITE_DISTANCE; + } } } break; @@ -2407,31 +2407,31 @@ get_max_match_length(Node* node, OnigDistance *max, ScanEnv* env) EncloseNode* en = NENCLOSE(node); switch (en->type) { case ENCLOSE_MEMORY: - if (IS_ENCLOSE_MAX_FIXED(en)) - *max = en->max_len; - else { - if (IS_ENCLOSE_MARK1(NENCLOSE(node))) - *max = ONIG_INFINITE_DISTANCE; - else { - SET_ENCLOSE_STATUS(node, NST_MARK1); - r = get_max_match_length(en->target, max, env); - CLEAR_ENCLOSE_STATUS(node, NST_MARK1); - if (r == 0) { - en->max_len = *max; - SET_ENCLOSE_STATUS(node, NST_MAX_FIXED); - } - } - } - break; + if (IS_ENCLOSE_MAX_FIXED(en)) + *max = en->max_len; + else { + if (IS_ENCLOSE_MARK1(NENCLOSE(node))) + *max = ONIG_INFINITE_DISTANCE; + else { + SET_ENCLOSE_STATUS(node, NST_MARK1); + r = get_max_match_length(en->target, max, env); + CLEAR_ENCLOSE_STATUS(node, NST_MARK1); + if (r == 0) { + en->max_len = *max; + SET_ENCLOSE_STATUS(node, NST_MAX_FIXED); + } + } + } + break; case ENCLOSE_OPTION: case ENCLOSE_STOP_BACKTRACK: case ENCLOSE_CONDITION: - r = get_max_match_length(en->target, max, env); - break; + r = get_max_match_length(en->target, max, env); + break; case ENCLOSE_ABSENT: - break; + break; } } break; @@ -2461,7 +2461,7 @@ get_char_length_tree1(Node* node, regex_t* reg, int* len, int level) do { r = get_char_length_tree1(NCAR(node), reg, &tlen, level); if (r == 0) - *len = (int )distance_add(*len, tlen); + *len = (int )distance_add(*len, tlen); } while (r == 0 && IS_NOT_NULL(node = NCDR(node))); break; @@ -2472,21 +2472,21 @@ get_char_length_tree1(Node* node, regex_t* reg, int* len, int level) r = get_char_length_tree1(NCAR(node), reg, &tlen, level); while (r == 0 && IS_NOT_NULL(node = NCDR(node))) { - r = get_char_length_tree1(NCAR(node), reg, &tlen2, level); - if (r == 0) { - if (tlen != tlen2) - varlen = 1; - } + r = get_char_length_tree1(NCAR(node), reg, &tlen2, level); + if (r == 0) { + if (tlen != tlen2) + varlen = 1; + } } if (r == 0) { - if (varlen != 0) { - if (level == 1) - r = GET_CHAR_LEN_TOP_ALT_VARLEN; - else - r = GET_CHAR_LEN_VARLEN; - } - else - *len = tlen; + if (varlen != 0) { + if (level == 1) + r = GET_CHAR_LEN_TOP_ALT_VARLEN; + else + r = GET_CHAR_LEN_VARLEN; + } + else + *len = tlen; } } break; @@ -2496,8 +2496,8 @@ get_char_length_tree1(Node* node, regex_t* reg, int* len, int level) StrNode* sn = NSTR(node); UChar *s = sn->s; while (s < sn->end) { - s += enclen(reg->enc, s, sn->end); - (*len)++; + s += enclen(reg->enc, s, sn->end); + (*len)++; } } break; @@ -2506,12 +2506,12 @@ get_char_length_tree1(Node* node, regex_t* reg, int* len, int level) { QtfrNode* qn = NQTFR(node); if (qn->lower == qn->upper) { - r = get_char_length_tree1(qn->target, reg, &tlen, level); - if (r == 0) - *len = (int )distance_multiply(tlen, qn->lower); + r = get_char_length_tree1(qn->target, reg, &tlen, level); + if (r == 0) + *len = (int )distance_multiply(tlen, qn->lower); } else - r = GET_CHAR_LEN_VARLEN; + r = GET_CHAR_LEN_VARLEN; } break; @@ -2539,25 +2539,25 @@ get_char_length_tree1(Node* node, regex_t* reg, int* len, int level) switch (en->type) { case ENCLOSE_MEMORY: #ifdef USE_SUBEXP_CALL - if (IS_ENCLOSE_CLEN_FIXED(en)) - *len = en->char_len; - else { - r = get_char_length_tree1(en->target, reg, len, level); - if (r == 0) { - en->char_len = *len; - SET_ENCLOSE_STATUS(node, NST_CLEN_FIXED); - } - } - break; + if (IS_ENCLOSE_CLEN_FIXED(en)) + *len = en->char_len; + else { + r = get_char_length_tree1(en->target, reg, len, level); + if (r == 0) { + en->char_len = *len; + SET_ENCLOSE_STATUS(node, NST_CLEN_FIXED); + } + } + break; #endif case ENCLOSE_OPTION: case ENCLOSE_STOP_BACKTRACK: case ENCLOSE_CONDITION: - r = get_char_length_tree1(en->target, reg, len, level); - break; + r = get_char_length_tree1(en->target, reg, len, level); + break; case ENCLOSE_ABSENT: default: - break; + break; } } break; @@ -2596,29 +2596,29 @@ is_not_included(Node* x, Node* y, regex_t* reg) { switch (ytype) { case NT_CTYPE: - if (NCTYPE(y)->ctype == NCTYPE(x)->ctype && - NCTYPE(y)->not != NCTYPE(x)->not && - NCTYPE(y)->ascii_range == NCTYPE(x)->ascii_range) - return 1; - else - return 0; - break; + if (NCTYPE(y)->ctype == NCTYPE(x)->ctype && + NCTYPE(y)->not != NCTYPE(x)->not && + NCTYPE(y)->ascii_range == NCTYPE(x)->ascii_range) + return 1; + else + return 0; + break; case NT_CCLASS: swap: - { - Node* tmp; - tmp = x; x = y; y = tmp; - goto retry; - } - break; + { + Node* tmp; + tmp = x; x = y; y = tmp; + goto retry; + } + break; case NT_STR: - goto swap; - break; + goto swap; + break; default: - break; + break; } } break; @@ -2628,80 +2628,80 @@ is_not_included(Node* x, Node* y, regex_t* reg) CClassNode* xc = NCCLASS(x); switch (ytype) { case NT_CTYPE: - switch (NCTYPE(y)->ctype) { - case ONIGENC_CTYPE_WORD: - if (NCTYPE(y)->not == 0) { - if (IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) { - for (i = 0; i < SINGLE_BYTE_SIZE; i++) { - if (BITSET_AT(xc->bs, i)) { - if (NCTYPE(y)->ascii_range) { - if (IS_CODE_SB_WORD(reg->enc, i)) return 0; - } - else { - if (ONIGENC_IS_CODE_WORD(reg->enc, i)) return 0; - } - } - } - return 1; - } - return 0; - } - else { - if (IS_NOT_NULL(xc->mbuf)) return 0; - for (i = 0; i < SINGLE_BYTE_SIZE; i++) { - int is_word; - if (NCTYPE(y)->ascii_range) - is_word = IS_CODE_SB_WORD(reg->enc, i); - else - is_word = ONIGENC_IS_CODE_WORD(reg->enc, i); - if (! is_word) { - if (!IS_NCCLASS_NOT(xc)) { - if (BITSET_AT(xc->bs, i)) - return 0; - } - else { - if (! BITSET_AT(xc->bs, i)) - return 0; - } - } - } - return 1; - } - break; - - default: - break; - } - break; + switch (NCTYPE(y)->ctype) { + case ONIGENC_CTYPE_WORD: + if (NCTYPE(y)->not == 0) { + if (IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) { + for (i = 0; i < SINGLE_BYTE_SIZE; i++) { + if (BITSET_AT(xc->bs, i)) { + if (NCTYPE(y)->ascii_range) { + if (IS_CODE_SB_WORD(reg->enc, i)) return 0; + } + else { + if (ONIGENC_IS_CODE_WORD(reg->enc, i)) return 0; + } + } + } + return 1; + } + return 0; + } + else { + if (IS_NOT_NULL(xc->mbuf)) return 0; + for (i = 0; i < SINGLE_BYTE_SIZE; i++) { + int is_word; + if (NCTYPE(y)->ascii_range) + is_word = IS_CODE_SB_WORD(reg->enc, i); + else + is_word = ONIGENC_IS_CODE_WORD(reg->enc, i); + if (! is_word) { + if (!IS_NCCLASS_NOT(xc)) { + if (BITSET_AT(xc->bs, i)) + return 0; + } + else { + if (! BITSET_AT(xc->bs, i)) + return 0; + } + } + } + return 1; + } + break; + + default: + break; + } + break; case NT_CCLASS: - { - int v; - CClassNode* yc = NCCLASS(y); - - for (i = 0; i < SINGLE_BYTE_SIZE; i++) { - v = BITSET_AT(xc->bs, i); - if ((v != 0 && !IS_NCCLASS_NOT(xc)) || - (v == 0 && IS_NCCLASS_NOT(xc))) { - v = BITSET_AT(yc->bs, i); - if ((v != 0 && !IS_NCCLASS_NOT(yc)) || - (v == 0 && IS_NCCLASS_NOT(yc))) - return 0; - } - } - if ((IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) || - (IS_NULL(yc->mbuf) && !IS_NCCLASS_NOT(yc))) - return 1; - return 0; - } - break; + { + int v; + CClassNode* yc = NCCLASS(y); + + for (i = 0; i < SINGLE_BYTE_SIZE; i++) { + v = BITSET_AT(xc->bs, i); + if ((v != 0 && !IS_NCCLASS_NOT(xc)) || + (v == 0 && IS_NCCLASS_NOT(xc))) { + v = BITSET_AT(yc->bs, i); + if ((v != 0 && !IS_NCCLASS_NOT(yc)) || + (v == 0 && IS_NCCLASS_NOT(yc))) + return 0; + } + } + if ((IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) || + (IS_NULL(yc->mbuf) && !IS_NCCLASS_NOT(yc))) + return 1; + return 0; + } + break; case NT_STR: - goto swap; - break; + goto swap; + break; default: - break; + break; } } break; @@ -2710,60 +2710,60 @@ is_not_included(Node* x, Node* y, regex_t* reg) { StrNode* xs = NSTR(x); if (NSTRING_LEN(x) == 0) - break; + break; switch (ytype) { case NT_CTYPE: - switch (NCTYPE(y)->ctype) { - case ONIGENC_CTYPE_WORD: - if (NCTYPE(y)->ascii_range) { - if (ONIGENC_IS_MBC_ASCII_WORD(reg->enc, xs->s, xs->end)) - return NCTYPE(y)->not; - else - return !(NCTYPE(y)->not); - } - else { - if (ONIGENC_IS_MBC_WORD(reg->enc, xs->s, xs->end)) - return NCTYPE(y)->not; - else - return !(NCTYPE(y)->not); - } - break; - default: - break; - } - break; + switch (NCTYPE(y)->ctype) { + case ONIGENC_CTYPE_WORD: + if (NCTYPE(y)->ascii_range) { + if (ONIGENC_IS_MBC_ASCII_WORD(reg->enc, xs->s, xs->end)) + return NCTYPE(y)->not; + else + return !(NCTYPE(y)->not); + } + else { + if (ONIGENC_IS_MBC_WORD(reg->enc, xs->s, xs->end)) + return NCTYPE(y)->not; + else + return !(NCTYPE(y)->not); + } + break; + default: + break; + } + break; case NT_CCLASS: - { - CClassNode* cc = NCCLASS(y); + { + CClassNode* cc = NCCLASS(y); - code = ONIGENC_MBC_TO_CODE(reg->enc, xs->s, - xs->s + ONIGENC_MBC_MAXLEN(reg->enc)); - return (onig_is_code_in_cc(reg->enc, code, cc) != 0 ? 0 : 1); - } - break; + code = ONIGENC_MBC_TO_CODE(reg->enc, xs->s, + xs->s + ONIGENC_MBC_MAXLEN(reg->enc)); + return (onig_is_code_in_cc(reg->enc, code, cc) != 0 ? 0 : 1); + } + break; case NT_STR: - { - UChar *q; - StrNode* ys = NSTR(y); - len = NSTRING_LEN(x); - if (len > NSTRING_LEN(y)) len = NSTRING_LEN(y); - if (NSTRING_IS_AMBIG(x) || NSTRING_IS_AMBIG(y)) { - /* tiny version */ - return 0; - } - else { - for (i = 0, p = ys->s, q = xs->s; (OnigDistance )i < len; i++, p++, q++) { - if (*p != *q) return 1; - } - } - } - break; + { + UChar *q; + StrNode* ys = NSTR(y); + len = NSTRING_LEN(x); + if (len > NSTRING_LEN(y)) len = NSTRING_LEN(y); + if (NSTRING_IS_AMBIG(x) || NSTRING_IS_AMBIG(y)) { + /* tiny version */ + return 0; + } + else { + for (i = 0, p = ys->s, q = xs->s; (OnigDistance )i < len; i++, p++, q++) { + if (*p != *q) return 1; + } + } + } + break; default: - break; + break; } } break; @@ -2805,13 +2805,13 @@ get_head_value_node(Node* node, int exact, regex_t* reg) StrNode* sn = NSTR(node); if (sn->end <= sn->s) - break; + break; if (exact != 0 && - !NSTRING_IS_RAW(node) && IS_IGNORECASE(reg->options)) { + !NSTRING_IS_RAW(node) && IS_IGNORECASE(reg->options)) { } else { - n = node; + n = node; } } break; @@ -2821,11 +2821,11 @@ get_head_value_node(Node* node, int exact, regex_t* reg) QtfrNode* qn = NQTFR(node); if (qn->lower > 0) { #ifdef USE_OP_PUSH_OR_JUMP_EXACT - if (IS_NOT_NULL(qn->head_exact)) - n = qn->head_exact; - else + if (IS_NOT_NULL(qn->head_exact)) + n = qn->head_exact; + else #endif - n = get_head_value_node(qn->target, exact, reg); + n = get_head_value_node(qn->target, exact, reg); } } break; @@ -2835,23 +2835,23 @@ get_head_value_node(Node* node, int exact, regex_t* reg) EncloseNode* en = NENCLOSE(node); switch (en->type) { case ENCLOSE_OPTION: - { - OnigOptionType options = reg->options; + { + OnigOptionType options = reg->options; - reg->options = NENCLOSE(node)->option; - n = get_head_value_node(NENCLOSE(node)->target, exact, reg); - reg->options = options; - } - break; + reg->options = NENCLOSE(node)->option; + n = get_head_value_node(NENCLOSE(node)->target, exact, reg); + reg->options = options; + } + break; case ENCLOSE_MEMORY: case ENCLOSE_STOP_BACKTRACK: case ENCLOSE_CONDITION: - n = get_head_value_node(en->target, exact, reg); - break; + n = get_head_value_node(en->target, exact, reg); + break; case ENCLOSE_ABSENT: - break; + break; } } break; @@ -2882,20 +2882,20 @@ check_type_tree(Node* node, int type_mask, int enclose_mask, int anchor_mask) case NT_ALT: do { r = check_type_tree(NCAR(node), type_mask, enclose_mask, - anchor_mask); + anchor_mask); } while (r == 0 && IS_NOT_NULL(node = NCDR(node))); break; case NT_QTFR: r = check_type_tree(NQTFR(node)->target, type_mask, enclose_mask, - anchor_mask); + anchor_mask); break; case NT_ENCLOSE: { EncloseNode* en = NENCLOSE(node); if ((en->type & enclose_mask) == 0) - return 1; + return 1; r = check_type_tree(en->target, type_mask, enclose_mask, anchor_mask); } @@ -2908,7 +2908,7 @@ check_type_tree(Node* node, int type_mask, int enclose_mask, int anchor_mask) if (NANCHOR(node)->target) r = check_type_tree(NANCHOR(node)->target, - type_mask, enclose_mask, anchor_mask); + type_mask, enclose_mask, anchor_mask); break; default: @@ -2938,14 +2938,14 @@ subexp_inf_recursive_check(Node* node, ScanEnv* env, int head) x = node; do { - ret = subexp_inf_recursive_check(NCAR(x), env, head); - if (ret < 0 || ret == RECURSION_INFINITE) return ret; - r |= ret; - if (head) { - ret = get_min_match_length(NCAR(x), &min, env); - if (ret != 0) return ret; - if (min != 0) head = 0; - } + ret = subexp_inf_recursive_check(NCAR(x), env, head); + if (ret < 0 || ret == RECURSION_INFINITE) return ret; + r |= ret; + if (head) { + ret = get_min_match_length(NCAR(x), &min, env); + if (ret != 0) return ret; + if (min != 0) head = 0; + } } while (IS_NOT_NULL(x = NCDR(x))); } break; @@ -2955,9 +2955,9 @@ subexp_inf_recursive_check(Node* node, ScanEnv* env, int head) int ret; r = RECURSION_EXIST; do { - ret = subexp_inf_recursive_check(NCAR(node), env, head); - if (ret < 0 || ret == RECURSION_INFINITE) return ret; - r &= ret; + ret = subexp_inf_recursive_check(NCAR(node), env, head); + if (ret < 0 || ret == RECURSION_INFINITE) return ret; + r &= ret; } while (IS_NOT_NULL(node = NCDR(node))); } break; @@ -2977,8 +2977,8 @@ subexp_inf_recursive_check(Node* node, ScanEnv* env, int head) case ANCHOR_PREC_READ_NOT: case ANCHOR_LOOK_BEHIND: case ANCHOR_LOOK_BEHIND_NOT: - r = subexp_inf_recursive_check(an->target, env, head); - break; + r = subexp_inf_recursive_check(an->target, env, head); + break; } } break; @@ -3033,8 +3033,8 @@ subexp_inf_recursive_check_trav(Node* node, ScanEnv* env) case ANCHOR_PREC_READ_NOT: case ANCHOR_LOOK_BEHIND: case ANCHOR_LOOK_BEHIND_NOT: - r = subexp_inf_recursive_check_trav(an->target, env); - break; + r = subexp_inf_recursive_check_trav(an->target, env); + break; } } break; @@ -3044,10 +3044,10 @@ subexp_inf_recursive_check_trav(Node* node, ScanEnv* env) EncloseNode* en = NENCLOSE(node); if (IS_ENCLOSE_RECURSION(en)) { - SET_ENCLOSE_STATUS(node, NST_MARK1); - r = subexp_inf_recursive_check(en->target, env, 1); - if (r > 0) return ONIGERR_NEVER_ENDING_RECURSION; - CLEAR_ENCLOSE_STATUS(node, NST_MARK1); + SET_ENCLOSE_STATUS(node, NST_MARK1); + r = subexp_inf_recursive_check(en->target, env, 1); + if (r > 0) return ONIGERR_NEVER_ENDING_RECURSION; + CLEAR_ENCLOSE_STATUS(node, NST_MARK1); } r = subexp_inf_recursive_check_trav(en->target, env); } @@ -3086,8 +3086,8 @@ subexp_recursive_check(Node* node) case ANCHOR_PREC_READ_NOT: case ANCHOR_LOOK_BEHIND: case ANCHOR_LOOK_BEHIND_NOT: - r = subexp_recursive_check(an->target); - break; + r = subexp_recursive_check(an->target); + break; } } break; @@ -3132,9 +3132,9 @@ subexp_recursive_check_trav(Node* node, ScanEnv* env) { int ret; do { - ret = subexp_recursive_check_trav(NCAR(node), env); - if (ret == FOUND_CALLED_NODE) r = FOUND_CALLED_NODE; - else if (ret < 0) return ret; + ret = subexp_recursive_check_trav(NCAR(node), env); + if (ret == FOUND_CALLED_NODE) r = FOUND_CALLED_NODE; + else if (ret < 0) return ret; } while (IS_NOT_NULL(node = NCDR(node))); } break; @@ -3143,7 +3143,7 @@ subexp_recursive_check_trav(Node* node, ScanEnv* env) r = subexp_recursive_check_trav(NQTFR(node)->target, env); if (NQTFR(node)->upper == 0) { if (r == FOUND_CALLED_NODE) - NQTFR(node)->is_referred = 1; + NQTFR(node)->is_referred = 1; } break; @@ -3155,8 +3155,8 @@ subexp_recursive_check_trav(Node* node, ScanEnv* env) case ANCHOR_PREC_READ_NOT: case ANCHOR_LOOK_BEHIND: case ANCHOR_LOOK_BEHIND_NOT: - r = subexp_recursive_check_trav(an->target, env); - break; + r = subexp_recursive_check_trav(an->target, env); + break; } } break; @@ -3166,16 +3166,16 @@ subexp_recursive_check_trav(Node* node, ScanEnv* env) EncloseNode* en = NENCLOSE(node); if (! IS_ENCLOSE_RECURSION(en)) { - if (IS_ENCLOSE_CALLED(en)) { - SET_ENCLOSE_STATUS(node, NST_MARK1); - r = subexp_recursive_check(en->target); - if (r != 0) SET_ENCLOSE_STATUS(node, NST_RECURSION); - CLEAR_ENCLOSE_STATUS(node, NST_MARK1); - } + if (IS_ENCLOSE_CALLED(en)) { + SET_ENCLOSE_STATUS(node, NST_MARK1); + r = subexp_recursive_check(en->target); + if (r != 0) SET_ENCLOSE_STATUS(node, NST_RECURSION); + CLEAR_ENCLOSE_STATUS(node, NST_MARK1); + } } r = subexp_recursive_check_trav(en->target, env); if (IS_ENCLOSE_CALLED(en)) - r |= FOUND_CALLED_NODE; + r |= FOUND_CALLED_NODE; } break; @@ -3219,60 +3219,60 @@ setup_subexp_call(Node* node, ScanEnv* env) Node** nodes = SCANENV_MEM_NODES(env); if (cn->group_num != 0) { - int gnum = cn->group_num; + int gnum = cn->group_num; # ifdef USE_NAMED_GROUP - if (env->num_named > 0 && - IS_SYNTAX_BV(env->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) && - !ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_CAPTURE_GROUP)) { - return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED; - } + if (env->num_named > 0 && + IS_SYNTAX_BV(env->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) && + !ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_CAPTURE_GROUP)) { + return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED; + } # endif - if (gnum > env->num_mem) { - onig_scan_env_set_error_string(env, - ONIGERR_UNDEFINED_GROUP_REFERENCE, cn->name, cn->name_end); - return ONIGERR_UNDEFINED_GROUP_REFERENCE; - } + if (gnum > env->num_mem) { + onig_scan_env_set_error_string(env, + ONIGERR_UNDEFINED_GROUP_REFERENCE, cn->name, cn->name_end); + return ONIGERR_UNDEFINED_GROUP_REFERENCE; + } # ifdef USE_NAMED_GROUP set_call_attr: # endif - cn->target = nodes[cn->group_num]; - if (IS_NULL(cn->target)) { - onig_scan_env_set_error_string(env, - ONIGERR_UNDEFINED_NAME_REFERENCE, cn->name, cn->name_end); - return ONIGERR_UNDEFINED_NAME_REFERENCE; - } - SET_ENCLOSE_STATUS(cn->target, NST_CALLED); - BIT_STATUS_ON_AT(env->bt_mem_start, cn->group_num); - cn->unset_addr_list = env->unset_addr_list; + cn->target = nodes[cn->group_num]; + if (IS_NULL(cn->target)) { + onig_scan_env_set_error_string(env, + ONIGERR_UNDEFINED_NAME_REFERENCE, cn->name, cn->name_end); + return ONIGERR_UNDEFINED_NAME_REFERENCE; + } + SET_ENCLOSE_STATUS(cn->target, NST_CALLED); + BIT_STATUS_ON_AT(env->bt_mem_start, cn->group_num); + cn->unset_addr_list = env->unset_addr_list; } # ifdef USE_NAMED_GROUP # ifdef USE_PERL_SUBEXP_CALL else if (cn->name == cn->name_end) { - goto set_call_attr; + goto set_call_attr; } # endif else { - int *refs; - - int n = onig_name_to_group_numbers(env->reg, cn->name, cn->name_end, - &refs); - if (n <= 0) { - onig_scan_env_set_error_string(env, - ONIGERR_UNDEFINED_NAME_REFERENCE, cn->name, cn->name_end); - return ONIGERR_UNDEFINED_NAME_REFERENCE; - } - else if (n > 1 && - ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME_CALL)) { - onig_scan_env_set_error_string(env, - ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL, cn->name, cn->name_end); - return ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL; - } - else { - cn->group_num = refs[0]; - goto set_call_attr; - } + int *refs; + + int n = onig_name_to_group_numbers(env->reg, cn->name, cn->name_end, + &refs); + if (n <= 0) { + onig_scan_env_set_error_string(env, + ONIGERR_UNDEFINED_NAME_REFERENCE, cn->name, cn->name_end); + return ONIGERR_UNDEFINED_NAME_REFERENCE; + } + else if (n > 1 && + ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME_CALL)) { + onig_scan_env_set_error_string(env, + ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL, cn->name, cn->name_end); + return ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL; + } + else { + cn->group_num = refs[0]; + goto set_call_attr; + } } # endif } @@ -3287,8 +3287,8 @@ setup_subexp_call(Node* node, ScanEnv* env) case ANCHOR_PREC_READ_NOT: case ANCHOR_LOOK_BEHIND: case ANCHOR_LOOK_BEHIND_NOT: - r = setup_subexp_call(an->target, env); - break; + r = setup_subexp_call(an->target, env); + break; } } break; @@ -3370,26 +3370,26 @@ next_setup(Node* node, Node* next_node, regex_t* reg) Node* n = get_head_value_node(next_node, 1, reg); /* '\0': for UTF-16BE etc... */ if (IS_NOT_NULL(n) && NSTR(n)->s[0] != '\0') { - qn->next_head_exact = n; + qn->next_head_exact = n; } #endif /* automatic possessification a*b ==> (?>a*)b */ if (qn->lower <= 1) { - int ttype = NTYPE(qn->target); - if (IS_NODE_TYPE_SIMPLE(ttype)) { - Node *x, *y; - x = get_head_value_node(qn->target, 0, reg); - if (IS_NOT_NULL(x)) { - y = get_head_value_node(next_node, 0, reg); - if (IS_NOT_NULL(y) && is_not_included(x, y, reg)) { - Node* en = onig_node_new_enclose(ENCLOSE_STOP_BACKTRACK); - CHECK_NULL_RETURN_MEMERR(en); - SET_ENCLOSE_STATUS(en, NST_STOP_BT_SIMPLE_REPEAT); - swap_node(node, en); - NENCLOSE(node)->target = en; - } - } - } + int ttype = NTYPE(qn->target); + if (IS_NODE_TYPE_SIMPLE(ttype)) { + Node *x, *y; + x = get_head_value_node(qn->target, 0, reg); + if (IS_NOT_NULL(x)) { + y = get_head_value_node(next_node, 0, reg); + if (IS_NOT_NULL(y) && is_not_included(x, y, reg)) { + Node* en = onig_node_new_enclose(ENCLOSE_STOP_BACKTRACK); + CHECK_NULL_RETURN_MEMERR(en); + SET_ENCLOSE_STATUS(en, NST_STOP_BT_SIMPLE_REPEAT); + swap_node(node, en); + NENCLOSE(node)->target = en; + } + } + } } } } @@ -3425,15 +3425,15 @@ update_string_node_case_fold(regex_t* reg, Node *node) len = ONIGENC_MBC_CASE_FOLD(reg->enc, reg->case_fold_flag, &p, end, buf); for (i = 0; i < len; i++) { if (sp >= ebuf) { - UChar* p = (UChar* )xrealloc(sbuf, sbuf_size * 2); - if (IS_NULL(p)) { - xfree(sbuf); - return ONIGERR_MEMORY; - } - sbuf = p; - sp = sbuf + sbuf_size; - sbuf_size *= 2; - ebuf = sbuf + sbuf_size; + UChar* p = (UChar* )xrealloc(sbuf, sbuf_size * 2); + if (IS_NULL(p)) { + xfree(sbuf); + return ONIGERR_MEMORY; + } + sbuf = p; + sp = sbuf + sbuf_size; + sbuf_size *= 2; + ebuf = sbuf + sbuf_size; } *sp++ = buf[i]; @@ -3448,7 +3448,7 @@ update_string_node_case_fold(regex_t* reg, Node *node) static int expand_case_fold_make_rem_string(Node** rnode, UChar *s, UChar *end, - regex_t* reg) + regex_t* reg) { int r; Node *node; @@ -3470,7 +3470,7 @@ expand_case_fold_make_rem_string(Node** rnode, UChar *s, UChar *end, static int is_case_fold_variable_len(int item_num, OnigCaseFoldCodeItem items[], - int slen) + int slen) { int i; @@ -3487,8 +3487,8 @@ is_case_fold_variable_len(int item_num, OnigCaseFoldCodeItem items[], static int expand_case_fold_string_alt(int item_num, OnigCaseFoldCodeItem items[], - UChar *p, int slen, UChar *end, - regex_t* reg, Node **rnode) + UChar *p, int slen, UChar *end, + regex_t* reg, Node **rnode) { int r, i, j, len, varlen; Node *anode, *var_anode, *snode, *xnode, *an; @@ -3533,8 +3533,8 @@ expand_case_fold_string_alt(int item_num, OnigCaseFoldCodeItem items[], for (j = 0; j < items[i].code_len; j++) { len = ONIGENC_CODE_TO_MBC(reg->enc, items[i].code[j], buf); if (len < 0) { - r = len; - goto mem_err2; + r = len; + goto mem_err2; } r = onig_node_str_cat(snode, buf, buf + len); @@ -3551,29 +3551,29 @@ expand_case_fold_string_alt(int item_num, OnigCaseFoldCodeItem items[], UChar *q = p + items[i].byte_len; if (q < end) { - r = expand_case_fold_make_rem_string(&rem, q, end, reg); - if (r != 0) { - onig_node_free(an); - goto mem_err2; - } - - xnode = onig_node_list_add(NULL_NODE, snode); - if (IS_NULL(xnode)) { - onig_node_free(an); - onig_node_free(rem); - goto mem_err2; - } - if (IS_NULL(onig_node_list_add(xnode, rem))) { - onig_node_free(an); - onig_node_free(xnode); - onig_node_free(rem); - goto mem_err; - } - - NCAR(an) = xnode; + r = expand_case_fold_make_rem_string(&rem, q, end, reg); + if (r != 0) { + onig_node_free(an); + goto mem_err2; + } + + xnode = onig_node_list_add(NULL_NODE, snode); + if (IS_NULL(xnode)) { + onig_node_free(an); + onig_node_free(rem); + goto mem_err2; + } + if (IS_NULL(onig_node_list_add(xnode, rem))) { + onig_node_free(an); + onig_node_free(xnode); + onig_node_free(rem); + goto mem_err; + } + + NCAR(an) = xnode; } else { - NCAR(an) = snode; + NCAR(an) = snode; } NCDR(var_anode) = an; @@ -3621,7 +3621,7 @@ expand_case_fold_string(Node* node, regex_t* reg) p = start; while (p < end) { n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(reg->enc, reg->case_fold_flag, - p, end, items); + p, end, items); if (n < 0) { r = n; goto err; @@ -3632,23 +3632,23 @@ expand_case_fold_string(Node* node, regex_t* reg) varlen = is_case_fold_variable_len(n, items, len); if (n == 0 || varlen == 0) { if (IS_NULL(snode)) { - if (IS_NULL(root) && IS_NOT_NULL(prev_node)) { + if (IS_NULL(root) && IS_NOT_NULL(prev_node)) { onig_node_free(top_root); - top_root = root = onig_node_list_add(NULL_NODE, prev_node); - if (IS_NULL(root)) { - onig_node_free(prev_node); - goto mem_err; - } - } - - prev_node = snode = onig_node_new_str(NULL, NULL); - if (IS_NULL(snode)) goto mem_err; - if (IS_NOT_NULL(root)) { - if (IS_NULL(onig_node_list_add(root, snode))) { - onig_node_free(snode); - goto mem_err; - } - } + top_root = root = onig_node_list_add(NULL_NODE, prev_node); + if (IS_NULL(root)) { + onig_node_free(prev_node); + goto mem_err; + } + } + + prev_node = snode = onig_node_new_str(NULL, NULL); + if (IS_NULL(snode)) goto mem_err; + if (IS_NOT_NULL(root)) { + if (IS_NULL(onig_node_list_add(root, snode))) { + onig_node_free(snode); + goto mem_err; + } + } } r = onig_node_str_cat(snode, p, p + len); @@ -3659,42 +3659,42 @@ expand_case_fold_string(Node* node, regex_t* reg) if (alt_num > THRESHOLD_CASE_FOLD_ALT_FOR_EXPANSION) break; if (IS_NOT_NULL(snode)) { - r = update_string_node_case_fold(reg, snode); - if (r == 0) { - NSTRING_SET_AMBIG(snode); - } + r = update_string_node_case_fold(reg, snode); + if (r == 0) { + NSTRING_SET_AMBIG(snode); + } } if (IS_NULL(root) && IS_NOT_NULL(prev_node)) { onig_node_free(top_root); - top_root = root = onig_node_list_add(NULL_NODE, prev_node); - if (IS_NULL(root)) { - onig_node_free(prev_node); - goto mem_err; - } + top_root = root = onig_node_list_add(NULL_NODE, prev_node); + if (IS_NULL(root)) { + onig_node_free(prev_node); + goto mem_err; + } } r = expand_case_fold_string_alt(n, items, p, len, end, reg, &prev_node); if (r < 0) goto mem_err; if (r == 1) { - if (IS_NULL(root)) { - top_root = prev_node; - } - else { - if (IS_NULL(onig_node_list_add(root, prev_node))) { - onig_node_free(prev_node); - goto mem_err; - } - } - - root = NCAR(prev_node); + if (IS_NULL(root)) { + top_root = prev_node; + } + else { + if (IS_NULL(onig_node_list_add(root, prev_node))) { + onig_node_free(prev_node); + goto mem_err; + } + } + + root = NCAR(prev_node); } else { /* r == 0 */ - if (IS_NOT_NULL(root)) { - if (IS_NULL(onig_node_list_add(root, prev_node))) { - onig_node_free(prev_node); - goto mem_err; - } - } + if (IS_NOT_NULL(root)) { + if (IS_NULL(onig_node_list_add(root, prev_node))) { + onig_node_free(prev_node); + goto mem_err; + } + } } snode = NULL_NODE; @@ -3719,9 +3719,9 @@ expand_case_fold_string(Node* node, regex_t* reg) onig_node_free(top_root); top_root = root = onig_node_list_add(NULL_NODE, prev_node); if (IS_NULL(root)) { - onig_node_free(srem); - onig_node_free(prev_node); - goto mem_err; + onig_node_free(srem); + onig_node_free(prev_node); + goto mem_err; } } @@ -3730,8 +3730,8 @@ expand_case_fold_string(Node* node, regex_t* reg) } else { if (IS_NULL(onig_node_list_add(root, srem))) { - onig_node_free(srem); - goto mem_err; + onig_node_free(srem); + goto mem_err; } } } @@ -3771,7 +3771,7 @@ setup_comb_exp_check(Node* node, int state, ScanEnv* env) case NT_LIST: { do { - r = setup_comb_exp_check(NCAR(node), r, env); + r = setup_comb_exp_check(NCAR(node), r, env); } while (r >= 0 && IS_NOT_NULL(node = NCDR(node))); } break; @@ -3780,8 +3780,8 @@ setup_comb_exp_check(Node* node, int state, ScanEnv* env) { int ret; do { - ret = setup_comb_exp_check(NCAR(node), state, env); - r |= ret; + ret = setup_comb_exp_check(NCAR(node), state, env); + r |= ret; } while (ret >= 0 && IS_NOT_NULL(node = NCDR(node))); } break; @@ -3795,55 +3795,55 @@ setup_comb_exp_check(Node* node, int state, ScanEnv* env) int var_num; if (! IS_REPEAT_INFINITE(qn->upper)) { - if (qn->upper > 1) { - /* {0,1}, {1,1} are allowed */ - child_state |= CEC_IN_FINITE_REPEAT; - - /* check (a*){n,m}, (a+){n,m} => (a*){n,n}, (a+){n,n} */ - if (env->backrefed_mem == 0) { - if (NTYPE(qn->target) == NT_ENCLOSE) { - EncloseNode* en = NENCLOSE(qn->target); - if (en->type == ENCLOSE_MEMORY) { - if (NTYPE(en->target) == NT_QTFR) { - QtfrNode* q = NQTFR(en->target); - if (IS_REPEAT_INFINITE(q->upper) - && q->greedy == qn->greedy) { - qn->upper = (qn->lower == 0 ? 1 : qn->lower); - if (qn->upper == 1) - child_state = state; - } - } - } - } - } - } + if (qn->upper > 1) { + /* {0,1}, {1,1} are allowed */ + child_state |= CEC_IN_FINITE_REPEAT; + + /* check (a*){n,m}, (a+){n,m} => (a*){n,n}, (a+){n,n} */ + if (env->backrefed_mem == 0) { + if (NTYPE(qn->target) == NT_ENCLOSE) { + EncloseNode* en = NENCLOSE(qn->target); + if (en->type == ENCLOSE_MEMORY) { + if (NTYPE(en->target) == NT_QTFR) { + QtfrNode* q = NQTFR(en->target); + if (IS_REPEAT_INFINITE(q->upper) + && q->greedy == qn->greedy) { + qn->upper = (qn->lower == 0 ? 1 : qn->lower); + if (qn->upper == 1) + child_state = state; + } + } + } + } + } + } } if (state & CEC_IN_FINITE_REPEAT) { - qn->comb_exp_check_num = -1; + qn->comb_exp_check_num = -1; } else { - if (IS_REPEAT_INFINITE(qn->upper)) { - var_num = CEC_INFINITE_NUM; - child_state |= CEC_IN_INFINITE_REPEAT; - } - else { - var_num = qn->upper - qn->lower; - } - - if (var_num >= CEC_THRES_NUM_BIG_REPEAT) - add_state |= CEC_CONT_BIG_REPEAT; - - if (((state & CEC_IN_INFINITE_REPEAT) != 0 && var_num != 0) || - ((state & CEC_CONT_BIG_REPEAT) != 0 && - var_num >= CEC_THRES_NUM_BIG_REPEAT)) { - if (qn->comb_exp_check_num == 0) { - env->num_comb_exp_check++; - qn->comb_exp_check_num = env->num_comb_exp_check; - if (env->curr_max_regnum > env->comb_exp_max_regnum) - env->comb_exp_max_regnum = env->curr_max_regnum; - } - } + if (IS_REPEAT_INFINITE(qn->upper)) { + var_num = CEC_INFINITE_NUM; + child_state |= CEC_IN_INFINITE_REPEAT; + } + else { + var_num = qn->upper - qn->lower; + } + + if (var_num >= CEC_THRES_NUM_BIG_REPEAT) + add_state |= CEC_CONT_BIG_REPEAT; + + if (((state & CEC_IN_INFINITE_REPEAT) != 0 && var_num != 0) || + ((state & CEC_CONT_BIG_REPEAT) != 0 && + var_num >= CEC_THRES_NUM_BIG_REPEAT)) { + if (qn->comb_exp_check_num == 0) { + env->num_comb_exp_check++; + qn->comb_exp_check_num = env->num_comb_exp_check; + if (env->curr_max_regnum > env->comb_exp_max_regnum) + env->comb_exp_max_regnum = env->curr_max_regnum; + } + } } r = setup_comb_exp_check(target, child_state, env); @@ -3857,17 +3857,17 @@ setup_comb_exp_check(Node* node, int state, ScanEnv* env) switch (en->type) { case ENCLOSE_MEMORY: - { - if (env->curr_max_regnum < en->regnum) - env->curr_max_regnum = en->regnum; + { + if (env->curr_max_regnum < en->regnum) + env->curr_max_regnum = en->regnum; - r = setup_comb_exp_check(en->target, state, env); - } - break; + r = setup_comb_exp_check(en->target, state, env); + } + break; default: - r = setup_comb_exp_check(en->target, state, env); - break; + r = setup_comb_exp_check(en->target, state, env); + break; } } break; @@ -3917,11 +3917,11 @@ restart: { Node* prev = NULL_NODE; do { - r = setup_tree(NCAR(node), reg, state, env); - if (IS_NOT_NULL(prev) && r == 0) { - r = next_setup(prev, NCAR(node), reg); - } - prev = NCAR(node); + r = setup_tree(NCAR(node), reg, state, env); + if (IS_NOT_NULL(prev) && r == 0) { + r = next_setup(prev, NCAR(node), reg); + } + prev = NCAR(node); } while (r == 0 && IS_NOT_NULL(node = NCDR(node))); } break; @@ -3958,15 +3958,15 @@ restart: BRefNode* br = NBREF(node); p = BACKREFS_P(br); for (i = 0; i < br->back_num; i++) { - if (p[i] > env->num_mem) return ONIGERR_INVALID_BACKREF; - BIT_STATUS_ON_AT(env->backrefed_mem, p[i]); - BIT_STATUS_ON_AT(env->bt_mem_start, p[i]); + if (p[i] > env->num_mem) return ONIGERR_INVALID_BACKREF; + BIT_STATUS_ON_AT(env->backrefed_mem, p[i]); + BIT_STATUS_ON_AT(env->bt_mem_start, p[i]); #ifdef USE_BACKREF_WITH_LEVEL - if (IS_BACKREF_NEST_LEVEL(br)) { - BIT_STATUS_ON_AT(env->bt_mem_end, p[i]); - } + if (IS_BACKREF_NEST_LEVEL(br)) { + BIT_STATUS_ON_AT(env->bt_mem_end, p[i]); + } #endif - SET_ENCLOSE_STATUS(nodes[p[i]], NST_MEM_BACKREFED); + SET_ENCLOSE_STATUS(nodes[p[i]], NST_MEM_BACKREFED); } } break; @@ -3978,100 +3978,100 @@ restart: Node* target = qn->target; if ((state & IN_REPEAT) != 0) { - qn->state |= NST_IN_REPEAT; + qn->state |= NST_IN_REPEAT; } if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 1) { - r = get_min_match_length(target, &d, env); - if (r) break; - if (d == 0) { - qn->target_empty_info = NQ_TARGET_IS_EMPTY; + r = get_min_match_length(target, &d, env); + if (r) break; + if (d == 0) { + qn->target_empty_info = NQ_TARGET_IS_EMPTY; #ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT - r = quantifiers_memory_node_info(target); - if (r < 0) break; - if (r > 0) { - qn->target_empty_info = r; - } + r = quantifiers_memory_node_info(target); + if (r < 0) break; + if (r > 0) { + qn->target_empty_info = r; + } #endif #if 0 - r = get_max_match_length(target, &d, env); - if (r == 0 && d == 0) { - /* ()* ==> ()?, ()+ ==> () */ - qn->upper = 1; - if (qn->lower > 1) qn->lower = 1; - if (NTYPE(target) == NT_STR) { - qn->upper = qn->lower = 0; /* /(?:)+/ ==> // */ - } - } + r = get_max_match_length(target, &d, env); + if (r == 0 && d == 0) { + /* ()* ==> ()?, ()+ ==> () */ + qn->upper = 1; + if (qn->lower > 1) qn->lower = 1; + if (NTYPE(target) == NT_STR) { + qn->upper = qn->lower = 0; /* /(?:)+/ ==> // */ + } + } #endif - } + } } state |= IN_REPEAT; if (qn->lower != qn->upper) - state |= IN_VAR_REPEAT; + state |= IN_VAR_REPEAT; r = setup_tree(target, reg, state, env); if (r) break; /* expand string */ #define EXPAND_STRING_MAX_LENGTH 100 if (NTYPE(target) == NT_STR) { - if (qn->lower > 1) { - int i, n = qn->lower; - OnigDistance len = NSTRING_LEN(target); - StrNode* sn = NSTR(target); - Node* np; - - np = onig_node_new_str(sn->s, sn->end); - if (IS_NULL(np)) return ONIGERR_MEMORY; - NSTR(np)->flag = sn->flag; - - for (i = 1; i < n && (i+1) * len <= EXPAND_STRING_MAX_LENGTH; i++) { - r = onig_node_str_cat(np, sn->s, sn->end); - if (r) { - onig_node_free(np); - return r; - } - } - if (i < qn->upper || IS_REPEAT_INFINITE(qn->upper)) { - Node *np1, *np2; - - qn->lower -= i; - if (! IS_REPEAT_INFINITE(qn->upper)) - qn->upper -= i; - - np1 = onig_node_new_list(np, NULL); - if (IS_NULL(np1)) { - onig_node_free(np); - return ONIGERR_MEMORY; - } - swap_node(np1, node); - np2 = onig_node_list_add(node, np1); - if (IS_NULL(np2)) { - onig_node_free(np1); - return ONIGERR_MEMORY; - } - } - else { - swap_node(np, node); - onig_node_free(np); - } - break; /* break case NT_QTFR: */ - } + if (qn->lower > 1) { + int i, n = qn->lower; + OnigDistance len = NSTRING_LEN(target); + StrNode* sn = NSTR(target); + Node* np; + + np = onig_node_new_str(sn->s, sn->end); + if (IS_NULL(np)) return ONIGERR_MEMORY; + NSTR(np)->flag = sn->flag; + + for (i = 1; i < n && (i+1) * len <= EXPAND_STRING_MAX_LENGTH; i++) { + r = onig_node_str_cat(np, sn->s, sn->end); + if (r) { + onig_node_free(np); + return r; + } + } + if (i < qn->upper || IS_REPEAT_INFINITE(qn->upper)) { + Node *np1, *np2; + + qn->lower -= i; + if (! IS_REPEAT_INFINITE(qn->upper)) + qn->upper -= i; + + np1 = onig_node_new_list(np, NULL); + if (IS_NULL(np1)) { + onig_node_free(np); + return ONIGERR_MEMORY; + } + swap_node(np1, node); + np2 = onig_node_list_add(node, np1); + if (IS_NULL(np2)) { + onig_node_free(np1); + return ONIGERR_MEMORY; + } + } + else { + swap_node(np, node); + onig_node_free(np); + } + break; /* break case NT_QTFR: */ + } } #ifdef USE_OP_PUSH_OR_JUMP_EXACT if (qn->greedy && (qn->target_empty_info != 0)) { - if (NTYPE(target) == NT_QTFR) { - QtfrNode* tqn = NQTFR(target); - if (IS_NOT_NULL(tqn->head_exact)) { - qn->head_exact = tqn->head_exact; - tqn->head_exact = NULL; - } - } - else { - qn->head_exact = get_head_value_node(qn->target, 1, reg); - } + if (NTYPE(target) == NT_QTFR) { + QtfrNode* tqn = NQTFR(target); + if (IS_NOT_NULL(tqn->head_exact)) { + qn->head_exact = tqn->head_exact; + tqn->head_exact = NULL; + } + } + else { + qn->head_exact = get_head_value_node(qn->target, 1, reg); + } } #endif } @@ -4083,61 +4083,61 @@ restart: switch (en->type) { case ENCLOSE_OPTION: - { - OnigOptionType options = reg->options; - reg->options = NENCLOSE(node)->option; - r = setup_tree(NENCLOSE(node)->target, reg, state, env); - reg->options = options; - } - break; + { + OnigOptionType options = reg->options; + reg->options = NENCLOSE(node)->option; + r = setup_tree(NENCLOSE(node)->target, reg, state, env); + reg->options = options; + } + break; case ENCLOSE_MEMORY: - if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT | IN_CALL)) != 0) { - BIT_STATUS_ON_AT(env->bt_mem_start, en->regnum); - /* SET_ENCLOSE_STATUS(node, NST_MEM_IN_ALT_NOT); */ - } - if (IS_ENCLOSE_CALLED(en)) - state |= IN_CALL; - if (IS_ENCLOSE_RECURSION(en)) - state |= IN_RECCALL; - else if ((state & IN_RECCALL) != 0) - SET_CALL_RECURSION(node); - r = setup_tree(en->target, reg, state, env); - break; + if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT | IN_CALL)) != 0) { + BIT_STATUS_ON_AT(env->bt_mem_start, en->regnum); + /* SET_ENCLOSE_STATUS(node, NST_MEM_IN_ALT_NOT); */ + } + if (IS_ENCLOSE_CALLED(en)) + state |= IN_CALL; + if (IS_ENCLOSE_RECURSION(en)) + state |= IN_RECCALL; + else if ((state & IN_RECCALL) != 0) + SET_CALL_RECURSION(node); + r = setup_tree(en->target, reg, state, env); + break; case ENCLOSE_STOP_BACKTRACK: - { - Node* target = en->target; - r = setup_tree(target, reg, state, env); - if (NTYPE(target) == NT_QTFR) { - QtfrNode* tqn = NQTFR(target); - if (IS_REPEAT_INFINITE(tqn->upper) && tqn->lower <= 1 && - tqn->greedy != 0) { /* (?>a*), a*+ etc... */ - int qtype = NTYPE(tqn->target); - if (IS_NODE_TYPE_SIMPLE(qtype)) - SET_ENCLOSE_STATUS(node, NST_STOP_BT_SIMPLE_REPEAT); - } - } - } - break; + { + Node* target = en->target; + r = setup_tree(target, reg, state, env); + if (NTYPE(target) == NT_QTFR) { + QtfrNode* tqn = NQTFR(target); + if (IS_REPEAT_INFINITE(tqn->upper) && tqn->lower <= 1 && + tqn->greedy != 0) { /* (?>a*), a*+ etc... */ + int qtype = NTYPE(tqn->target); + if (IS_NODE_TYPE_SIMPLE(qtype)) + SET_ENCLOSE_STATUS(node, NST_STOP_BT_SIMPLE_REPEAT); + } + } + } + break; case ENCLOSE_CONDITION: #ifdef USE_NAMED_GROUP - if (! IS_ENCLOSE_NAME_REF(NENCLOSE(node)) && - env->num_named > 0 && - IS_SYNTAX_BV(env->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) && - !ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_CAPTURE_GROUP)) { - return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED; - } + if (! IS_ENCLOSE_NAME_REF(NENCLOSE(node)) && + env->num_named > 0 && + IS_SYNTAX_BV(env->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) && + !ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_CAPTURE_GROUP)) { + return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED; + } #endif - if (NENCLOSE(node)->regnum > env->num_mem) - return ONIGERR_INVALID_BACKREF; - r = setup_tree(NENCLOSE(node)->target, reg, state, env); - break; + if (NENCLOSE(node)->regnum > env->num_mem) + return ONIGERR_INVALID_BACKREF; + r = setup_tree(NENCLOSE(node)->target, reg, state, env); + break; case ENCLOSE_ABSENT: - r = setup_tree(NENCLOSE(node)->target, reg, state, env); - break; + r = setup_tree(NENCLOSE(node)->target, reg, state, env); + break; } } break; @@ -4148,11 +4148,11 @@ restart: switch (an->type) { case ANCHOR_PREC_READ: - r = setup_tree(an->target, reg, state, env); - break; + r = setup_tree(an->target, reg, state, env); + break; case ANCHOR_PREC_READ_NOT: - r = setup_tree(an->target, reg, (state | IN_NOT), env); - break; + r = setup_tree(an->target, reg, (state | IN_NOT), env); + break; /* allowed node types in look-behind */ #define ALLOWED_TYPE_IN_LB \ @@ -4174,30 +4174,30 @@ restart: ANCHOR_WORD_BEGIN | ANCHOR_WORD_END ) case ANCHOR_LOOK_BEHIND: - { - r = check_type_tree(an->target, ALLOWED_TYPE_IN_LB, - ALLOWED_ENCLOSE_IN_LB, ALLOWED_ANCHOR_IN_LB); - if (r < 0) return r; - if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; - if (NTYPE(node) != NT_ANCHOR) goto restart; - r = setup_tree(an->target, reg, state, env); - if (r != 0) return r; - r = setup_look_behind(node, reg, env); - } - break; + { + r = check_type_tree(an->target, ALLOWED_TYPE_IN_LB, + ALLOWED_ENCLOSE_IN_LB, ALLOWED_ANCHOR_IN_LB); + if (r < 0) return r; + if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; + if (NTYPE(node) != NT_ANCHOR) goto restart; + r = setup_tree(an->target, reg, state, env); + if (r != 0) return r; + r = setup_look_behind(node, reg, env); + } + break; case ANCHOR_LOOK_BEHIND_NOT: - { - r = check_type_tree(an->target, ALLOWED_TYPE_IN_LB, - ALLOWED_ENCLOSE_IN_LB_NOT, ALLOWED_ANCHOR_IN_LB_NOT); - if (r < 0) return r; - if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; - if (NTYPE(node) != NT_ANCHOR) goto restart; - r = setup_tree(an->target, reg, (state | IN_NOT), env); - if (r != 0) return r; - r = setup_look_behind(node, reg, env); - } - break; + { + r = check_type_tree(an->target, ALLOWED_TYPE_IN_LB, + ALLOWED_ENCLOSE_IN_LB_NOT, ALLOWED_ANCHOR_IN_LB_NOT); + if (r < 0) return r; + if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN; + if (NTYPE(node) != NT_ANCHOR) goto restart; + r = setup_tree(an->target, reg, (state | IN_NOT), env); + if (r != 0) return r; + r = setup_look_behind(node, reg, env); + } + break; } } break; @@ -4213,7 +4213,7 @@ restart: /* set skip map for Boyer-Moore search */ static int set_bm_skip(UChar* s, UChar* end, regex_t* reg, - UChar skip[], int** int_skip, int ignore_case) + UChar skip[], int** int_skip, int ignore_case) { OnigDistance i, len; int clen, flen, n, j, k; @@ -4229,24 +4229,24 @@ set_bm_skip(UChar* s, UChar* end, regex_t* reg, for (i = 0; i < len - 1; i += clen) { p = s + i; if (ignore_case) - n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, - p, end, items); + n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, + p, end, items); clen = enclen(enc, p, end); if (p + clen > end) - clen = (int )(end - p); + clen = (int )(end - p); for (j = 0; j < n; j++) { - if ((items[j].code_len != 1) || (items[j].byte_len != clen)) - return 1; /* different length isn't supported. */ - flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); - if (flen != clen) - return 1; /* different length isn't supported. */ + if ((items[j].code_len != 1) || (items[j].byte_len != clen)) + return 1; /* different length isn't supported. */ + flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); + if (flen != clen) + return 1; /* different length isn't supported. */ } for (j = 0; j < clen; j++) { - skip[s[i + j]] = (UChar )(len - 1 - i - j); - for (k = 0; k < n; k++) { - skip[buf[k][j]] = (UChar )(len - 1 - i - j); - } + skip[s[i + j]] = (UChar )(len - 1 - i - j); + for (k = 0; k < n; k++) { + skip[buf[k][j]] = (UChar )(len - 1 - i - j); + } } } } @@ -4265,24 +4265,24 @@ set_bm_skip(UChar* s, UChar* end, regex_t* reg, for (i = 0; i < len - 1; i += clen) { p = s + i; if (ignore_case) - n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, - p, end, items); + n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, + p, end, items); clen = enclen(enc, p, end); if (p + clen > end) - clen = (int )(end - p); + clen = (int )(end - p); for (j = 0; j < n; j++) { - if ((items[j].code_len != 1) || (items[j].byte_len != clen)) - return 1; /* different length isn't supported. */ - flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); - if (flen != clen) - return 1; /* different length isn't supported. */ + if ((items[j].code_len != 1) || (items[j].byte_len != clen)) + return 1; /* different length isn't supported. */ + flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); + if (flen != clen) + return 1; /* different length isn't supported. */ } for (j = 0; j < clen; j++) { - (*int_skip)[s[i + j]] = (int )(len - 1 - i - j); - for (k = 0; k < n; k++) { - (*int_skip)[buf[k][j]] = (int )(len - 1 - i - j); - } + (*int_skip)[s[i + j]] = (int )(len - 1 - i - j); + for (k = 0; k < n; k++) { + (*int_skip)[buf[k][j]] = (int )(len - 1 - i - j); + } } } # endif @@ -4295,7 +4295,7 @@ set_bm_skip(UChar* s, UChar* end, regex_t* reg, /* set skip map for Sunday's quick search */ static int set_bm_skip(UChar* s, UChar* end, regex_t* reg, - UChar skip[], int** int_skip, int ignore_case) + UChar skip[], int** int_skip, int ignore_case) { OnigDistance i, len; int clen, flen, n, j, k; @@ -4311,24 +4311,24 @@ set_bm_skip(UChar* s, UChar* end, regex_t* reg, for (i = 0; i < len; i += clen) { p = s + i; if (ignore_case) - n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, - p, end, items); + n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, + p, end, items); clen = enclen(enc, p, end); if (p + clen > end) - clen = (int )(end - p); + clen = (int )(end - p); for (j = 0; j < n; j++) { - if ((items[j].code_len != 1) || (items[j].byte_len != clen)) - return 1; /* different length isn't supported. */ - flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); - if (flen != clen) - return 1; /* different length isn't supported. */ + if ((items[j].code_len != 1) || (items[j].byte_len != clen)) + return 1; /* different length isn't supported. */ + flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); + if (flen != clen) + return 1; /* different length isn't supported. */ } for (j = 0; j < clen; j++) { - skip[s[i + j]] = (UChar )(len - i - j); - for (k = 0; k < n; k++) { - skip[buf[k][j]] = (UChar )(len - i - j); - } + skip[s[i + j]] = (UChar )(len - i - j); + for (k = 0; k < n; k++) { + skip[buf[k][j]] = (UChar )(len - i - j); + } } } } @@ -4347,24 +4347,24 @@ set_bm_skip(UChar* s, UChar* end, regex_t* reg, for (i = 0; i < len; i += clen) { p = s + i; if (ignore_case) - n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, - p, end, items); + n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, + p, end, items); clen = enclen(enc, p, end); if (p + clen > end) - clen = (int )(end - p); + clen = (int )(end - p); for (j = 0; j < n; j++) { - if ((items[j].code_len != 1) || (items[j].byte_len != clen)) - return 1; /* different length isn't supported. */ - flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); - if (flen != clen) - return 1; /* different length isn't supported. */ + if ((items[j].code_len != 1) || (items[j].byte_len != clen)) + return 1; /* different length isn't supported. */ + flen = ONIGENC_CODE_TO_MBC(enc, items[j].code[0], buf[j]); + if (flen != clen) + return 1; /* different length isn't supported. */ } for (j = 0; j < clen; j++) { - (*int_skip)[s[i + j]] = (int )(len - i - j); - for (k = 0; k < n; k++) { - (*int_skip)[buf[k][j]] = (int )(len - i - j); - } + (*int_skip)[s[i + j]] = (int )(len - i - j); + for (k = 0; k < n; k++) { + (*int_skip)[buf[k][j]] = (int )(len - i - j); + } } } # endif @@ -4562,7 +4562,7 @@ copy_opt_anc_info(OptAncInfo* to, OptAncInfo* from) static void concat_opt_anc_info(OptAncInfo* to, OptAncInfo* left, OptAncInfo* right, - OnigDistance left_len, OnigDistance right_len) + OnigDistance left_len, OnigDistance right_len) { clear_opt_anc_info(to); @@ -4678,7 +4678,7 @@ concat_opt_exact_info(OptExactInfo* to, OptExactInfo* add, OnigEncoding enc) static void concat_opt_exact_info_str(OptExactInfo* to, UChar* s, UChar* end, - int raw ARG_UNUSED, OnigEncoding enc) + int raw ARG_UNUSED, OnigEncoding enc) { int i, j, len; UChar *p; @@ -4923,7 +4923,7 @@ concat_left_node_opt_info(OnigEncoding enc, NodeOptInfo* to, NodeOptInfo* add) if (add->exb.len > 0 && to->len.max == 0) { concat_opt_anc_info(&tanc, &to->anc, &add->exb.anc, - to->len.max, add->len.max); + to->len.max, add->len.max); copy_opt_anc_info(&add->exb.anc, &tanc); } @@ -4954,12 +4954,12 @@ concat_left_node_opt_info(OnigEncoding enc, NodeOptInfo* to, NodeOptInfo* add) if (to->expr.len > 0) { if (add->len.max > 0) { if (to->expr.len > (int )add->len.max) - to->expr.len = (int )add->len.max; + to->expr.len = (int )add->len.max; if (to->expr.mmd.max == 0) - select_opt_exact_info(enc, &to->exb, &to->expr); + select_opt_exact_info(enc, &to->exb, &to->expr); else - select_opt_exact_info(enc, &to->exm, &to->expr); + select_opt_exact_info(enc, &to->exm, &to->expr); } } else if (add->expr.len > 0) { @@ -5005,11 +5005,11 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) copy_opt_env(&nenv, env); do { - r = optimize_node_left(NCAR(nd), &nopt, &nenv); - if (r == 0) { - add_mml(&nenv.mmd, &nopt.len); - concat_left_node_opt_info(env->enc, opt, &nopt); - } + r = optimize_node_left(NCAR(nd), &nopt, &nenv); + if (r == 0) { + add_mml(&nenv.mmd, &nopt.len); + concat_left_node_opt_info(env->enc, opt, &nopt); + } } while (r == 0 && IS_NOT_NULL(nd = NCDR(nd))); } break; @@ -5020,11 +5020,11 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) Node* nd = node; do { - r = optimize_node_left(NCAR(nd), &nopt, env); - if (r == 0) { - if (nd == node) copy_node_opt_info(opt, &nopt); - else alt_merge_node_opt_info(opt, &nopt, env); - } + r = optimize_node_left(NCAR(nd), &nopt, env); + if (r == 0) { + if (nd == node) copy_node_opt_info(opt, &nopt); + else alt_merge_node_opt_info(opt, &nopt, env); + } } while ((r == 0) && IS_NOT_NULL(nd = NCDR(nd))); } break; @@ -5036,40 +5036,40 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) int is_raw = NSTRING_IS_RAW(node); if (! NSTRING_IS_AMBIG(node)) { - concat_opt_exact_info_str(&opt->exb, sn->s, sn->end, - is_raw, env->enc); - opt->exb.ignore_case = 0; - if (slen > 0) { - add_char_opt_map_info(&opt->map, *(sn->s), env->enc); - } - set_mml(&opt->len, slen, slen); + concat_opt_exact_info_str(&opt->exb, sn->s, sn->end, + is_raw, env->enc); + opt->exb.ignore_case = 0; + if (slen > 0) { + add_char_opt_map_info(&opt->map, *(sn->s), env->enc); + } + set_mml(&opt->len, slen, slen); } else { - OnigDistance max; - - if (NSTRING_IS_DONT_GET_OPT_INFO(node)) { - int n = onigenc_strlen(env->enc, sn->s, sn->end); - max = ONIGENC_MBC_MAXLEN_DIST(env->enc) * (OnigDistance)n; - } - else { - concat_opt_exact_info_str(&opt->exb, sn->s, sn->end, - is_raw, env->enc); - opt->exb.ignore_case = 1; - - if (slen > 0) { - r = add_char_amb_opt_map_info(&opt->map, sn->s, sn->end, - env->enc, env->case_fold_flag); - if (r != 0) break; - } - - max = slen; - } - - set_mml(&opt->len, slen, max); + OnigDistance max; + + if (NSTRING_IS_DONT_GET_OPT_INFO(node)) { + int n = onigenc_strlen(env->enc, sn->s, sn->end); + max = ONIGENC_MBC_MAXLEN_DIST(env->enc) * (OnigDistance)n; + } + else { + concat_opt_exact_info_str(&opt->exb, sn->s, sn->end, + is_raw, env->enc); + opt->exb.ignore_case = 1; + + if (slen > 0) { + r = add_char_amb_opt_map_info(&opt->map, sn->s, sn->end, + env->enc, env->case_fold_flag); + if (r != 0) break; + } + + max = slen; + } + + set_mml(&opt->len, slen, max); } if ((OnigDistance )opt->exb.len == slen) - opt->exb.reach_end = 1; + opt->exb.reach_end = 1; } break; @@ -5081,19 +5081,19 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) /* no need to check ignore case. (set in setup_tree()) */ if (IS_NOT_NULL(cc->mbuf) || IS_NCCLASS_NOT(cc)) { - OnigDistance min = ONIGENC_MBC_MINLEN(env->enc); - OnigDistance max = ONIGENC_MBC_MAXLEN_DIST(env->enc); + OnigDistance min = ONIGENC_MBC_MINLEN(env->enc); + OnigDistance max = ONIGENC_MBC_MAXLEN_DIST(env->enc); - set_mml(&opt->len, min, max); + set_mml(&opt->len, min, max); } else { - for (i = 0; i < SINGLE_BYTE_SIZE; i++) { - z = BITSET_AT(cc->bs, i); - if ((z && !IS_NCCLASS_NOT(cc)) || (!z && IS_NCCLASS_NOT(cc))) { - add_char_opt_map_info(&opt->map, (UChar )i, env->enc); - } - } - set_mml(&opt->len, 1, 1); + for (i = 0; i < SINGLE_BYTE_SIZE; i++) { + z = BITSET_AT(cc->bs, i); + if ((z && !IS_NCCLASS_NOT(cc)) || (!z && IS_NCCLASS_NOT(cc))) { + add_char_opt_map_info(&opt->map, (UChar )i, env->enc); + } + } + set_mml(&opt->len, 1, 1); } } break; @@ -5106,30 +5106,30 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) max = ONIGENC_MBC_MAXLEN_DIST(env->enc); if (max == 1) { - min = 1; - - maxcode = NCTYPE(node)->ascii_range ? 0x80 : SINGLE_BYTE_SIZE; - switch (NCTYPE(node)->ctype) { - case ONIGENC_CTYPE_WORD: - if (NCTYPE(node)->not != 0) { - for (i = 0; i < SINGLE_BYTE_SIZE; i++) { - if (! ONIGENC_IS_CODE_WORD(env->enc, i) || i >= maxcode) { - add_char_opt_map_info(&opt->map, (UChar )i, env->enc); - } - } - } - else { - for (i = 0; i < maxcode; i++) { - if (ONIGENC_IS_CODE_WORD(env->enc, i)) { - add_char_opt_map_info(&opt->map, (UChar )i, env->enc); - } - } - } - break; - } + min = 1; + + maxcode = NCTYPE(node)->ascii_range ? 0x80 : SINGLE_BYTE_SIZE; + switch (NCTYPE(node)->ctype) { + case ONIGENC_CTYPE_WORD: + if (NCTYPE(node)->not != 0) { + for (i = 0; i < SINGLE_BYTE_SIZE; i++) { + if (! ONIGENC_IS_CODE_WORD(env->enc, i) || i >= maxcode) { + add_char_opt_map_info(&opt->map, (UChar )i, env->enc); + } + } + } + else { + for (i = 0; i < maxcode; i++) { + if (ONIGENC_IS_CODE_WORD(env->enc, i)) { + add_char_opt_map_info(&opt->map, (UChar )i, env->enc); + } + } + } + break; + } } else { - min = ONIGENC_MBC_MINLEN(env->enc); + min = ONIGENC_MBC_MINLEN(env->enc); } set_mml(&opt->len, min, max); } @@ -5158,20 +5158,20 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) case ANCHOR_PREC_READ: { - NodeOptInfo nopt; + NodeOptInfo nopt; - r = optimize_node_left(NANCHOR(node)->target, &nopt, env); - if (r == 0) { - if (nopt.exb.len > 0) - copy_opt_exact_info(&opt->expr, &nopt.exb); - else if (nopt.exm.len > 0) - copy_opt_exact_info(&opt->expr, &nopt.exm); + r = optimize_node_left(NANCHOR(node)->target, &nopt, env); + if (r == 0) { + if (nopt.exb.len > 0) + copy_opt_exact_info(&opt->expr, &nopt.exb); + else if (nopt.exm.len > 0) + copy_opt_exact_info(&opt->expr, &nopt.exm); - opt->expr.reach_end = 0; + opt->expr.reach_end = 0; - if (nopt.map.value > 0) - copy_opt_map_info(&opt->map, &nopt.map); - } + if (nopt.map.value > 0) + copy_opt_map_info(&opt->map, &nopt.map); + } } break; @@ -5189,8 +5189,8 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) BRefNode* br = NBREF(node); if (br->state & NST_RECURSION) { - set_mml(&opt->len, 0, ONIG_INFINITE_DISTANCE); - break; + set_mml(&opt->len, 0, ONIG_INFINITE_DISTANCE); + break; } backs = BACKREFS_P(br); r = get_min_match_length(nodes[backs[0]], &min, env->scan_env); @@ -5198,12 +5198,12 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) r = get_max_match_length(nodes[backs[0]], &max, env->scan_env); if (r != 0) break; for (i = 1; i < br->back_num; i++) { - r = get_min_match_length(nodes[backs[i]], &tmin, env->scan_env); - if (r != 0) break; - r = get_max_match_length(nodes[backs[i]], &tmax, env->scan_env); - if (r != 0) break; - if (min > tmin) min = tmin; - if (max < tmax) max = tmax; + r = get_min_match_length(nodes[backs[i]], &tmin, env->scan_env); + if (r != 0) break; + r = get_max_match_length(nodes[backs[i]], &tmax, env->scan_env); + if (r != 0) break; + if (min > tmin) min = tmin; + if (max < tmax) max = tmax; } if (r == 0) set_mml(&opt->len, min, max); } @@ -5233,44 +5233,44 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) if (r) break; if (/*qn->lower == 0 &&*/ IS_REPEAT_INFINITE(qn->upper)) { - if (env->mmd.max == 0 && - NTYPE(qn->target) == NT_CANY && qn->greedy) { - if (IS_MULTILINE(env->options)) - /* implicit anchor: /.*a/ ==> /\A.*a/ */ - add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_ML); - else - add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR); - } + if (env->mmd.max == 0 && + NTYPE(qn->target) == NT_CANY && qn->greedy) { + if (IS_MULTILINE(env->options)) + /* implicit anchor: /.*a/ ==> /\A.*a/ */ + add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_ML); + else + add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR); + } } else { - if (qn->lower > 0) { - copy_node_opt_info(opt, &nopt); - if (nopt.exb.len > 0) { - if (nopt.exb.reach_end) { - for (i = 2; i <= qn->lower && - ! is_full_opt_exact_info(&opt->exb); i++) { - concat_opt_exact_info(&opt->exb, &nopt.exb, env->enc); - } - if (i < qn->lower) { - opt->exb.reach_end = 0; - } - } - } - - if (qn->lower != qn->upper) { - opt->exb.reach_end = 0; - opt->exm.reach_end = 0; - } - if (qn->lower > 1) - opt->exm.reach_end = 0; - } + if (qn->lower > 0) { + copy_node_opt_info(opt, &nopt); + if (nopt.exb.len > 0) { + if (nopt.exb.reach_end) { + for (i = 2; i <= qn->lower && + ! is_full_opt_exact_info(&opt->exb); i++) { + concat_opt_exact_info(&opt->exb, &nopt.exb, env->enc); + } + if (i < qn->lower) { + opt->exb.reach_end = 0; + } + } + } + + if (qn->lower != qn->upper) { + opt->exb.reach_end = 0; + opt->exm.reach_end = 0; + } + if (qn->lower > 1) + opt->exm.reach_end = 0; + } } min = distance_multiply(nopt.len.min, qn->lower); if (IS_REPEAT_INFINITE(qn->upper)) - max = (nopt.len.max > 0 ? ONIG_INFINITE_DISTANCE : 0); + max = (nopt.len.max > 0 ? ONIG_INFINITE_DISTANCE : 0); else - max = distance_multiply(nopt.len.max, qn->upper); + max = distance_multiply(nopt.len.max, qn->upper); set_mml(&opt->len, min, max); } @@ -5282,47 +5282,47 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) switch (en->type) { case ENCLOSE_OPTION: - { - OnigOptionType save = env->options; + { + OnigOptionType save = env->options; - env->options = en->option; - r = optimize_node_left(en->target, opt, env); - env->options = save; - } - break; + env->options = en->option; + r = optimize_node_left(en->target, opt, env); + env->options = save; + } + break; case ENCLOSE_MEMORY: #ifdef USE_SUBEXP_CALL - en->opt_count++; - if (en->opt_count > MAX_NODE_OPT_INFO_REF_COUNT) { - OnigDistance min, max; - - min = 0; - max = ONIG_INFINITE_DISTANCE; - if (IS_ENCLOSE_MIN_FIXED(en)) min = en->min_len; - if (IS_ENCLOSE_MAX_FIXED(en)) max = en->max_len; - set_mml(&opt->len, min, max); - } - else + en->opt_count++; + if (en->opt_count > MAX_NODE_OPT_INFO_REF_COUNT) { + OnigDistance min, max; + + min = 0; + max = ONIG_INFINITE_DISTANCE; + if (IS_ENCLOSE_MIN_FIXED(en)) min = en->min_len; + if (IS_ENCLOSE_MAX_FIXED(en)) max = en->max_len; + set_mml(&opt->len, min, max); + } + else #endif - { - r = optimize_node_left(en->target, opt, env); + { + r = optimize_node_left(en->target, opt, env); - if (is_set_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_MASK)) { - if (BIT_STATUS_AT(env->scan_env->backrefed_mem, en->regnum)) - remove_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_MASK); - } - } - break; + if (is_set_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_MASK)) { + if (BIT_STATUS_AT(env->scan_env->backrefed_mem, en->regnum)) + remove_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_MASK); + } + } + break; case ENCLOSE_STOP_BACKTRACK: case ENCLOSE_CONDITION: - r = optimize_node_left(en->target, opt, env); - break; + r = optimize_node_left(en->target, opt, env); + break; case ENCLOSE_ABSENT: - set_mml(&opt->len, 0, ONIG_INFINITE_DISTANCE); - break; + set_mml(&opt->len, 0, ONIG_INFINITE_DISTANCE); + break; } } break; @@ -5330,7 +5330,7 @@ optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env) default: #ifdef ONIG_DEBUG fprintf(stderr, "optimize_node_left: undefined node type %d\n", - NTYPE(node)); + NTYPE(node)); #endif r = ONIGERR_TYPE_BUG; break; @@ -5353,18 +5353,18 @@ set_optimize_exact_info(regex_t* reg, OptExactInfo* e) reg->exact_end = reg->exact + e->len; allow_reverse = - ONIGENC_IS_ALLOWED_REVERSE_MATCH(reg->enc, reg->exact, reg->exact_end); + ONIGENC_IS_ALLOWED_REVERSE_MATCH(reg->enc, reg->exact, reg->exact_end); if (e->ignore_case > 0) { if (e->len >= 3 || (e->len >= 2 && allow_reverse)) { r = set_bm_skip(reg->exact, reg->exact_end, reg, - reg->map, &(reg->int_map), 1); + reg->map, &(reg->int_map), 1); if (r == 0) { - reg->optimize = (allow_reverse != 0 - ? ONIG_OPTIMIZE_EXACT_BM_IC : ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC); + reg->optimize = (allow_reverse != 0 + ? ONIG_OPTIMIZE_EXACT_BM_IC : ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC); } else { - reg->optimize = ONIG_OPTIMIZE_EXACT_IC; + reg->optimize = ONIG_OPTIMIZE_EXACT_IC; } } else { @@ -5374,13 +5374,13 @@ set_optimize_exact_info(regex_t* reg, OptExactInfo* e) else { if (e->len >= 3 || (e->len >= 2 && allow_reverse)) { r = set_bm_skip(reg->exact, reg->exact_end, reg, - reg->map, &(reg->int_map), 0); + reg->map, &(reg->int_map), 0); if (r == 0) { - reg->optimize = (allow_reverse != 0 - ? ONIG_OPTIMIZE_EXACT_BM : ONIG_OPTIMIZE_EXACT_BM_NOT_REV); + reg->optimize = (allow_reverse != 0 + ? ONIG_OPTIMIZE_EXACT_BM : ONIG_OPTIMIZE_EXACT_BM_NOT_REV); } else { - reg->optimize = ONIG_OPTIMIZE_EXACT; + reg->optimize = ONIG_OPTIMIZE_EXACT; } } else { @@ -5451,7 +5451,7 @@ set_optimize_info_from_tree(Node* node, regex_t* reg, ScanEnv* scan_env) reg->anchor &= ~ANCHOR_ANYCHAR_STAR_ML; reg->anchor |= opt.anc.right_anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF | - ANCHOR_PREC_READ_NOT); + ANCHOR_PREC_READ_NOT); if (reg->anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF)) { reg->anchor_dmin = opt.len.min; @@ -5461,7 +5461,7 @@ set_optimize_info_from_tree(Node* node, regex_t* reg, ScanEnv* scan_env) if (opt.exb.len > 0 || opt.exm.len > 0) { select_opt_exact_info(reg->enc, &opt.exb, &opt.exm); if (opt.map.value > 0 && - comp_opt_exact_or_map_info(&opt.exb, &opt.map) > 0) { + comp_opt_exact_or_map_info(&opt.exb, &opt.map) > 0) { goto set_map; } else { @@ -5505,7 +5505,7 @@ clear_optimize_info(regex_t* reg) #ifdef ONIG_DEBUG static void print_enc_string(FILE* fp, OnigEncoding enc, - const UChar *s, const UChar *end) + const UChar *s, const UChar *end) { fprintf(fp, "\nPATTERN: /"); @@ -5517,10 +5517,10 @@ static void print_enc_string(FILE* fp, OnigEncoding enc, while (p < end) { code = ONIGENC_MBC_TO_CODE(enc, p, end); if (code >= 0x80) { - fprintf(fp, " 0x%04x ", (int )code); + fprintf(fp, " 0x%04x ", (int )code); } else { - fputc((int )code, fp); + fputc((int )code, fp); } p += enclen(enc, p, end); @@ -5641,15 +5641,15 @@ print_optimize_info(FILE* f, regex_t* reg) c = 0; fputc('[', f); for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++) { - if (reg->map[i] != 0) { - if (c > 0) fputs(", ", f); - c++; - if (ONIGENC_MBC_MAXLEN(reg->enc) == 1 && - ONIGENC_IS_CODE_PRINT(reg->enc, (OnigCodePoint )i)) - fputc(i, f); - else - fprintf(f, "%d", i); - } + if (reg->map[i] != 0) { + if (c > 0) fputs(", ", f); + c++; + if (ONIGENC_MBC_MAXLEN(reg->enc) == 1 && + ONIGENC_IS_CODE_PRINT(reg->enc, (OnigCodePoint )i)) + fputc(i, f); + else + fprintf(f, "%d", i); + } } fprintf(f, "]\n"); } @@ -5808,7 +5808,7 @@ static void print_tree(FILE* f, Node* node); #ifdef RUBY extern int onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, - OnigErrorInfo* einfo) + OnigErrorInfo* einfo) { return onig_compile_ruby(reg, pattern, pattern_end, einfo, NULL, 0); } @@ -5817,11 +5817,11 @@ onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, #ifdef RUBY extern int onig_compile_ruby(regex_t* reg, const UChar* pattern, const UChar* pattern_end, - OnigErrorInfo* einfo, const char *sourcefile, int sourceline) + OnigErrorInfo* einfo, const char *sourcefile, int sourceline) #else extern int onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, - OnigErrorInfo* einfo) + OnigErrorInfo* einfo) #endif { #define COMPILE_INIT_SIZE 20 @@ -5938,10 +5938,10 @@ onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, if (scan_env.comb_exp_max_regnum > 0) { int i; for (i = 1; i <= scan_env.comb_exp_max_regnum; i++) { - if (BIT_STATUS_AT(scan_env.backrefed_mem, i) != 0) { - scan_env.num_comb_exp_check = 0; - break; - } + if (BIT_STATUS_AT(scan_env.backrefed_mem, i) != 0) { + scan_env.num_comb_exp_check = 0; + break; + } } } } @@ -5975,9 +5975,9 @@ onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, reg->stack_pop_level = STACK_POP_LEVEL_ALL; else { if (reg->bt_mem_start != 0) - reg->stack_pop_level = STACK_POP_LEVEL_MEM_START; + reg->stack_pop_level = STACK_POP_LEVEL_MEM_START; else - reg->stack_pop_level = STACK_POP_LEVEL_FREE; + reg->stack_pop_level = STACK_POP_LEVEL_FREE; } } #ifdef USE_SUBEXP_CALL @@ -6023,8 +6023,8 @@ static int onig_inited = 0; extern int onig_reg_init(regex_t* reg, OnigOptionType option, - OnigCaseFoldType case_fold_flag, - OnigEncoding enc, const OnigSyntaxType* syntax) + OnigCaseFoldType case_fold_flag, + OnigEncoding enc, const OnigSyntaxType* syntax) { if (! onig_inited) onig_init(); @@ -6084,8 +6084,8 @@ onig_new_without_alloc(regex_t* reg, const UChar* pattern, extern int onig_new(regex_t** reg, const UChar* pattern, const UChar* pattern_end, - OnigOptionType option, OnigEncoding enc, const OnigSyntaxType* syntax, - OnigErrorInfo* einfo) + OnigOptionType option, OnigEncoding enc, const OnigSyntaxType* syntax, + OnigErrorInfo* einfo) { *reg = (regex_t* )xmalloc(sizeof(regex_t)); if (IS_NULL(*reg)) return ONIGERR_MEMORY; @@ -6438,9 +6438,9 @@ onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, break; case ARG_OPTION: { - OnigOptionType option = *((OnigOptionType* )bp); - bp += SIZE_OPTION; - fprintf(f, ":%d", option); + OnigOptionType option = *((OnigOptionType* )bp); + bp += SIZE_OPTION; + fprintf(f, ":%d", option); } break; @@ -6489,13 +6489,13 @@ onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, break; case OP_EXACTMBN: { - int mb_len; + int mb_len; - GET_LENGTH_INC(mb_len, bp); - GET_LENGTH_INC(len, bp); - fprintf(f, ":%d:%d:", mb_len, len); - n = len * mb_len; - while (n-- > 0) { fputc(*bp++, f); } + GET_LENGTH_INC(mb_len, bp); + GET_LENGTH_INC(len, bp); + fprintf(f, ":%d:%d:", mb_len, len); + n = len * mb_len; + while (n-- > 0) { fputc(*bp++, f); } } break; @@ -6559,40 +6559,40 @@ onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, fputs(" ", f); GET_LENGTH_INC(len, bp); for (i = 0; i < len; i++) { - GET_MEMNUM_INC(mem, bp); - if (i > 0) fputs(", ", f); - fprintf(f, "%d", mem); + GET_MEMNUM_INC(mem, bp); + if (i > 0) fputs(", ", f); + fprintf(f, "%d", mem); } break; case OP_BACKREF_WITH_LEVEL: { - OnigOptionType option; - LengthType level; - - GET_OPTION_INC(option, bp); - fprintf(f, ":%d", option); - GET_LENGTH_INC(level, bp); - fprintf(f, ":%d", level); - - fputs(" ", f); - GET_LENGTH_INC(len, bp); - for (i = 0; i < len; i++) { - GET_MEMNUM_INC(mem, bp); - if (i > 0) fputs(", ", f); - fprintf(f, "%d", mem); - } + OnigOptionType option; + LengthType level; + + GET_OPTION_INC(option, bp); + fprintf(f, ":%d", option); + GET_LENGTH_INC(level, bp); + fprintf(f, ":%d", level); + + fputs(" ", f); + GET_LENGTH_INC(len, bp); + for (i = 0; i < len; i++) { + GET_MEMNUM_INC(mem, bp); + if (i > 0) fputs(", ", f); + fprintf(f, "%d", mem); + } } break; case OP_REPEAT: case OP_REPEAT_NG: { - mem = *((MemNumType* )bp); - bp += SIZE_MEMNUM; - addr = *((RelAddrType* )bp); - bp += SIZE_RELADDR; - fprintf(f, ":%d:%d", mem, addr); + mem = *((MemNumType* )bp); + bp += SIZE_MEMNUM; + addr = *((RelAddrType* )bp); + bp += SIZE_RELADDR; + fprintf(f, ":%d:%d", mem, addr); } break; @@ -6633,7 +6633,7 @@ onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, default: fprintf(stderr, "onig_print_compiled_byte_code: undefined code %d\n", - bp[-1]); + bp[-1]); } } fputs("]", f); @@ -6690,8 +6690,8 @@ print_indent_tree(FILE* f, Node* node, int indent) print_indent_tree(f, NCAR(node), indent + add); while (IS_NOT_NULL(node = NCDR(node))) { if (NTYPE(node) != type) { - fprintf(f, "ERROR: list/alt right is not a cons. %d\n", NTYPE(node)); - exit(0); + fprintf(f, "ERROR: list/alt right is not a cons. %d\n", NTYPE(node)); + exit(0); } print_indent_tree(f, NCAR(node), indent + add); } @@ -6699,12 +6699,12 @@ print_indent_tree(FILE* f, Node* node, int indent) case NT_STR: fprintf(f, "<string%s:%"PRIxPTR">", - (NSTRING_IS_RAW(node) ? "-raw" : ""), (intptr_t )node); + (NSTRING_IS_RAW(node) ? "-raw" : ""), (intptr_t )node); for (p = NSTR(node)->s; p < NSTR(node)->end; p++) { if (*p >= 0x20 && *p < 0x7f) - fputc(*p, f); + fputc(*p, f); else { - fprintf(f, " 0x%02x", *p); + fprintf(f, " 0x%02x", *p); } } break; @@ -6718,8 +6718,8 @@ print_indent_tree(FILE* f, Node* node, int indent) OnigCodePoint* end = (OnigCodePoint* )(bbuf->p + bbuf->used); fprintf(f, "%d", *data++); for (; data < end; data+=2) { - fprintf(f, ","); - fprintf(f, "%04x-%04x", data[0], data[1]); + fprintf(f, ","); + fprintf(f, "%04x-%04x", data[0], data[1]); } } break; @@ -6729,9 +6729,9 @@ print_indent_tree(FILE* f, Node* node, int indent) switch (NCTYPE(node)->ctype) { case ONIGENC_CTYPE_WORD: if (NCTYPE(node)->not != 0) - fputs("not word", f); + fputs("not word", f); else - fputs("word", f); + fputs("word", f); break; default: @@ -6779,8 +6779,8 @@ print_indent_tree(FILE* f, Node* node, int indent) p = BACKREFS_P(br); fprintf(f, "<backref:%"PRIxPTR">", (intptr_t )node); for (i = 0; i < br->back_num; i++) { - if (i > 0) fputs(", ", f); - fprintf(f, "%d", p[i]); + if (i > 0) fputs(", ", f); + fprintf(f, "%d", p[i]); } } break; @@ -6797,8 +6797,8 @@ print_indent_tree(FILE* f, Node* node, int indent) case NT_QTFR: fprintf(f, "<quantifier:%"PRIxPTR">{%d,%d}%s\n", (intptr_t )node, - NQTFR(node)->lower, NQTFR(node)->upper, - (NQTFR(node)->greedy ? "" : "?")); + NQTFR(node)->lower, NQTFR(node)->upper, + (NQTFR(node)->greedy ? "" : "?")); print_indent_tree(f, NQTFR(node)->target, indent + add); break; @@ -89,7 +89,7 @@ onigenc_get_right_adjust_char_head(OnigEncoding enc, const UChar* start, const U extern UChar* onigenc_get_right_adjust_char_head_with_prev(OnigEncoding enc, - const UChar* start, const UChar* s, const UChar* end, const UChar** prev) + const UChar* start, const UChar* s, const UChar* end, const UChar** prev) { UChar* p = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, start, s, end); @@ -425,8 +425,8 @@ const OnigPairCaseFoldCodes OnigAsciiLowerMap[] = { extern int onigenc_ascii_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED, - OnigApplyAllCaseFoldFunc f, void* arg, - OnigEncoding enc ARG_UNUSED) + OnigApplyAllCaseFoldFunc f, void* arg, + OnigEncoding enc ARG_UNUSED) { OnigCodePoint code; int i, r; @@ -446,8 +446,8 @@ onigenc_ascii_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED, extern int onigenc_ascii_get_case_fold_codes_by_str(OnigCaseFoldType flag ARG_UNUSED, - const OnigUChar* p, const OnigUChar* end ARG_UNUSED, - OnigCaseFoldCodeItem items[], OnigEncoding enc ARG_UNUSED) + const OnigUChar* p, const OnigUChar* end ARG_UNUSED, + OnigCaseFoldCodeItem items[], OnigEncoding enc ARG_UNUSED) { if (0x41 <= *p && *p <= 0x5a) { items[0].byte_len = 1; @@ -467,7 +467,7 @@ onigenc_ascii_get_case_fold_codes_by_str(OnigCaseFoldType flag ARG_UNUSED, static int ss_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED, - OnigApplyAllCaseFoldFunc f, void* arg) + OnigApplyAllCaseFoldFunc f, void* arg) { OnigCodePoint ss[] = { 0x73, 0x73 }; @@ -513,7 +513,7 @@ onigenc_get_case_fold_codes_by_str_with_map(int map_size, items[0].code_len = 1; items[0].code[0] = (OnigCodePoint )(*p + 0x20); if (*p == 0x53 && ess_tsett_flag != 0 && end > p + 1 - && (*(p+1) == 0x53 || *(p+1) == 0x73)) { + && (*(p+1) == 0x53 || *(p+1) == 0x73)) { /* SS */ items[1].byte_len = 2; items[1].code_len = 1; @@ -528,7 +528,7 @@ onigenc_get_case_fold_codes_by_str_with_map(int map_size, items[0].code_len = 1; items[0].code[0] = (OnigCodePoint )(*p - 0x20); if (*p == 0x73 && ess_tsett_flag != 0 && end > p + 1 - && (*(p+1) == 0x73 || *(p+1) == 0x53)) { + && (*(p+1) == 0x73 || *(p+1) == 0x53)) { /* ss */ items[1].byte_len = 2; items[1].code_len = 1; @@ -566,16 +566,16 @@ onigenc_get_case_fold_codes_by_str_with_map(int map_size, for (i = 0; i < map_size; i++) { if (*p == map[i].from) { - items[0].byte_len = 1; - items[0].code_len = 1; - items[0].code[0] = map[i].to; - return 1; + items[0].byte_len = 1; + items[0].code_len = 1; + items[0].code[0] = map[i].to; + return 1; } else if (*p == map[i].to) { - items[0].byte_len = 1; - items[0].code_len = 1; - items[0].code[0] = map[i].from; - return 1; + items[0].byte_len = 1; + items[0].code_len = 1; + items[0].code[0] = map[i].from; + return 1; } } } @@ -586,9 +586,9 @@ onigenc_get_case_fold_codes_by_str_with_map(int map_size, extern int onigenc_not_support_get_ctype_code_range(OnigCtype ctype ARG_UNUSED, - OnigCodePoint* sb_out ARG_UNUSED, - const OnigCodePoint* ranges[] ARG_UNUSED, - OnigEncoding enc) + OnigCodePoint* sb_out ARG_UNUSED, + const OnigCodePoint* ranges[] ARG_UNUSED, + OnigEncoding enc) { return ONIG_NO_SUPPORT_CONFIG; } @@ -605,7 +605,7 @@ onigenc_is_mbc_newline_0x0a(const UChar* p, const UChar* end, OnigEncoding enc A /* for single byte encodings */ extern int onigenc_ascii_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED, const UChar** p, - const UChar* end, UChar* lower, OnigEncoding enc ARG_UNUSED) + const UChar* end, UChar* lower, OnigEncoding enc ARG_UNUSED) { *lower = ONIGENC_ASCII_CODE_TO_LOWER_CASE(**p); @@ -616,7 +616,7 @@ onigenc_ascii_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED, const UChar** p, #if 0 extern int onigenc_ascii_is_mbc_ambiguous(OnigCaseFoldType flag ARG_UNUSED, - const UChar** pp, const UChar* end ARG_UNUSED) + const UChar** pp, const UChar* end ARG_UNUSED) { const UChar* p = *pp; @@ -627,14 +627,14 @@ onigenc_ascii_is_mbc_ambiguous(OnigCaseFoldType flag ARG_UNUSED, extern int onigenc_single_byte_mbc_enc_len(const UChar* p ARG_UNUSED, const UChar* e ARG_UNUSED, - OnigEncoding enc ARG_UNUSED) + OnigEncoding enc ARG_UNUSED) { return 1; } extern OnigCodePoint onigenc_single_byte_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED, - OnigEncoding enc ARG_UNUSED) + OnigEncoding enc ARG_UNUSED) { return (OnigCodePoint )(*p); } @@ -658,25 +658,25 @@ onigenc_single_byte_code_to_mbc(OnigCodePoint code, UChar *buf, OnigEncoding enc extern UChar* onigenc_single_byte_left_adjust_char_head(const UChar* start ARG_UNUSED, - const UChar* s, - const UChar* end ARG_UNUSED, - OnigEncoding enc ARG_UNUSED) + const UChar* s, + const UChar* end ARG_UNUSED, + OnigEncoding enc ARG_UNUSED) { return (UChar* )s; } extern int onigenc_always_true_is_allowed_reverse_match(const UChar* s ARG_UNUSED, - const UChar* end ARG_UNUSED, - OnigEncoding enc ARG_UNUSED) + const UChar* end ARG_UNUSED, + OnigEncoding enc ARG_UNUSED) { return TRUE; } extern int onigenc_always_false_is_allowed_reverse_match(const UChar* s ARG_UNUSED, - const UChar* end ARG_UNUSED, - OnigEncoding enc ARG_UNUSED) + const UChar* end ARG_UNUSED, + OnigEncoding enc ARG_UNUSED) { return FALSE; } @@ -712,7 +712,7 @@ onigenc_mbn_mbc_to_code(OnigEncoding enc, const UChar* p, const UChar* end) extern int onigenc_mbn_mbc_case_fold(OnigEncoding enc, OnigCaseFoldType flag ARG_UNUSED, const UChar** pp, const UChar* end ARG_UNUSED, - UChar* lower) + UChar* lower) { int len; const UChar *p = *pp; @@ -843,7 +843,7 @@ onigenc_minimum_property_name_to_ctype(OnigEncoding enc, const UChar* p, const U extern int onigenc_mb2_is_code_ctype(OnigEncoding enc, OnigCodePoint code, - unsigned int ctype) + unsigned int ctype) { if (code < 128) return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype); @@ -858,7 +858,7 @@ onigenc_mb2_is_code_ctype(OnigEncoding enc, OnigCodePoint code, extern int onigenc_mb4_is_code_ctype(OnigEncoding enc, OnigCodePoint code, - unsigned int ctype) + unsigned int ctype) { if (code < 128) return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype); @@ -961,14 +961,14 @@ onigenc_property_list_add_property(UChar* name, const OnigCodePoint* prop, *pnum = *pnum + 1; onig_st_insert_strend(*table, name, name + strlen((char* )name), - (hash_data_type )(*pnum + ONIGENC_MAX_STD_CTYPE)); + (hash_data_type )(*pnum + ONIGENC_MAX_STD_CTYPE)); return 0; } #endif extern int onigenc_ascii_only_case_map(OnigCaseFoldType* flagP, const OnigUChar** pp, const OnigUChar* end, - OnigUChar* to, OnigUChar* to_end, const struct OnigEncodingTypeST* enc) + OnigUChar* to, OnigUChar* to_end, const struct OnigEncodingTypeST* enc) { OnigCodePoint code; OnigUChar *to_start = to; @@ -985,8 +985,9 @@ onigenc_ascii_only_case_map(OnigCaseFoldType* flagP, const OnigUChar** pp, const if (code >= 'a' && code <= 'z' && (flags & ONIGENC_CASE_UPCASE)) { flags |= ONIGENC_CASE_MODIFIED; code += 'A' - 'a'; - } else if (code >= 'A' && code <= 'Z' && - (flags & (ONIGENC_CASE_DOWNCASE | ONIGENC_CASE_FOLD))) { + } + else if (code >= 'A' && code <= 'Z' && + (flags & (ONIGENC_CASE_DOWNCASE | ONIGENC_CASE_FOLD))) { flags |= ONIGENC_CASE_MODIFIED; code += 'a' - 'A'; } @@ -1000,8 +1001,8 @@ onigenc_ascii_only_case_map(OnigCaseFoldType* flagP, const OnigUChar** pp, const extern int onigenc_single_byte_ascii_only_case_map(OnigCaseFoldType* flagP, const OnigUChar** pp, - const OnigUChar* end, OnigUChar* to, OnigUChar* to_end, - const struct OnigEncodingTypeST* enc) + const OnigUChar* end, OnigUChar* to, OnigUChar* to_end, + const struct OnigEncodingTypeST* enc) { OnigCodePoint code; OnigUChar *to_start = to; @@ -1013,8 +1014,9 @@ onigenc_single_byte_ascii_only_case_map(OnigCaseFoldType* flagP, const OnigUChar if (code >= 'a' && code <= 'z' && (flags & ONIGENC_CASE_UPCASE)) { flags |= ONIGENC_CASE_MODIFIED; code += 'A' - 'a'; - } else if (code >= 'A' && code <= 'Z' && - (flags & (ONIGENC_CASE_DOWNCASE | ONIGENC_CASE_FOLD))) { + } + else if (code >= 'A' && code <= 'Z' && + (flags & (ONIGENC_CASE_DOWNCASE | ONIGENC_CASE_FOLD))) { flags |= ONIGENC_CASE_MODIFIED; code += 'a' - 'A'; } @@ -192,8 +192,8 @@ ONIG_EXTERN int onigenc_unicode_apply_all_case_fold(OnigCaseFoldType flag, OnigA #define UTF16_IS_SURROGATE_SECOND(c) (((c) & 0xfc) == 0xdc) #define UTF16_IS_SURROGATE(c) (((c) & 0xf8) == 0xd8) #define UNICODE_VALID_CODEPOINT_P(c) ( \ - ((c) <= 0x10ffff) && \ - !((c) < 0x10000 && UTF16_IS_SURROGATE((c) >> 8))) + ((c) <= 0x10ffff) && \ + !((c) < 0x10000 && UTF16_IS_SURROGATE((c) >> 8))) #define ONIGENC_ISO_8859_1_TO_LOWER_CASE(c) \ OnigEncISO_8859_1_ToLowerCaseTable[c] @@ -239,8 +239,8 @@ extern int ONIG_ENC_REGISTER(const char *, OnigEncoding); # define OnigEncodingDefine(f,n) \ OnigEncodingDeclare(n); \ void Init_##f(void) { \ - ONIG_ENC_REGISTER(OnigEncodingName(n).name, \ - &OnigEncodingName(n)); \ + ONIG_ENC_REGISTER(OnigEncodingName(n).name, \ + &OnigEncodingName(n)); \ } \ OnigEncodingDeclare(n) #else diff --git a/regerror.c b/regerror.c index b18fc2e88b..599af698af 100644 --- a/regerror.c +++ b/regerror.c @@ -194,7 +194,7 @@ static void sprint_byte_with_x(char* s, unsigned int v) } static int to_ascii(OnigEncoding enc, UChar *s, UChar *end, - UChar buf[], int buf_size, int *is_over) + UChar buf[], int buf_size, int *is_over) { int len; UChar *p; @@ -206,24 +206,24 @@ static int to_ascii(OnigEncoding enc, UChar *s, UChar *end, while (p < end) { code = ONIGENC_MBC_TO_CODE(enc, p, end); if (code >= 0x80) { - if (code > 0xffff && len + 10 <= buf_size) { - sprint_byte_with_x((char*)(&(buf[len])), (unsigned int)(code >> 24)); - sprint_byte((char*)(&(buf[len+4])), (unsigned int)(code >> 16)); - sprint_byte((char*)(&(buf[len+6])), (unsigned int)(code >> 8)); - sprint_byte((char*)(&(buf[len+8])), (unsigned int)code); - len += 10; - } - else if (len + 6 <= buf_size) { - sprint_byte_with_x((char*)(&(buf[len])), (unsigned int)(code >> 8)); - sprint_byte((char*)(&(buf[len+4])), (unsigned int)code); - len += 6; - } - else { - break; - } + if (code > 0xffff && len + 10 <= buf_size) { + sprint_byte_with_x((char*)(&(buf[len])), (unsigned int)(code >> 24)); + sprint_byte((char*)(&(buf[len+4])), (unsigned int)(code >> 16)); + sprint_byte((char*)(&(buf[len+6])), (unsigned int)(code >> 8)); + sprint_byte((char*)(&(buf[len+8])), (unsigned int)code); + len += 10; + } + else if (len + 6 <= buf_size) { + sprint_byte_with_x((char*)(&(buf[len])), (unsigned int)(code >> 8)); + sprint_byte((char*)(&(buf[len+4])), (unsigned int)code); + len += 6; + } + else { + break; + } } else { - buf[len++] = (UChar )code; + buf[len++] = (UChar )code; } p += enclen(enc, p, end); @@ -267,27 +267,27 @@ onig_error_code_to_str(UChar* s, OnigPosition code, ...) case ONIGERR_INVALID_CHAR_PROPERTY_NAME: einfo = va_arg(vargs, OnigErrorInfo*); len = to_ascii(einfo->enc, einfo->par, einfo->par_end, - parbuf, MAX_ERROR_PAR_LEN - 3, &is_over); + parbuf, MAX_ERROR_PAR_LEN - 3, &is_over); q = onig_error_code_to_format(code); p = s; while (*q != '\0') { if (*q == '%') { - q++; - if (*q == 'n') { /* '%n': name */ - xmemcpy(p, parbuf, len); - p += len; - if (is_over != 0) { - xmemcpy(p, "...", 3); - p += 3; - } - q++; - } - else - goto normal_char; + q++; + if (*q == 'n') { /* '%n': name */ + xmemcpy(p, parbuf, len); + p += len; + if (is_over != 0) { + xmemcpy(p, "...", 3); + p += 3; + } + q++; + } + else + goto normal_char; } else { normal_char: - *p++ = *q++; + *p++ = *q++; } } *p = '\0'; @@ -347,24 +347,24 @@ onig_vsnprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, } } else if (*p == '\\') { - *s++ = *p++; - len = enclen(enc, p, pat_end); - while (len-- > 0) *s++ = *p++; + *s++ = *p++; + len = enclen(enc, p, pat_end); + while (len-- > 0) *s++ = *p++; } else if (*p == '/') { - *s++ = (unsigned char )'\\'; - *s++ = *p++; + *s++ = (unsigned char )'\\'; + *s++ = *p++; } else if (!ONIGENC_IS_CODE_PRINT(enc, *p) && - (!ONIGENC_IS_CODE_SPACE(enc, *p) || + (!ONIGENC_IS_CODE_SPACE(enc, *p) || ONIGENC_IS_CODE_CNTRL(enc, *p))) { - sprint_byte_with_x((char* )bs, (unsigned int )(*p++)); - len = onigenc_str_bytelen_null(ONIG_ENCODING_ASCII, bs); + sprint_byte_with_x((char* )bs, (unsigned int )(*p++)); + len = onigenc_str_bytelen_null(ONIG_ENCODING_ASCII, bs); bp = bs; - while (len-- > 0) *s++ = *bp++; + while (len-- > 0) *s++ = *bp++; } else { - *s++ = *p++; + *s++ = *p++; } } @@ -381,7 +381,7 @@ onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, va_list args; va_start(args, fmt); onig_vsnprintf_with_pattern(buf, bufsize, enc, - pat, pat_end, fmt, args); + pat, pat_end, fmt, args); va_end(args); } #endif @@ -55,7 +55,7 @@ rb_enc_asciicompat(OnigEncoding enc) # define ONIGENC_IS_MBC_ASCII_WORD(enc,s,end) \ (rb_enc_asciicompat(enc) ? (ISALNUM(*s) || *s=='_') : \ onigenc_ascii_is_code_ctype( \ - ONIGENC_MBC_TO_CODE(enc,s,end),ONIGENC_CTYPE_WORD,enc)) + ONIGENC_MBC_TO_CODE(enc,s,end),ONIGENC_CTYPE_WORD,enc)) #endif /* RUBY */ #ifdef USE_CRNL_AS_LINE_TERMINATOR @@ -66,28 +66,28 @@ rb_enc_asciicompat(OnigEncoding enc) is_mbc_newline_ex((enc),(p),(start),(end),(option),(check_prev)) static int is_mbc_newline_ex(OnigEncoding enc, const UChar *p, const UChar *start, - const UChar *end, OnigOptionType option, int check_prev) + const UChar *end, OnigOptionType option, int check_prev) { if (IS_NEWLINE_CRLF(option)) { if (ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0a) { if (check_prev) { - const UChar *prev = onigenc_get_prev_char_head(enc, start, p, end); - if ((prev != NULL) && ONIGENC_MBC_TO_CODE(enc, prev, end) == 0x0d) - return 0; - else - return 1; + const UChar *prev = onigenc_get_prev_char_head(enc, start, p, end); + if ((prev != NULL) && ONIGENC_MBC_TO_CODE(enc, prev, end) == 0x0d) + return 0; + else + return 1; } else - return 1; + return 1; } else { const UChar *pnext = p + enclen(enc, p, end); if (pnext < end && - ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0d && - ONIGENC_MBC_TO_CODE(enc, pnext, end) == 0x0a) - return 1; + ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0d && + ONIGENC_MBC_TO_CODE(enc, pnext, end) == 0x0a) + return 1; if (ONIGENC_IS_MBC_NEWLINE(enc, p, end)) - return 1; + return 1; return 0; } } @@ -111,7 +111,7 @@ history_tree_clear(OnigCaptureTreeNode* node) if (IS_NOT_NULL(node)) { for (i = 0; i < node->num_childs; i++) { if (IS_NOT_NULL(node->childs[i])) { - history_tree_free(node->childs[i]); + history_tree_free(node->childs[i]); } } for (i = 0; i < node->allocated; i++) { @@ -170,18 +170,18 @@ history_tree_add_child(OnigCaptureTreeNode* parent, OnigCaptureTreeNode* child) if (IS_NULL(parent->childs)) { n = HISTORY_TREE_INIT_ALLOC_SIZE; parent->childs = - (OnigCaptureTreeNode** )xmalloc(sizeof(OnigCaptureTreeNode*) * n); + (OnigCaptureTreeNode** )xmalloc(sizeof(OnigCaptureTreeNode*) * n); CHECK_NULL_RETURN_MEMERR(parent->childs); } else { OnigCaptureTreeNode** tmp; n = parent->allocated * 2; tmp = - (OnigCaptureTreeNode** )xrealloc(parent->childs, - sizeof(OnigCaptureTreeNode*) * n); + (OnigCaptureTreeNode** )xrealloc(parent->childs, + sizeof(OnigCaptureTreeNode*) * n); if (tmp == 0) { - history_tree_clear(parent); - return ONIGERR_MEMORY; + history_tree_clear(parent); + return ONIGERR_MEMORY; } parent->childs = tmp; } @@ -276,7 +276,7 @@ static OnigPosition count_num_cache_opcodes_inner( switch (*p++) { case OP_FINISH: case OP_END: - break; + break; case OP_EXACT1: p++; break; case OP_EXACT2: p += 2; break; @@ -284,50 +284,50 @@ static OnigPosition count_num_cache_opcodes_inner( case OP_EXACT4: p += 4; break; case OP_EXACT5: p += 5; break; case OP_EXACTN: - GET_LENGTH_INC(len, p); p += len; break; + GET_LENGTH_INC(len, p); p += len; break; case OP_EXACTMB2N1: p += 2; break; case OP_EXACTMB2N2: p += 4; break; case OP_EXACTMB2N3: p += 6; break; case OP_EXACTMB2N: - GET_LENGTH_INC(len, p); p += len * 2; break; + GET_LENGTH_INC(len, p); p += len * 2; break; case OP_EXACTMB3N: - GET_LENGTH_INC(len, p); p += len * 3; break; + GET_LENGTH_INC(len, p); p += len * 3; break; case OP_EXACTMBN: - { - int mb_len; - GET_LENGTH_INC(mb_len, p); - GET_LENGTH_INC(len, p); - p += mb_len * len; - } - break; + { + int mb_len; + GET_LENGTH_INC(mb_len, p); + GET_LENGTH_INC(len, p); + p += mb_len * len; + } + break; case OP_EXACT1_IC: - len = enclen(enc, p, pend); p += len; break; + len = enclen(enc, p, pend); p += len; break; case OP_EXACTN_IC: - GET_LENGTH_INC(len, p); p += len; break; + GET_LENGTH_INC(len, p); p += len; break; case OP_CCLASS: case OP_CCLASS_NOT: - p += SIZE_BITSET; break; + p += SIZE_BITSET; break; case OP_CCLASS_MB: case OP_CCLASS_MB_NOT: - GET_LENGTH_INC(len, p); p += len; break; + GET_LENGTH_INC(len, p); p += len; break; case OP_CCLASS_MIX: case OP_CCLASS_MIX_NOT: - p += SIZE_BITSET; - GET_LENGTH_INC(len, p); - p += len; - break; + p += SIZE_BITSET; + GET_LENGTH_INC(len, p); + p += len; + break; case OP_ANYCHAR: case OP_ANYCHAR_ML: - break; + break; case OP_ANYCHAR_STAR: case OP_ANYCHAR_ML_STAR: - num_cache_opcodes++; break; + num_cache_opcodes++; break; case OP_ANYCHAR_STAR_PEEK_NEXT: case OP_ANYCHAR_ML_STAR_PEEK_NEXT: - p++; num_cache_opcodes++; break; + p++; num_cache_opcodes++; break; case OP_WORD: case OP_NOT_WORD: @@ -335,7 +335,7 @@ static OnigPosition count_num_cache_opcodes_inner( case OP_NOT_WORD_BOUND: case OP_WORD_BEGIN: case OP_WORD_END: - break; + break; case OP_ASCII_WORD: case OP_NOT_ASCII_WORD: @@ -343,7 +343,7 @@ static OnigPosition count_num_cache_opcodes_inner( case OP_NOT_ASCII_WORD_BOUND: case OP_ASCII_WORD_BEGIN: case OP_ASCII_WORD_END: - break; + break; case OP_BEGIN_BUF: case OP_END_BUF: @@ -351,7 +351,7 @@ static OnigPosition count_num_cache_opcodes_inner( case OP_END_LINE: case OP_SEMI_END_BUF: case OP_BEGIN_POSITION: - break; + break; case OP_BACKREF1: case OP_BACKREF2: @@ -360,7 +360,7 @@ static OnigPosition count_num_cache_opcodes_inner( case OP_BACKREF_MULTI: case OP_BACKREF_MULTI_IC: case OP_BACKREF_WITH_LEVEL: - goto impossible; + goto impossible; case OP_MEMORY_START: case OP_MEMORY_START_PUSH: @@ -368,152 +368,152 @@ static OnigPosition count_num_cache_opcodes_inner( case OP_MEMORY_END_PUSH_REC: case OP_MEMORY_END: case OP_MEMORY_END_REC: - p += SIZE_MEMNUM; - // A memory (capture) in look-around is found. - if (lookaround_nesting != 0) { - goto impossible; + p += SIZE_MEMNUM; + // A memory (capture) in look-around is found. + if (lookaround_nesting != 0) { + goto impossible; } - break; + break; case OP_KEEP: - break; + break; case OP_FAIL: - break; + break; case OP_JUMP: - p += SIZE_RELADDR; - break; + p += SIZE_RELADDR; + break; case OP_PUSH: - p += SIZE_RELADDR; - num_cache_opcodes++; - break; + p += SIZE_RELADDR; + num_cache_opcodes++; + break; case OP_POP: - break; + break; case OP_PUSH_OR_JUMP_EXACT1: case OP_PUSH_IF_PEEK_NEXT: - p += SIZE_RELADDR + 1; num_cache_opcodes++; break; + p += SIZE_RELADDR + 1; num_cache_opcodes++; break; case OP_REPEAT: case OP_REPEAT_NG: - if (current_repeat_mem != -1) { - // A nested OP_REPEAT is not yet supported. - goto impossible; - } - GET_MEMNUM_INC(repeat_mem, p); - p += SIZE_RELADDR; - if (reg->repeat_range[repeat_mem].lower == 0) { - num_cache_opcodes++; - } - result = count_num_cache_opcodes_inner(reg, repeat_mem, lookaround_nesting, &p, &num_cache_opcodes); - if (result < 0 || num_cache_opcodes < 0) { - goto fail; - } + if (current_repeat_mem != -1) { + // A nested OP_REPEAT is not yet supported. + goto impossible; + } + GET_MEMNUM_INC(repeat_mem, p); + p += SIZE_RELADDR; + if (reg->repeat_range[repeat_mem].lower == 0) { + num_cache_opcodes++; + } + result = count_num_cache_opcodes_inner(reg, repeat_mem, lookaround_nesting, &p, &num_cache_opcodes); + if (result < 0 || num_cache_opcodes < 0) { + goto fail; + } { - OnigRepeatRange *repeat_range = ®->repeat_range[repeat_mem]; - if (repeat_range->lower < repeat_range->upper) { - num_cache_opcodes++; - } + OnigRepeatRange *repeat_range = ®->repeat_range[repeat_mem]; + if (repeat_range->lower < repeat_range->upper) { + num_cache_opcodes++; + } } - break; + break; case OP_REPEAT_INC: case OP_REPEAT_INC_NG: - GET_MEMNUM_INC(repeat_mem, p); - if (repeat_mem != current_repeat_mem) { - // A lone or invalid OP_REPEAT_INC is found. - goto impossible; - } - goto exit; + GET_MEMNUM_INC(repeat_mem, p); + if (repeat_mem != current_repeat_mem) { + // A lone or invalid OP_REPEAT_INC is found. + goto impossible; + } + goto exit; case OP_REPEAT_INC_SG: case OP_REPEAT_INC_NG_SG: - goto impossible; + goto impossible; case OP_NULL_CHECK_START: - p += SIZE_MEMNUM; - break; + p += SIZE_MEMNUM; + break; case OP_NULL_CHECK_END: case OP_NULL_CHECK_END_MEMST_PUSH: - p += SIZE_MEMNUM; - break; + p += SIZE_MEMNUM; + break; case OP_NULL_CHECK_END_MEMST: - p += SIZE_MEMNUM; - break; + p += SIZE_MEMNUM; + break; case OP_PUSH_POS: - if (lookaround_nesting < 0) { - // A look-around nested in a atomic grouping is found. - goto impossible; - } - result = count_num_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &p, &num_cache_opcodes); - if (result < 0 || num_cache_opcodes < 0) { - goto fail; - } - break; + if (lookaround_nesting < 0) { + // A look-around nested in a atomic grouping is found. + goto impossible; + } + result = count_num_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &p, &num_cache_opcodes); + if (result < 0 || num_cache_opcodes < 0) { + goto fail; + } + break; case OP_PUSH_POS_NOT: - if (lookaround_nesting < 0) { - // A look-around nested in a atomic grouping is found. - goto impossible; - } - p += SIZE_RELADDR; - result = count_num_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &p, &num_cache_opcodes); - if (result < 0 || num_cache_opcodes < 0) { - goto fail; - } - break; + if (lookaround_nesting < 0) { + // A look-around nested in a atomic grouping is found. + goto impossible; + } + p += SIZE_RELADDR; + result = count_num_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &p, &num_cache_opcodes); + if (result < 0 || num_cache_opcodes < 0) { + goto fail; + } + break; case OP_PUSH_LOOK_BEHIND_NOT: - if (lookaround_nesting < 0) { - // A look-around nested in a atomic grouping is found. - goto impossible; - } - p += SIZE_RELADDR; - p += SIZE_LENGTH; - result = count_num_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &p, &num_cache_opcodes); - if (result < 0 || num_cache_opcodes < 0) { - goto fail; - } - break; + if (lookaround_nesting < 0) { + // A look-around nested in a atomic grouping is found. + goto impossible; + } + p += SIZE_RELADDR; + p += SIZE_LENGTH; + result = count_num_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &p, &num_cache_opcodes); + if (result < 0 || num_cache_opcodes < 0) { + goto fail; + } + break; case OP_PUSH_STOP_BT: - if (lookaround_nesting != 0) { - // A nested atomic grouping is found. - goto impossible; - } - result = count_num_cache_opcodes_inner(reg, current_repeat_mem, -1, &p, &num_cache_opcodes); - if (result < 0 || num_cache_opcodes < 0) { - goto fail; - } - break; + if (lookaround_nesting != 0) { + // A nested atomic grouping is found. + goto impossible; + } + result = count_num_cache_opcodes_inner(reg, current_repeat_mem, -1, &p, &num_cache_opcodes); + if (result < 0 || num_cache_opcodes < 0) { + goto fail; + } + break; case OP_POP_POS: case OP_FAIL_POS: case OP_FAIL_LOOK_BEHIND_NOT: case OP_POP_STOP_BT: - goto exit; + goto exit; case OP_LOOK_BEHIND: - p += SIZE_LENGTH; - break; + p += SIZE_LENGTH; + break; case OP_PUSH_ABSENT_POS: case OP_ABSENT_END: case OP_ABSENT: - goto impossible; + goto impossible; case OP_CALL: case OP_RETURN: - goto impossible; + goto impossible; case OP_CONDITION: - goto impossible; + goto impossible; case OP_STATE_CHECK_PUSH: case OP_STATE_CHECK_PUSH_OR_JUMP: case OP_STATE_CHECK: case OP_STATE_CHECK_ANYCHAR_STAR: case OP_STATE_CHECK_ANYCHAR_ML_STAR: - goto impossible; + goto impossible; case OP_SET_OPTION_PUSH: case OP_SET_OPTION: - p += SIZE_OPTION; - break; + p += SIZE_OPTION; + break; default: - goto bytecode_error; + goto bytecode_error; } } @@ -582,7 +582,7 @@ init_cache_opcodes_inner( switch (*p++) { case OP_FINISH: case OP_END: - break; + break; case OP_EXACT1: p++; break; case OP_EXACT2: p += 2; break; @@ -590,53 +590,53 @@ init_cache_opcodes_inner( case OP_EXACT4: p += 4; break; case OP_EXACT5: p += 5; break; case OP_EXACTN: - GET_LENGTH_INC(len, p); p += len; break; + GET_LENGTH_INC(len, p); p += len; break; case OP_EXACTMB2N1: p += 2; break; case OP_EXACTMB2N2: p += 4; break; case OP_EXACTMB2N3: p += 6; break; case OP_EXACTMB2N: - GET_LENGTH_INC(len, p); p += len * 2; break; + GET_LENGTH_INC(len, p); p += len * 2; break; case OP_EXACTMB3N: - GET_LENGTH_INC(len, p); p += len * 3; break; + GET_LENGTH_INC(len, p); p += len * 3; break; case OP_EXACTMBN: - { - int mb_len; - GET_LENGTH_INC(mb_len, p); - GET_LENGTH_INC(len, p); - p += mb_len * len; - } - break; + { + int mb_len; + GET_LENGTH_INC(mb_len, p); + GET_LENGTH_INC(len, p); + p += mb_len * len; + } + break; case OP_EXACT1_IC: - len = enclen(enc, p, pend); p += len; break; + len = enclen(enc, p, pend); p += len; break; case OP_EXACTN_IC: - GET_LENGTH_INC(len, p); p += len; break; + GET_LENGTH_INC(len, p); p += len; break; case OP_CCLASS: case OP_CCLASS_NOT: - p += SIZE_BITSET; break; + p += SIZE_BITSET; break; case OP_CCLASS_MB: case OP_CCLASS_MB_NOT: - GET_LENGTH_INC(len, p); p += len; break; + GET_LENGTH_INC(len, p); p += len; break; case OP_CCLASS_MIX: case OP_CCLASS_MIX_NOT: - p += SIZE_BITSET; - GET_LENGTH_INC(len, p); - p += len; - break; + p += SIZE_BITSET; + GET_LENGTH_INC(len, p); + p += len; + break; case OP_ANYCHAR: case OP_ANYCHAR_ML: - break; + break; case OP_ANYCHAR_STAR: case OP_ANYCHAR_ML_STAR: - INC_CACHE_OPCODES; - break; + INC_CACHE_OPCODES; + break; case OP_ANYCHAR_STAR_PEEK_NEXT: case OP_ANYCHAR_ML_STAR_PEEK_NEXT: - p++; - INC_CACHE_OPCODES; - break; + p++; + INC_CACHE_OPCODES; + break; case OP_WORD: case OP_NOT_WORD: @@ -644,7 +644,7 @@ init_cache_opcodes_inner( case OP_NOT_WORD_BOUND: case OP_WORD_BEGIN: case OP_WORD_END: - break; + break; case OP_ASCII_WORD: case OP_NOT_ASCII_WORD: @@ -652,7 +652,7 @@ init_cache_opcodes_inner( case OP_NOT_ASCII_WORD_BOUND: case OP_ASCII_WORD_BEGIN: case OP_ASCII_WORD_END: - break; + break; case OP_BEGIN_BUF: case OP_END_BUF: @@ -660,7 +660,7 @@ init_cache_opcodes_inner( case OP_END_LINE: case OP_SEMI_END_BUF: case OP_BEGIN_POSITION: - break; + break; case OP_BACKREF1: case OP_BACKREF2: @@ -669,7 +669,7 @@ init_cache_opcodes_inner( case OP_BACKREF_MULTI: case OP_BACKREF_MULTI_IC: case OP_BACKREF_WITH_LEVEL: - goto unexpected_bytecode_error; + goto unexpected_bytecode_error; case OP_MEMORY_START: case OP_MEMORY_START_PUSH: @@ -677,149 +677,149 @@ init_cache_opcodes_inner( case OP_MEMORY_END_PUSH_REC: case OP_MEMORY_END: case OP_MEMORY_END_REC: - p += SIZE_MEMNUM; - if (lookaround_nesting != 0) { - goto unexpected_bytecode_error; - } - break; + p += SIZE_MEMNUM; + if (lookaround_nesting != 0) { + goto unexpected_bytecode_error; + } + break; case OP_KEEP: - break; + break; case OP_FAIL: - break; + break; case OP_JUMP: - p += SIZE_RELADDR; - break; + p += SIZE_RELADDR; + break; case OP_PUSH: - p += SIZE_RELADDR; - INC_CACHE_OPCODES; - break; + p += SIZE_RELADDR; + INC_CACHE_OPCODES; + break; case OP_POP: - break; + break; case OP_PUSH_OR_JUMP_EXACT1: case OP_PUSH_IF_PEEK_NEXT: - p += SIZE_RELADDR + 1; - INC_CACHE_OPCODES; - break; + p += SIZE_RELADDR + 1; + INC_CACHE_OPCODES; + break; case OP_REPEAT: case OP_REPEAT_NG: - GET_MEMNUM_INC(repeat_mem, p); - p += SIZE_RELADDR; - if (reg->repeat_range[repeat_mem].lower == 0) { - INC_CACHE_OPCODES; - } - { - long num_cache_points_in_repeat = 0; + GET_MEMNUM_INC(repeat_mem, p); + p += SIZE_RELADDR; + if (reg->repeat_range[repeat_mem].lower == 0) { + INC_CACHE_OPCODES; + } + { + long num_cache_points_in_repeat = 0; long num_cache_points_at_repeat = cache_point; - OnigCacheOpcode* cache_opcodes_in_repeat = cache_opcodes; - result = init_cache_opcodes_inner(reg, repeat_mem, lookaround_nesting, &cache_opcodes, &p, &num_cache_points_in_repeat); - if (result != 0) { - goto fail; - } - OnigRepeatRange *repeat_range = ®->repeat_range[repeat_mem]; - if (repeat_range->lower < repeat_range->upper) { - INC_CACHE_OPCODES; - cache_point -= lookaround_nesting != 0 ? 2 : 1; - } - int repeat_bounds = repeat_range->upper == 0x7fffffff ? 1 : repeat_range->upper - repeat_range->lower; - cache_point += num_cache_points_in_repeat * repeat_range->lower + (num_cache_points_in_repeat + (lookaround_nesting != 0 ? 2 : 1)) * repeat_bounds; - for (; cache_opcodes_in_repeat < cache_opcodes; cache_opcodes_in_repeat++) { - cache_opcodes_in_repeat->num_cache_points_at_outer_repeat = num_cache_points_at_repeat; - cache_opcodes_in_repeat->num_cache_points_in_outer_repeat = num_cache_points_in_repeat; - } - } - break; + OnigCacheOpcode* cache_opcodes_in_repeat = cache_opcodes; + result = init_cache_opcodes_inner(reg, repeat_mem, lookaround_nesting, &cache_opcodes, &p, &num_cache_points_in_repeat); + if (result != 0) { + goto fail; + } + OnigRepeatRange *repeat_range = ®->repeat_range[repeat_mem]; + if (repeat_range->lower < repeat_range->upper) { + INC_CACHE_OPCODES; + cache_point -= lookaround_nesting != 0 ? 2 : 1; + } + int repeat_bounds = repeat_range->upper == 0x7fffffff ? 1 : repeat_range->upper - repeat_range->lower; + cache_point += num_cache_points_in_repeat * repeat_range->lower + (num_cache_points_in_repeat + (lookaround_nesting != 0 ? 2 : 1)) * repeat_bounds; + for (; cache_opcodes_in_repeat < cache_opcodes; cache_opcodes_in_repeat++) { + cache_opcodes_in_repeat->num_cache_points_at_outer_repeat = num_cache_points_at_repeat; + cache_opcodes_in_repeat->num_cache_points_in_outer_repeat = num_cache_points_in_repeat; + } + } + break; case OP_REPEAT_INC: case OP_REPEAT_INC_NG: - p += SIZE_MEMNUM; + p += SIZE_MEMNUM; goto exit; case OP_REPEAT_INC_SG: case OP_REPEAT_INC_NG_SG: - goto unexpected_bytecode_error; + goto unexpected_bytecode_error; case OP_NULL_CHECK_START: - p += SIZE_MEMNUM; - break; + p += SIZE_MEMNUM; + break; case OP_NULL_CHECK_END: case OP_NULL_CHECK_END_MEMST_PUSH: - p += SIZE_MEMNUM; - break; + p += SIZE_MEMNUM; + break; case OP_NULL_CHECK_END_MEMST: - p += SIZE_MEMNUM; - break; + p += SIZE_MEMNUM; + break; case OP_PUSH_POS: - lookaround: - { - OnigCacheOpcode* cache_opcodes_in_lookaround = cache_opcodes; - result = init_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &cache_opcodes, &p, &cache_point); - if (result != 0) { - goto fail; - } - UChar* match_addr = p - 1; - for (; cache_opcodes_in_lookaround < cache_opcodes; cache_opcodes_in_lookaround++) { - if (cache_opcodes_in_lookaround->match_addr == NULL) { - cache_opcodes_in_lookaround->match_addr = match_addr; - } - } - } - break; + lookaround: + { + OnigCacheOpcode* cache_opcodes_in_lookaround = cache_opcodes; + result = init_cache_opcodes_inner(reg, current_repeat_mem, lookaround_nesting + 1, &cache_opcodes, &p, &cache_point); + if (result != 0) { + goto fail; + } + UChar* match_addr = p - 1; + for (; cache_opcodes_in_lookaround < cache_opcodes; cache_opcodes_in_lookaround++) { + if (cache_opcodes_in_lookaround->match_addr == NULL) { + cache_opcodes_in_lookaround->match_addr = match_addr; + } + } + } + break; case OP_PUSH_POS_NOT: - p += SIZE_RELADDR; + p += SIZE_RELADDR; goto lookaround; case OP_PUSH_LOOK_BEHIND_NOT: - p += SIZE_RELADDR; - p += SIZE_LENGTH; + p += SIZE_RELADDR; + p += SIZE_LENGTH; goto lookaround; case OP_PUSH_STOP_BT: - { - OnigCacheOpcode* cache_opcodes_in_atomic = cache_opcodes; - result = init_cache_opcodes_inner(reg, current_repeat_mem, -1, &cache_opcodes, &p, &cache_point); - if (result != 0) { - goto fail; - } - UChar* match_addr = p - 1; - for (; cache_opcodes_in_atomic < cache_opcodes; cache_opcodes_in_atomic++) { - if (cache_opcodes_in_atomic->match_addr == NULL) { - cache_opcodes_in_atomic->match_addr = match_addr; - } - } - } - break; + { + OnigCacheOpcode* cache_opcodes_in_atomic = cache_opcodes; + result = init_cache_opcodes_inner(reg, current_repeat_mem, -1, &cache_opcodes, &p, &cache_point); + if (result != 0) { + goto fail; + } + UChar* match_addr = p - 1; + for (; cache_opcodes_in_atomic < cache_opcodes; cache_opcodes_in_atomic++) { + if (cache_opcodes_in_atomic->match_addr == NULL) { + cache_opcodes_in_atomic->match_addr = match_addr; + } + } + } + break; case OP_POP_POS: case OP_FAIL_POS: case OP_FAIL_LOOK_BEHIND_NOT: case OP_POP_STOP_BT: - goto exit; + goto exit; case OP_LOOK_BEHIND: - p += SIZE_LENGTH; - break; + p += SIZE_LENGTH; + break; case OP_ABSENT_END: case OP_ABSENT: - goto unexpected_bytecode_error; + goto unexpected_bytecode_error; case OP_CALL: case OP_RETURN: - goto unexpected_bytecode_error; + goto unexpected_bytecode_error; case OP_CONDITION: - goto unexpected_bytecode_error; + goto unexpected_bytecode_error; case OP_STATE_CHECK_PUSH: case OP_STATE_CHECK_PUSH_OR_JUMP: case OP_STATE_CHECK: case OP_STATE_CHECK_ANYCHAR_STAR: case OP_STATE_CHECK_ANYCHAR_ML_STAR: - goto unexpected_bytecode_error; + goto unexpected_bytecode_error; case OP_SET_OPTION_PUSH: case OP_SET_OPTION: - p += SIZE_OPTION; - break; + p += SIZE_OPTION; + break; default: - goto bytecode_error; + goto bytecode_error; } } @@ -1175,7 +1175,7 @@ onig_region_copy(OnigRegion* to, const OnigRegion* from) }\ else {\ alloc_addr = (char* )xalloca(sizeof(OnigStackIndex) * (ptr_num)\ - + sizeof(OnigStackType) * (stack_num));\ + + sizeof(OnigStackType) * (stack_num));\ heap_addr = NULL;\ stk_alloc = (OnigStackType* )(alloc_addr + sizeof(OnigStackIndex) * (ptr_num));\ stk_base = stk_alloc;\ @@ -1208,7 +1208,7 @@ onig_set_match_stack_limit_size(unsigned int size) static int stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end, - OnigStackType** arg_stk, OnigStackType* stk_alloc, OnigMatchArg* msa) + OnigStackType** arg_stk, OnigStackType* stk_alloc, OnigMatchArg* msa) { size_t n; OnigStackType *x, *stk_base, *stk_end, *stk; @@ -1232,9 +1232,9 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end, n *= 2; if (limit_size != 0 && n > limit_size) { if ((unsigned int )(stk_end - stk_base) == limit_size) - return ONIGERR_MATCH_STACK_LIMIT_OVER; + return ONIGERR_MATCH_STACK_LIMIT_OVER; else - n = limit_size; + n = limit_size; } x = (OnigStackType* )xrealloc(stk_base, sizeof(OnigStackType) * n); if (IS_NULL(x)) { @@ -1883,7 +1883,7 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end, } while(0) static int string_cmp_ic(OnigEncoding enc, int case_fold_flag, - UChar* s1, UChar** ps2, OnigDistance mblen, const UChar* text_end) + UChar* s1, UChar** ps2, OnigDistance mblen, const UChar* text_end) { UChar buf1[ONIGENC_MBC_CASE_FOLD_MAXLEN]; UChar buf2[ONIGENC_MBC_CASE_FOLD_MAXLEN]; @@ -1970,29 +1970,29 @@ make_capture_history_tree(OnigCaptureTreeNode* node, OnigStackType** kp, if (k->type == STK_MEM_START) { n = k->u.mem.num; if (n <= ONIG_MAX_CAPTURE_HISTORY_GROUP && - BIT_STATUS_AT(reg->capture_history, n) != 0) { - child = history_node_new(); - CHECK_NULL_RETURN_MEMERR(child); - child->group = n; - child->beg = k->u.mem.pstr - str; - r = history_tree_add_child(node, child); - if (r != 0) { - history_tree_free(child); - return r; - } - *kp = (k + 1); - r = make_capture_history_tree(child, kp, stk_top, str, reg); - if (r != 0) return r; - - k = *kp; - child->end = k->u.mem.pstr - str; + BIT_STATUS_AT(reg->capture_history, n) != 0) { + child = history_node_new(); + CHECK_NULL_RETURN_MEMERR(child); + child->group = n; + child->beg = k->u.mem.pstr - str; + r = history_tree_add_child(node, child); + if (r != 0) { + history_tree_free(child); + return r; + } + *kp = (k + 1); + r = make_capture_history_tree(child, kp, stk_top, str, reg); + if (r != 0) return r; + + k = *kp; + child->end = k->u.mem.pstr - str; } } else if (k->type == STK_MEM_END) { if (k->u.mem.num == node->group) { - node->end = k->u.mem.pstr - str; - *kp = k; - return 0; + node->end = k->u.mem.pstr - str; + *kp = k; + return 0; } } k++; @@ -2017,9 +2017,9 @@ mem_is_in_memp(int mem, int num, UChar* memp) } static int backref_match_at_nested_level(regex_t* reg, - OnigStackType* top, OnigStackType* stk_base, - int ignore_case, int case_fold_flag, - int nest, int mem_num, UChar* memp, UChar** s, const UChar* send) + OnigStackType* top, OnigStackType* stk_base, + int ignore_case, int case_fold_flag, + int nest, int mem_num, UChar* memp, UChar** s, const UChar* send) { UChar *ss, *p, *pstart, *pend = NULL_UCHARP; int level; @@ -2037,33 +2037,33 @@ static int backref_match_at_nested_level(regex_t* reg, } else if (level == nest) { if (k->type == STK_MEM_START) { - if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) { - pstart = k->u.mem.pstr; - if (pend != NULL_UCHARP) { - if (pend - pstart > send - *s) return 0; /* or goto next_mem; */ - p = pstart; - ss = *s; - - if (ignore_case != 0) { - if (string_cmp_ic(reg->enc, case_fold_flag, - pstart, &ss, pend - pstart, send) == 0) - return 0; /* or goto next_mem; */ - } - else { - while (p < pend) { - if (*p++ != *ss++) return 0; /* or goto next_mem; */ - } - } - - *s = ss; - return 1; - } - } + if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) { + pstart = k->u.mem.pstr; + if (pend != NULL_UCHARP) { + if (pend - pstart > send - *s) return 0; /* or goto next_mem; */ + p = pstart; + ss = *s; + + if (ignore_case != 0) { + if (string_cmp_ic(reg->enc, case_fold_flag, + pstart, &ss, pend - pstart, send) == 0) + return 0; /* or goto next_mem; */ + } + else { + while (p < pend) { + if (*p++ != *ss++) return 0; /* or goto next_mem; */ + } + } + + *s = ss; + return 1; + } + } } else if (k->type == STK_MEM_END) { - if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) { - pend = k->u.mem.pstr; - } + if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) { + pend = k->u.mem.pstr; + } } } k--; @@ -2081,7 +2081,7 @@ static int backref_match_at_nested_level(regex_t* reg, static LARGE_INTEGER ts, te, freq; # define GETTIME(t) QueryPerformanceCounter(&(t)) # define TIMEDIFF(te,ts) (unsigned long )(((te).QuadPart - (ts).QuadPart) \ - * 1000000 / freq.QuadPart) + * 1000000 / freq.QuadPart) # else /* _WIN32 */ # define USE_TIMEOFDAY @@ -2147,7 +2147,7 @@ onig_print_statistics(FILE* f) fprintf(f, " count prev time\n"); for (i = 0; OnigOpInfo[i].opcode >= 0; i++) { fprintf(f, "%8d: %8d: %10lu: %s\n", - OpCounter[i], OpPrevCounter[i], OpTime[i], OnigOpInfo[i].name); + OpCounter[i], OpPrevCounter[i], OpTime[i], OnigOpInfo[i].name); } fprintf(f, "\nmax stack depth: %d\n", MaxStackDepth); } @@ -2286,9 +2286,9 @@ static void memoize_extended_match_cache_point(uint8_t *match_cache_buf, long ma static OnigPosition match_at(regex_t* reg, const UChar* str, const UChar* end, #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE - const UChar* right_range, + const UChar* right_range, #endif - const UChar* sstart, UChar* sprev, OnigMatchArg* msa) + const UChar* sstart, UChar* sprev, OnigMatchArg* msa) { static const UChar FinishCode[] = { OP_FINISH }; @@ -2538,16 +2538,16 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, } #ifndef USE_SUBEXP_CALL mem_start_stk--; /* for index start from 1, - mem_start_stk[1]..mem_start_stk[num_mem] */ + mem_start_stk[1]..mem_start_stk[num_mem] */ mem_end_stk--; /* for index start from 1, - mem_end_stk[1]..mem_end_stk[num_mem] */ + mem_end_stk[1]..mem_end_stk[num_mem] */ #endif #ifdef ONIG_DEBUG_MATCH fprintf(stderr, "match_at: str: %"PRIuPTR" (%p), end: %"PRIuPTR" (%p), start: %"PRIuPTR" (%p), sprev: %"PRIuPTR" (%p)\n", - (uintptr_t )str, str, (uintptr_t )end, end, (uintptr_t )sstart, sstart, (uintptr_t )sprev, sprev); + (uintptr_t )str, str, (uintptr_t )end, end, (uintptr_t )sstart, sstart, (uintptr_t )sprev, sprev); fprintf(stderr, "size: %d, start offset: %d\n", - (int )(end - str), (int )(sstart - str)); + (int )(end - str), (int )(sstart - str)); fprintf(stderr, "\n ofs> str stk:type addr:opcode\n"); #endif @@ -2567,10 +2567,10 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, bp = buf; \ q = s; \ if (*op != OP_FINISH) { /* s may not be a valid pointer if OP_FINISH. */ \ - for (i = 0; i < 7 && q < end; i++) { \ - len = enclen(encode, q, end); \ - while (len-- > 0) *bp++ = *q++; \ - } \ + for (i = 0; i < 7 && q < end; i++) { \ + len = enclen(encode, q, end); \ + while (len-- > 0) *bp++ = *q++; \ + } \ if (q < end) { xmemcpy(bp, "...", 3); bp += 3; } \ } \ xmemcpy(bp, "\"", 1); bp += 1; \ @@ -2578,9 +2578,9 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, fputs((char* )buf, stderr); \ for (i = 0; i < 20 - (bp - buf); i++) fputc(' ', stderr); \ fprintf(stderr, "%4"PRIdPTR":%s %4"PRIdPTR":", \ - stk - stk_base - 1, \ - (stk > stk_base) ? stack_type_str(stk[-1].type) : " ", \ - (op == FinishCode) ? (ptrdiff_t )-1 : op - reg->p); \ + stk - stk_base - 1, \ + (stk > stk_base) ? stack_type_str(stk[-1].type) : " ", \ + (op == FinishCode) ? (ptrdiff_t )-1 : op - reg->p); \ onig_print_compiled_byte_code(stderr, op, reg->p+reg->used, NULL, encode); \ fprintf(stderr, "\n"); \ } @@ -2609,16 +2609,16 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, uint8_t match_cache_point_mask = 1 << (match_cache_point & 7);\ MATCH_CACHE_DEBUG;\ if (msa->match_cache_buf[match_cache_point_index] & match_cache_point_mask) {\ - MATCH_CACHE_DEBUG_HIT; MATCH_CACHE_HIT;\ - if (cache_opcode->lookaround_nesting == 0) goto fail;\ - else if (cache_opcode->lookaround_nesting < 0) {\ - if (check_extended_match_cache_point(msa->match_cache_buf, match_cache_point_index, match_cache_point_mask)) {\ + MATCH_CACHE_DEBUG_HIT; MATCH_CACHE_HIT;\ + if (cache_opcode->lookaround_nesting == 0) goto fail;\ + else if (cache_opcode->lookaround_nesting < 0) {\ + if (check_extended_match_cache_point(msa->match_cache_buf, match_cache_point_index, match_cache_point_mask)) {\ STACK_STOP_BT_FAIL;\ goto fail;\ } else goto fail;\ } else {\ - if (check_extended_match_cache_point(msa->match_cache_buf, match_cache_point_index, match_cache_point_mask)) {\ - p = cache_opcode->match_addr;\ + if (check_extended_match_cache_point(msa->match_cache_buf, match_cache_point_index, match_cache_point_mask)) {\ + p = cache_opcode->match_addr;\ MOP_OUT;\ JUMP;\ } else goto fail;\ @@ -2636,66 +2636,66 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_END) MOP_IN(OP_END); n = s - sstart; if (n > best_len) { - OnigRegion* region; + OnigRegion* region; #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE - if (IS_FIND_LONGEST(option)) { - if (n > msa->best_len) { - msa->best_len = n; - msa->best_s = (UChar* )sstart; - } - else - goto end_best_len; - } + if (IS_FIND_LONGEST(option)) { + if (n > msa->best_len) { + msa->best_len = n; + msa->best_s = (UChar* )sstart; + } + else + goto end_best_len; + } #endif - best_len = n; - region = msa->region; - if (region) { - region->beg[0] = ((pkeep > s) ? s : pkeep) - str; - region->end[0] = s - str; - for (i = 1; i <= num_mem; i++) { - if (mem_end_stk[i] != INVALID_STACK_INDEX) { - if (BIT_STATUS_AT(reg->bt_mem_start, i)) - region->beg[i] = STACK_AT(mem_start_stk[i])->u.mem.pstr - str; - else - region->beg[i] = (UChar* )((void* )mem_start_stk[i]) - str; - - region->end[i] = (BIT_STATUS_AT(reg->bt_mem_end, i) - ? STACK_AT(mem_end_stk[i])->u.mem.pstr - : (UChar* )((void* )mem_end_stk[i])) - str; - } - else { - region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS; - } - } + best_len = n; + region = msa->region; + if (region) { + region->beg[0] = ((pkeep > s) ? s : pkeep) - str; + region->end[0] = s - str; + for (i = 1; i <= num_mem; i++) { + if (mem_end_stk[i] != INVALID_STACK_INDEX) { + if (BIT_STATUS_AT(reg->bt_mem_start, i)) + region->beg[i] = STACK_AT(mem_start_stk[i])->u.mem.pstr - str; + else + region->beg[i] = (UChar* )((void* )mem_start_stk[i]) - str; + + region->end[i] = (BIT_STATUS_AT(reg->bt_mem_end, i) + ? STACK_AT(mem_end_stk[i])->u.mem.pstr + : (UChar* )((void* )mem_end_stk[i])) - str; + } + else { + region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS; + } + } #ifdef USE_CAPTURE_HISTORY - if (reg->capture_history != 0) { - int r; - OnigCaptureTreeNode* node; - - if (IS_NULL(region->history_root)) { - region->history_root = node = history_node_new(); - CHECK_NULL_RETURN_MEMERR(node); - } - else { - node = region->history_root; - history_tree_clear(node); - } - - node->group = 0; - node->beg = ((pkeep > s) ? s : pkeep) - str; - node->end = s - str; - - stkp = stk_base; - r = make_capture_history_tree(region->history_root, &stkp, - stk, (UChar* )str, reg); - if (r < 0) { - best_len = r; /* error code */ - goto finish; - } - } + if (reg->capture_history != 0) { + int r; + OnigCaptureTreeNode* node; + + if (IS_NULL(region->history_root)) { + region->history_root = node = history_node_new(); + CHECK_NULL_RETURN_MEMERR(node); + } + else { + node = region->history_root; + history_tree_clear(node); + } + + node->group = 0; + node->beg = ((pkeep > s) ? s : pkeep) - str; + node->end = s - str; + + stkp = stk_base; + r = make_capture_history_tree(region->history_root, &stkp, + stk, (UChar* )str, reg); + if (r < 0) { + best_len = r; /* error code */ + goto finish; + } + } #endif /* USE_CAPTURE_HISTORY */ - } /* if (region) */ + } /* if (region) */ } /* n > best_len */ #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE @@ -2704,13 +2704,13 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, MOP_OUT; if (IS_FIND_CONDITION(option)) { - if (IS_FIND_NOT_EMPTY(option) && s == sstart) { - best_len = ONIG_MISMATCH; - goto fail; /* for retry */ - } - if (IS_FIND_LONGEST(option) && DATA_ENSURE_CHECK1) { - goto fail; /* for retry */ - } + if (IS_FIND_NOT_EMPTY(option) && s == sstart) { + best_len = ONIG_MISMATCH; + goto fail; /* for retry */ + } + if (IS_FIND_LONGEST(option) && DATA_ENSURE_CHECK1) { + goto fail; /* for retry */ + } } /* default behavior: return first-matching result. */ @@ -2726,22 +2726,22 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_EXACT1_IC) MOP_IN(OP_EXACT1_IC); { - int len; - UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN]; - - DATA_ENSURE(1); - len = ONIGENC_MBC_CASE_FOLD(encode, - /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */ - case_fold_flag, - &s, end, lowbuf); - DATA_ENSURE(0); - q = lowbuf; - while (len-- > 0) { - if (*p != *q) { - goto fail; - } - p++; q++; - } + int len; + UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN]; + + DATA_ENSURE(1); + len = ONIGENC_MBC_CASE_FOLD(encode, + /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */ + case_fold_flag, + &s, end, lowbuf); + DATA_ENSURE(0); + q = lowbuf; + while (len-- > 0) { + if (*p != *q) { + goto fail; + } + p++; q++; + } } MOP_OUT; NEXT; @@ -2802,7 +2802,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_LENGTH_INC(tlen, p); DATA_ENSURE(tlen); while (tlen-- > 0) { - if (*p++ != *s++) goto fail; + if (*p++ != *s++) goto fail; } sprev = s - 1; MOP_OUT; @@ -2810,26 +2810,26 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_EXACTN_IC) MOP_IN(OP_EXACTN_IC); { - int len; - UChar *q, *endp, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN]; - - GET_LENGTH_INC(tlen, p); - endp = p + tlen; - - while (p < endp) { - sprev = s; - DATA_ENSURE(1); - len = ONIGENC_MBC_CASE_FOLD(encode, - /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */ - case_fold_flag, - &s, end, lowbuf); - DATA_ENSURE(0); - q = lowbuf; - while (len-- > 0) { - if (*p != *q) goto fail; - p++; q++; - } - } + int len; + UChar *q, *endp, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN]; + + GET_LENGTH_INC(tlen, p); + endp = p + tlen; + + while (p < endp) { + sprev = s; + DATA_ENSURE(1); + len = ONIGENC_MBC_CASE_FOLD(encode, + /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */ + case_fold_flag, + &s, end, lowbuf); + DATA_ENSURE(0); + q = lowbuf; + while (len-- > 0) { + if (*p != *q) goto fail; + p++; q++; + } + } } MOP_OUT; @@ -2880,10 +2880,10 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_LENGTH_INC(tlen, p); DATA_ENSURE(tlen * 2); while (tlen-- > 0) { - if (*p != *s) goto fail; - p++; s++; - if (*p != *s) goto fail; - p++; s++; + if (*p != *s) goto fail; + p++; s++; + if (*p != *s) goto fail; + p++; s++; } sprev = s - 2; MOP_OUT; @@ -2893,12 +2893,12 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_LENGTH_INC(tlen, p); DATA_ENSURE(tlen * 3); while (tlen-- > 0) { - if (*p != *s) goto fail; - p++; s++; - if (*p != *s) goto fail; - p++; s++; - if (*p != *s) goto fail; - p++; s++; + if (*p != *s) goto fail; + p++; s++; + if (*p != *s) goto fail; + p++; s++; + if (*p != *s) goto fail; + p++; s++; } sprev = s - 3; MOP_OUT; @@ -2910,8 +2910,8 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, tlen2 *= tlen; DATA_ENSURE(tlen2); while (tlen2-- > 0) { - if (*p != *s) goto fail; - p++; s++; + if (*p != *s) goto fail; + p++; s++; } sprev = s - tlen; MOP_OUT; @@ -2931,23 +2931,23 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, cclass_mb: GET_LENGTH_INC(tlen, p); { - OnigCodePoint code; - UChar *ss; - int mb_len; + OnigCodePoint code; + UChar *ss; + int mb_len; - DATA_ENSURE(1); - mb_len = enclen_approx(encode, s, end); - DATA_ENSURE(mb_len); - ss = s; - s += mb_len; - code = ONIGENC_MBC_TO_CODE(encode, ss, s); + DATA_ENSURE(1); + mb_len = enclen_approx(encode, s, end); + DATA_ENSURE(mb_len); + ss = s; + s += mb_len; + code = ONIGENC_MBC_TO_CODE(encode, ss, s); #ifdef PLATFORM_UNALIGNED_WORD_ACCESS - if (! onig_is_in_code_range(p, code)) goto fail; + if (! onig_is_in_code_range(p, code)) goto fail; #else - q = p; - ALIGNMENT_RIGHT(q); - if (! onig_is_in_code_range(q, code)) goto fail; + q = p; + ALIGNMENT_RIGHT(q); + if (! onig_is_in_code_range(q, code)) goto fail; #endif } p += tlen; @@ -2957,17 +2957,17 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_CCLASS_MIX) MOP_IN(OP_CCLASS_MIX); DATA_ENSURE(1); if (ONIGENC_IS_MBC_HEAD(encode, s, end)) { - p += SIZE_BITSET; - goto cclass_mb; + p += SIZE_BITSET; + goto cclass_mb; } else { - if (BITSET_AT(((BitSetRef )p), *s) == 0) - goto fail; + if (BITSET_AT(((BitSetRef )p), *s) == 0) + goto fail; - p += SIZE_BITSET; - GET_LENGTH_INC(tlen, p); - p += tlen; - s++; + p += SIZE_BITSET; + GET_LENGTH_INC(tlen, p); + p += tlen; + s++; } MOP_OUT; NEXT; @@ -2983,36 +2983,36 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_CCLASS_MB_NOT) MOP_IN(OP_CCLASS_MB_NOT); DATA_ENSURE(1); if (! ONIGENC_IS_MBC_HEAD(encode, s, end)) { - s++; - GET_LENGTH_INC(tlen, p); - p += tlen; - goto cc_mb_not_success; + s++; + GET_LENGTH_INC(tlen, p); + p += tlen; + goto cc_mb_not_success; } cclass_mb_not: GET_LENGTH_INC(tlen, p); { - OnigCodePoint code; - UChar *ss; - int mb_len = enclen(encode, s, end); - - if (! DATA_ENSURE_CHECK(mb_len)) { - DATA_ENSURE(1); - s = (UChar* )end; - p += tlen; - goto cc_mb_not_success; - } + OnigCodePoint code; + UChar *ss; + int mb_len = enclen(encode, s, end); + + if (! DATA_ENSURE_CHECK(mb_len)) { + DATA_ENSURE(1); + s = (UChar* )end; + p += tlen; + goto cc_mb_not_success; + } - ss = s; - s += mb_len; - code = ONIGENC_MBC_TO_CODE(encode, ss, s); + ss = s; + s += mb_len; + code = ONIGENC_MBC_TO_CODE(encode, ss, s); #ifdef PLATFORM_UNALIGNED_WORD_ACCESS - if (onig_is_in_code_range(p, code)) goto fail; + if (onig_is_in_code_range(p, code)) goto fail; #else - q = p; - ALIGNMENT_RIGHT(q); - if (onig_is_in_code_range(q, code)) goto fail; + q = p; + ALIGNMENT_RIGHT(q); + if (onig_is_in_code_range(q, code)) goto fail; #endif } p += tlen; @@ -3024,17 +3024,17 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_CCLASS_MIX_NOT) MOP_IN(OP_CCLASS_MIX_NOT); DATA_ENSURE(1); if (ONIGENC_IS_MBC_HEAD(encode, s, end)) { - p += SIZE_BITSET; - goto cclass_mb_not; + p += SIZE_BITSET; + goto cclass_mb_not; } else { - if (BITSET_AT(((BitSetRef )p), *s) != 0) - goto fail; + if (BITSET_AT(((BitSetRef )p), *s) != 0) + goto fail; - p += SIZE_BITSET; - GET_LENGTH_INC(tlen, p); - p += tlen; - s++; + p += SIZE_BITSET; + GET_LENGTH_INC(tlen, p); + p += tlen; + s++; } MOP_OUT; NEXT; @@ -3058,52 +3058,52 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_ANYCHAR_STAR) MOP_IN(OP_ANYCHAR_STAR); while (DATA_ENSURE_CHECK1) { - CHECK_MATCH_CACHE; - STACK_PUSH_ALT(p, s, sprev, pkeep); - n = enclen_approx(encode, s, end); - DATA_ENSURE(n); - if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail; - sprev = s; - s += n; + CHECK_MATCH_CACHE; + STACK_PUSH_ALT(p, s, sprev, pkeep); + n = enclen_approx(encode, s, end); + DATA_ENSURE(n); + if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail; + sprev = s; + s += n; } MOP_OUT; JUMP; CASE(OP_ANYCHAR_ML_STAR) MOP_IN(OP_ANYCHAR_ML_STAR); while (DATA_ENSURE_CHECK1) { - CHECK_MATCH_CACHE; - STACK_PUSH_ALT(p, s, sprev, pkeep); - n = enclen_approx(encode, s, end); - if (n > 1) { - DATA_ENSURE(n); - sprev = s; - s += n; - } - else { - sprev = s; - s++; - } + CHECK_MATCH_CACHE; + STACK_PUSH_ALT(p, s, sprev, pkeep); + n = enclen_approx(encode, s, end); + if (n > 1) { + DATA_ENSURE(n); + sprev = s; + s += n; + } + else { + sprev = s; + s++; + } } MOP_OUT; JUMP; CASE(OP_ANYCHAR_STAR_PEEK_NEXT) MOP_IN(OP_ANYCHAR_STAR_PEEK_NEXT); while (DATA_ENSURE_CHECK1) { - CHECK_MATCH_CACHE; - if (*p == *s) { - STACK_PUSH_ALT(p + 1, s, sprev, pkeep); - } else { + CHECK_MATCH_CACHE; + if (*p == *s) { + STACK_PUSH_ALT(p + 1, s, sprev, pkeep); + } else { #ifdef USE_MATCH_CACHE - /* We need to increment num_fails here, for invoking a cache optimization correctly. */ - /* Actually, the matching will be failed if we use `OP_ANYCHAR_STAR` simply in this case.*/ - msa->num_fails++; + /* We need to increment num_fails here, for invoking a cache optimization correctly. */ + /* Actually, the matching will be failed if we use `OP_ANYCHAR_STAR` simply in this case.*/ + msa->num_fails++; #endif - } - n = enclen_approx(encode, s, end); - DATA_ENSURE(n); - if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail; - sprev = s; - s += n; + } + n = enclen_approx(encode, s, end); + DATA_ENSURE(n); + if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail; + sprev = s; + s += n; } p++; MOP_OUT; @@ -3111,26 +3111,26 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_ANYCHAR_ML_STAR_PEEK_NEXT)MOP_IN(OP_ANYCHAR_ML_STAR_PEEK_NEXT); while (DATA_ENSURE_CHECK1) { - CHECK_MATCH_CACHE; - if (*p == *s) { - STACK_PUSH_ALT(p + 1, s, sprev, pkeep); - } else { + CHECK_MATCH_CACHE; + if (*p == *s) { + STACK_PUSH_ALT(p + 1, s, sprev, pkeep); + } else { #ifdef USE_MATCH_CACHE - /* We need to increment num_fails here, for invoking a cache optimization correctly. */ - /* Actually, the matching will be failed if we use `OP_ANYCHAR_STAR_ML` simply in this case.*/ - msa->num_fails++; + /* We need to increment num_fails here, for invoking a cache optimization correctly. */ + /* Actually, the matching will be failed if we use `OP_ANYCHAR_STAR_ML` simply in this case.*/ + msa->num_fails++; #endif - } - n = enclen_approx(encode, s, end); - if (n > 1) { - DATA_ENSURE(n); - sprev = s; - s += n; - } - else { - sprev = s; - s++; - } + } + n = enclen_approx(encode, s, end); + if (n > 1) { + DATA_ENSURE(n); + sprev = s; + s += n; + } + else { + sprev = s; + s++; + } } p++; MOP_OUT; @@ -3140,15 +3140,15 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_STATE_CHECK_ANYCHAR_STAR) MOP_IN(OP_STATE_CHECK_ANYCHAR_STAR); GET_STATE_CHECK_NUM_INC(mem, p); while (DATA_ENSURE_CHECK1) { - STATE_CHECK_VAL(scv, mem); - if (scv) goto fail; - - STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep); - n = enclen_approx(encode, s, end); - DATA_ENSURE(n); - if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail; - sprev = s; - s += n; + STATE_CHECK_VAL(scv, mem); + if (scv) goto fail; + + STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep); + n = enclen_approx(encode, s, end); + DATA_ENSURE(n); + if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail; + sprev = s; + s += n; } MOP_OUT; NEXT; @@ -3158,20 +3158,20 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_STATE_CHECK_NUM_INC(mem, p); while (DATA_ENSURE_CHECK1) { - STATE_CHECK_VAL(scv, mem); - if (scv) goto fail; - - STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep); - n = enclen_approx(encode, s, end); - if (n > 1) { - DATA_ENSURE(n); - sprev = s; - s += n; - } - else { - sprev = s; - s++; - } + STATE_CHECK_VAL(scv, mem); + if (scv) goto fail; + + STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep); + n = enclen_approx(encode, s, end); + if (n > 1) { + DATA_ENSURE(n); + sprev = s; + s += n; + } + else { + sprev = s; + s++; + } } MOP_OUT; NEXT; @@ -3180,7 +3180,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_WORD) MOP_IN(OP_WORD); DATA_ENSURE(1); if (! ONIGENC_IS_MBC_WORD(encode, s, end)) - goto fail; + goto fail; s += enclen(encode, s, end); MOP_OUT; @@ -3189,7 +3189,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_ASCII_WORD) MOP_IN(OP_ASCII_WORD); DATA_ENSURE(1); if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) - goto fail; + goto fail; s += enclen(encode, s, end); MOP_OUT; @@ -3198,7 +3198,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_NOT_WORD) MOP_IN(OP_NOT_WORD); DATA_ENSURE(1); if (ONIGENC_IS_MBC_WORD(encode, s, end)) - goto fail; + goto fail; s += enclen(encode, s, end); MOP_OUT; @@ -3207,7 +3207,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_NOT_ASCII_WORD) MOP_IN(OP_NOT_ASCII_WORD); DATA_ENSURE(1); if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) - goto fail; + goto fail; s += enclen(encode, s, end); MOP_OUT; @@ -3215,70 +3215,70 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_WORD_BOUND) MOP_IN(OP_WORD_BOUND); if (ON_STR_BEGIN(s)) { - DATA_ENSURE(1); - if (! ONIGENC_IS_MBC_WORD(encode, s, end)) - goto fail; + DATA_ENSURE(1); + if (! ONIGENC_IS_MBC_WORD(encode, s, end)) + goto fail; } else if (ON_STR_END(s)) { - if (! ONIGENC_IS_MBC_WORD(encode, sprev, end)) - goto fail; + if (! ONIGENC_IS_MBC_WORD(encode, sprev, end)) + goto fail; } else { - if (ONIGENC_IS_MBC_WORD(encode, s, end) - == ONIGENC_IS_MBC_WORD(encode, sprev, end)) - goto fail; + if (ONIGENC_IS_MBC_WORD(encode, s, end) + == ONIGENC_IS_MBC_WORD(encode, sprev, end)) + goto fail; } MOP_OUT; JUMP; CASE(OP_ASCII_WORD_BOUND) MOP_IN(OP_ASCII_WORD_BOUND); if (ON_STR_BEGIN(s)) { - DATA_ENSURE(1); - if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) - goto fail; + DATA_ENSURE(1); + if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) + goto fail; } else if (ON_STR_END(s)) { - if (! ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) - goto fail; + if (! ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) + goto fail; } else { - if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end) - == ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) - goto fail; + if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end) + == ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) + goto fail; } MOP_OUT; JUMP; CASE(OP_NOT_WORD_BOUND) MOP_IN(OP_NOT_WORD_BOUND); if (ON_STR_BEGIN(s)) { - if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end)) - goto fail; + if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end)) + goto fail; } else if (ON_STR_END(s)) { - if (ONIGENC_IS_MBC_WORD(encode, sprev, end)) - goto fail; + if (ONIGENC_IS_MBC_WORD(encode, sprev, end)) + goto fail; } else { - if (ONIGENC_IS_MBC_WORD(encode, s, end) - != ONIGENC_IS_MBC_WORD(encode, sprev, end)) - goto fail; + if (ONIGENC_IS_MBC_WORD(encode, s, end) + != ONIGENC_IS_MBC_WORD(encode, sprev, end)) + goto fail; } MOP_OUT; JUMP; CASE(OP_NOT_ASCII_WORD_BOUND) MOP_IN(OP_NOT_ASCII_WORD_BOUND); if (ON_STR_BEGIN(s)) { - if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) - goto fail; + if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) + goto fail; } else if (ON_STR_END(s)) { - if (ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) - goto fail; + if (ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) + goto fail; } else { - if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end) - != ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) - goto fail; + if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end) + != ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) + goto fail; } MOP_OUT; JUMP; @@ -3286,40 +3286,40 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, #ifdef USE_WORD_BEGIN_END CASE(OP_WORD_BEGIN) MOP_IN(OP_WORD_BEGIN); if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end)) { - if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_WORD(encode, sprev, end)) { - MOP_OUT; - JUMP; - } + if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_WORD(encode, sprev, end)) { + MOP_OUT; + JUMP; + } } goto fail; NEXT; CASE(OP_ASCII_WORD_BEGIN) MOP_IN(OP_ASCII_WORD_BEGIN); if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) { - if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) { - MOP_OUT; - JUMP; - } + if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) { + MOP_OUT; + JUMP; + } } goto fail; NEXT; CASE(OP_WORD_END) MOP_IN(OP_WORD_END); if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_WORD(encode, sprev, end)) { - if (ON_STR_END(s) || !ONIGENC_IS_MBC_WORD(encode, s, end)) { - MOP_OUT; - JUMP; - } + if (ON_STR_END(s) || !ONIGENC_IS_MBC_WORD(encode, s, end)) { + MOP_OUT; + JUMP; + } } goto fail; NEXT; CASE(OP_ASCII_WORD_END) MOP_IN(OP_ASCII_WORD_END); if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) { - if (ON_STR_END(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) { - MOP_OUT; - JUMP; - } + if (ON_STR_END(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) { + MOP_OUT; + JUMP; + } } goto fail; NEXT; @@ -3341,18 +3341,18 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_BEGIN_LINE) MOP_IN(OP_BEGIN_LINE); if (ON_STR_BEGIN(s)) { - if (IS_NOTBOL(msa->options)) goto fail; - MOP_OUT; - JUMP; + if (IS_NOTBOL(msa->options)) goto fail; + MOP_OUT; + JUMP; } else if (ONIGENC_IS_MBC_NEWLINE(encode, sprev, end) #ifdef USE_CRNL_AS_LINE_TERMINATOR - && !(IS_NEWLINE_CRLF(option) - && ONIGENC_IS_MBC_CRNL(encode, sprev, end)) + && !(IS_NEWLINE_CRLF(option) + && ONIGENC_IS_MBC_CRNL(encode, sprev, end)) #endif - && !ON_STR_END(s)) { - MOP_OUT; - JUMP; + && !ON_STR_END(s)) { + MOP_OUT; + JUMP; } goto fail; NEXT; @@ -3360,18 +3360,18 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_END_LINE) MOP_IN(OP_END_LINE); if (ON_STR_END(s)) { #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE - if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) { + if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) { #endif - if (IS_NOTEOL(msa->options)) goto fail; - MOP_OUT; - JUMP; + if (IS_NOTEOL(msa->options)) goto fail; + MOP_OUT; + JUMP; #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE - } + } #endif } else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) { - MOP_OUT; - JUMP; + MOP_OUT; + JUMP; } goto fail; NEXT; @@ -3379,30 +3379,30 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_SEMI_END_BUF) MOP_IN(OP_SEMI_END_BUF); if (ON_STR_END(s)) { #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE - if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) { + if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) { #endif - if (IS_NOTEOL(msa->options)) goto fail; - MOP_OUT; - JUMP; + if (IS_NOTEOL(msa->options)) goto fail; + MOP_OUT; + JUMP; #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE - } + } #endif } else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) { - UChar* ss = s + enclen(encode, s, end); - if (ON_STR_END(ss)) { - MOP_OUT; - JUMP; - } + UChar* ss = s + enclen(encode, s, end); + if (ON_STR_END(ss)) { + MOP_OUT; + JUMP; + } #ifdef USE_CRNL_AS_LINE_TERMINATOR - else if (IS_NEWLINE_CRLF(option) - && ONIGENC_IS_MBC_CRNL(encode, s, end)) { - ss += enclen(encode, ss, end); - if (ON_STR_END(ss)) { - MOP_OUT; - JUMP; - } - } + else if (IS_NEWLINE_CRLF(option) + && ONIGENC_IS_MBC_CRNL(encode, s, end)) { + ss += enclen(encode, ss, end); + if (ON_STR_END(ss)) { + MOP_OUT; + JUMP; + } + } #endif } goto fail; @@ -3410,7 +3410,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_BEGIN_POSITION) MOP_IN(OP_BEGIN_POSITION); if (s != msa->gpos) - goto fail; + goto fail; MOP_OUT; JUMP; @@ -3460,9 +3460,9 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, STACK_GET_MEM_START(mem, stkp); if (BIT_STATUS_AT(reg->bt_mem_start, mem)) - mem_start_stk[mem] = GET_STACK_INDEX(stkp); + mem_start_stk[mem] = GET_STACK_INDEX(stkp); else - mem_start_stk[mem] = (OnigStackIndex )((void* )stkp->u.mem.pstr); + mem_start_stk[mem] = (OnigStackIndex )((void* )stkp->u.mem.pstr); STACK_PUSH_MEM_END_MARK(mem); MOP_OUT; @@ -3483,167 +3483,167 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_MEMNUM_INC(mem, p); backref: { - int len; - UChar *pstart, *pend; - - /* if you want to remove following line, - you should check in parse and compile time. */ - if (mem > num_mem) goto fail; - if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail; - if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail; - - if (BIT_STATUS_AT(reg->bt_mem_start, mem)) - pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; - else - pstart = (UChar* )((void* )mem_start_stk[mem]); - - pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) - ? STACK_AT(mem_end_stk[mem])->u.mem.pstr - : (UChar* )((void* )mem_end_stk[mem])); - n = pend - pstart; - DATA_ENSURE(n); - sprev = s; - STRING_CMP(pstart, s, n); - while (sprev + (len = enclen_approx(encode, sprev, end)) < s) - sprev += len; - - MOP_OUT; - JUMP; + int len; + UChar *pstart, *pend; + + /* if you want to remove following line, + you should check in parse and compile time. */ + if (mem > num_mem) goto fail; + if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail; + if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail; + + if (BIT_STATUS_AT(reg->bt_mem_start, mem)) + pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; + else + pstart = (UChar* )((void* )mem_start_stk[mem]); + + pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) + ? STACK_AT(mem_end_stk[mem])->u.mem.pstr + : (UChar* )((void* )mem_end_stk[mem])); + n = pend - pstart; + DATA_ENSURE(n); + sprev = s; + STRING_CMP(pstart, s, n); + while (sprev + (len = enclen_approx(encode, sprev, end)) < s) + sprev += len; + + MOP_OUT; + JUMP; } CASE(OP_BACKREFN_IC) MOP_IN(OP_BACKREFN_IC); GET_MEMNUM_INC(mem, p); { - int len; - UChar *pstart, *pend; - - /* if you want to remove following line, - you should check in parse and compile time. */ - if (mem > num_mem) goto fail; - if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail; - if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail; - - if (BIT_STATUS_AT(reg->bt_mem_start, mem)) - pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; - else - pstart = (UChar* )((void* )mem_start_stk[mem]); - - pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) - ? STACK_AT(mem_end_stk[mem])->u.mem.pstr - : (UChar* )((void* )mem_end_stk[mem])); - n = pend - pstart; - DATA_ENSURE(n); - sprev = s; - STRING_CMP_IC(case_fold_flag, pstart, &s, n, end); - while (sprev + (len = enclen_approx(encode, sprev, end)) < s) - sprev += len; - - MOP_OUT; - JUMP; + int len; + UChar *pstart, *pend; + + /* if you want to remove following line, + you should check in parse and compile time. */ + if (mem > num_mem) goto fail; + if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail; + if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail; + + if (BIT_STATUS_AT(reg->bt_mem_start, mem)) + pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; + else + pstart = (UChar* )((void* )mem_start_stk[mem]); + + pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) + ? STACK_AT(mem_end_stk[mem])->u.mem.pstr + : (UChar* )((void* )mem_end_stk[mem])); + n = pend - pstart; + DATA_ENSURE(n); + sprev = s; + STRING_CMP_IC(case_fold_flag, pstart, &s, n, end); + while (sprev + (len = enclen_approx(encode, sprev, end)) < s) + sprev += len; + + MOP_OUT; + JUMP; } NEXT; CASE(OP_BACKREF_MULTI) MOP_IN(OP_BACKREF_MULTI); { - int len, is_fail; - UChar *pstart, *pend, *swork; - - GET_LENGTH_INC(tlen, p); - for (i = 0; i < tlen; i++) { - GET_MEMNUM_INC(mem, p); - - if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue; - if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue; - - if (BIT_STATUS_AT(reg->bt_mem_start, mem)) - pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; - else - pstart = (UChar* )((void* )mem_start_stk[mem]); - - pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) - ? STACK_AT(mem_end_stk[mem])->u.mem.pstr - : (UChar* )((void* )mem_end_stk[mem])); - n = pend - pstart; - DATA_ENSURE_CONTINUE(n); - sprev = s; - swork = s; - STRING_CMP_VALUE(pstart, swork, n, is_fail); - if (is_fail) continue; - s = swork; - while (sprev + (len = enclen_approx(encode, sprev, end)) < s) - sprev += len; - - p += (SIZE_MEMNUM * (tlen - i - 1)); - break; /* success */ - } - if (i == tlen) goto fail; - MOP_OUT; - JUMP; + int len, is_fail; + UChar *pstart, *pend, *swork; + + GET_LENGTH_INC(tlen, p); + for (i = 0; i < tlen; i++) { + GET_MEMNUM_INC(mem, p); + + if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue; + if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue; + + if (BIT_STATUS_AT(reg->bt_mem_start, mem)) + pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; + else + pstart = (UChar* )((void* )mem_start_stk[mem]); + + pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) + ? STACK_AT(mem_end_stk[mem])->u.mem.pstr + : (UChar* )((void* )mem_end_stk[mem])); + n = pend - pstart; + DATA_ENSURE_CONTINUE(n); + sprev = s; + swork = s; + STRING_CMP_VALUE(pstart, swork, n, is_fail); + if (is_fail) continue; + s = swork; + while (sprev + (len = enclen_approx(encode, sprev, end)) < s) + sprev += len; + + p += (SIZE_MEMNUM * (tlen - i - 1)); + break; /* success */ + } + if (i == tlen) goto fail; + MOP_OUT; + JUMP; } NEXT; CASE(OP_BACKREF_MULTI_IC) MOP_IN(OP_BACKREF_MULTI_IC); { - int len, is_fail; - UChar *pstart, *pend, *swork; - - GET_LENGTH_INC(tlen, p); - for (i = 0; i < tlen; i++) { - GET_MEMNUM_INC(mem, p); - - if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue; - if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue; - - if (BIT_STATUS_AT(reg->bt_mem_start, mem)) - pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; - else - pstart = (UChar* )((void* )mem_start_stk[mem]); - - pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) - ? STACK_AT(mem_end_stk[mem])->u.mem.pstr - : (UChar* )((void* )mem_end_stk[mem])); - n = pend - pstart; - DATA_ENSURE_CONTINUE(n); - sprev = s; - swork = s; - STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, end, is_fail); - if (is_fail) continue; - s = swork; - while (sprev + (len = enclen(encode, sprev, end)) < s) - sprev += len; - - p += (SIZE_MEMNUM * (tlen - i - 1)); - break; /* success */ - } - if (i == tlen) goto fail; - MOP_OUT; - JUMP; + int len, is_fail; + UChar *pstart, *pend, *swork; + + GET_LENGTH_INC(tlen, p); + for (i = 0; i < tlen; i++) { + GET_MEMNUM_INC(mem, p); + + if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue; + if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue; + + if (BIT_STATUS_AT(reg->bt_mem_start, mem)) + pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr; + else + pstart = (UChar* )((void* )mem_start_stk[mem]); + + pend = (BIT_STATUS_AT(reg->bt_mem_end, mem) + ? STACK_AT(mem_end_stk[mem])->u.mem.pstr + : (UChar* )((void* )mem_end_stk[mem])); + n = pend - pstart; + DATA_ENSURE_CONTINUE(n); + sprev = s; + swork = s; + STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, end, is_fail); + if (is_fail) continue; + s = swork; + while (sprev + (len = enclen(encode, sprev, end)) < s) + sprev += len; + + p += (SIZE_MEMNUM * (tlen - i - 1)); + break; /* success */ + } + if (i == tlen) goto fail; + MOP_OUT; + JUMP; } #ifdef USE_BACKREF_WITH_LEVEL CASE(OP_BACKREF_WITH_LEVEL) { - int len; - OnigOptionType ic; - LengthType level; - - GET_OPTION_INC(ic, p); - GET_LENGTH_INC(level, p); - GET_LENGTH_INC(tlen, p); - - sprev = s; - if (backref_match_at_nested_level(reg, stk, stk_base, ic, - case_fold_flag, (int )level, (int )tlen, p, &s, end)) { - while (sprev + (len = enclen(encode, sprev, end)) < s) - sprev += len; - - p += (SIZE_MEMNUM * tlen); - } - else - goto fail; - - MOP_OUT; - JUMP; + int len; + OnigOptionType ic; + LengthType level; + + GET_OPTION_INC(ic, p); + GET_LENGTH_INC(level, p); + GET_LENGTH_INC(tlen, p); + + sprev = s; + if (backref_match_at_nested_level(reg, stk, stk_base, ic, + case_fold_flag, (int )level, (int )tlen, p, &s, end)) { + while (sprev + (len = enclen(encode, sprev, end)) < s) + sprev += len; + + p += (SIZE_MEMNUM * tlen); + } + else + goto fail; + + MOP_OUT; + JUMP; } #endif @@ -3670,33 +3670,33 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_NULL_CHECK_END) MOP_IN(OP_NULL_CHECK_END); { - int isnull; + int isnull; - GET_MEMNUM_INC(mem, p); /* mem: null check id */ - STACK_NULL_CHECK(isnull, mem, s); - if (isnull) { + GET_MEMNUM_INC(mem, p); /* mem: null check id */ + STACK_NULL_CHECK(isnull, mem, s); + if (isnull) { #ifdef ONIG_DEBUG_MATCH - fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%"PRIuPTR" (%p)\n", - (int )mem, (uintptr_t )s, s); + fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%"PRIuPTR" (%p)\n", + (int )mem, (uintptr_t )s, s); #endif - null_check_found: - /* empty loop founded, skip next instruction */ - switch (*p++) { - case OP_JUMP: - case OP_PUSH: - p += SIZE_RELADDR; - break; - case OP_REPEAT_INC: - case OP_REPEAT_INC_NG: - case OP_REPEAT_INC_SG: - case OP_REPEAT_INC_NG_SG: - p += SIZE_MEMNUM; - break; - default: - goto unexpected_bytecode_error; - break; - } - } + null_check_found: + /* empty loop founded, skip next instruction */ + switch (*p++) { + case OP_JUMP: + case OP_PUSH: + p += SIZE_RELADDR; + break; + case OP_REPEAT_INC: + case OP_REPEAT_INC_NG: + case OP_REPEAT_INC_SG: + case OP_REPEAT_INC_NG_SG: + p += SIZE_MEMNUM; + break; + default: + goto unexpected_bytecode_error; + break; + } + } } MOP_OUT; JUMP; @@ -3704,18 +3704,18 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, #ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT CASE(OP_NULL_CHECK_END_MEMST) MOP_IN(OP_NULL_CHECK_END_MEMST); { - int isnull; + int isnull; - GET_MEMNUM_INC(mem, p); /* mem: null check id */ - STACK_NULL_CHECK_MEMST(isnull, mem, s, reg); - if (isnull) { + GET_MEMNUM_INC(mem, p); /* mem: null check id */ + STACK_NULL_CHECK_MEMST(isnull, mem, s, reg); + if (isnull) { # ifdef ONIG_DEBUG_MATCH - fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%"PRIuPTR" (%p)\n", - (int )mem, (uintptr_t )s, s); + fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%"PRIuPTR" (%p)\n", + (int )mem, (uintptr_t )s, s); # endif - if (isnull == -1) goto fail; - goto null_check_found; - } + if (isnull == -1) goto fail; + goto null_check_found; + } } MOP_OUT; JUMP; @@ -3725,25 +3725,25 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_NULL_CHECK_END_MEMST_PUSH) MOP_IN(OP_NULL_CHECK_END_MEMST_PUSH); { - int isnull; + int isnull; - GET_MEMNUM_INC(mem, p); /* mem: null check id */ + GET_MEMNUM_INC(mem, p); /* mem: null check id */ # ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT - STACK_NULL_CHECK_MEMST_REC(isnull, mem, s, reg); + STACK_NULL_CHECK_MEMST_REC(isnull, mem, s, reg); # else - STACK_NULL_CHECK_REC(isnull, mem, s); + STACK_NULL_CHECK_REC(isnull, mem, s); # endif - if (isnull) { + if (isnull) { # ifdef ONIG_DEBUG_MATCH - fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%"PRIuPTR" (%p)\n", - (int )mem, (uintptr_t )s, s); + fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%"PRIuPTR" (%p)\n", + (int )mem, (uintptr_t )s, s); # endif - if (isnull == -1) goto fail; - goto null_check_found; - } - else { - STACK_PUSH_NULL_CHECK_END(mem); - } + if (isnull == -1) goto fail; + goto null_check_found; + } + else { + STACK_PUSH_NULL_CHECK_END(mem); + } } MOP_OUT; JUMP; @@ -3779,10 +3779,10 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_RELADDR_INC(addr, p); STATE_CHECK_VAL(scv, mem); if (scv) { - p += addr; + p += addr; } else { - STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep); + STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep); } MOP_OUT; JUMP; @@ -3811,11 +3811,11 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_PUSH_OR_JUMP_EXACT1) MOP_IN(OP_PUSH_OR_JUMP_EXACT1); GET_RELADDR_INC(addr, p); if (*p == *s && DATA_ENSURE_CHECK1) { - p++; - CHECK_MATCH_CACHE; - STACK_PUSH_ALT(p + addr, s, sprev, pkeep); - MOP_OUT; - JUMP; + p++; + CHECK_MATCH_CACHE; + STACK_PUSH_ALT(p + addr, s, sprev, pkeep); + MOP_OUT; + JUMP; } p += (addr + 1); MOP_OUT; @@ -3826,10 +3826,10 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_RELADDR_INC(addr, p); CHECK_MATCH_CACHE; if (*p == *s) { - p++; - STACK_PUSH_ALT(p + addr, s, sprev, pkeep); - MOP_OUT; - JUMP; + p++; + STACK_PUSH_ALT(p + addr, s, sprev, pkeep); + MOP_OUT; + JUMP; } p++; INC_NUM_FAILS; @@ -3838,35 +3838,35 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_REPEAT) MOP_IN(OP_REPEAT); { - GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ - GET_RELADDR_INC(addr, p); + GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ + GET_RELADDR_INC(addr, p); - STACK_ENSURE(1); - repeat_stk[mem] = GET_STACK_INDEX(stk); - STACK_PUSH_REPEAT(mem, p); + STACK_ENSURE(1); + repeat_stk[mem] = GET_STACK_INDEX(stk); + STACK_PUSH_REPEAT(mem, p); - if (reg->repeat_range[mem].lower == 0) { - CHECK_MATCH_CACHE; - STACK_PUSH_ALT(p + addr, s, sprev, pkeep); - } + if (reg->repeat_range[mem].lower == 0) { + CHECK_MATCH_CACHE; + STACK_PUSH_ALT(p + addr, s, sprev, pkeep); + } } MOP_OUT; JUMP; CASE(OP_REPEAT_NG) MOP_IN(OP_REPEAT_NG); { - GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ - GET_RELADDR_INC(addr, p); - - STACK_ENSURE(1); - repeat_stk[mem] = GET_STACK_INDEX(stk); - STACK_PUSH_REPEAT(mem, p); - - if (reg->repeat_range[mem].lower == 0) { - CHECK_MATCH_CACHE; - STACK_PUSH_ALT(p, s, sprev, pkeep); - p += addr; - } + GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */ + GET_RELADDR_INC(addr, p); + + STACK_ENSURE(1); + repeat_stk[mem] = GET_STACK_INDEX(stk); + STACK_PUSH_REPEAT(mem, p); + + if (reg->repeat_range[mem].lower == 0) { + CHECK_MATCH_CACHE; + STACK_PUSH_ALT(p, s, sprev, pkeep); + p += addr; + } } MOP_OUT; JUMP; @@ -3879,23 +3879,23 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, repeat_inc: stkp->u.repeat.count++; if (stkp->u.repeat.count >= reg->repeat_range[mem].upper) { - /* end of repeat. Nothing to do. */ + /* end of repeat. Nothing to do. */ } else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { #ifdef USE_MATCH_CACHE - if (*pbegin == OP_REPEAT_INC) { + if (*pbegin == OP_REPEAT_INC) { #undef MATCH_CACHE_HIT #define MATCH_CACHE_HIT stkp->u.repeat.count--; - CHECK_MATCH_CACHE; + CHECK_MATCH_CACHE; #undef MATCH_CACHE_HIT #define MATCH_CACHE_HIT ((void) 0) - } + } #endif - STACK_PUSH_ALT(p, s, sprev, pkeep); - p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */ + STACK_PUSH_ALT(p, s, sprev, pkeep); + p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */ } else { - p = stkp->u.repeat.pcode; + p = stkp->u.repeat.pcode; } STACK_PUSH_REPEAT_INC(si); MOP_OUT; @@ -3917,22 +3917,22 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, repeat_inc_ng: stkp->u.repeat.count++; if (stkp->u.repeat.count < reg->repeat_range[mem].upper) { - if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { - UChar* pcode = stkp->u.repeat.pcode; - - STACK_PUSH_REPEAT_INC(si); - if (*pbegin == OP_REPEAT_INC_NG) { - CHECK_MATCH_CACHE; - } - STACK_PUSH_ALT(pcode, s, sprev, pkeep); - } - else { - p = stkp->u.repeat.pcode; - STACK_PUSH_REPEAT_INC(si); - } + if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) { + UChar* pcode = stkp->u.repeat.pcode; + + STACK_PUSH_REPEAT_INC(si); + if (*pbegin == OP_REPEAT_INC_NG) { + CHECK_MATCH_CACHE; + } + STACK_PUSH_ALT(pcode, s, sprev, pkeep); + } + else { + p = stkp->u.repeat.pcode; + STACK_PUSH_REPEAT_INC(si); + } } else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) { - STACK_PUSH_REPEAT_INC(si); + STACK_PUSH_REPEAT_INC(si); } MOP_OUT; CHECK_INTERRUPT_IN_MATCH_AT; @@ -3952,9 +3952,9 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_POP_POS) MOP_IN(OP_POP_POS); { - STACK_POS_END(stkp); - s = stkp->u.state.pstr; - sprev = stkp->u.state.pstr_prev; + STACK_POS_END(stkp); + s = stkp->u.state.pstr; + sprev = stkp->u.state.pstr_prev; } MOP_OUT; JUMP; @@ -3993,15 +3993,15 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_LENGTH_INC(tlen, p); q = (UChar* )ONIGENC_STEP_BACK(encode, str, s, end, (int )tlen); if (IS_NULL(q)) { - /* too short case -> success. ex. /(?<!XXX)a/.match("a") - If you want to change to fail, replace following line. */ - p += addr; - /* goto fail; */ + /* too short case -> success. ex. /(?<!XXX)a/.match("a") + If you want to change to fail, replace following line. */ + p += addr; + /* goto fail; */ } else { - STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev, pkeep); - s = q; - sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end); + STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev, pkeep); + s = q; + sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end); } MOP_OUT; JUMP; @@ -4019,44 +4019,44 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_ABSENT) MOP_IN(OP_ABSENT); { - const UChar* aend = ABSENT_END_POS; - UChar* absent; - UChar* selfp = p - 1; + const UChar* aend = ABSENT_END_POS; + UChar* absent; + UChar* selfp = p - 1; - STACK_POP_ABSENT_POS(absent, ABSENT_END_POS); /* Restore end-pos. */ - GET_RELADDR_INC(addr, p); + STACK_POP_ABSENT_POS(absent, ABSENT_END_POS); /* Restore end-pos. */ + GET_RELADDR_INC(addr, p); #ifdef ONIG_DEBUG_MATCH - fprintf(stderr, "ABSENT: s:%p, end:%p, absent:%p, aend:%p\n", s, end, absent, aend); + fprintf(stderr, "ABSENT: s:%p, end:%p, absent:%p, aend:%p\n", s, end, absent, aend); #endif - if ((absent > aend) && (s > absent)) { - /* An empty match occurred in (?~...) at the start point. - * Never match. */ - STACK_POP; - goto fail; - } - else if ((s >= aend) && (s > absent)) { - if (s > aend) { - /* Only one (or less) character matched in the last iteration. - * This is not a possible point. */ - goto fail; - } - /* All possible points were found. Try matching after (?~...). */ - DATA_ENSURE(0); - p += addr; - } - else if (s == end) { - /* At the end of the string, just match with it */ - DATA_ENSURE(0); - p += addr; - } - else { - STACK_PUSH_ALT(p + addr, s, sprev, pkeep); /* Push possible point. */ - n = enclen(encode, s, end); - STACK_PUSH_ABSENT_POS(absent, ABSENT_END_POS); /* Save the original pos. */ - STACK_PUSH_ALT(selfp, s + n, s, pkeep); /* Next iteration. */ - STACK_PUSH_ABSENT; - ABSENT_END_POS = aend; - } + if ((absent > aend) && (s > absent)) { + /* An empty match occurred in (?~...) at the start point. + * Never match. */ + STACK_POP; + goto fail; + } + else if ((s >= aend) && (s > absent)) { + if (s > aend) { + /* Only one (or less) character matched in the last iteration. + * This is not a possible point. */ + goto fail; + } + /* All possible points were found. Try matching after (?~...). */ + DATA_ENSURE(0); + p += addr; + } + else if (s == end) { + /* At the end of the string, just match with it */ + DATA_ENSURE(0); + p += addr; + } + else { + STACK_PUSH_ALT(p + addr, s, sprev, pkeep); /* Push possible point. */ + n = enclen(encode, s, end); + STACK_PUSH_ABSENT_POS(absent, ABSENT_END_POS); /* Save the original pos. */ + STACK_PUSH_ALT(selfp, s + n, s, pkeep); /* Next iteration. */ + STACK_PUSH_ABSENT; + ABSENT_END_POS = aend; + } } MOP_OUT; JUMP; @@ -4065,7 +4065,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, /* The pattern inside (?~...) was matched. * Set the end-pos temporary and go to next iteration. */ if (sprev < ABSENT_END_POS) - ABSENT_END_POS = sprev; + ABSENT_END_POS = sprev; #ifdef ONIG_DEBUG_MATCH fprintf(stderr, "ABSENT_END: end:%p\n", ABSENT_END_POS); #endif @@ -4092,9 +4092,9 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, GET_MEMNUM_INC(mem, p); GET_RELADDR_INC(addr, p); if ((mem > num_mem) || - (mem_end_stk[mem] == INVALID_STACK_INDEX) || - (mem_start_stk[mem] == INVALID_STACK_INDEX)) { - p += addr; + (mem_end_stk[mem] == INVALID_STACK_INDEX) || + (mem_start_stk[mem] == INVALID_STACK_INDEX)) { + p += addr; } MOP_OUT; JUMP; @@ -4105,9 +4105,9 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, CASE(OP_FAIL) if (0) { - /* fall */ + /* fall */ fail: - MOP_OUT; + MOP_OUT; } MOP_IN(OP_FAIL); STACK_POP; @@ -4118,71 +4118,71 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, #ifdef USE_MATCH_CACHE if ( - msa->match_cache_status != MATCH_CACHE_STATUS_DISABLED && - ++msa->num_fails >= (long)(end - str) * msa->num_cache_opcodes + msa->match_cache_status != MATCH_CACHE_STATUS_DISABLED && + ++msa->num_fails >= (long)(end - str) * msa->num_cache_opcodes ) { - if (msa->match_cache_status == MATCH_CACHE_STATUS_UNINIT) { - msa->match_cache_status = MATCH_CACHE_STATUS_INIT; - OnigPosition r = count_num_cache_opcodes(reg, &msa->num_cache_opcodes); - if (r < 0) goto bytecode_error; - } - if (msa->num_cache_opcodes == NUM_CACHE_OPCODES_IMPOSSIBLE || msa->num_cache_opcodes == 0) { - msa->match_cache_status = MATCH_CACHE_STATUS_DISABLED; - goto fail_match_cache; - } - if (msa->num_fails < (long)(end - str) * msa->num_cache_opcodes) { - goto fail_match_cache; - } - if (msa->cache_opcodes == NULL) { - msa->match_cache_status = MATCH_CACHE_STATUS_ENABLED; - OnigCacheOpcode* cache_opcodes = (OnigCacheOpcode*)xmalloc(msa->num_cache_opcodes * sizeof(OnigCacheOpcode)); - if (cache_opcodes == NULL) { - return ONIGERR_MEMORY; - } - OnigPosition r = init_cache_opcodes(reg, cache_opcodes, &msa->num_cache_points); - if (r < 0) { - if (r == ONIGERR_UNEXPECTED_BYTECODE) goto unexpected_bytecode_error; - else goto bytecode_error; - } - msa->cache_opcodes = cache_opcodes; + if (msa->match_cache_status == MATCH_CACHE_STATUS_UNINIT) { + msa->match_cache_status = MATCH_CACHE_STATUS_INIT; + OnigPosition r = count_num_cache_opcodes(reg, &msa->num_cache_opcodes); + if (r < 0) goto bytecode_error; + } + if (msa->num_cache_opcodes == NUM_CACHE_OPCODES_IMPOSSIBLE || msa->num_cache_opcodes == 0) { + msa->match_cache_status = MATCH_CACHE_STATUS_DISABLED; + goto fail_match_cache; + } + if (msa->num_fails < (long)(end - str) * msa->num_cache_opcodes) { + goto fail_match_cache; + } + if (msa->cache_opcodes == NULL) { + msa->match_cache_status = MATCH_CACHE_STATUS_ENABLED; + OnigCacheOpcode* cache_opcodes = (OnigCacheOpcode*)xmalloc(msa->num_cache_opcodes * sizeof(OnigCacheOpcode)); + if (cache_opcodes == NULL) { + return ONIGERR_MEMORY; + } + OnigPosition r = init_cache_opcodes(reg, cache_opcodes, &msa->num_cache_points); + if (r < 0) { + if (r == ONIGERR_UNEXPECTED_BYTECODE) goto unexpected_bytecode_error; + else goto bytecode_error; + } + msa->cache_opcodes = cache_opcodes; #ifdef ONIG_DEBUG_MATCH_CACHE - fprintf(stderr, "MATCH CACHE: #cache opcodes = %ld\n", msa->num_cache_opcodes); - fprintf(stderr, "MATCH CACHE: #cache points = %ld\n", msa->num_cache_points); - fprintf(stderr, "MATCH CACHE: cache opcodes (%p):\n", msa->cache_opcodes); - for (int i = 0; i < msa->num_cache_opcodes; i++) { - fprintf(stderr, "MATCH CACHE: [%p] cache_point=%ld outer_repeat_mem=%d num_cache_opcodes_at_outer_repeat=%ld num_cache_opcodes_in_outer_repeat=%ld lookaround_nesting=%d match_addr=%p\n", msa->cache_opcodes[i].addr, msa->cache_opcodes[i].cache_point, msa->cache_opcodes[i].outer_repeat_mem, msa->cache_opcodes[i].num_cache_points_at_outer_repeat, msa->cache_opcodes[i].num_cache_points_in_outer_repeat, msa->cache_opcodes[i].lookaround_nesting, msa->cache_opcodes[i].match_addr); - } + fprintf(stderr, "MATCH CACHE: #cache opcodes = %ld\n", msa->num_cache_opcodes); + fprintf(stderr, "MATCH CACHE: #cache points = %ld\n", msa->num_cache_points); + fprintf(stderr, "MATCH CACHE: cache opcodes (%p):\n", msa->cache_opcodes); + for (int i = 0; i < msa->num_cache_opcodes; i++) { + fprintf(stderr, "MATCH CACHE: [%p] cache_point=%ld outer_repeat_mem=%d num_cache_opcodes_at_outer_repeat=%ld num_cache_opcodes_in_outer_repeat=%ld lookaround_nesting=%d match_addr=%p\n", msa->cache_opcodes[i].addr, msa->cache_opcodes[i].cache_point, msa->cache_opcodes[i].outer_repeat_mem, msa->cache_opcodes[i].num_cache_points_at_outer_repeat, msa->cache_opcodes[i].num_cache_points_in_outer_repeat, msa->cache_opcodes[i].lookaround_nesting, msa->cache_opcodes[i].match_addr); + } #endif - } - if (msa->match_cache_buf == NULL) { - size_t length = (end - str) + 1; - size_t num_match_cache_points = (size_t)msa->num_cache_points * length; + } + if (msa->match_cache_buf == NULL) { + size_t length = (end - str) + 1; + size_t num_match_cache_points = (size_t)msa->num_cache_points * length; #ifdef ONIG_DEBUG_MATCH_CACHE - fprintf(stderr, "MATCH CACHE: #match cache points = %"PRIuSIZE" (length = %"PRIuSIZE")\n", num_match_cache_points, length); + fprintf(stderr, "MATCH CACHE: #match cache points = %"PRIuSIZE" (length = %"PRIuSIZE")\n", num_match_cache_points, length); #endif - /* Overflow check */ - if (num_match_cache_points / length != (size_t)msa->num_cache_points) { - return ONIGERR_MEMORY; - } - if (num_match_cache_points >= LONG_MAX_LIMIT) { - return ONIGERR_MEMORY; - } - size_t match_cache_buf_length = (num_match_cache_points >> 3) + (num_match_cache_points & 7 ? 1 : 0) + 1; - uint8_t* match_cache_buf = (uint8_t*)xmalloc(match_cache_buf_length * sizeof(uint8_t)); - if (match_cache_buf == NULL) { - return ONIGERR_MEMORY; - } - xmemset(match_cache_buf, 0, match_cache_buf_length * sizeof(uint8_t)); - msa->match_cache_buf = match_cache_buf; - } + /* Overflow check */ + if (num_match_cache_points / length != (size_t)msa->num_cache_points) { + return ONIGERR_MEMORY; + } + if (num_match_cache_points >= LONG_MAX_LIMIT) { + return ONIGERR_MEMORY; + } + size_t match_cache_buf_length = (num_match_cache_points >> 3) + (num_match_cache_points & 7 ? 1 : 0) + 1; + uint8_t* match_cache_buf = (uint8_t*)xmalloc(match_cache_buf_length * sizeof(uint8_t)); + if (match_cache_buf == NULL) { + return ONIGERR_MEMORY; + } + xmemset(match_cache_buf, 0, match_cache_buf_length * sizeof(uint8_t)); + msa->match_cache_buf = match_cache_buf; + } } fail_match_cache: #endif #ifdef USE_COMBINATION_EXPLOSION_CHECK if (stk->u.state.state_check != 0) { - stk->type = STK_STATE_CHECK_MARK; - stk++; + stk->type = STK_STATE_CHECK_MARK; + stk++; } #endif @@ -4225,7 +4225,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end, static UChar* slow_search(OnigEncoding enc, UChar* target, UChar* target_end, - const UChar* text, const UChar* text_end, UChar* text_range) + const UChar* text, const UChar* text_end, UChar* text_range) { UChar *t, *p, *s, *end; @@ -4241,10 +4241,10 @@ slow_search(OnigEncoding enc, UChar* target, UChar* target_end, while (s < end) { if (*s == *target) { - p = s + 1; - t = target + 1; - if (target_end == t || memcmp(t, p, target_end - t) == 0) - return s; + p = s + 1; + t = target + 1; + if (target_end == t || memcmp(t, p, target_end - t) == 0) + return s; } s += n; } @@ -4255,7 +4255,7 @@ slow_search(OnigEncoding enc, UChar* target, UChar* target_end, p = s + 1; t = target + 1; if (target_end == t || memcmp(t, p, target_end - t) == 0) - return s; + return s; } s += enclen(enc, s, text_end); } @@ -4265,8 +4265,8 @@ slow_search(OnigEncoding enc, UChar* target, UChar* target_end, static int str_lower_case_match(OnigEncoding enc, int case_fold_flag, - const UChar* t, const UChar* tend, - const UChar* p, const UChar* end) + const UChar* t, const UChar* tend, + const UChar* p, const UChar* end) { int lowlen; UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN]; @@ -4285,8 +4285,8 @@ str_lower_case_match(OnigEncoding enc, int case_fold_flag, static UChar* slow_search_ic(OnigEncoding enc, int case_fold_flag, - UChar* target, UChar* target_end, - const UChar* text, const UChar* text_end, UChar* text_range) + UChar* target, UChar* target_end, + const UChar* text, const UChar* text_end, UChar* text_range) { UChar *s, *end; @@ -4299,7 +4299,7 @@ slow_search_ic(OnigEncoding enc, int case_fold_flag, while (s < end) { if (str_lower_case_match(enc, case_fold_flag, target, target_end, - s, text_end)) + s, text_end)) return s; s += enclen(enc, s, text_end); @@ -4310,8 +4310,8 @@ slow_search_ic(OnigEncoding enc, int case_fold_flag, static UChar* slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end, - const UChar* text, const UChar* adjust_text, - const UChar* text_end, const UChar* text_start) + const UChar* text, const UChar* adjust_text, + const UChar* text_end, const UChar* text_start) { UChar *t, *p, *s; @@ -4327,12 +4327,12 @@ slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end, p = s + 1; t = target + 1; while (t < target_end) { - if (*t != *p++) - break; - t++; + if (*t != *p++) + break; + t++; } if (t == target_end) - return s; + return s; } s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s, text_end); } @@ -4342,9 +4342,9 @@ slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end, static UChar* slow_search_backward_ic(OnigEncoding enc, int case_fold_flag, - UChar* target, UChar* target_end, - const UChar* text, const UChar* adjust_text, - const UChar* text_end, const UChar* text_start) + UChar* target, UChar* target_end, + const UChar* text, const UChar* adjust_text, + const UChar* text_end, const UChar* text_start) { UChar *s; @@ -4357,7 +4357,7 @@ slow_search_backward_ic(OnigEncoding enc, int case_fold_flag, while (s >= text) { if (str_lower_case_match(enc, case_fold_flag, - target, target_end, s, text_end)) + target, target_end, s, text_end)) return s; s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s, text_end); @@ -4370,8 +4370,8 @@ slow_search_backward_ic(OnigEncoding enc, int case_fold_flag, /* Boyer-Moore-Horspool search applied to a multibyte string */ static UChar* bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, - const UChar* text_range) + const UChar* text, const UChar* text_end, + const UChar* text_range) { const UChar *s, *se, *t, *p, *end; const UChar *tail; @@ -4379,7 +4379,7 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n", - (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); + (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); # endif tail = target_end - 1; @@ -4395,13 +4395,13 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, p = se = s + tlen1; t = tail; while (*p == *t) { - if (t == target) return (UChar* )s; - p--; t--; + if (t == target) return (UChar* )s; + p--; t--; } skip = reg->map[*se]; t = s; do { - s += enclen(reg->enc, s, end); + s += enclen(reg->enc, s, end); } while ((s - t) < skip && s < end); } } @@ -4411,13 +4411,13 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, p = se = s + tlen1; t = tail; while (*p == *t) { - if (t == target) return (UChar* )s; - p--; t--; + if (t == target) return (UChar* )s; + p--; t--; } skip = reg->int_map[*se]; t = s; do { - s += enclen(reg->enc, s, end); + s += enclen(reg->enc, s, end); } while ((s - t) < skip && s < end); } # endif @@ -4429,14 +4429,14 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, /* Boyer-Moore-Horspool search */ static UChar* bm_search(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, const UChar* text_range) + const UChar* text, const UChar* text_end, const UChar* text_range) { const UChar *s, *t, *p, *end; const UChar *tail; # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n", - (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); + (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); # endif end = text_range + (target_end - target) - 1; @@ -4451,11 +4451,11 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, t = tail; # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_loop: pos: %"PRIdPTR" %s\n", - (intptr_t )(s - text), s); + (intptr_t )(s - text), s); # endif while (*p == *t) { - if (t == target) return (UChar* )p; - p--; t--; + if (t == target) return (UChar* )p; + p--; t--; } s += reg->map[*s]; } @@ -4466,8 +4466,8 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, p = s; t = tail; while (*p == *t) { - if (t == target) return (UChar* )p; - p--; t--; + if (t == target) return (UChar* )p; + p--; t--; } s += reg->int_map[*s]; } @@ -4479,8 +4479,8 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, /* Boyer-Moore-Horspool search applied to a multibyte string (ignore case) */ static UChar* bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, - const UChar* text_range) + const UChar* text, const UChar* text_end, + const UChar* text_range) { const UChar *s, *se, *t, *end; const UChar *tail; @@ -4490,7 +4490,7 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_notrev_ic: text: %d (%p), text_end: %d (%p), text_range: %d (%p)\n", - (int )text, text, (int )text_end, text_end, (int )text_range, text_range); + (int )text, text, (int )text_end, text_end, (int )text_range, text_range); # endif tail = target_end - 1; @@ -4505,12 +4505,12 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { se = s + tlen1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - s, se + 1)) - return (UChar* )s; + s, se + 1)) + return (UChar* )s; skip = reg->map[*se]; t = s; do { - s += enclen(reg->enc, s, end); + s += enclen(reg->enc, s, end); } while ((s - t) < skip && s < end); } } @@ -4519,12 +4519,12 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { se = s + tlen1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - s, se + 1)) - return (UChar* )s; + s, se + 1)) + return (UChar* )s; skip = reg->int_map[*se]; t = s; do { - s += enclen(reg->enc, s, end); + s += enclen(reg->enc, s, end); } while ((s - t) < skip && s < end); } # endif @@ -4536,7 +4536,7 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, /* Boyer-Moore-Horspool search (ignore case) */ static UChar* bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, const UChar* text_range) + const UChar* text, const UChar* text_end, const UChar* text_range) { const UChar *s, *p, *end; const UChar *tail; @@ -4545,7 +4545,7 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_ic: text: %d (%p), text_end: %d (%p), text_range: %d (%p)\n", - (int )text, text, (int )text_end, text_end, (int )text_range, text_range); + (int )text, text, (int )text_end, text_end, (int )text_range, text_range); # endif end = text_range + (target_end - target) - 1; @@ -4558,8 +4558,8 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { p = s - (target_end - target) + 1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - p, s + 1)) - return (UChar* )p; + p, s + 1)) + return (UChar* )p; s += reg->map[*s]; } } @@ -4568,8 +4568,8 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { p = s - (target_end - target) + 1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - p, s + 1)) - return (UChar* )p; + p, s + 1)) + return (UChar* )p; s += reg->int_map[*s]; } # endif @@ -4582,8 +4582,8 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, /* Sunday's quick search applied to a multibyte string */ static UChar* bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, - const UChar* text_range) + const UChar* text, const UChar* text_end, + const UChar* text_range) { const UChar *s, *se, *t, *p, *end; const UChar *tail; @@ -4592,7 +4592,7 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n", - (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); + (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); # endif tail = target_end - 1; @@ -4608,14 +4608,14 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, p = se = s + tlen1; t = tail; while (*p == *t) { - if (t == target) return (UChar* )s; - p--; t--; + if (t == target) return (UChar* )s; + p--; t--; } if (s + 1 >= end) break; skip = reg->map[se[1]]; t = s; do { - s += enclen(enc, s, end); + s += enclen(enc, s, end); } while ((s - t) < skip && s < end); } } @@ -4625,14 +4625,14 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, p = se = s + tlen1; t = tail; while (*p == *t) { - if (t == target) return (UChar* )s; - p--; t--; + if (t == target) return (UChar* )s; + p--; t--; } if (s + 1 >= end) break; skip = reg->int_map[se[1]]; t = s; do { - s += enclen(enc, s, end); + s += enclen(enc, s, end); } while ((s - t) < skip && s < end); } # endif @@ -4644,7 +4644,7 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end, /* Sunday's quick search */ static UChar* bm_search(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, const UChar* text_range) + const UChar* text, const UChar* text_end, const UChar* text_range) { const UChar *s, *t, *p, *end; const UChar *tail; @@ -4652,7 +4652,7 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n", - (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); + (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); # endif tail = target_end - 1; @@ -4667,8 +4667,8 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, p = s; t = tail; while (*p == *t) { - if (t == target) return (UChar* )p; - p--; t--; + if (t == target) return (UChar* )p; + p--; t--; } if (s + 1 >= end) break; s += reg->map[s[1]]; @@ -4680,8 +4680,8 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, p = s; t = tail; while (*p == *t) { - if (t == target) return (UChar* )p; - p--; t--; + if (t == target) return (UChar* )p; + p--; t--; } if (s + 1 >= end) break; s += reg->int_map[s[1]]; @@ -4694,8 +4694,8 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end, /* Sunday's quick search applied to a multibyte string (ignore case) */ static UChar* bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, - const UChar* text_range) + const UChar* text, const UChar* text_end, + const UChar* text_range) { const UChar *s, *se, *t, *end; const UChar *tail; @@ -4705,7 +4705,7 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_notrev_ic: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n", - (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); + (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); # endif tail = target_end - 1; @@ -4720,13 +4720,13 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { se = s + tlen1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - s, se + 1)) - return (UChar* )s; + s, se + 1)) + return (UChar* )s; if (s + 1 >= end) break; skip = reg->map[se[1]]; t = s; do { - s += enclen(enc, s, end); + s += enclen(enc, s, end); } while ((s - t) < skip && s < end); } } @@ -4735,13 +4735,13 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { se = s + tlen1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - s, se + 1)) - return (UChar* )s; + s, se + 1)) + return (UChar* )s; if (s + 1 >= end) break; skip = reg->int_map[se[1]]; t = s; do { - s += enclen(enc, s, end); + s += enclen(enc, s, end); } while ((s - t) < skip && s < end); } # endif @@ -4753,7 +4753,7 @@ bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end, /* Sunday's quick search (ignore case) */ static UChar* bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* text_end, const UChar* text_range) + const UChar* text, const UChar* text_end, const UChar* text_range) { const UChar *s, *p, *end; const UChar *tail; @@ -4763,7 +4763,7 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, # ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "bm_search_ic: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n", - (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); + (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range); # endif tail = target_end - 1; @@ -4777,8 +4777,8 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { p = s - tlen1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - p, s + 1)) - return (UChar* )p; + p, s + 1)) + return (UChar* )p; if (s + 1 >= end) break; s += reg->map[s[1]]; } @@ -4788,8 +4788,8 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, while (s < end) { p = s - tlen1; if (str_lower_case_match(enc, case_fold_flag, target, target_end, - p, s + 1)) - return (UChar* )p; + p, s + 1)) + return (UChar* )p; if (s + 1 >= end) break; s += reg->int_map[s[1]]; } @@ -4802,7 +4802,7 @@ bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end, #ifdef USE_INT_MAP_BACKWARD static int set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED, - int** skip) + int** skip) { int i, len; @@ -4823,8 +4823,8 @@ set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED, static UChar* bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end, - const UChar* text, const UChar* adjust_text, - const UChar* text_end, const UChar* text_start) + const UChar* text, const UChar* adjust_text, + const UChar* text_end, const UChar* text_start) { const UChar *s, *t, *p; @@ -4853,7 +4853,7 @@ bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end, static UChar* map_search(OnigEncoding enc, UChar map[], - const UChar* text, const UChar* text_range, const UChar* text_end) + const UChar* text, const UChar* text_range, const UChar* text_end) { const UChar *s = text; @@ -4867,8 +4867,8 @@ map_search(OnigEncoding enc, UChar map[], static UChar* map_search_backward(OnigEncoding enc, UChar map[], - const UChar* text, const UChar* adjust_text, - const UChar* text_start, const UChar* text_end) + const UChar* text, const UChar* adjust_text, + const UChar* text_start, const UChar* text_end) { const UChar *s = text_start; @@ -4882,7 +4882,7 @@ map_search_backward(OnigEncoding enc, UChar map[], extern OnigPosition onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, OnigRegion* region, - OnigOptionType option) + OnigOptionType option) { ptrdiff_t r; UChar *prev; @@ -4906,9 +4906,9 @@ onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, On prev = (UChar* )onigenc_get_prev_char_head(reg->enc, str, at, end); r = match_at(reg, str, end, #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE - end, + end, #endif - at, prev, &msa); + at, prev, &msa); } MATCH_ARG_FREE(msa); @@ -4917,14 +4917,14 @@ onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, On static int forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s, - UChar* range, UChar** low, UChar** high, UChar** low_prev) + UChar* range, UChar** low, UChar** high, UChar** low_prev) { UChar *p, *pprev = (UChar* )NULL; size_t input_len = end - str; #ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "forward_search_range: str: %"PRIuPTR" (%p), end: %"PRIuPTR" (%p), s: %"PRIuPTR" (%p), range: %"PRIuPTR" (%p)\n", - (uintptr_t )str, str, (uintptr_t )end, end, (uintptr_t )s, s, (uintptr_t )range, range); + (uintptr_t )str, str, (uintptr_t )end, end, (uintptr_t )s, s, (uintptr_t )range, range); #endif if (reg->dmin > input_len) { @@ -4951,7 +4951,7 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s, break; case ONIG_OPTIMIZE_EXACT_IC: p = slow_search_ic(reg->enc, reg->case_fold_flag, - reg->exact, reg->exact_end, p, end, range); + reg->exact, reg->exact_end, p, end, range); break; case ONIG_OPTIMIZE_EXACT_BM: @@ -4988,62 +4988,62 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s, switch (reg->sub_anchor) { case ANCHOR_BEGIN_LINE: - if (!ON_STR_BEGIN(p)) { - prev = onigenc_get_prev_char_head(reg->enc, - (pprev ? pprev : str), p, end); - if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) - goto retry_gate; - } - break; + if (!ON_STR_BEGIN(p)) { + prev = onigenc_get_prev_char_head(reg->enc, + (pprev ? pprev : str), p, end); + if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) + goto retry_gate; + } + break; case ANCHOR_END_LINE: - if (ON_STR_END(p)) { + if (ON_STR_END(p)) { #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE - prev = (UChar* )onigenc_get_prev_char_head(reg->enc, - (pprev ? pprev : str), p); - if (prev && ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1)) - goto retry_gate; + prev = (UChar* )onigenc_get_prev_char_head(reg->enc, + (pprev ? pprev : str), p); + if (prev && ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1)) + goto retry_gate; #endif - } - else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1)) - goto retry_gate; - break; + } + else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1)) + goto retry_gate; + break; } } if (reg->dmax == 0) { *low = p; if (low_prev) { - if (*low > s) - *low_prev = onigenc_get_prev_char_head(reg->enc, s, p, end); - else - *low_prev = onigenc_get_prev_char_head(reg->enc, - (pprev ? pprev : str), p, end); + if (*low > s) + *low_prev = onigenc_get_prev_char_head(reg->enc, s, p, end); + else + *low_prev = onigenc_get_prev_char_head(reg->enc, + (pprev ? pprev : str), p, end); } *high = p; } else { if (reg->dmax != ONIG_INFINITE_DISTANCE) { - if ((OnigDistance)(p - str) < reg->dmax) { - *low = (UChar* )str; - if (low_prev) - *low_prev = onigenc_get_prev_char_head(reg->enc, str, *low, end); - } - else { - *low = p - reg->dmax; - if (*low > s) { - *low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s, - *low, end, (const UChar** )low_prev); - if (low_prev && IS_NULL(*low_prev)) - *low_prev = onigenc_get_prev_char_head(reg->enc, - (pprev ? pprev : s), *low, end); - } - else { - if (low_prev) - *low_prev = onigenc_get_prev_char_head(reg->enc, - (pprev ? pprev : str), *low, end); - } - } + if ((OnigDistance)(p - str) < reg->dmax) { + *low = (UChar* )str; + if (low_prev) + *low_prev = onigenc_get_prev_char_head(reg->enc, str, *low, end); + } + else { + *low = p - reg->dmax; + if (*low > s) { + *low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s, + *low, end, (const UChar** )low_prev); + if (low_prev && IS_NULL(*low_prev)) + *low_prev = onigenc_get_prev_char_head(reg->enc, + (pprev ? pprev : s), *low, end); + } + else { + if (low_prev) + *low_prev = onigenc_get_prev_char_head(reg->enc, + (pprev ? pprev : str), *low, end); + } + } } /* no needs to adjust *high, *high is used as range check only */ if ((OnigDistance)(p - str) < reg->dmin) @@ -5055,7 +5055,7 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s, #ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "forward_search_range success: low: %"PRIdPTR", high: %"PRIdPTR", dmin: %"PRIdPTR", dmax: %"PRIdPTR"\n", - *low - str, *high - str, reg->dmin, reg->dmax); + *low - str, *high - str, reg->dmin, reg->dmax); #endif return 1; /* success */ } @@ -5067,8 +5067,8 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s, static int backward_search_range(regex_t* reg, const UChar* str, const UChar* end, - UChar* s, const UChar* range, UChar* adjrange, - UChar** low, UChar** high) + UChar* s, const UChar* range, UChar* adjrange, + UChar** low, UChar** high) { UChar *p; size_t input_len = end - str; @@ -5084,15 +5084,15 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end, case ONIG_OPTIMIZE_EXACT: exact_method: p = slow_search_backward(reg->enc, reg->exact, reg->exact_end, - range, adjrange, end, p); + range, adjrange, end, p); break; case ONIG_OPTIMIZE_EXACT_IC: case ONIG_OPTIMIZE_EXACT_BM_IC: case ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC: p = slow_search_backward_ic(reg->enc, reg->case_fold_flag, - reg->exact, reg->exact_end, - range, adjrange, end, p); + reg->exact, reg->exact_end, + range, adjrange, end, p); break; case ONIG_OPTIMIZE_EXACT_BM: @@ -5101,14 +5101,14 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end, if (IS_NULL(reg->int_map_backward)) { int r; if (s - range < BM_BACKWARD_SEARCH_LENGTH_THRESHOLD) - goto exact_method; + goto exact_method; r = set_bm_backward_skip(reg->exact, reg->exact_end, reg->enc, - &(reg->int_map_backward)); + &(reg->int_map_backward)); if (r) return r; } p = bm_search_backward(reg, reg->exact, reg->exact_end, range, adjrange, - end, p); + end, p); #else goto exact_method; #endif @@ -5125,49 +5125,49 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end, switch (reg->sub_anchor) { case ANCHOR_BEGIN_LINE: - if (!ON_STR_BEGIN(p)) { - prev = onigenc_get_prev_char_head(reg->enc, str, p, end); - if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) { - p = prev; - goto retry; - } - } - break; + if (!ON_STR_BEGIN(p)) { + prev = onigenc_get_prev_char_head(reg->enc, str, p, end); + if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) { + p = prev; + goto retry; + } + } + break; case ANCHOR_END_LINE: - if (ON_STR_END(p)) { + if (ON_STR_END(p)) { #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE - prev = onigenc_get_prev_char_head(reg->enc, adjrange, p); - if (IS_NULL(prev)) goto fail; - if (ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1)) { - p = prev; - goto retry; - } + prev = onigenc_get_prev_char_head(reg->enc, adjrange, p); + if (IS_NULL(prev)) goto fail; + if (ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1)) { + p = prev; + goto retry; + } #endif - } - else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1)) { - p = onigenc_get_prev_char_head(reg->enc, adjrange, p, end); - if (IS_NULL(p)) goto fail; - goto retry; - } - break; + } + else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1)) { + p = onigenc_get_prev_char_head(reg->enc, adjrange, p, end); + if (IS_NULL(p)) goto fail; + goto retry; + } + break; } } if (reg->dmax != ONIG_INFINITE_DISTANCE) { if ((OnigDistance)(p - str) < reg->dmax) - *low = (UChar* )str; + *low = (UChar* )str; else - *low = p - reg->dmax; + *low = p - reg->dmax; if (reg->dmin != 0) { - if ((OnigDistance)(p - str) < reg->dmin) - *high = (UChar* )str; - else - *high = p - reg->dmin; + if ((OnigDistance)(p - str) < reg->dmin) + *high = (UChar* )str; + else + *high = p - reg->dmin; } else { - *high = p; + *high = p; } *high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high, end); @@ -5175,7 +5175,7 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end, #ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "backward_search_range: low: %d, high: %d\n", - (int )(*low - str), (int )(*high - str)); + (int )(*low - str), (int )(*high - str)); #endif return 1; /* success */ } @@ -5190,15 +5190,15 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end, extern OnigPosition onig_search(regex_t* reg, const UChar* str, const UChar* end, - const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option) + const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option) { return onig_search_gpos(reg, str, end, start, start, range, region, option); } extern OnigPosition onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, - const UChar* global_pos, - const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option) + const UChar* global_pos, + const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option) { ptrdiff_t r; UChar *s, *prev; @@ -5298,30 +5298,30 @@ onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, begin_position: if (range > start) { - if (global_pos > start) - { - if (global_pos < range) - range = global_pos + 1; - } - else - range = start + 1; + if (global_pos > start) + { + if (global_pos < range) + range = global_pos + 1; + } + else + range = start + 1; } else - range = start; + range = start; } else if (reg->anchor & ANCHOR_BEGIN_BUF) { /* search str-position only */ if (range > start) { - if (start != str) goto mismatch_no_msa; - range = str + 1; + if (start != str) goto mismatch_no_msa; + range = str + 1; } else { - if (range <= str) { - start = str; - range = str; - } - else - goto mismatch_no_msa; + if (range <= str) { + start = str; + range = str; + } + else + goto mismatch_no_msa; } } else if (reg->anchor & ANCHOR_END_BUF) { @@ -5329,38 +5329,38 @@ onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, end_buf: if ((OnigDistance)(max_semi_end - str) < reg->anchor_dmin) - goto mismatch_no_msa; + goto mismatch_no_msa; if (range > start) { - if ((OnigDistance)(min_semi_end - start) > reg->anchor_dmax) { - start = min_semi_end - reg->anchor_dmax; - if (start < end) - start = onigenc_get_right_adjust_char_head(reg->enc, str, start, end); - } - if ((OnigDistance)(max_semi_end - (range - 1)) < reg->anchor_dmin) { - if ((OnigDistance)(max_semi_end - str + 1) < reg->anchor_dmin) - goto mismatch_no_msa; - else - range = max_semi_end - reg->anchor_dmin + 1; - } - - if (start > range) goto mismatch_no_msa; - /* If start == range, match with empty at end. - Backward search is used. */ + if ((OnigDistance)(min_semi_end - start) > reg->anchor_dmax) { + start = min_semi_end - reg->anchor_dmax; + if (start < end) + start = onigenc_get_right_adjust_char_head(reg->enc, str, start, end); + } + if ((OnigDistance)(max_semi_end - (range - 1)) < reg->anchor_dmin) { + if ((OnigDistance)(max_semi_end - str + 1) < reg->anchor_dmin) + goto mismatch_no_msa; + else + range = max_semi_end - reg->anchor_dmin + 1; + } + + if (start > range) goto mismatch_no_msa; + /* If start == range, match with empty at end. + Backward search is used. */ } else { - if ((OnigDistance)(min_semi_end - range) > reg->anchor_dmax) { - range = min_semi_end - reg->anchor_dmax; - } - if ((OnigDistance)(max_semi_end - start) < reg->anchor_dmin) { - if ((OnigDistance)(max_semi_end - str) < reg->anchor_dmin) - goto mismatch_no_msa; - else { - start = max_semi_end - reg->anchor_dmin; - start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start, end); - } - } - if (range > start) goto mismatch_no_msa; + if ((OnigDistance)(min_semi_end - range) > reg->anchor_dmax) { + range = min_semi_end - reg->anchor_dmax; + } + if ((OnigDistance)(max_semi_end - start) < reg->anchor_dmin) { + if ((OnigDistance)(max_semi_end - str) < reg->anchor_dmin) + goto mismatch_no_msa; + else { + start = max_semi_end - reg->anchor_dmin; + start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start, end); + } + } + if (range > start) goto mismatch_no_msa; } } else if (reg->anchor & ANCHOR_SEMI_END_BUF) { @@ -5368,23 +5368,23 @@ onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, max_semi_end = (UChar* )end; if (ONIGENC_IS_MBC_NEWLINE(reg->enc, pre_end, end)) { - min_semi_end = pre_end; + min_semi_end = pre_end; #ifdef USE_CRNL_AS_LINE_TERMINATOR - pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, end, 1); - if (IS_NOT_NULL(pre_end) && - IS_NEWLINE_CRLF(reg->options) && - ONIGENC_IS_MBC_CRNL(reg->enc, pre_end, end)) { - min_semi_end = pre_end; - } + pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, end, 1); + if (IS_NOT_NULL(pre_end) && + IS_NEWLINE_CRLF(reg->options) && + ONIGENC_IS_MBC_CRNL(reg->enc, pre_end, end)) { + min_semi_end = pre_end; + } #endif - if (min_semi_end > str && start <= min_semi_end) { - goto end_buf; - } + if (min_semi_end > str && start <= min_semi_end) { + goto end_buf; + } } else { - min_semi_end = (UChar* )end; - goto end_buf; + min_semi_end = (UChar* )end; + goto end_buf; } } else if ((reg->anchor & ANCHOR_ANYCHAR_STAR_ML)) { @@ -5416,7 +5416,7 @@ onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, #ifdef ONIG_DEBUG_SEARCH fprintf(stderr, "onig_search(apply anchor): end: %d, start: %d, range: %d\n", - (int )(end - str), (int )(start - str), (int )(range - str)); + (int )(end - str), (int )(start - str), (int )(range - str)); #endif MATCH_ARG_INIT(msa, option, region, start, global_pos); @@ -5438,58 +5438,58 @@ onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, UChar *sch_range, *low, *high, *low_prev; if (reg->dmax != 0) { - if (reg->dmax == ONIG_INFINITE_DISTANCE) - sch_range = (UChar* )end; - else { - if ((OnigDistance)(end - range) < reg->dmax) - sch_range = (UChar* )end; - else { - sch_range = (UChar* )range + reg->dmax; - } - } + if (reg->dmax == ONIG_INFINITE_DISTANCE) + sch_range = (UChar* )end; + else { + if ((OnigDistance)(end - range) < reg->dmax) + sch_range = (UChar* )end; + else { + sch_range = (UChar* )range + reg->dmax; + } + } } else - sch_range = (UChar* )range; + sch_range = (UChar* )range; if ((end - start) < reg->threshold_len) - goto mismatch; + goto mismatch; if (reg->dmax != ONIG_INFINITE_DISTANCE) { - do { - if (! forward_search_range(reg, str, end, s, sch_range, - &low, &high, &low_prev)) goto mismatch; - if (s < low) { - s = low; - prev = low_prev; - } - while (s <= high) { - MATCH_AND_RETURN_CHECK(orig_range); - prev = s; - s += enclen(reg->enc, s, end); - } - } while (s < range); - goto mismatch; + do { + if (! forward_search_range(reg, str, end, s, sch_range, + &low, &high, &low_prev)) goto mismatch; + if (s < low) { + s = low; + prev = low_prev; + } + while (s <= high) { + MATCH_AND_RETURN_CHECK(orig_range); + prev = s; + s += enclen(reg->enc, s, end); + } + } while (s < range); + goto mismatch; } else { /* check only. */ - if (! forward_search_range(reg, str, end, s, sch_range, - &low, &high, (UChar** )NULL)) goto mismatch; - - if ((reg->anchor & ANCHOR_ANYCHAR_STAR) != 0) { - do { - MATCH_AND_RETURN_CHECK(orig_range); - prev = s; - s += enclen(reg->enc, s, end); - - if ((reg->anchor & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) == 0) { - while (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0) - && s < range) { - prev = s; - s += enclen(reg->enc, s, end); - } - } - } while (s < range); - goto mismatch; - } + if (! forward_search_range(reg, str, end, s, sch_range, + &low, &high, (UChar** )NULL)) goto mismatch; + + if ((reg->anchor & ANCHOR_ANYCHAR_STAR) != 0) { + do { + MATCH_AND_RETURN_CHECK(orig_range); + prev = s; + s += enclen(reg->enc, s, end); + + if ((reg->anchor & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) == 0) { + while (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0) + && s < range) { + prev = s; + s += enclen(reg->enc, s, end); + } + } + } while (s < range); + goto mismatch; + } } } @@ -5509,58 +5509,58 @@ onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end, const UChar *min_range; if (range < end) - adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range, end); + adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range, end); else - adjrange = (UChar* )end; + adjrange = (UChar* )end; if ((OnigDistance)(end - range) > reg->dmin) - min_range = range + reg->dmin; + min_range = range + reg->dmin; else - min_range = end; + min_range = end; if (reg->dmax != ONIG_INFINITE_DISTANCE && - end - range >= reg->threshold_len) { - do { - if ((OnigDistance)(end - s) > reg->dmax) - sch_start = s + reg->dmax; - else - sch_start = (UChar* )end; - - if (backward_search_range(reg, str, end, sch_start, min_range, adjrange, - &low, &high) <= 0) - goto mismatch; - - if (s > high) - s = high; - - while (s >= low) { - prev = onigenc_get_prev_char_head(reg->enc, str, s, end); - MATCH_AND_RETURN_CHECK(orig_start); - s = prev; - } - } while (s >= range); - goto mismatch; + end - range >= reg->threshold_len) { + do { + if ((OnigDistance)(end - s) > reg->dmax) + sch_start = s + reg->dmax; + else + sch_start = (UChar* )end; + + if (backward_search_range(reg, str, end, sch_start, min_range, adjrange, + &low, &high) <= 0) + goto mismatch; + + if (s > high) + s = high; + + while (s >= low) { + prev = onigenc_get_prev_char_head(reg->enc, str, s, end); + MATCH_AND_RETURN_CHECK(orig_start); + s = prev; + } + } while (s >= range); + goto mismatch; } else { /* check only. */ - if (end - range < reg->threshold_len) goto mismatch; - - if (reg->dmax != 0) { - if (reg->dmax == ONIG_INFINITE_DISTANCE) - sch_start = (UChar* )end; - else { - if ((OnigDistance)(end - s) > reg->dmax) { - sch_start = s + reg->dmax; - sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, - start, sch_start, end); - } else - sch_start = (UChar* )end; - } - } - else - sch_start = (UChar* )s; - - if (backward_search_range(reg, str, end, sch_start, min_range, adjrange, - &low, &high) <= 0) goto mismatch; + if (end - range < reg->threshold_len) goto mismatch; + + if (reg->dmax != 0) { + if (reg->dmax == ONIG_INFINITE_DISTANCE) + sch_start = (UChar* )end; + else { + if ((OnigDistance)(end - s) > reg->dmax) { + sch_start = s + reg->dmax; + sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, + start, sch_start, end); + } else + sch_start = (UChar* )end; + } + } + else + sch_start = (UChar* )s; + + if (backward_search_range(reg, str, end, sch_start, min_range, adjrange, + &low, &high) <= 0) goto mismatch; } } @@ -5617,9 +5617,9 @@ timeout: extern OnigPosition onig_scan(regex_t* reg, const UChar* str, const UChar* end, - OnigRegion* region, OnigOptionType option, - int (*scan_callback)(OnigPosition, OnigPosition, OnigRegion*, void*), - void* callback_arg) + OnigRegion* region, OnigOptionType option, + int (*scan_callback)(OnigPosition, OnigPosition, OnigRegion*, void*), + void* callback_arg) { OnigPosition r; OnigPosition n; @@ -5634,17 +5634,17 @@ onig_scan(regex_t* reg, const UChar* str, const UChar* end, rs = scan_callback(n, r, region, callback_arg); n++; if (rs != 0) - return rs; + return rs; if (region->end[0] == start - str) { - if (start >= end) break; - start += enclen(reg->enc, start, end); + if (start >= end) break; + start += enclen(reg->enc, start, end); } else - start = str + region->end[0]; + start = str + region->end[0]; if (start > end) - break; + break; } else if (r == ONIG_MISMATCH) { break; diff --git a/regparse.c b/regparse.c index 57ecd78dee..d9ae640a3f 100644 --- a/regparse.c +++ b/regparse.c @@ -329,7 +329,7 @@ strdup_with_null(OnigEncoding enc, const UChar* s, const UChar* end) static UChar* strcat_capa(UChar* dest, UChar* dest_end, const UChar* src, const UChar* src_end, - size_t capa) + size_t capa) { UChar* r; @@ -346,7 +346,7 @@ strcat_capa(UChar* dest, UChar* dest_end, const UChar* src, const UChar* src_end /* dest on static area */ static UChar* strcat_capa_from_static(UChar* dest, UChar* dest_end, - const UChar* src, const UChar* src_end, size_t capa) + const UChar* src, const UChar* src_end, size_t capa) { UChar* r; @@ -426,7 +426,7 @@ onig_st_init_strend_table_with_size(st_index_t size) extern int onig_st_lookup_strend(hash_table_type* table, const UChar* str_key, - const UChar* end_key, hash_data_type *value) + const UChar* end_key, hash_data_type *value) { st_str_end_key key; @@ -438,7 +438,7 @@ onig_st_lookup_strend(hash_table_type* table, const UChar* str_key, extern int onig_st_insert_strend(hash_table_type* table, const UChar* str_key, - const UChar* end_key, hash_data_type value) + const UChar* end_key, hash_data_type value) { st_str_end_key* key; int result; @@ -630,10 +630,10 @@ i_names(HashDataType key_ ARG_UNUSED, HashDataType e_, HashDataType arg_) NameEntry* e = (NameEntry *)e_; INamesArg* arg = (INamesArg *)arg_; int r = (*(arg->func))(e->name, - e->name + e->name_len, - e->back_num, - (e->back_num > 1 ? e->back_refs : &(e->back_ref1)), - arg->reg, arg->arg); + e->name + e->name_len, + e->back_num, + (e->back_num > 1 ? e->back_refs : &(e->back_ref1)), + arg->reg, arg->arg); if (r != 0) { arg->ret = r; return ST_STOP; @@ -725,16 +725,16 @@ onig_print_names(FILE* fp, regex_t* reg) e = &(t->e[i]); fprintf(fp, "%s: ", e->name); if (e->back_num == 0) { - fputs("-", fp); + fputs("-", fp); } else if (e->back_num == 1) { - fprintf(fp, "%d", e->back_ref1); + fprintf(fp, "%d", e->back_ref1); } else { - for (j = 0; j < e->back_num; j++) { - if (j > 0) fprintf(fp, ", "); - fprintf(fp, "%d", e->back_refs[j]); - } + for (j = 0; j < e->back_num; j++) { + if (j > 0) fprintf(fp, ", "); + fprintf(fp, "%d", e->back_refs[j]); + } } fputs("\n", fp); } @@ -755,13 +755,13 @@ names_clear(regex_t* reg) for (i = 0; i < t->num; i++) { e = &(t->e[i]); if (IS_NOT_NULL(e->name)) { - xfree(e->name); - e->name = NULL; - e->name_len = 0; - e->back_num = 0; - e->back_alloc = 0; - xfree(e->back_refs); - e->back_refs = (int* )NULL; + xfree(e->name); + e->name = NULL; + e->name_len = 0; + e->back_num = 0; + e->back_alloc = 0; + xfree(e->back_refs); + e->back_refs = (int* )NULL; } } if (IS_NOT_NULL(t->e)) { @@ -842,7 +842,7 @@ name_find(regex_t* reg, const UChar* name, const UChar* name_end) for (i = 0; i < t->num; i++) { e = &(t->e[i]); if (len == e->name_len && onig_strncmp(name, e->name, len) == 0) - return e; + return e; } } return (NameEntry* )NULL; @@ -860,8 +860,8 @@ onig_foreach_name(regex_t* reg, for (i = 0; i < t->num; i++) { e = &(t->e[i]); r = (*func)(e->name, e->name + e->name_len, e->back_num, - (e->back_num > 1 ? e->back_refs : &(e->back_ref1)), - reg, arg); + (e->back_num > 1 ? e->back_refs : &(e->back_ref1)), + reg, arg); if (r != 0) return r; } } @@ -950,8 +950,8 @@ name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env) t->e = (NameEntry* )xmalloc(sizeof(NameEntry) * alloc); if (IS_NULL(t->e)) { - xfree(t); - return ONIGERR_MEMORY; + xfree(t); + return ONIGERR_MEMORY; } t->alloc = alloc; reg->name_table = t; @@ -969,11 +969,11 @@ name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env) clear: for (i = t->num; i < t->alloc; i++) { - t->e[i].name = NULL; - t->e[i].name_len = 0; - t->e[i].back_num = 0; - t->e[i].back_alloc = 0; - t->e[i].back_refs = (int* )NULL; + t->e[i].name = NULL; + t->e[i].name_len = 0; + t->e[i].back_num = 0; + t->e[i].back_alloc = 0; + t->e[i].back_refs = (int* )NULL; } } e = &(t->e[t->num]); @@ -987,7 +987,7 @@ name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env) if (e->back_num >= 1 && ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME)) { onig_scan_env_set_error_string(env, ONIGERR_MULTIPLEX_DEFINED_NAME, - name, name_end); + name, name_end); return ONIGERR_MULTIPLEX_DEFINED_NAME; } @@ -1006,12 +1006,12 @@ name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env) } else { if (e->back_num > e->back_alloc) { - int* p; - alloc = e->back_alloc * 2; - p = (int* )xrealloc(e->back_refs, sizeof(int) * alloc); - CHECK_NULL_RETURN_MEMERR(p); - e->back_refs = p; - e->back_alloc = alloc; + int* p; + alloc = e->back_alloc * 2; + p = (int* )xrealloc(e->back_refs, sizeof(int) * alloc); + CHECK_NULL_RETURN_MEMERR(p); + e->back_refs = p; + e->back_alloc = alloc; } e->back_refs[e->back_num - 1] = backref; } @@ -1022,7 +1022,7 @@ name_add(regex_t* reg, UChar* name, UChar* name_end, int backref, ScanEnv* env) extern int onig_name_to_group_numbers(regex_t* reg, const UChar* name, - const UChar* name_end, int** nums) + const UChar* name_end, int** nums) { NameEntry* e = name_find(reg, name, name_end); @@ -1044,7 +1044,7 @@ onig_name_to_group_numbers(regex_t* reg, const UChar* name, extern int onig_name_to_backref_number(regex_t* reg, const UChar* name, - const UChar* name_end, const OnigRegion *region) + const UChar* name_end, const OnigRegion *region) { int i, n, *nums; @@ -1058,8 +1058,8 @@ onig_name_to_backref_number(regex_t* reg, const UChar* name, else { if (IS_NOT_NULL(region)) { for (i = n - 1; i >= 0; i--) { - if (region->beg[nums[i]] != ONIG_REGION_NOTPOS) - return nums[i]; + if (region->beg[nums[i]] != ONIG_REGION_NOTPOS) + return nums[i]; } } return nums[n - 1]; @@ -1070,14 +1070,14 @@ onig_name_to_backref_number(regex_t* reg, const UChar* name, extern int onig_name_to_group_numbers(regex_t* reg, const UChar* name, - const UChar* name_end, int** nums) + const UChar* name_end, int** nums) { return ONIG_NO_SUPPORT_CONFIG; } extern int onig_name_to_backref_number(regex_t* reg, const UChar* name, - const UChar* name_end, const OnigRegion* region) + const UChar* name_end, const OnigRegion* region) { return ONIG_NO_SUPPORT_CONFIG; } @@ -1166,20 +1166,20 @@ scan_env_add_mem_entry(ScanEnv* env) if (need >= SCANENV_MEMNODES_SIZE) { if (env->mem_alloc <= need) { if (IS_NULL(env->mem_nodes_dynamic)) { - alloc = INIT_SCANENV_MEMNODES_ALLOC_SIZE; - p = (Node** )xmalloc(sizeof(Node*) * alloc); - CHECK_NULL_RETURN_MEMERR(p); - xmemcpy(p, env->mem_nodes_static, - sizeof(Node*) * SCANENV_MEMNODES_SIZE); + alloc = INIT_SCANENV_MEMNODES_ALLOC_SIZE; + p = (Node** )xmalloc(sizeof(Node*) * alloc); + CHECK_NULL_RETURN_MEMERR(p); + xmemcpy(p, env->mem_nodes_static, + sizeof(Node*) * SCANENV_MEMNODES_SIZE); } else { - alloc = env->mem_alloc * 2; - p = (Node** )xrealloc(env->mem_nodes_dynamic, sizeof(Node*) * alloc); - CHECK_NULL_RETURN_MEMERR(p); + alloc = env->mem_alloc * 2; + p = (Node** )xrealloc(env->mem_nodes_dynamic, sizeof(Node*) * alloc); + CHECK_NULL_RETURN_MEMERR(p); } for (i = env->num_mem + 1; i < alloc; i++) - p[i] = NULL_NODE; + p[i] = NULL_NODE; env->mem_nodes_dynamic = p; env->mem_alloc = alloc; @@ -1210,7 +1210,7 @@ onig_node_free(Node* node) switch (NTYPE(node)) { case NT_STR: if (NSTR(node)->capa != 0 && - IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) { + IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) { xfree(NSTR(node)->s); } break; @@ -1373,9 +1373,9 @@ onig_node_new_anchor(int type) static Node* node_new_backref(int back_num, int* backrefs, int by_name, #ifdef USE_BACKREF_WITH_LEVEL - int exist_level, int nest_level, + int exist_level, int nest_level, #endif - ScanEnv* env) + ScanEnv* env) { int i; Node* node = node_new(); @@ -1398,7 +1398,7 @@ node_new_backref(int back_num, int* backrefs, int by_name, for (i = 0; i < back_num; i++) { if (backrefs[i] <= env->num_mem && - IS_NULL(SCANENV_MEM_NODES(env)[backrefs[i]])) { + IS_NULL(SCANENV_MEM_NODES(env)[backrefs[i]])) { NBREF(node)->state |= NST_RECURSION; /* /...(\1).../ */ break; } @@ -1523,18 +1523,18 @@ onig_node_str_cat(Node* node, const UChar* s, const UChar* end) ptrdiff_t capa = len + addlen + NODE_STR_MARGIN; if (capa <= NSTR(node)->capa) { - onig_strcpy(NSTR(node)->s + len, s, end); + onig_strcpy(NSTR(node)->s + len, s, end); } else { - if (NSTR(node)->s == NSTR(node)->buf) - p = strcat_capa_from_static(NSTR(node)->s, NSTR(node)->end, - s, end, capa); - else - p = strcat_capa(NSTR(node)->s, NSTR(node)->end, s, end, capa); + if (NSTR(node)->s == NSTR(node)->buf) + p = strcat_capa_from_static(NSTR(node)->s, NSTR(node)->end, + s, end, capa); + else + p = strcat_capa(NSTR(node)->s, NSTR(node)->end, s, end, capa); - CHECK_NULL_RETURN_MEMERR(p); - NSTR(node)->s = p; - NSTR(node)->capa = (int )capa; + CHECK_NULL_RETURN_MEMERR(p); + NSTR(node)->s = p; + NSTR(node)->capa = (int )capa; } } else { @@ -1656,7 +1656,7 @@ str_node_split_last_char(StrNode* sn, OnigEncoding enc) if (p && p > sn->s) { /* can be split. */ n = node_new_str(p, sn->end); if (IS_NOT_NULL(n) && (sn->flag & NSTR_RAW) != 0) - NSTRING_SET_RAW(n); + NSTRING_SET_RAW(n); sn->end = (UChar* )p; } } @@ -1704,7 +1704,7 @@ onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc) if (ONIGENC_IS_CODE_DIGIT(enc, c)) { val = (unsigned int )DIGITVAL(c); if ((INT_MAX_LIMIT - val) / 10UL < num) - return -1; /* overflow */ + return -1; /* overflow */ num = num * 10 + val; } @@ -1719,7 +1719,7 @@ onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc) static int scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int minlen, - int maxlen, OnigEncoding enc) + int maxlen, OnigEncoding enc) { OnigCodePoint c; unsigned int num, val; @@ -1734,7 +1734,7 @@ scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int minlen, if (ONIGENC_IS_CODE_XDIGIT(enc, c)) { val = (unsigned int )XDIGITVAL(enc,c); if ((INT_MAX_LIMIT - val) / 16UL < num) - return -1; /* overflow */ + return -1; /* overflow */ num = (num << 4) + XDIGITVAL(enc,c); } @@ -1752,7 +1752,7 @@ scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int minlen, static int scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen, - OnigEncoding enc) + OnigEncoding enc) { OnigCodePoint c; unsigned int num, val; @@ -1765,7 +1765,7 @@ scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen, if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') { val = ODIGITVAL(c); if ((INT_MAX_LIMIT - val) / 8UL < num) - return -1; /* overflow */ + return -1; /* overflow */ num = (num << 3) + val; } @@ -1806,7 +1806,7 @@ new_code_range(BBuf** pbuf) static int add_code_range_to_buf0(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to, - int checkdup) + int checkdup) { int r, inc_n, pos; OnigCodePoint low, high, bound, x; @@ -1857,7 +1857,7 @@ add_code_range_to_buf0(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePo if (inc_n != 1) { if (checkdup && from <= data[low*2+1] - && (data[low*2] <= from || data[low*2+1] <= to)) + && (data[low*2] <= from || data[low*2+1] <= to)) CC_DUP_WARN(env, from, to); if (from > data[low*2]) from = data[low*2]; @@ -1871,8 +1871,8 @@ add_code_range_to_buf0(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePo if (inc_n > 0) { if (high < n) { - int size = (n - high) * 2 * SIZE_CODE_POINT; - BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size); + int size = (n - high) * 2 * SIZE_CODE_POINT; + BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size); } } else { @@ -1982,10 +1982,10 @@ or_code_range_buf(OnigEncoding enc, BBuf* bbuf1, int not1, } else { if (not2 == 0) { - return bbuf_clone(pbuf, bbuf2); + return bbuf_clone(pbuf, bbuf2); } else { - return not_code_range_buf(enc, bbuf2, pbuf, env); + return not_code_range_buf(enc, bbuf2, pbuf, env); } } } @@ -2016,7 +2016,7 @@ or_code_range_buf(OnigEncoding enc, BBuf* bbuf1, int not1, static int and_code_range1(BBuf** pbuf, ScanEnv* env, OnigCodePoint from1, OnigCodePoint to1, - OnigCodePoint* data, int n) + OnigCodePoint* data, int n) { int i, r; OnigCodePoint from2, to2; @@ -2027,19 +2027,19 @@ and_code_range1(BBuf** pbuf, ScanEnv* env, OnigCodePoint from1, OnigCodePoint to if (from2 < from1) { if (to2 < from1) continue; else { - from1 = to2 + 1; + from1 = to2 + 1; } } else if (from2 <= to1) { if (to2 < to1) { - if (from1 <= from2 - 1) { - r = add_code_range_to_buf(pbuf, env, from1, from2-1); - if (r != 0) return r; - } - from1 = to2 + 1; + if (from1 <= from2 - 1) { + r = add_code_range_to_buf(pbuf, env, from1, from2-1); + if (r != 0) return r; + } + from1 = to2 + 1; } else { - to1 = from2 - 1; + to1 = from2 - 1; } } else { @@ -2088,14 +2088,14 @@ and_code_range_buf(BBuf* bbuf1, int not1, BBuf* bbuf2, int not2, BBuf** pbuf, Sc from1 = data1[i*2]; to1 = data1[i*2+1]; for (j = 0; j < n2; j++) { - from2 = data2[j*2]; - to2 = data2[j*2+1]; - if (from2 > to1) break; - if (to2 < from1) continue; - from = MAX(from1, from2); - to = MIN(to1, to2); - r = add_code_range_to_buf(pbuf, env, from, to); - if (r != 0) return r; + from2 = data2[j*2]; + to2 = data2[j*2+1]; + if (from2 > to1) break; + if (to2 < from1) continue; + from = MAX(from1, from2); + to = MIN(to1, to2); + r = add_code_range_to_buf(pbuf, env, from, to); + if (r != 0) return r; } } } @@ -2151,10 +2151,10 @@ and_cclass(CClassNode* dest, CClassNode* cc, ScanEnv* env) else { r = and_code_range_buf(buf1, not1, buf2, not2, &pbuf, env); if (r == 0 && not1 != 0) { - BBuf *tbuf = 0; - r = not_code_range_buf(enc, pbuf, &tbuf, env); - bbuf_free(pbuf); - pbuf = tbuf; + BBuf *tbuf = 0; + r = not_code_range_buf(enc, pbuf, &tbuf, env); + bbuf_free(pbuf); + pbuf = tbuf; } } if (r != 0) { @@ -2209,10 +2209,10 @@ or_cclass(CClassNode* dest, CClassNode* cc, ScanEnv* env) else { r = or_code_range_buf(enc, buf1, not1, buf2, not2, &pbuf, env); if (r == 0 && not1 != 0) { - BBuf *tbuf = 0; - r = not_code_range_buf(enc, pbuf, &tbuf, env); - bbuf_free(pbuf); - pbuf = tbuf; + BBuf *tbuf = 0; + r = not_code_range_buf(enc, pbuf, &tbuf, env); + bbuf_free(pbuf); + pbuf = tbuf; } } if (r != 0) { @@ -2244,12 +2244,12 @@ conv_backslash_value(OnigCodePoint c, ScanEnv* env) case 'e': return '\033'; case 'v': if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_V_VTAB)) - return '\v'; + return '\v'; break; default: if (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z')) - UNKNOWN_ESC_WARN(env, c); + UNKNOWN_ESC_WARN(env, c); break; } } @@ -2508,7 +2508,7 @@ fetch_range_quantifier(UChar** src, UChar* end, OnigToken* tok, ScanEnv* env) if (p == prev) { if (non_low != 0) - goto invalid; + goto invalid; up = REPEAT_INFINITE; /* {n,} : {n,infinite} */ } } @@ -2568,8 +2568,8 @@ fetch_escaped_value(UChar** src, UChar* end, ScanEnv* env, OnigCodePoint* val) if (PEND) return ONIGERR_END_PATTERN_AT_META; PFETCH_S(c); if (c == MC_ESC(env->syntax)) { - v = fetch_escaped_value(&p, end, env, &c); - if (v < 0) return v; + v = fetch_escaped_value(&p, end, env, &c); + if (v < 0) return v; } c = ((c & 0xff) | 0x80); } @@ -2593,14 +2593,14 @@ fetch_escaped_value(UChar** src, UChar* end, ScanEnv* env, OnigCodePoint* val) if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL; PFETCH_S(c); if (c == '?') { - c = 0177; + c = 0177; } else { - if (c == MC_ESC(env->syntax)) { - v = fetch_escaped_value(&p, end, env, &c); - if (v < 0) return v; - } - c &= 0x9f; + if (c == MC_ESC(env->syntax)) { + v = fetch_escaped_value(&p, end, env, &c); + if (v < 0) return v; + } + c &= 0x9f; } break; } @@ -2651,8 +2651,8 @@ get_name_end_code_point(OnigCodePoint start) */ static int fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end, - UChar** rname_end, ScanEnv* env, - int* rback_num, int* rlevel) + UChar** rname_end, ScanEnv* env, + int* rback_num, int* rlevel) { int r, sign, is_num, exist_level; OnigCodePoint end_code; @@ -2703,11 +2703,11 @@ fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end, if (is_num != 0) { if (ONIGENC_IS_CODE_DIGIT(enc, c)) { - is_num = 1; + is_num = 1; } else { - r = ONIGERR_INVALID_GROUP_NAME; - is_num = 0; + r = ONIGERR_INVALID_GROUP_NAME; + is_num = 0; } } else if (!ONIGENC_IS_CODE_NAME(enc, c)) { @@ -2721,8 +2721,8 @@ fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end, int flag = (c == '-' ? -1 : 1); if (PEND) { - r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME; - goto end; + r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME; + goto end; } PFETCH(c); if (! ONIGENC_IS_CODE_DIGIT(enc, c)) goto err; @@ -2733,9 +2733,9 @@ fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end, exist_level = 1; if (!PEND) { - PFETCH(c); - if (c == end_code) - goto end; + PFETCH(c); + if (c == end_code) + goto end; } } @@ -2771,7 +2771,7 @@ fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end, */ static int fetch_name(OnigCodePoint start_code, UChar** src, UChar* end, - UChar** rname_end, ScanEnv* env, int* rback_num, int ref) + UChar** rname_end, ScanEnv* env, int* rback_num, int ref) { int r, is_num, sign; OnigCodePoint end_code; @@ -2800,21 +2800,21 @@ fetch_name(OnigCodePoint start_code, UChar** src, UChar* end, if (ONIGENC_IS_CODE_DIGIT(enc, c)) { if (ref == 1) - is_num = 1; + is_num = 1; else { - r = ONIGERR_INVALID_GROUP_NAME; - is_num = 0; + r = ONIGERR_INVALID_GROUP_NAME; + is_num = 0; } } else if (c == '-') { if (ref == 1) { - is_num = 2; - sign = -1; - pnum_head = p; + is_num = 2; + sign = -1; + pnum_head = p; } else { - r = ONIGERR_INVALID_GROUP_NAME; - is_num = 0; + r = ONIGERR_INVALID_GROUP_NAME; + is_num = 0; } } else if (!ONIGENC_IS_CODE_NAME(enc, c)) { @@ -2827,30 +2827,30 @@ fetch_name(OnigCodePoint start_code, UChar** src, UChar* end, name_end = p; PFETCH_S(c); if (c == end_code || c == ')') { - if (is_num == 2) { - r = ONIGERR_INVALID_GROUP_NAME; - goto teardown; - } - break; + if (is_num == 2) { + r = ONIGERR_INVALID_GROUP_NAME; + goto teardown; + } + break; } if (is_num != 0) { - if (ONIGENC_IS_CODE_DIGIT(enc, c)) { - is_num = 1; - } - else { - if (!ONIGENC_IS_CODE_WORD(enc, c)) - r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME; - else - r = ONIGERR_INVALID_GROUP_NAME; - goto teardown; - } + if (ONIGENC_IS_CODE_DIGIT(enc, c)) { + is_num = 1; + } + else { + if (!ONIGENC_IS_CODE_WORD(enc, c)) + r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME; + else + r = ONIGERR_INVALID_GROUP_NAME; + goto teardown; + } } else { - if (!ONIGENC_IS_CODE_NAME(enc, c)) { - r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME; - goto teardown; - } + if (!ONIGENC_IS_CODE_NAME(enc, c)) { + r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME; + goto teardown; + } } } @@ -2864,8 +2864,8 @@ fetch_name(OnigCodePoint start_code, UChar** src, UChar* end, *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc); if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER; else if (*rback_num == 0) { - r = ONIGERR_INVALID_GROUP_NAME; - goto err; + r = ONIGERR_INVALID_GROUP_NAME; + goto err; } *rback_num *= sign; @@ -2881,7 +2881,7 @@ teardown: name_end = p; PFETCH_S(c); if (c == end_code || c == ')') - break; + break; } if (PEND) name_end = end; @@ -2894,7 +2894,7 @@ teardown: #else static int fetch_name(OnigCodePoint start_code, UChar** src, UChar* end, - UChar** rname_end, ScanEnv* env, int* rback_num, int ref) + UChar** rname_end, ScanEnv* env, int* rback_num, int ref) { int r, is_num, sign; OnigCodePoint end_code; @@ -2982,8 +2982,8 @@ onig_syntax_warn(ScanEnv *env, const char *fmt, ...) UChar buf[WARN_BUFSIZE]; va_start(args, fmt); onig_vsnprintf_with_pattern(buf, WARN_BUFSIZE, env->enc, - env->pattern, env->pattern_end, - fmt, args); + env->pattern, env->pattern_end, + fmt, args); va_end(args); #ifdef RUBY if (env->sourcefile == NULL) @@ -3045,7 +3045,7 @@ UNKNOWN_ESC_WARN(ScanEnv *env, int c) static UChar* find_str_position(OnigCodePoint s[], int n, UChar* from, UChar* to, - UChar **next, OnigEncoding enc) + UChar **next, OnigEncoding enc) { int i; OnigCodePoint x; @@ -3057,14 +3057,14 @@ find_str_position(OnigCodePoint s[], int n, UChar* from, UChar* to, q = p + enclen(enc, p, to); if (x == s[0]) { for (i = 1; i < n && q < to; i++) { - x = ONIGENC_MBC_TO_CODE(enc, q, to); - if (x != s[i]) break; - q += enclen(enc, q, to); + x = ONIGENC_MBC_TO_CODE(enc, q, to); + if (x != s[i]) break; + q += enclen(enc, q, to); } if (i >= n) { - if (IS_NOT_NULL(next)) - *next = q; - return p; + if (IS_NOT_NULL(next)) + *next = q; + return p; } } p = q; @@ -3074,7 +3074,7 @@ find_str_position(OnigCodePoint s[], int n, UChar* from, UChar* to, static int str_exist_check_with_esc(OnigCodePoint s[], int n, UChar* from, UChar* to, - OnigCodePoint bad, OnigEncoding enc, const OnigSyntaxType* syn) + OnigCodePoint bad, OnigEncoding enc, const OnigSyntaxType* syn) { int i, in_esc; OnigCodePoint x; @@ -3091,19 +3091,19 @@ str_exist_check_with_esc(OnigCodePoint s[], int n, UChar* from, UChar* to, x = ONIGENC_MBC_TO_CODE(enc, p, to); q = p + enclen(enc, p, to); if (x == s[0]) { - for (i = 1; i < n && q < to; i++) { - x = ONIGENC_MBC_TO_CODE(enc, q, to); - if (x != s[i]) break; - q += enclen(enc, q, to); - } - if (i >= n) return 1; - p += enclen(enc, p, to); + for (i = 1; i < n && q < to; i++) { + x = ONIGENC_MBC_TO_CODE(enc, q, to); + if (x != s[i]) break; + q += enclen(enc, q, to); + } + if (i >= n) return 1; + p += enclen(enc, p, to); } else { - x = ONIGENC_MBC_TO_CODE(enc, p, to); - if (x == bad) return 0; - else if (x == MC_ESC(syn)) in_esc = 1; - p = q; + x = ONIGENC_MBC_TO_CODE(enc, p, to); + if (x == bad) return 0; + else if (x == MC_ESC(syn)) in_esc = 1; + p = q; } } } @@ -3197,22 +3197,22 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) c2 = PPEEK; if (c2 == '{' && - IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) { - PINC; - tok->type = TK_CHAR_PROPERTY; - tok->u.prop.not = (c == 'P' ? 1 : 0); - - if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) { - PFETCH(c2); - if (c2 == '^') { - tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0); - } - else - PUNFETCH; - } + IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) { + PINC; + tok->type = TK_CHAR_PROPERTY; + tok->u.prop.not = (c == 'P' ? 1 : 0); + + if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) { + PFETCH(c2); + if (c2 == '^') { + tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0); + } + else + PUNFETCH; + } } else { - onig_syntax_warn(env, "invalid Unicode Property \\%c", c); + onig_syntax_warn(env, "invalid Unicode Property \\%c", c); } break; @@ -3221,35 +3221,35 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) { - PINC; - num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc); - if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; - if (!PEND) { - c2 = PPEEK; - if (ONIGENC_IS_CODE_XDIGIT(enc, c2)) - return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; - } + PINC; + num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc); + if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; + if (!PEND) { + c2 = PPEEK; + if (ONIGENC_IS_CODE_XDIGIT(enc, c2)) + return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; + } - if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS('}'))) { - PINC; - tok->type = TK_CODE_POINT; - tok->base = 16; - tok->u.code = (OnigCodePoint )num; - } - else { - /* can't read nothing or invalid format */ - p = prev; - } + if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS('}'))) { + PINC; + tok->type = TK_CODE_POINT; + tok->base = 16; + tok->u.code = (OnigCodePoint )num; + } + else { + /* can't read nothing or invalid format */ + p = prev; + } } else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) { - num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc); - if (num < 0) return ONIGERR_TOO_BIG_NUMBER; - if (p == prev) { /* can't read nothing. */ - num = 0; /* but, it's not error */ - } - tok->type = TK_RAW_BYTE; - tok->base = 16; - tok->u.c = num; + num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc); + if (num < 0) return ONIGERR_TOO_BIG_NUMBER; + if (p == prev) { /* can't read nothing. */ + num = 0; /* but, it's not error */ + } + tok->type = TK_RAW_BYTE; + tok->base = 16; + tok->u.c = num; } break; @@ -3258,15 +3258,15 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) { - num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc); - if (num < -1) return ONIGERR_TOO_SHORT_DIGITS; - else if (num < 0) return ONIGERR_TOO_BIG_NUMBER; - if (p == prev) { /* can't read nothing. */ - num = 0; /* but, it's not error */ - } - tok->type = TK_CODE_POINT; - tok->base = 16; - tok->u.code = (OnigCodePoint )num; + num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc); + if (num < -1) return ONIGERR_TOO_SHORT_DIGITS; + else if (num < 0) return ONIGERR_TOO_BIG_NUMBER; + if (p == prev) { /* can't read nothing. */ + num = 0; /* but, it's not error */ + } + tok->type = TK_CODE_POINT; + tok->base = 16; + tok->u.code = (OnigCodePoint )num; } break; @@ -3275,41 +3275,41 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) { - PINC; - num = scan_unsigned_octal_number(&p, end, 11, enc); - if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; - if (!PEND) { - c2 = PPEEK; - if (ONIGENC_IS_CODE_DIGIT(enc, c2) && c2 < '8') - return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; - } + PINC; + num = scan_unsigned_octal_number(&p, end, 11, enc); + if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; + if (!PEND) { + c2 = PPEEK; + if (ONIGENC_IS_CODE_DIGIT(enc, c2) && c2 < '8') + return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; + } - if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS('}'))) { - PINC; - tok->type = TK_CODE_POINT; - tok->base = 8; - tok->u.code = (OnigCodePoint )num; - } - else { - /* can't read nothing or invalid format */ - p = prev; - } + if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS('}'))) { + PINC; + tok->type = TK_CODE_POINT; + tok->base = 8; + tok->u.code = (OnigCodePoint )num; + } + else { + /* can't read nothing or invalid format */ + p = prev; + } } break; case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) { - PUNFETCH; - prev = p; - num = scan_unsigned_octal_number(&p, end, 3, enc); - if (num < 0 || 0xff < num) return ONIGERR_TOO_BIG_NUMBER; - if (p == prev) { /* can't read nothing. */ - num = 0; /* but, it's not error */ - } - tok->type = TK_RAW_BYTE; - tok->base = 8; - tok->u.c = num; + PUNFETCH; + prev = p; + num = scan_unsigned_octal_number(&p, end, 3, enc); + if (num < 0 || 0xff < num) return ONIGERR_TOO_BIG_NUMBER; + if (p == prev) { /* can't read nothing. */ + num = 0; /* but, it's not error */ + } + tok->type = TK_RAW_BYTE; + tok->base = 8; + tok->u.c = num; } break; @@ -3318,8 +3318,8 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) num = fetch_escaped_value(&p, end, env, &c2); if (num < 0) return num; if ((OnigCodePoint )tok->u.c != c2) { - tok->u.code = (OnigCodePoint )c2; - tok->type = TK_CODE_POINT; + tok->u.code = (OnigCodePoint )c2; + tok->type = TK_CODE_POINT; } break; } @@ -3331,26 +3331,26 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) PINC; if (str_exist_check_with_esc(send, 2, p, end, (OnigCodePoint )']', enc, syn)) { - tok->type = TK_POSIX_BRACKET_OPEN; + tok->type = TK_POSIX_BRACKET_OPEN; } else { - PUNFETCH; - goto cc_in_cc; + PUNFETCH; + goto cc_in_cc; } } else { cc_in_cc: if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP)) { - tok->type = TK_CC_CC_OPEN; + tok->type = TK_CC_CC_OPEN; } else { - CC_ESC_WARN(env, (UChar* )"["); + CC_ESC_WARN(env, (UChar* )"["); } } } else if (c == '&') { if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP) && - !PEND && (PPEEK_IS('&'))) { + !PEND && (PPEEK_IS('&'))) { PINC; tok->type = TK_CC_AND; } @@ -3364,7 +3364,7 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) #ifdef USE_NAMED_GROUP static int fetch_named_backref_token(OnigCodePoint c, OnigToken* tok, UChar** src, - UChar* end, ScanEnv* env) + UChar* end, ScanEnv* env) { int r, num; const OnigSyntaxType* syn = env->syntax; @@ -3379,7 +3379,7 @@ fetch_named_backref_token(OnigCodePoint c, OnigToken* tok, UChar** src, # ifdef USE_BACKREF_WITH_LEVEL name_end = NULL_UCHARP; /* no need. escape gcc warning. */ r = fetch_name_with_level(c, &p, end, &name_end, - env, &back_num, &tok->u.backref.level); + env, &back_num, &tok->u.backref.level); if (r == 1) tok->u.backref.exist_level = 1; else tok->u.backref.exist_level = 0; # else @@ -3391,13 +3391,13 @@ fetch_named_backref_token(OnigCodePoint c, OnigToken* tok, UChar** src, if (back_num < 0) { back_num = BACKREF_REL_TO_ABS(back_num, env); if (back_num <= 0) - return ONIGERR_INVALID_BACKREF; + return ONIGERR_INVALID_BACKREF; } if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) { if (back_num > env->num_mem || - IS_NULL(SCANENV_MEM_NODES(env)[back_num])) - return ONIGERR_INVALID_BACKREF; + IS_NULL(SCANENV_MEM_NODES(env)[back_num])) + return ONIGERR_INVALID_BACKREF; } tok->type = TK_BACKREF; tok->u.backref.by_name = 0; @@ -3408,15 +3408,15 @@ fetch_named_backref_token(OnigCodePoint c, OnigToken* tok, UChar** src, num = onig_name_to_group_numbers(env->reg, prev, name_end, &backs); if (num <= 0) { onig_scan_env_set_error_string(env, - ONIGERR_UNDEFINED_NAME_REFERENCE, prev, name_end); + ONIGERR_UNDEFINED_NAME_REFERENCE, prev, name_end); return ONIGERR_UNDEFINED_NAME_REFERENCE; } if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) { int i; for (i = 0; i < num; i++) { - if (backs[i] > env->num_mem || - IS_NULL(SCANENV_MEM_NODES(env)[backs[i]])) - return ONIGERR_INVALID_BACKREF; + if (backs[i] > env->num_mem || + IS_NULL(SCANENV_MEM_NODES(env)[backs[i]])) + return ONIGERR_INVALID_BACKREF; } } @@ -3490,26 +3490,26 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) tok->u.repeat.upper = 1; greedy_check: if (!PEND && PPEEK_IS('?') && - IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_NON_GREEDY)) { - PFETCH(c); - tok->u.repeat.greedy = 0; - tok->u.repeat.possessive = 0; + IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_NON_GREEDY)) { + PFETCH(c); + tok->u.repeat.greedy = 0; + tok->u.repeat.possessive = 0; } else { possessive_check: - if (!PEND && PPEEK_IS('+') && - ((IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT) && - tok->type != TK_INTERVAL) || - (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL) && - tok->type == TK_INTERVAL))) { - PFETCH(c); - tok->u.repeat.greedy = 1; - tok->u.repeat.possessive = 1; - } - else { - tok->u.repeat.greedy = 1; - tok->u.repeat.possessive = 0; - } + if (!PEND && PPEEK_IS('+') && + ((IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT) && + tok->type != TK_INTERVAL) || + (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL) && + tok->type == TK_INTERVAL))) { + PFETCH(c); + tok->u.repeat.greedy = 1; + tok->u.repeat.possessive = 1; + } + else { + tok->u.repeat.greedy = 1; + tok->u.repeat.possessive = 0; + } } break; @@ -3519,10 +3519,10 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) if (r < 0) return r; /* error */ if (r == 0) goto greedy_check; else if (r == 2) { /* {n} */ - if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY)) - goto possessive_check; + if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY)) + goto possessive_check; - goto greedy_check; + goto greedy_check; } /* r == 1 : normal char */ break; @@ -3561,7 +3561,7 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) tok->type = TK_ANCHOR; tok->u.anchor.subtype = ANCHOR_WORD_BOUND; tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option) - && ! IS_WORD_BOUND_ALL_RANGE(env->option); + && ! IS_WORD_BOUND_ALL_RANGE(env->option); break; case 'B': @@ -3569,7 +3569,7 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) tok->type = TK_ANCHOR; tok->u.anchor.subtype = ANCHOR_NOT_WORD_BOUND; tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option) - && ! IS_WORD_BOUND_ALL_RANGE(env->option); + && ! IS_WORD_BOUND_ALL_RANGE(env->option); break; #ifdef USE_WORD_BEGIN_END @@ -3671,33 +3671,33 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) { - PINC; - num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc); - if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; - if (!PEND) { - if (ONIGENC_IS_CODE_XDIGIT(enc, PPEEK)) - return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; - } + PINC; + num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc); + if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; + if (!PEND) { + if (ONIGENC_IS_CODE_XDIGIT(enc, PPEEK)) + return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; + } - if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS('}')) { - PINC; - tok->type = TK_CODE_POINT; - tok->u.code = (OnigCodePoint )num; - } - else { - /* can't read nothing or invalid format */ - p = prev; - } + if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS('}')) { + PINC; + tok->type = TK_CODE_POINT; + tok->u.code = (OnigCodePoint )num; + } + else { + /* can't read nothing or invalid format */ + p = prev; + } } else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) { - num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc); - if (num < 0) return ONIGERR_TOO_BIG_NUMBER; - if (p == prev) { /* can't read nothing. */ - num = 0; /* but, it's not error */ - } - tok->type = TK_RAW_BYTE; - tok->base = 16; - tok->u.c = num; + num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc); + if (num < 0) return ONIGERR_TOO_BIG_NUMBER; + if (p == prev) { /* can't read nothing. */ + num = 0; /* but, it's not error */ + } + tok->type = TK_RAW_BYTE; + tok->base = 16; + tok->u.c = num; } break; @@ -3706,15 +3706,15 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) { - num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc); - if (num < -1) return ONIGERR_TOO_SHORT_DIGITS; - else if (num < 0) return ONIGERR_TOO_BIG_NUMBER; - if (p == prev) { /* can't read nothing. */ - num = 0; /* but, it's not error */ - } - tok->type = TK_CODE_POINT; - tok->base = 16; - tok->u.code = (OnigCodePoint )num; + num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc); + if (num < -1) return ONIGERR_TOO_SHORT_DIGITS; + else if (num < 0) return ONIGERR_TOO_BIG_NUMBER; + if (p == prev) { /* can't read nothing. */ + num = 0; /* but, it's not error */ + } + tok->type = TK_CODE_POINT; + tok->base = 16; + tok->u.code = (OnigCodePoint )num; } break; @@ -3723,24 +3723,24 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) { - PINC; - num = scan_unsigned_octal_number(&p, end, 11, enc); - if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; - if (!PEND) { - OnigCodePoint c = PPEEK; - if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') - return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; - } + PINC; + num = scan_unsigned_octal_number(&p, end, 11, enc); + if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE; + if (!PEND) { + OnigCodePoint c = PPEEK; + if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') + return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE; + } - if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS('}')) { - PINC; - tok->type = TK_CODE_POINT; - tok->u.code = (OnigCodePoint )num; - } - else { - /* can't read nothing or invalid format */ - p = prev; - } + if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS('}')) { + PINC; + tok->type = TK_CODE_POINT; + tok->u.code = (OnigCodePoint )num; + } + else { + /* can't read nothing or invalid format */ + p = prev; + } } break; @@ -3750,64 +3750,64 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) prev = p; num = onig_scan_unsigned_number(&p, end, enc); if (num < 0 || num > ONIG_MAX_BACKREF_NUM) { - goto skip_backref; + goto skip_backref; } if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_DECIMAL_BACKREF) && - (num <= env->num_mem || num <= 9)) { /* This spec. from GNU regex */ - if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) { - if (num > env->num_mem || IS_NULL(SCANENV_MEM_NODES(env)[num])) - return ONIGERR_INVALID_BACKREF; - } + (num <= env->num_mem || num <= 9)) { /* This spec. from GNU regex */ + if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) { + if (num > env->num_mem || IS_NULL(SCANENV_MEM_NODES(env)[num])) + return ONIGERR_INVALID_BACKREF; + } - tok->type = TK_BACKREF; - tok->u.backref.num = 1; - tok->u.backref.ref1 = num; - tok->u.backref.by_name = 0; + tok->type = TK_BACKREF; + tok->u.backref.num = 1; + tok->u.backref.ref1 = num; + tok->u.backref.by_name = 0; #ifdef USE_BACKREF_WITH_LEVEL - tok->u.backref.exist_level = 0; + tok->u.backref.exist_level = 0; #endif - break; + break; } skip_backref: if (c == '8' || c == '9') { - /* normal char */ - p = prev; PINC; - break; + /* normal char */ + p = prev; PINC; + break; } p = prev; /* fall through */ case '0': if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) { - prev = p; - num = scan_unsigned_octal_number(&p, end, (c == '0' ? 2:3), enc); - if (num < 0 || 0xff < num) return ONIGERR_TOO_BIG_NUMBER; - if (p == prev) { /* can't read nothing. */ - num = 0; /* but, it's not error */ - } - tok->type = TK_RAW_BYTE; - tok->base = 8; - tok->u.c = num; + prev = p; + num = scan_unsigned_octal_number(&p, end, (c == '0' ? 2:3), enc); + if (num < 0 || 0xff < num) return ONIGERR_TOO_BIG_NUMBER; + if (p == prev) { /* can't read nothing. */ + num = 0; /* but, it's not error */ + } + tok->type = TK_RAW_BYTE; + tok->base = 8; + tok->u.c = num; } else if (c != '0') { - PINC; + PINC; } break; #ifdef USE_NAMED_GROUP case 'k': if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_K_NAMED_BACKREF)) { - PFETCH(c); - if (c == '<' || c == '\'') { - r = fetch_named_backref_token(c, tok, &p, end, env); - if (r < 0) return r; - } - else { - PUNFETCH; - onig_syntax_warn(env, "invalid back reference"); - } + PFETCH(c); + if (c == '<' || c == '\'') { + r = fetch_named_backref_token(c, tok, &p, end, env); + if (r < 0) return r; + } + else { + PUNFETCH; + onig_syntax_warn(env, "invalid back reference"); + } } break; #endif @@ -3816,52 +3816,52 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) case 'g': # ifdef USE_NAMED_GROUP if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_BRACE_BACKREF)) { - PFETCH(c); - if (c == '{') { - r = fetch_named_backref_token(c, tok, &p, end, env); - if (r < 0) return r; - } - else - PUNFETCH; + PFETCH(c); + if (c == '{') { + r = fetch_named_backref_token(c, tok, &p, end, env); + if (r < 0) return r; + } + else + PUNFETCH; } # endif # ifdef USE_SUBEXP_CALL if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_SUBEXP_CALL)) { - PFETCH(c); - if (c == '<' || c == '\'') { - int gnum = -1, rel = 0; - UChar* name_end; - OnigCodePoint cnext; - - cnext = PPEEK; - if (cnext == '0') { - PINC; - if (PPEEK_IS(get_name_end_code_point(c))) { /* \g<0>, \g'0' */ - PINC; - name_end = p; - gnum = 0; - } - } - else if (cnext == '+') { - PINC; - rel = 1; - } - prev = p; - if (gnum < 0) { - r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &gnum, 1); - if (r < 0) return r; - } + PFETCH(c); + if (c == '<' || c == '\'') { + int gnum = -1, rel = 0; + UChar* name_end; + OnigCodePoint cnext; + + cnext = PPEEK; + if (cnext == '0') { + PINC; + if (PPEEK_IS(get_name_end_code_point(c))) { /* \g<0>, \g'0' */ + PINC; + name_end = p; + gnum = 0; + } + } + else if (cnext == '+') { + PINC; + rel = 1; + } + prev = p; + if (gnum < 0) { + r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &gnum, 1); + if (r < 0) return r; + } - tok->type = TK_CALL; - tok->u.call.name = prev; - tok->u.call.name_end = name_end; - tok->u.call.gnum = gnum; - tok->u.call.rel = rel; - } - else { - onig_syntax_warn(env, "invalid subexp call"); - PUNFETCH; - } + tok->type = TK_CALL; + tok->u.call.name = prev; + tok->u.call.name_end = name_end; + tok->u.call.gnum = gnum; + tok->u.call.rel = rel; + } + else { + onig_syntax_warn(env, "invalid subexp call"); + PUNFETCH; + } } # endif break; @@ -3869,65 +3869,65 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) case 'Q': if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)) { - tok->type = TK_QUOTE_OPEN; + tok->type = TK_QUOTE_OPEN; } break; case 'p': case 'P': if (PPEEK_IS('{') && - IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) { - PINC; - tok->type = TK_CHAR_PROPERTY; - tok->u.prop.not = (c == 'P' ? 1 : 0); - - if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) { - PFETCH(c); - if (c == '^') { - tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0); - } - else - PUNFETCH; - } + IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) { + PINC; + tok->type = TK_CHAR_PROPERTY; + tok->u.prop.not = (c == 'P' ? 1 : 0); + + if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) { + PFETCH(c); + if (c == '^') { + tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0); + } + else + PUNFETCH; + } } else { - onig_syntax_warn(env, "invalid Unicode Property \\%c", c); + onig_syntax_warn(env, "invalid Unicode Property \\%c", c); } break; case 'R': if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK)) { - tok->type = TK_LINEBREAK; + tok->type = TK_LINEBREAK; } break; case 'X': if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER)) { - tok->type = TK_EXTENDED_GRAPHEME_CLUSTER; + tok->type = TK_EXTENDED_GRAPHEME_CLUSTER; } break; case 'K': if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP)) { - tok->type = TK_KEEP; + tok->type = TK_KEEP; } break; default: { - OnigCodePoint c2; - - PUNFETCH; - num = fetch_escaped_value(&p, end, env, &c2); - if (num < 0) return num; - /* set_raw: */ - if ((OnigCodePoint )tok->u.c != c2) { - tok->type = TK_CODE_POINT; - tok->u.code = (OnigCodePoint )c2; - } - else { /* string */ - p = tok->backp + enclen(enc, tok->backp, end); - } + OnigCodePoint c2; + + PUNFETCH; + num = fetch_escaped_value(&p, end, env, &c2); + if (num < 0) return num; + /* set_raw: */ + if ((OnigCodePoint )tok->u.c != c2) { + tok->type = TK_CODE_POINT; + tok->u.code = (OnigCodePoint )c2; + } + else { /* string */ + p = tok->backp + enclen(enc, tok->backp, end); + } } break; } @@ -3938,18 +3938,18 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) #ifdef USE_VARIABLE_META_CHARS if ((c != ONIG_INEFFECTIVE_META_CHAR) && - IS_SYNTAX_OP(syn, ONIG_SYN_OP_VARIABLE_META_CHARACTERS)) { + IS_SYNTAX_OP(syn, ONIG_SYN_OP_VARIABLE_META_CHARACTERS)) { if (c == MC_ANYCHAR(syn)) - goto any_char; + goto any_char; else if (c == MC_ANYTIME(syn)) - goto anytime; + goto anytime; else if (c == MC_ZERO_OR_ONE_TIME(syn)) - goto zero_or_one_time; + goto zero_or_one_time; else if (c == MC_ONE_OR_MORE_TIME(syn)) - goto one_or_more_time; + goto one_or_more_time; else if (c == MC_ANYCHAR_ANYTIME(syn)) { - tok->type = TK_ANYCHAR_ANYTIME; - goto out; + tok->type = TK_ANYCHAR_ANYTIME; + goto out; } } #endif @@ -4002,10 +4002,10 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) if (r < 0) return r; /* error */ if (r == 0) goto greedy_check; else if (r == 2) { /* {n} */ - if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY)) - goto possessive_check; + if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY)) + goto possessive_check; - goto greedy_check; + goto greedy_check; } /* r == 1 : normal char */ break; @@ -4017,114 +4017,114 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) case '(': if (PPEEK_IS('?') && - IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) { - PINC; - if (PPEEK_IS('#')) { - PFETCH(c); - while (1) { - if (PEND) return ONIGERR_END_PATTERN_IN_GROUP; - PFETCH(c); - if (c == MC_ESC(syn)) { - if (!PEND) PFETCH(c); - } - else { - if (c == ')') break; - } - } - goto start; - } + IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) { + PINC; + if (PPEEK_IS('#')) { + PFETCH(c); + while (1) { + if (PEND) return ONIGERR_END_PATTERN_IN_GROUP; + PFETCH(c); + if (c == MC_ESC(syn)) { + if (!PEND) PFETCH(c); + } + else { + if (c == ')') break; + } + } + goto start; + } #ifdef USE_PERL_SUBEXP_CALL - /* (?&name), (?n), (?R), (?0), (?+n), (?-n) */ - c = PPEEK; - if ((c == '&' || c == 'R' || ONIGENC_IS_CODE_DIGIT(enc, c)) && - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) { - /* (?&name), (?n), (?R), (?0) */ - int gnum; - UChar *name; - UChar *name_end; - - if (c == 'R' || c == '0') { - PINC; /* skip 'R' / '0' */ - if (!PPEEK_IS(')')) return ONIGERR_INVALID_GROUP_NAME; - PINC; /* skip ')' */ - name_end = name = p; - gnum = 0; - } - else { - int numref = 1; - if (c == '&') { /* (?&name) */ - PINC; - numref = 0; /* don't allow number name */ - } - name = p; - r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, numref); - if (r < 0) return r; - } + /* (?&name), (?n), (?R), (?0), (?+n), (?-n) */ + c = PPEEK; + if ((c == '&' || c == 'R' || ONIGENC_IS_CODE_DIGIT(enc, c)) && + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) { + /* (?&name), (?n), (?R), (?0) */ + int gnum; + UChar *name; + UChar *name_end; + + if (c == 'R' || c == '0') { + PINC; /* skip 'R' / '0' */ + if (!PPEEK_IS(')')) return ONIGERR_INVALID_GROUP_NAME; + PINC; /* skip ')' */ + name_end = name = p; + gnum = 0; + } + else { + int numref = 1; + if (c == '&') { /* (?&name) */ + PINC; + numref = 0; /* don't allow number name */ + } + name = p; + r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, numref); + if (r < 0) return r; + } - tok->type = TK_CALL; - tok->u.call.name = name; - tok->u.call.name_end = name_end; - tok->u.call.gnum = gnum; - tok->u.call.rel = 0; - break; - } - else if ((c == '-' || c == '+') && - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) { - /* (?+n), (?-n) */ - int gnum; - UChar *name; - UChar *name_end; - OnigCodePoint cnext; - PFETCH_READY; - - PINC; /* skip '-' / '+' */ - cnext = PPEEK; - if (ONIGENC_IS_CODE_DIGIT(enc, cnext)) { - if (c == '-') PUNFETCH; - name = p; - r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, 1); - if (r < 0) return r; - - tok->type = TK_CALL; - tok->u.call.name = name; - tok->u.call.name_end = name_end; - tok->u.call.gnum = gnum; - tok->u.call.rel = 1; - break; - } - } + tok->type = TK_CALL; + tok->u.call.name = name; + tok->u.call.name_end = name_end; + tok->u.call.gnum = gnum; + tok->u.call.rel = 0; + break; + } + else if ((c == '-' || c == '+') && + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) { + /* (?+n), (?-n) */ + int gnum; + UChar *name; + UChar *name_end; + OnigCodePoint cnext; + PFETCH_READY; + + PINC; /* skip '-' / '+' */ + cnext = PPEEK; + if (ONIGENC_IS_CODE_DIGIT(enc, cnext)) { + if (c == '-') PUNFETCH; + name = p; + r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, 1); + if (r < 0) return r; + + tok->type = TK_CALL; + tok->u.call.name = name; + tok->u.call.name_end = name_end; + tok->u.call.gnum = gnum; + tok->u.call.rel = 1; + break; + } + } #endif /* USE_PERL_SUBEXP_CALL */ #ifdef USE_CAPITAL_P_NAMED_GROUP - if (PPEEK_IS('P') && - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) { - int gnum; - UChar *name; - UChar *name_end; - PFETCH_READY; - - PINC; /* skip 'P' */ - if (PEND) return ONIGERR_UNDEFINED_GROUP_OPTION; - PFETCH(c); - if (c == '=') { /* (?P=name): backref */ - r = fetch_named_backref_token((OnigCodePoint )'(', tok, &p, end, env); - if (r < 0) return r; - break; - } - else if (c == '>') { /* (?P>name): subexp call */ - name = p; - r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, 0); - if (r < 0) return r; - - tok->type = TK_CALL; - tok->u.call.name = name; - tok->u.call.name_end = name_end; - tok->u.call.gnum = gnum; - tok->u.call.rel = 0; - break; - } - } + if (PPEEK_IS('P') && + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) { + int gnum; + UChar *name; + UChar *name_end; + PFETCH_READY; + + PINC; /* skip 'P' */ + if (PEND) return ONIGERR_UNDEFINED_GROUP_OPTION; + PFETCH(c); + if (c == '=') { /* (?P=name): backref */ + r = fetch_named_backref_token((OnigCodePoint )'(', tok, &p, end, env); + if (r < 0) return r; + break; + } + else if (c == '>') { /* (?P>name): subexp call */ + name = p; + r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum, 0); + if (r < 0) return r; + + tok->type = TK_CALL; + tok->u.call.name = name; + tok->u.call.name_end = name_end; + tok->u.call.gnum = gnum; + tok->u.call.rel = 0; + break; + } + } #endif /* USE_CAPITAL_P_NAMED_GROUP */ - PUNFETCH; + PUNFETCH; } if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP)) break; @@ -4140,14 +4140,14 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break; tok->type = TK_ANCHOR; tok->u.anchor.subtype = (IS_SINGLELINE(env->option) - ? ANCHOR_BEGIN_BUF : ANCHOR_BEGIN_LINE); + ? ANCHOR_BEGIN_BUF : ANCHOR_BEGIN_LINE); break; case '$': if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR)) break; tok->type = TK_ANCHOR; tok->u.anchor.subtype = (IS_SINGLELINE(env->option) - ? ANCHOR_SEMI_END_BUF : ANCHOR_END_LINE); + ? ANCHOR_SEMI_END_BUF : ANCHOR_END_LINE); break; case '[': @@ -4157,24 +4157,24 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env) case ']': if (*src > env->pattern) /* /].../ is allowed. */ - CLOSE_BRACKET_WITHOUT_ESC_WARN(env, (UChar* )"]"); + CLOSE_BRACKET_WITHOUT_ESC_WARN(env, (UChar* )"]"); break; case '#': if (IS_EXTEND(env->option)) { - while (!PEND) { - PFETCH(c); - if (ONIGENC_IS_CODE_NEWLINE(enc, c)) - break; - } - goto start; - break; + while (!PEND) { + PFETCH(c); + if (ONIGENC_IS_CODE_NEWLINE(enc, c)) + break; + } + goto start; + break; } break; case ' ': case '\t': case '\n': case '\r': case '\f': if (IS_EXTEND(env->option)) - goto start; + goto start; break; default: @@ -4203,18 +4203,18 @@ add_ctype_to_cc_by_range(CClassNode* cc, int ctype ARG_UNUSED, int not, if (not == 0) { for (i = 0; i < n; i++) { for (j = ONIGENC_CODE_RANGE_FROM(mbr, i); - j <= ONIGENC_CODE_RANGE_TO(mbr, i); j++) { - if (j >= sb_out) { - if (j > ONIGENC_CODE_RANGE_FROM(mbr, i)) { - r = add_code_range_to_buf(&(cc->mbuf), env, j, - ONIGENC_CODE_RANGE_TO(mbr, i)); - if (r != 0) return r; - i++; - } + j <= ONIGENC_CODE_RANGE_TO(mbr, i); j++) { + if (j >= sb_out) { + if (j > ONIGENC_CODE_RANGE_FROM(mbr, i)) { + r = add_code_range_to_buf(&(cc->mbuf), env, j, + ONIGENC_CODE_RANGE_TO(mbr, i)); + if (r != 0) return r; + i++; + } - goto sb_end; - } - BITSET_SET_BIT_CHKDUP(cc->bs, j); + goto sb_end; + } + BITSET_SET_BIT_CHKDUP(cc->bs, j); } } @@ -4231,11 +4231,11 @@ add_ctype_to_cc_by_range(CClassNode* cc, int ctype ARG_UNUSED, int not, for (i = 0; i < n; i++) { for (j = prev; - j < ONIGENC_CODE_RANGE_FROM(mbr, i); j++) { - if (j >= sb_out) { - goto sb_end2; - } - BITSET_SET_BIT_CHKDUP(cc->bs, j); + j < ONIGENC_CODE_RANGE_FROM(mbr, i); j++) { + if (j >= sb_out) { + goto sb_end2; + } + BITSET_SET_BIT_CHKDUP(cc->bs, j); } prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1; } @@ -4248,9 +4248,9 @@ add_ctype_to_cc_by_range(CClassNode* cc, int ctype ARG_UNUSED, int not, for (i = 0; i < n; i++) { if (prev < ONIGENC_CODE_RANGE_FROM(mbr, i)) { - r = add_code_range_to_buf(&(cc->mbuf), env, prev, + r = add_code_range_to_buf(&(cc->mbuf), env, prev, ONIGENC_CODE_RANGE_FROM(mbr, i) - 1); - if (r != 0) return r; + if (r != 0) return r; } prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1; } @@ -4278,30 +4278,30 @@ add_ctype_to_cc(CClassNode* cc, int ctype, int not, int ascii_range, ScanEnv* en CClassNode ccwork; initialize_cclass(&ccwork); r = add_ctype_to_cc_by_range(&ccwork, ctype, not, env, sb_out, - ranges); + ranges); if (r == 0) { - if (not) { - r = add_code_range_to_buf0(&(ccwork.mbuf), env, 0x80, ONIG_LAST_CODE_POINT, FALSE); - } - else { - CClassNode ccascii; - initialize_cclass(&ccascii); - if (ONIGENC_MBC_MINLEN(env->enc) > 1) { - r = add_code_range(&(ccascii.mbuf), env, 0x00, 0x7F); - } - else { - bitset_set_range(env, ccascii.bs, 0x00, 0x7F); - r = 0; - } - if (r == 0) { - r = and_cclass(&ccwork, &ccascii, env); - } - if (IS_NOT_NULL(ccascii.mbuf)) bbuf_free(ccascii.mbuf); - } - if (r == 0) { - r = or_cclass(cc, &ccwork, env); - } - if (IS_NOT_NULL(ccwork.mbuf)) bbuf_free(ccwork.mbuf); + if (not) { + r = add_code_range_to_buf0(&(ccwork.mbuf), env, 0x80, ONIG_LAST_CODE_POINT, FALSE); + } + else { + CClassNode ccascii; + initialize_cclass(&ccascii); + if (ONIGENC_MBC_MINLEN(env->enc) > 1) { + r = add_code_range(&(ccascii.mbuf), env, 0x00, 0x7F); + } + else { + bitset_set_range(env, ccascii.bs, 0x00, 0x7F); + r = 0; + } + if (r == 0) { + r = and_cclass(&ccwork, &ccascii, env); + } + if (IS_NOT_NULL(ccascii.mbuf)) bbuf_free(ccascii.mbuf); + } + if (r == 0) { + r = or_cclass(cc, &ccwork, env); + } + if (IS_NOT_NULL(ccwork.mbuf)) bbuf_free(ccwork.mbuf); } } else { @@ -4329,15 +4329,15 @@ add_ctype_to_cc(CClassNode* cc, int ctype, int not, int ascii_range, ScanEnv* en case ONIGENC_CTYPE_ALNUM: if (not != 0) { for (c = 0; c < SINGLE_BYTE_SIZE; c++) { - if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)) - BITSET_SET_BIT_CHKDUP(cc->bs, c); + if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)) + BITSET_SET_BIT_CHKDUP(cc->bs, c); } ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); } else { for (c = 0; c < SINGLE_BYTE_SIZE; c++) { - if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)) - BITSET_SET_BIT_CHKDUP(cc->bs, c); + if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)) + BITSET_SET_BIT_CHKDUP(cc->bs, c); } } break; @@ -4346,39 +4346,39 @@ add_ctype_to_cc(CClassNode* cc, int ctype, int not, int ascii_range, ScanEnv* en case ONIGENC_CTYPE_PRINT: if (not != 0) { for (c = 0; c < SINGLE_BYTE_SIZE; c++) { - if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype) - || c >= maxcode) - BITSET_SET_BIT_CHKDUP(cc->bs, c); + if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype) + || c >= maxcode) + BITSET_SET_BIT_CHKDUP(cc->bs, c); } if (ascii_range) - ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); + ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); } else { for (c = 0; c < maxcode; c++) { - if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)) - BITSET_SET_BIT_CHKDUP(cc->bs, c); + if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)) + BITSET_SET_BIT_CHKDUP(cc->bs, c); } if (! ascii_range) - ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); + ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); } break; case ONIGENC_CTYPE_WORD: if (not == 0) { for (c = 0; c < maxcode; c++) { - if (ONIGENC_IS_CODE_WORD(enc, c)) BITSET_SET_BIT_CHKDUP(cc->bs, c); + if (ONIGENC_IS_CODE_WORD(enc, c)) BITSET_SET_BIT_CHKDUP(cc->bs, c); } if (! ascii_range) - ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); + ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); } else { for (c = 0; c < SINGLE_BYTE_SIZE; c++) { - if ((ONIGENC_CODE_TO_MBCLEN(enc, c) > 0) /* check invalid code point */ - && (! ONIGENC_IS_CODE_WORD(enc, c) || c >= maxcode)) - BITSET_SET_BIT_CHKDUP(cc->bs, c); + if ((ONIGENC_CODE_TO_MBCLEN(enc, c) > 0) /* check invalid code point */ + && (! ONIGENC_IS_CODE_WORD(enc, c) || c >= maxcode)) + BITSET_SET_BIT_CHKDUP(cc->bs, c); } if (ascii_range) - ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); + ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf); } break; @@ -4392,7 +4392,7 @@ add_ctype_to_cc(CClassNode* cc, int ctype, int not, int ascii_range, ScanEnv* en static int parse_posix_bracket(CClassNode* cc, CClassNode* asc_cc, - UChar** src, UChar* end, ScanEnv* env) + UChar** src, UChar* end, ScanEnv* env) { #define POSIX_BRACKET_CHECK_LIMIT_LENGTH 20 #define POSIX_BRACKET_NAME_MIN_LEN 4 @@ -4432,22 +4432,22 @@ parse_posix_bracket(CClassNode* cc, CClassNode* asc_cc, goto not_posix_bracket; ascii_range = IS_ASCII_RANGE(env->option) && - ! IS_POSIX_BRACKET_ALL_RANGE(env->option); + ! IS_POSIX_BRACKET_ALL_RANGE(env->option); for (pb = PBS; pb < PBS + numberof(PBS); pb++) { if (onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0) { p = (UChar* )onigenc_step(enc, p, end, pb->len); if (onigenc_with_ascii_strncmp(enc, p, end, (UChar* )":]", 2) != 0) - return ONIGERR_INVALID_POSIX_BRACKET_TYPE; + return ONIGERR_INVALID_POSIX_BRACKET_TYPE; r = add_ctype_to_cc(cc, pb->ctype, not, ascii_range, env); if (r != 0) return r; if (IS_NOT_NULL(asc_cc)) { - if (pb->ctype != ONIGENC_CTYPE_WORD && - pb->ctype != ONIGENC_CTYPE_ASCII && - !ascii_range) - r = add_ctype_to_cc(asc_cc, pb->ctype, not, ascii_range, env); - if (r != 0) return r; + if (pb->ctype != ONIGENC_CTYPE_WORD && + pb->ctype != ONIGENC_CTYPE_ASCII && + !ascii_range) + r = add_ctype_to_cc(asc_cc, pb->ctype, not, ascii_range, env); + if (r != 0) return r; } PINC_S; PINC_S; @@ -4468,7 +4468,7 @@ parse_posix_bracket(CClassNode* cc, CClassNode* asc_cc, if (! PEND) { PFETCH_S(c); if (c == ']') - return ONIGERR_INVALID_POSIX_BRACKET_TYPE; + return ONIGERR_INVALID_POSIX_BRACKET_TYPE; } } @@ -4509,7 +4509,7 @@ static int cclass_case_fold(Node** np, CClassNode* cc, CClassNode* asc_cc, ScanE static int parse_char_property(Node** np, OnigToken* tok, UChar** src, UChar* end, - ScanEnv* env) + ScanEnv* env) { int r, ctype; CClassNode* cc; @@ -4547,8 +4547,8 @@ enum CCVALTYPE { static int next_state_class(CClassNode* cc, CClassNode* asc_cc, - OnigCodePoint* vs, enum CCVALTYPE* type, - enum CCSTATE* state, ScanEnv* env) + OnigCodePoint* vs, enum CCVALTYPE* type, + enum CCSTATE* state, ScanEnv* env) { int r; @@ -4559,14 +4559,14 @@ next_state_class(CClassNode* cc, CClassNode* asc_cc, if (*type == CCV_SB) { BITSET_SET_BIT_CHKDUP(cc->bs, (int )(*vs)); if (IS_NOT_NULL(asc_cc)) - BITSET_SET_BIT(asc_cc->bs, (int )(*vs)); + BITSET_SET_BIT(asc_cc->bs, (int )(*vs)); } else if (*type == CCV_CODE_POINT) { r = add_code_range(&(cc->mbuf), env, *vs, *vs); if (r < 0) return r; if (IS_NOT_NULL(asc_cc)) { - r = add_code_range0(&(asc_cc->mbuf), env, *vs, *vs, 0); - if (r < 0) return r; + r = add_code_range0(&(asc_cc->mbuf), env, *vs, *vs, 0); + if (r < 0) return r; } } } @@ -4578,10 +4578,10 @@ next_state_class(CClassNode* cc, CClassNode* asc_cc, static int next_state_val(CClassNode* cc, CClassNode* asc_cc, - OnigCodePoint *from, OnigCodePoint to, - int* from_israw, int to_israw, - enum CCVALTYPE intype, enum CCVALTYPE* type, - enum CCSTATE* state, ScanEnv* env) + OnigCodePoint *from, OnigCodePoint to, + int* from_israw, int to_israw, + enum CCVALTYPE intype, enum CCVALTYPE* type, + enum CCSTATE* state, ScanEnv* env) { int r; @@ -4590,14 +4590,14 @@ next_state_val(CClassNode* cc, CClassNode* asc_cc, if (*type == CCV_SB) { BITSET_SET_BIT_CHKDUP(cc->bs, (int )(*from)); if (IS_NOT_NULL(asc_cc)) - BITSET_SET_BIT(asc_cc->bs, (int )(*from)); + BITSET_SET_BIT(asc_cc->bs, (int )(*from)); } else if (*type == CCV_CODE_POINT) { r = add_code_range(&(cc->mbuf), env, *from, *from); if (r < 0) return r; if (IS_NOT_NULL(asc_cc)) { - r = add_code_range0(&(asc_cc->mbuf), env, *from, *from, 0); - if (r < 0) return r; + r = add_code_range0(&(asc_cc->mbuf), env, *from, *from, 0); + if (r < 0) return r; } } break; @@ -4605,42 +4605,42 @@ next_state_val(CClassNode* cc, CClassNode* asc_cc, case CCS_RANGE: if (intype == *type) { if (intype == CCV_SB) { - if (*from > 0xff || to > 0xff) - return ONIGERR_INVALID_CODE_POINT_VALUE; - - if (*from > to) { - if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC)) - goto ccs_range_end; - else - return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS; - } - bitset_set_range(env, cc->bs, (int )*from, (int )to); - if (IS_NOT_NULL(asc_cc)) - bitset_set_range(env, asc_cc->bs, (int )*from, (int )to); + if (*from > 0xff || to > 0xff) + return ONIGERR_INVALID_CODE_POINT_VALUE; + + if (*from > to) { + if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC)) + goto ccs_range_end; + else + return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS; + } + bitset_set_range(env, cc->bs, (int )*from, (int )to); + if (IS_NOT_NULL(asc_cc)) + bitset_set_range(env, asc_cc->bs, (int )*from, (int )to); } else { - r = add_code_range(&(cc->mbuf), env, *from, to); - if (r < 0) return r; - if (IS_NOT_NULL(asc_cc)) { - r = add_code_range0(&(asc_cc->mbuf), env, *from, to, 0); - if (r < 0) return r; - } + r = add_code_range(&(cc->mbuf), env, *from, to); + if (r < 0) return r; + if (IS_NOT_NULL(asc_cc)) { + r = add_code_range0(&(asc_cc->mbuf), env, *from, to, 0); + if (r < 0) return r; + } } } else { if (*from > to) { - if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC)) - goto ccs_range_end; - else - return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS; + if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC)) + goto ccs_range_end; + else + return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS; } bitset_set_range(env, cc->bs, (int )*from, (int )(to < 0xff ? to : 0xff)); r = add_code_range(&(cc->mbuf), env, (OnigCodePoint )*from, to); if (r < 0) return r; if (IS_NOT_NULL(asc_cc)) { - bitset_set_range(env, asc_cc->bs, (int )*from, (int )(to < 0xff ? to : 0xff)); - r = add_code_range0(&(asc_cc->mbuf), env, (OnigCodePoint )*from, to, 0); - if (r < 0) return r; + bitset_set_range(env, asc_cc->bs, (int )*from, (int )(to < 0xff ? to : 0xff)); + r = add_code_range0(&(asc_cc->mbuf), env, (OnigCodePoint )*from, to, 0); + if (r < 0) return r; } } ccs_range_end: @@ -4664,7 +4664,7 @@ next_state_val(CClassNode* cc, CClassNode* asc_cc, static int code_exist_check(OnigCodePoint c, UChar* from, UChar* end, int ignore_escaped, - ScanEnv* env) + ScanEnv* env) { int in_esc; OnigCodePoint code; @@ -4687,7 +4687,7 @@ code_exist_check(OnigCodePoint c, UChar* from, UChar* end, int ignore_escaped, static int parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* end, - ScanEnv* env) + ScanEnv* env) { int r, neg, len, fetched, and_start; OnigCodePoint v, vs; @@ -4748,16 +4748,16 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e switch (r) { case TK_CHAR: if ((tok->u.code >= SINGLE_BYTE_SIZE) || - (len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c)) > 1) { - in_type = CCV_CODE_POINT; + (len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c)) > 1) { + in_type = CCV_CODE_POINT; } else if (len < 0) { - r = len; - goto err; + r = len; + goto err; } else { sb_char: - in_type = CCV_SB; + in_type = CCV_SB; } v = (OnigCodePoint )tok->u.c; in_israw = 0; @@ -4767,54 +4767,54 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e case TK_RAW_BYTE: /* tok->base != 0 : octal or hexadec. */ if (! ONIGENC_IS_SINGLEBYTE(env->enc) && tok->base != 0) { - UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN]; - UChar* bufe = buf + ONIGENC_CODE_TO_MBC_MAXLEN; - UChar* psave = p; - int i, base = tok->base; - - buf[0] = (UChar )tok->u.c; - for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) { - r = fetch_token_in_cc(tok, &p, end, env); - if (r < 0) goto err; - if (r != TK_RAW_BYTE || tok->base != base) { - fetched = 1; - break; - } - buf[i] = (UChar )tok->u.c; - } + UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN]; + UChar* bufe = buf + ONIGENC_CODE_TO_MBC_MAXLEN; + UChar* psave = p; + int i, base = tok->base; + + buf[0] = (UChar )tok->u.c; + for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) { + r = fetch_token_in_cc(tok, &p, end, env); + if (r < 0) goto err; + if (r != TK_RAW_BYTE || tok->base != base) { + fetched = 1; + break; + } + buf[i] = (UChar )tok->u.c; + } - if (i < ONIGENC_MBC_MINLEN(env->enc)) { - r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING; - goto err; - } + if (i < ONIGENC_MBC_MINLEN(env->enc)) { + r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING; + goto err; + } - len = enclen(env->enc, buf, buf + i); - if (i < len) { - r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING; - goto err; - } - else if (i > len) { /* fetch back */ - p = psave; - for (i = 1; i < len; i++) { - (void)fetch_token_in_cc(tok, &p, end, env); - /* no need to check the return value (already checked above) */ - } - fetched = 0; - } + len = enclen(env->enc, buf, buf + i); + if (i < len) { + r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING; + goto err; + } + else if (i > len) { /* fetch back */ + p = psave; + for (i = 1; i < len; i++) { + (void)fetch_token_in_cc(tok, &p, end, env); + /* no need to check the return value (already checked above) */ + } + fetched = 0; + } - if (i == 1) { - v = (OnigCodePoint )buf[0]; - goto raw_single; - } - else { - v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe); - in_type = CCV_CODE_POINT; - } + if (i == 1) { + v = (OnigCodePoint )buf[0]; + goto raw_single; + } + else { + v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe); + in_type = CCV_CODE_POINT; + } } else { - v = (OnigCodePoint )tok->u.c; + v = (OnigCodePoint )tok->u.c; raw_single: - in_type = CCV_SB; + in_type = CCV_SB; } in_israw = 1; goto val_entry2; @@ -4826,13 +4826,13 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e val_entry: len = ONIGENC_CODE_TO_MBCLEN(env->enc, v); if (len < 0) { - r = len; - goto err; + r = len; + goto err; } in_type = (len == 1 ? CCV_SB : CCV_CODE_POINT); val_entry2: r = next_state_val(cc, asc_cc, &vs, v, &val_israw, in_israw, in_type, &val_type, - &state, env); + &state, env); if (r != 0) goto err; break; @@ -4840,24 +4840,24 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e r = parse_posix_bracket(cc, asc_cc, &p, end, env); if (r < 0) goto err; if (r == 1) { /* is not POSIX bracket */ - CC_ESC_WARN(env, (UChar* )"["); - p = tok->backp; - v = (OnigCodePoint )tok->u.c; - in_israw = 0; - goto val_entry; + CC_ESC_WARN(env, (UChar* )"["); + p = tok->backp; + v = (OnigCodePoint )tok->u.c; + in_israw = 0; + goto val_entry; } goto next_class; break; case TK_CHAR_TYPE: r = add_ctype_to_cc(cc, tok->u.prop.ctype, tok->u.prop.not, - IS_ASCII_RANGE(env->option), env); + IS_ASCII_RANGE(env->option), env); if (r != 0) return r; if (IS_NOT_NULL(asc_cc)) { - if (tok->u.prop.ctype != ONIGENC_CTYPE_WORD) - r = add_ctype_to_cc(asc_cc, tok->u.prop.ctype, tok->u.prop.not, - IS_ASCII_RANGE(env->option), env); - if (r != 0) return r; + if (tok->u.prop.ctype != ONIGENC_CTYPE_WORD) + r = add_ctype_to_cc(asc_cc, tok->u.prop.ctype, tok->u.prop.not, + IS_ASCII_RANGE(env->option), env); + if (r != 0) return r; } next_class: @@ -4867,133 +4867,133 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e case TK_CHAR_PROPERTY: { - int ctype; - - ctype = fetch_char_property_to_ctype(&p, end, env); - if (ctype < 0) return ctype; - r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, 0, env); - if (r != 0) return r; - if (IS_NOT_NULL(asc_cc)) { - if (ctype != ONIGENC_CTYPE_ASCII) - r = add_ctype_to_cc(asc_cc, ctype, tok->u.prop.not, 0, env); - if (r != 0) return r; - } - goto next_class; + int ctype; + + ctype = fetch_char_property_to_ctype(&p, end, env); + if (ctype < 0) return ctype; + r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, 0, env); + if (r != 0) return r; + if (IS_NOT_NULL(asc_cc)) { + if (ctype != ONIGENC_CTYPE_ASCII) + r = add_ctype_to_cc(asc_cc, ctype, tok->u.prop.not, 0, env); + if (r != 0) return r; + } + goto next_class; } break; case TK_CC_RANGE: if (state == CCS_VALUE) { - r = fetch_token_in_cc(tok, &p, end, env); - if (r < 0) goto err; - fetched = 1; - if (r == TK_CC_CLOSE) { /* allow [x-] */ - range_end_val: - v = (OnigCodePoint )'-'; - in_israw = 0; - goto val_entry; - } - else if (r == TK_CC_AND) { - CC_ESC_WARN(env, (UChar* )"-"); - goto range_end_val; - } + r = fetch_token_in_cc(tok, &p, end, env); + if (r < 0) goto err; + fetched = 1; + if (r == TK_CC_CLOSE) { /* allow [x-] */ + range_end_val: + v = (OnigCodePoint )'-'; + in_israw = 0; + goto val_entry; + } + else if (r == TK_CC_AND) { + CC_ESC_WARN(env, (UChar* )"-"); + goto range_end_val; + } - if (val_type == CCV_CLASS) { - r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS; - goto err; - } + if (val_type == CCV_CLASS) { + r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS; + goto err; + } - state = CCS_RANGE; + state = CCS_RANGE; } else if (state == CCS_START) { - /* [-xa] is allowed */ - v = (OnigCodePoint )tok->u.c; - in_israw = 0; + /* [-xa] is allowed */ + v = (OnigCodePoint )tok->u.c; + in_israw = 0; - r = fetch_token_in_cc(tok, &p, end, env); - if (r < 0) goto err; - fetched = 1; - /* [--x] or [a&&-x] is warned. */ - if (r == TK_CC_RANGE || and_start != 0) - CC_ESC_WARN(env, (UChar* )"-"); + r = fetch_token_in_cc(tok, &p, end, env); + if (r < 0) goto err; + fetched = 1; + /* [--x] or [a&&-x] is warned. */ + if (r == TK_CC_RANGE || and_start != 0) + CC_ESC_WARN(env, (UChar* )"-"); - goto val_entry; + goto val_entry; } else if (state == CCS_RANGE) { - CC_ESC_WARN(env, (UChar* )"-"); - goto sb_char; /* [!--x] is allowed */ + CC_ESC_WARN(env, (UChar* )"-"); + goto sb_char; /* [!--x] is allowed */ } else { /* CCS_COMPLETE */ - r = fetch_token_in_cc(tok, &p, end, env); - if (r < 0) goto err; - fetched = 1; - if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */ - else if (r == TK_CC_AND) { - CC_ESC_WARN(env, (UChar* )"-"); - goto range_end_val; - } + r = fetch_token_in_cc(tok, &p, end, env); + if (r < 0) goto err; + fetched = 1; + if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */ + else if (r == TK_CC_AND) { + CC_ESC_WARN(env, (UChar* )"-"); + goto range_end_val; + } - if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) { - CC_ESC_WARN(env, (UChar* )"-"); - goto range_end_val; /* [0-9-a] is allowed as [0-9\-a] */ - } - r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS; - goto err; + if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) { + CC_ESC_WARN(env, (UChar* )"-"); + goto range_end_val; /* [0-9-a] is allowed as [0-9\-a] */ + } + r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS; + goto err; } break; case TK_CC_CC_OPEN: /* [ */ { - Node *anode, *aasc_node; - CClassNode* acc; + Node *anode, *aasc_node; + CClassNode* acc; - r = parse_char_class(&anode, &aasc_node, tok, &p, end, env); - if (r == 0) { - acc = NCCLASS(anode); - r = or_cclass(cc, acc, env); - } - if (r == 0 && IS_NOT_NULL(aasc_node)) { - acc = NCCLASS(aasc_node); - r = or_cclass(asc_cc, acc, env); - } - onig_node_free(anode); - onig_node_free(aasc_node); - if (r != 0) goto err; + r = parse_char_class(&anode, &aasc_node, tok, &p, end, env); + if (r == 0) { + acc = NCCLASS(anode); + r = or_cclass(cc, acc, env); + } + if (r == 0 && IS_NOT_NULL(aasc_node)) { + acc = NCCLASS(aasc_node); + r = or_cclass(asc_cc, acc, env); + } + onig_node_free(anode); + onig_node_free(aasc_node); + if (r != 0) goto err; } break; case TK_CC_AND: /* && */ { - if (state == CCS_VALUE) { - r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type, - &val_type, &state, env); - if (r != 0) goto err; - } - /* initialize local variables */ - and_start = 1; - state = CCS_START; - - if (IS_NOT_NULL(prev_cc)) { - r = and_cclass(prev_cc, cc, env); - if (r != 0) goto err; - bbuf_free(cc->mbuf); - if (IS_NOT_NULL(asc_cc)) { - r = and_cclass(asc_prev_cc, asc_cc, env); - if (r != 0) goto err; - bbuf_free(asc_cc->mbuf); - } - } - else { - prev_cc = cc; - cc = &work_cc; - if (IS_NOT_NULL(asc_cc)) { - asc_prev_cc = asc_cc; - asc_cc = &asc_work_cc; - } - } - initialize_cclass(cc); - if (IS_NOT_NULL(asc_cc)) - initialize_cclass(asc_cc); + if (state == CCS_VALUE) { + r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type, + &val_type, &state, env); + if (r != 0) goto err; + } + /* initialize local variables */ + and_start = 1; + state = CCS_START; + + if (IS_NOT_NULL(prev_cc)) { + r = and_cclass(prev_cc, cc, env); + if (r != 0) goto err; + bbuf_free(cc->mbuf); + if (IS_NOT_NULL(asc_cc)) { + r = and_cclass(asc_prev_cc, asc_cc, env); + if (r != 0) goto err; + bbuf_free(asc_cc->mbuf); + } + } + else { + prev_cc = cc; + cc = &work_cc; + if (IS_NOT_NULL(asc_cc)) { + asc_prev_cc = asc_cc; + asc_cc = &asc_work_cc; + } + } + initialize_cclass(cc); + if (IS_NOT_NULL(asc_cc)) + initialize_cclass(asc_cc); } break; @@ -5017,7 +5017,7 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e if (state == CCS_VALUE) { r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type, - &val_type, &state, env); + &val_type, &state, env); if (r != 0) goto err; } @@ -5056,12 +5056,12 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e #define NEWLINE_CODE 0x0a if (ONIGENC_IS_CODE_NEWLINE(env->enc, NEWLINE_CODE)) { - if (ONIGENC_CODE_TO_MBCLEN(env->enc, NEWLINE_CODE) == 1) - BITSET_SET_BIT_CHKDUP(cc->bs, NEWLINE_CODE); - else { - r = add_code_range(&(cc->mbuf), env, NEWLINE_CODE, NEWLINE_CODE); - if (r < 0) goto err; - } + if (ONIGENC_CODE_TO_MBCLEN(env->enc, NEWLINE_CODE) == 1) + BITSET_SET_BIT_CHKDUP(cc->bs, NEWLINE_CODE); + else { + r = add_code_range(&(cc->mbuf), env, NEWLINE_CODE, NEWLINE_CODE); + if (r < 0) goto err; + } } } } @@ -5078,11 +5078,11 @@ parse_char_class(Node** np, Node** asc_np, OnigToken* tok, UChar** src, UChar* e } static int parse_subexp(Node** top, OnigToken* tok, int term, - UChar** src, UChar* end, ScanEnv* env); + UChar** src, UChar* end, ScanEnv* env); static int parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end, - ScanEnv* env) + ScanEnv* env) { int r = 0, num; Node *target, *work1 = NULL, *work2 = NULL; @@ -5129,28 +5129,28 @@ parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end, break; case '~': /* (?~...) absent operator */ if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_TILDE_ABSENT)) { - *np = node_new_enclose(ENCLOSE_ABSENT); + *np = node_new_enclose(ENCLOSE_ABSENT); } else { - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; } break; #ifdef USE_NAMED_GROUP case '\'': if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) { - goto named_group1; + goto named_group1; } else - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; break; # ifdef USE_CAPITAL_P_NAMED_GROUP case 'P': /* (?P<name>...) */ if (!PEND && - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) { - PFETCH(c); - if (c == '<') goto named_group1; + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) { + PFETCH(c); + if (c == '<') goto named_group1; } return ONIGERR_UNDEFINED_GROUP_OPTION; break; @@ -5161,49 +5161,49 @@ parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end, if (PEND) return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS; PFETCH(c); if (c == '=') - *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND); + *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND); else if (c == '!') - *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND_NOT); + *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND_NOT); #ifdef USE_NAMED_GROUP else { /* (?<name>...) */ - if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) { - UChar *name; - UChar *name_end; + if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) { + UChar *name; + UChar *name_end; - PUNFETCH; - c = '<'; + PUNFETCH; + c = '<'; - named_group1: - list_capture = 0; + named_group1: + list_capture = 0; # ifdef USE_CAPTURE_HISTORY - named_group2: + named_group2: # endif - name = p; - r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num, 0); - if (r < 0) return r; - - num = scan_env_add_mem_entry(env); - if (num < 0) return num; - if (list_capture != 0 && num >= (int )BIT_STATUS_BITS_NUM) - return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY; - - r = name_add(env->reg, name, name_end, num, env); - if (r != 0) return r; - *np = node_new_enclose_memory(env->option, 1); - CHECK_NULL_RETURN_MEMERR(*np); - NENCLOSE(*np)->regnum = num; - if (list_capture != 0) - BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num); - env->num_named++; - } - else { - return ONIGERR_UNDEFINED_GROUP_OPTION; - } + name = p; + r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num, 0); + if (r < 0) return r; + + num = scan_env_add_mem_entry(env); + if (num < 0) return num; + if (list_capture != 0 && num >= (int )BIT_STATUS_BITS_NUM) + return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY; + + r = name_add(env->reg, name, name_end, num, env); + if (r != 0) return r; + *np = node_new_enclose_memory(env->option, 1); + CHECK_NULL_RETURN_MEMERR(*np); + NENCLOSE(*np)->regnum = num; + if (list_capture != 0) + BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num); + env->num_named++; + } + else { + return ONIGERR_UNDEFINED_GROUP_OPTION; + } } #else else { - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; } #endif break; @@ -5212,122 +5212,122 @@ parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end, case '@': if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY)) { # ifdef USE_NAMED_GROUP - if (!PEND && - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) { - PFETCH(c); - if (c == '<' || c == '\'') { - list_capture = 1; - goto named_group2; /* (?@<name>...) */ - } - PUNFETCH; - } + if (!PEND && + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) { + PFETCH(c); + if (c == '<' || c == '\'') { + list_capture = 1; + goto named_group2; /* (?@<name>...) */ + } + PUNFETCH; + } # endif - *np = node_new_enclose_memory(env->option, 0); - CHECK_NULL_RETURN_MEMERR(*np); - num = scan_env_add_mem_entry(env); - if (num < 0) return num; - if (num >= (int )BIT_STATUS_BITS_NUM) - return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY; + *np = node_new_enclose_memory(env->option, 0); + CHECK_NULL_RETURN_MEMERR(*np); + num = scan_env_add_mem_entry(env); + if (num < 0) return num; + if (num >= (int )BIT_STATUS_BITS_NUM) + return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY; - NENCLOSE(*np)->regnum = num; - BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num); + NENCLOSE(*np)->regnum = num; + BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num); } else { - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; } break; #endif /* USE_CAPTURE_HISTORY */ case '(': /* conditional expression: (?(cond)yes), (?(cond)yes|no) */ if (!PEND && - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LPAREN_CONDITION)) { - UChar *name = NULL; - UChar *name_end; - PFETCH(c); - if (ONIGENC_IS_CODE_DIGIT(enc, c)) { /* (n) */ - PUNFETCH; - r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &num, 1); - if (r < 0) return r; + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LPAREN_CONDITION)) { + UChar *name = NULL; + UChar *name_end; + PFETCH(c); + if (ONIGENC_IS_CODE_DIGIT(enc, c)) { /* (n) */ + PUNFETCH; + r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &num, 1); + if (r < 0) return r; #if 0 - /* Relative number is not currently supported. (same as Perl) */ - if (num < 0) { - num = BACKREF_REL_TO_ABS(num, env); - if (num <= 0) - return ONIGERR_INVALID_BACKREF; - } + /* Relative number is not currently supported. (same as Perl) */ + if (num < 0) { + num = BACKREF_REL_TO_ABS(num, env); + if (num <= 0) + return ONIGERR_INVALID_BACKREF; + } #endif - if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_STRICT_CHECK_BACKREF)) { - if (num > env->num_mem || - IS_NULL(SCANENV_MEM_NODES(env)[num])) - return ONIGERR_INVALID_BACKREF; - } - } + if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_STRICT_CHECK_BACKREF)) { + if (num > env->num_mem || + IS_NULL(SCANENV_MEM_NODES(env)[num])) + return ONIGERR_INVALID_BACKREF; + } + } #ifdef USE_NAMED_GROUP - else if (c == '<' || c == '\'') { /* (<name>), ('name') */ - name = p; - r = fetch_named_backref_token(c, tok, &p, end, env); - if (r < 0) return r; - if (!PPEEK_IS(')')) return ONIGERR_UNDEFINED_GROUP_OPTION; - PINC; - - if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP)) { - num = tok->u.backref.ref1; - } - else { - /* FIXME: - * Use left most named group for now. This is the same as Perl. - * However this should use the same strategy as normal back- - * references on Ruby syntax; search right to left. */ - int len = tok->u.backref.num; - num = len > 1 ? tok->u.backref.refs[0] : tok->u.backref.ref1; - } - } + else if (c == '<' || c == '\'') { /* (<name>), ('name') */ + name = p; + r = fetch_named_backref_token(c, tok, &p, end, env); + if (r < 0) return r; + if (!PPEEK_IS(')')) return ONIGERR_UNDEFINED_GROUP_OPTION; + PINC; + + if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP)) { + num = tok->u.backref.ref1; + } + else { + /* FIXME: + * Use left most named group for now. This is the same as Perl. + * However this should use the same strategy as normal back- + * references on Ruby syntax; search right to left. */ + int len = tok->u.backref.num; + num = len > 1 ? tok->u.backref.refs[0] : tok->u.backref.ref1; + } + } #endif - else - return ONIGERR_INVALID_CONDITION_PATTERN; - *np = node_new_enclose(ENCLOSE_CONDITION); - CHECK_NULL_RETURN_MEMERR(*np); - NENCLOSE(*np)->regnum = num; - if (IS_NOT_NULL(name)) NENCLOSE(*np)->state |= NST_NAME_REF; + else + return ONIGERR_INVALID_CONDITION_PATTERN; + *np = node_new_enclose(ENCLOSE_CONDITION); + CHECK_NULL_RETURN_MEMERR(*np); + NENCLOSE(*np)->regnum = num; + if (IS_NOT_NULL(name)) NENCLOSE(*np)->state |= NST_NAME_REF; } else - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; break; #if 0 case '|': /* branch reset: (?|...) */ if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_VBAR_BRANCH_RESET)) { - /* TODO */ + /* TODO */ } else - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; break; #endif case '^': /* loads default options */ if (!PEND && IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) { - /* d-imsx */ - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); - ONOFF(option, ONIG_OPTION_IGNORECASE, 1); - ONOFF(option, ONIG_OPTION_SINGLELINE, 0); - ONOFF(option, ONIG_OPTION_MULTILINE, 1); - ONOFF(option, ONIG_OPTION_EXTEND, 1); - PFETCH(c); + /* d-imsx */ + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); + ONOFF(option, ONIG_OPTION_IGNORECASE, 1); + ONOFF(option, ONIG_OPTION_SINGLELINE, 0); + ONOFF(option, ONIG_OPTION_MULTILINE, 1); + ONOFF(option, ONIG_OPTION_EXTEND, 1); + PFETCH(c); } #if 0 else if (!PEND && IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) { - /* d-imx */ - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0); - ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0); - ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0); - ONOFF(option, ONIG_OPTION_IGNORECASE, 1); - ONOFF(option, ONIG_OPTION_MULTILINE, 1); - ONOFF(option, ONIG_OPTION_EXTEND, 1); - PFETCH(c); + /* d-imx */ + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0); + ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0); + ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0); + ONOFF(option, ONIG_OPTION_IGNORECASE, 1); + ONOFF(option, ONIG_OPTION_MULTILINE, 1); + ONOFF(option, ONIG_OPTION_EXTEND, 1); + PFETCH(c); } #endif else { - return ONIGERR_UNDEFINED_GROUP_OPTION; + return ONIGERR_UNDEFINED_GROUP_OPTION; } /* fall through */ #ifdef USE_POSIXLINE_OPTION @@ -5336,120 +5336,120 @@ parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end, case '-': case 'i': case 'm': case 's': case 'x': case 'a': case 'd': case 'l': case 'u': { - int neg = 0; - - while (1) { - switch (c) { - case ':': - case ')': - break; - - case '-': neg = 1; break; - case 'x': ONOFF(option, ONIG_OPTION_EXTEND, neg); break; - case 'i': ONOFF(option, ONIG_OPTION_IGNORECASE, neg); break; - case 's': - if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) { - ONOFF(option, ONIG_OPTION_MULTILINE, neg); - } - else - return ONIGERR_UNDEFINED_GROUP_OPTION; - break; - - case 'm': - if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) { - ONOFF(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0)); - } - else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) { - ONOFF(option, ONIG_OPTION_MULTILINE, neg); - } - else - return ONIGERR_UNDEFINED_GROUP_OPTION; - break; + int neg = 0; + + while (1) { + switch (c) { + case ':': + case ')': + break; + + case '-': neg = 1; break; + case 'x': ONOFF(option, ONIG_OPTION_EXTEND, neg); break; + case 'i': ONOFF(option, ONIG_OPTION_IGNORECASE, neg); break; + case 's': + if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) { + ONOFF(option, ONIG_OPTION_MULTILINE, neg); + } + else + return ONIGERR_UNDEFINED_GROUP_OPTION; + break; + + case 'm': + if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) { + ONOFF(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0)); + } + else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) { + ONOFF(option, ONIG_OPTION_MULTILINE, neg); + } + else + return ONIGERR_UNDEFINED_GROUP_OPTION; + break; #ifdef USE_POSIXLINE_OPTION - case 'p': - ONOFF(option, ONIG_OPTION_MULTILINE|ONIG_OPTION_SINGLELINE, neg); - break; + case 'p': + ONOFF(option, ONIG_OPTION_MULTILINE|ONIG_OPTION_SINGLELINE, neg); + break; #endif - case 'a': /* limits \d, \s, \w and POSIX brackets to ASCII range */ - if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) || - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) && - (neg == 0)) { - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0); - ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1); - ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1); - } - else - return ONIGERR_UNDEFINED_GROUP_OPTION; - break; - - case 'u': - if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) || - IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) && - (neg == 0)) { - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); - ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1); - ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1); - } - else - return ONIGERR_UNDEFINED_GROUP_OPTION; - break; - - case 'd': - if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) && - (neg == 0)) { - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); - } - else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY) && - (neg == 0)) { - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0); - ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0); - ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0); - } - else - return ONIGERR_UNDEFINED_GROUP_OPTION; - break; - - case 'l': - if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) && (neg == 0)) { - ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); - } - else - return ONIGERR_UNDEFINED_GROUP_OPTION; - break; - - default: - return ONIGERR_UNDEFINED_GROUP_OPTION; - } + case 'a': /* limits \d, \s, \w and POSIX brackets to ASCII range */ + if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) || + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) && + (neg == 0)) { + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0); + ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1); + ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1); + } + else + return ONIGERR_UNDEFINED_GROUP_OPTION; + break; + + case 'u': + if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) || + IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) && + (neg == 0)) { + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); + ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1); + ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1); + } + else + return ONIGERR_UNDEFINED_GROUP_OPTION; + break; + + case 'd': + if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) && + (neg == 0)) { + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); + } + else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY) && + (neg == 0)) { + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0); + ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0); + ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0); + } + else + return ONIGERR_UNDEFINED_GROUP_OPTION; + break; - if (c == ')') { - *np = node_new_option(option); - CHECK_NULL_RETURN_MEMERR(*np); - *src = p; - return 2; /* option only */ - } - else if (c == ':') { - OnigOptionType prev = env->option; - - env->option = option; - r = fetch_token(tok, &p, end, env); - if (r < 0) { - env->option = prev; - return r; - } - r = parse_subexp(&target, tok, term, &p, end, env); - env->option = prev; - if (r < 0) return r; - *np = node_new_option(option); - CHECK_NULL_RETURN_MEMERR(*np); - NENCLOSE(*np)->target = target; - *src = p; - return 0; - } + case 'l': + if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) && (neg == 0)) { + ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1); + } + else + return ONIGERR_UNDEFINED_GROUP_OPTION; + break; - if (PEND) return ONIGERR_END_PATTERN_IN_GROUP; - PFETCH(c); - } + default: + return ONIGERR_UNDEFINED_GROUP_OPTION; + } + + if (c == ')') { + *np = node_new_option(option); + CHECK_NULL_RETURN_MEMERR(*np); + *src = p; + return 2; /* option only */ + } + else if (c == ':') { + OnigOptionType prev = env->option; + + env->option = option; + r = fetch_token(tok, &p, end, env); + if (r < 0) { + env->option = prev; + return r; + } + r = parse_subexp(&target, tok, term, &p, end, env); + env->option = prev; + if (r < 0) return r; + *np = node_new_option(option); + CHECK_NULL_RETURN_MEMERR(*np); + NENCLOSE(*np)->target = target; + *src = p; + return 0; + } + + if (PEND) return ONIGERR_END_PATTERN_IN_GROUP; + PFETCH(c); + } } break; @@ -5488,14 +5488,14 @@ parse_enclose(Node** np, OnigToken* tok, int term, UChar** src, UChar* end, } else if (NENCLOSE(*np)->type == ENCLOSE_CONDITION) { if (NTYPE(target) != NT_ALT) { - /* convert (?(cond)yes) to (?(cond)yes|empty) */ - work1 = node_new_empty(); - if (IS_NULL(work1)) goto err; - work2 = onig_node_new_alt(work1, NULL_NODE); - if (IS_NULL(work2)) goto err; - work1 = onig_node_new_alt(target, work2); - if (IS_NULL(work1)) goto err; - NENCLOSE(*np)->target = work1; + /* convert (?(cond)yes) to (?(cond)yes|empty) */ + work1 = node_new_empty(); + if (IS_NULL(work1)) goto err; + work2 = onig_node_new_alt(work1, NULL_NODE); + if (IS_NULL(work2)) goto err; + work1 = onig_node_new_alt(target, work2); + if (IS_NULL(work1)) goto err; + NENCLOSE(*np)->target = work1; } } } @@ -5534,11 +5534,11 @@ set_quantifier(Node* qnode, Node* target, int group, ScanEnv* env) if (! group) { StrNode* sn = NSTR(target); if (str_node_can_be_split(sn, env->enc)) { - Node* n = str_node_split_last_char(sn, env->enc); - if (IS_NOT_NULL(n)) { - qn->target = n; - return 2; - } + Node* n = str_node_split_last_char(sn, env->enc); + if (IS_NOT_NULL(n)) { + qn->target = n; + return 2; + } } } break; @@ -5552,43 +5552,43 @@ set_quantifier(Node* qnode, Node* target, int group, ScanEnv* env) #ifdef USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR if (nestq_num >= 0 && targetq_num >= 0 && - IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) { - switch (ReduceTypeTable[targetq_num][nestq_num]) { - case RQ_ASIS: - break; - - case RQ_DEL: - if (onig_warn != onig_null_warn) { - onig_syntax_warn(env, "regular expression has redundant nested repeat operator '%s'", - PopularQStr[targetq_num]); - } - goto warn_exit; - break; - - default: - if (onig_warn != onig_null_warn) { - onig_syntax_warn(env, "nested repeat operator '%s' and '%s' was replaced with '%s' in regular expression", - PopularQStr[targetq_num], PopularQStr[nestq_num], - ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]); - } - goto warn_exit; - break; - } + IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) { + switch (ReduceTypeTable[targetq_num][nestq_num]) { + case RQ_ASIS: + break; + + case RQ_DEL: + if (onig_warn != onig_null_warn) { + onig_syntax_warn(env, "regular expression has redundant nested repeat operator '%s'", + PopularQStr[targetq_num]); + } + goto warn_exit; + break; + + default: + if (onig_warn != onig_null_warn) { + onig_syntax_warn(env, "nested repeat operator '%s' and '%s' was replaced with '%s' in regular expression", + PopularQStr[targetq_num], PopularQStr[nestq_num], + ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]); + } + goto warn_exit; + break; + } } warn_exit: #endif if (targetq_num >= 0) { - if (nestq_num >= 0) { - onig_reduce_nested_quantifier(qnode, target); - goto q_exit; - } - else if (targetq_num == 1 || targetq_num == 2) { /* * or + */ - /* (?:a*){n,m}, (?:a+){n,m} => (?:a*){n,n}, (?:a+){n,n} */ - if (! IS_REPEAT_INFINITE(qn->upper) && qn->upper > 1 && qn->greedy) { - qn->upper = (qn->lower == 0 ? 1 : qn->lower); - } - } + if (nestq_num >= 0) { + onig_reduce_nested_quantifier(qnode, target); + goto q_exit; + } + else if (targetq_num == 1 || targetq_num == 2) { /* * or + */ + /* (?:a*){n,m}, (?:a+){n,m} => (?:a*){n,n}, (?:a+){n,n} */ + if (! IS_REPEAT_INFINITE(qn->upper) && qn->upper > 1 && qn->greedy) { + qn->upper = (qn->lower == 0 ? 1 : qn->lower); + } + } } } break; @@ -5638,7 +5638,7 @@ typedef struct { static int i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[], - int to_len, void* arg) + int to_len, void* arg) { IApplyCaseFoldArg* iarg; ScanEnv* env; @@ -5669,7 +5669,7 @@ i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[], int is_in = onig_is_code_in_cc(env->enc, from, cc); #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS if ((is_in != 0 && !IS_NCCLASS_NOT(cc)) || - (is_in == 0 && IS_NCCLASS_NOT(cc))) { + (is_in == 0 && IS_NCCLASS_NOT(cc))) { if (add_flag) { if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) { r = add_code_range0(&(cc->mbuf), env, *to, *to, 0); @@ -5707,26 +5707,26 @@ i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[], if (onig_is_code_in_cc(env->enc, from, cc) #ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS - && !IS_NCCLASS_NOT(cc) + && !IS_NCCLASS_NOT(cc) #endif - ) { + ) { for (i = 0; i < to_len; i++) { - len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf); - if (i == 0) { - snode = onig_node_new_str(buf, buf + len); - CHECK_NULL_RETURN_MEMERR(snode); - - /* char-class expanded multi-char only - compare with string folded at match time. */ - NSTRING_SET_AMBIG(snode); - } - else { - r = onig_node_str_cat(snode, buf, buf + len); - if (r < 0) { - onig_node_free(snode); - return r; - } - } + len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf); + if (i == 0) { + snode = onig_node_new_str(buf, buf + len); + CHECK_NULL_RETURN_MEMERR(snode); + + /* char-class expanded multi-char only + compare with string folded at match time. */ + NSTRING_SET_AMBIG(snode); + } + else { + r = onig_node_str_cat(snode, buf, buf + len); + if (r < 0) { + onig_node_free(snode); + return r; + } + } } *(iarg->ptail) = onig_node_new_alt(snode, NULL_NODE); @@ -5751,7 +5751,7 @@ cclass_case_fold(Node** np, CClassNode* cc, CClassNode* asc_cc, ScanEnv* env) iarg.ptail = &(iarg.alt_root); r = ONIGENC_APPLY_ALL_CASE_FOLD(env->enc, env->case_fold_flag, - i_apply_case_fold, &iarg); + i_apply_case_fold, &iarg); if (r != 0) { onig_node_free(iarg.alt_root); return r; @@ -6214,8 +6214,8 @@ is_onechar_cclass(CClassNode* cc, OnigCodePoint* code) /* only one char found in the bbuf, save the code point. */ c = data[0]; if (((c < SINGLE_BYTE_SIZE) && BITSET_AT(cc->bs, c))) { - /* skip if c is included in the bitset */ - c = not_found; + /* skip if c is included in the bitset */ + c = not_found; } } else { @@ -6228,9 +6228,10 @@ is_onechar_cclass(CClassNode* cc, OnigCodePoint* code) Bits b1 = cc->bs[i]; if (b1 != 0) { if (((b1 & (b1 - 1)) == 0) && (c == not_found)) { - c = BITS_IN_ROOM * i + countbits(b1 - 1); - } else { - return 0; /* the character class contains multiple chars */ + c = BITS_IN_ROOM * i + countbits(b1 - 1); + } + else { + return 0; /* the character class contains multiple chars */ } } } @@ -6247,7 +6248,7 @@ is_onechar_cclass(CClassNode* cc, OnigCodePoint* code) static int parse_exp(Node** np, OnigToken* tok, int term, - UChar** src, UChar* end, ScanEnv* env) + UChar** src, UChar* end, ScanEnv* env) { int r, len, group = 0; Node* qn; @@ -6276,14 +6277,14 @@ parse_exp(Node** np, OnigToken* tok, int term, env->option = NENCLOSE(*np)->option; r = fetch_token(tok, src, end, env); if (r < 0) { - env->option = prev; - return r; + env->option = prev; + return r; } r = parse_subexp(&target, tok, term, src, end, env); env->option = prev; if (r < 0) { - onig_node_free(target); - return r; + onig_node_free(target); + return r; } NENCLOSE(*np)->target = target; return tok->type; @@ -6321,20 +6322,20 @@ parse_exp(Node** np, OnigToken* tok, int term, string_loop: while (1) { - r = fetch_token(tok, src, end, env); - if (r < 0) return r; - if (r == TK_STRING) { - r = onig_node_str_cat(*np, tok->backp, *src); - } + r = fetch_token(tok, src, end, env); + if (r < 0) return r; + if (r == TK_STRING) { + r = onig_node_str_cat(*np, tok->backp, *src); + } #ifndef NUMBERED_CHAR_IS_NOT_CASE_AMBIG - else if (r == TK_CODE_POINT) { - r = node_str_cat_codepoint(*np, env->enc, tok->u.code); - } + else if (r == TK_CODE_POINT) { + r = node_str_cat_codepoint(*np, env->enc, tok->u.code); + } #endif - else { - break; - } - if (r < 0) return r; + else { + break; + } + if (r < 0) return r; } string_end: @@ -6350,36 +6351,36 @@ parse_exp(Node** np, OnigToken* tok, int term, CHECK_NULL_RETURN_MEMERR(*np); len = 1; while (1) { - if (len >= ONIGENC_MBC_MINLEN(env->enc)) { - if (len == enclen(env->enc, NSTR(*np)->s, NSTR(*np)->end)) { - r = fetch_token(tok, src, end, env); - NSTRING_CLEAR_RAW(*np); - goto string_end; - } - } + if (len >= ONIGENC_MBC_MINLEN(env->enc)) { + if (len == enclen(env->enc, NSTR(*np)->s, NSTR(*np)->end)) { + r = fetch_token(tok, src, end, env); + NSTRING_CLEAR_RAW(*np); + goto string_end; + } + } - r = fetch_token(tok, src, end, env); - if (r < 0) return r; - if (r != TK_RAW_BYTE) { - /* Don't use this, it is wrong for little endian encodings. */ + r = fetch_token(tok, src, end, env); + if (r < 0) return r; + if (r != TK_RAW_BYTE) { + /* Don't use this, it is wrong for little endian encodings. */ #ifdef USE_PAD_TO_SHORT_BYTE_CHAR - int rem; - if (len < ONIGENC_MBC_MINLEN(env->enc)) { - rem = ONIGENC_MBC_MINLEN(env->enc) - len; - (void )node_str_head_pad(NSTR(*np), rem, (UChar )0); - if (len + rem == enclen(env->enc, NSTR(*np)->s)) { - NSTRING_CLEAR_RAW(*np); - goto string_end; - } - } + int rem; + if (len < ONIGENC_MBC_MINLEN(env->enc)) { + rem = ONIGENC_MBC_MINLEN(env->enc) - len; + (void )node_str_head_pad(NSTR(*np), rem, (UChar )0); + if (len + rem == enclen(env->enc, NSTR(*np)->s)) { + NSTRING_CLEAR_RAW(*np); + goto string_end; + } + } #endif - return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING; - } + return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING; + } - r = node_str_cat_char(*np, (UChar )tok->u.c); - if (r < 0) return r; + r = node_str_cat_char(*np, (UChar )tok->u.c); + if (r < 0) return r; - len++; + len++; } } break; @@ -6408,7 +6409,7 @@ parse_exp(Node** np, OnigToken* tok, int term, qstart = *src; qend = find_str_position(end_op, 2, qstart, end, &nextp, env->enc); if (IS_NULL(qend)) { - nextp = qend = end; + nextp = qend = end; } *np = node_new_str(qstart, qend); CHECK_NULL_RETURN_MEMERR(*np); @@ -6420,30 +6421,30 @@ parse_exp(Node** np, OnigToken* tok, int term, { switch (tok->u.prop.ctype) { case ONIGENC_CTYPE_WORD: - *np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not, - IS_ASCII_RANGE(env->option)); - CHECK_NULL_RETURN_MEMERR(*np); - break; + *np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not, + IS_ASCII_RANGE(env->option)); + CHECK_NULL_RETURN_MEMERR(*np); + break; case ONIGENC_CTYPE_SPACE: case ONIGENC_CTYPE_DIGIT: case ONIGENC_CTYPE_XDIGIT: - { - CClassNode* cc; - - *np = node_new_cclass(); - CHECK_NULL_RETURN_MEMERR(*np); - cc = NCCLASS(*np); - r = add_ctype_to_cc(cc, tok->u.prop.ctype, 0, - IS_ASCII_RANGE(env->option), env); - if (r != 0) return r; - if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc); - } - break; + { + CClassNode* cc; + + *np = node_new_cclass(); + CHECK_NULL_RETURN_MEMERR(*np); + cc = NCCLASS(*np); + r = add_ctype_to_cc(cc, tok->u.prop.ctype, 0, + IS_ASCII_RANGE(env->option), env); + if (r != 0) return r; + if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc); + } + break; default: - return ONIGERR_PARSER_BUG; - break; + return ONIGERR_PARSER_BUG; + break; } } break; @@ -6461,26 +6462,26 @@ parse_exp(Node** np, OnigToken* tok, int term, r = parse_char_class(np, &asc_node, tok, src, end, env); if (r != 0) { - onig_node_free(asc_node); - return r; + onig_node_free(asc_node); + return r; } cc = NCCLASS(*np); if (is_onechar_cclass(cc, &code)) { - onig_node_free(*np); - onig_node_free(asc_node); - *np = node_new_empty(); - CHECK_NULL_RETURN_MEMERR(*np); - r = node_str_cat_codepoint(*np, env->enc, code); - if (r != 0) return r; - goto string_loop; + onig_node_free(*np); + onig_node_free(asc_node); + *np = node_new_empty(); + CHECK_NULL_RETURN_MEMERR(*np); + r = node_str_cat_codepoint(*np, env->enc, code); + if (r != 0) return r; + goto string_loop; } if (IS_IGNORECASE(env->option)) { - r = cclass_case_fold(np, cc, NCCLASS(asc_node), env); - if (r != 0) { - onig_node_free(asc_node); - return r; - } + r = cclass_case_fold(np, cc, NCCLASS(asc_node), env); + if (r != 0) { + onig_node_free(asc_node); + return r; + } } onig_node_free(asc_node); } @@ -6503,13 +6504,13 @@ parse_exp(Node** np, OnigToken* tok, int term, case TK_BACKREF: len = tok->u.backref.num; *np = node_new_backref(len, - (len > 1 ? tok->u.backref.refs : &(tok->u.backref.ref1)), - tok->u.backref.by_name, + (len > 1 ? tok->u.backref.refs : &(tok->u.backref.ref1)), + tok->u.backref.by_name, #ifdef USE_BACKREF_WITH_LEVEL - tok->u.backref.exist_level, - tok->u.backref.level, + tok->u.backref.exist_level, + tok->u.backref.level, #endif - env); + env); CHECK_NULL_RETURN_MEMERR(*np); break; @@ -6519,10 +6520,10 @@ parse_exp(Node** np, OnigToken* tok, int term, int gnum = tok->u.call.gnum; if (gnum < 0 || tok->u.call.rel != 0) { - if (gnum > 0) gnum--; - gnum = BACKREF_REL_TO_ABS(gnum, env); - if (gnum <= 0) - return ONIGERR_INVALID_BACKREF; + if (gnum > 0) gnum--; + gnum = BACKREF_REL_TO_ABS(gnum, env); + if (gnum <= 0) + return ONIGERR_INVALID_BACKREF; } *np = node_new_call(tok->u.call.name, tok->u.call.name_end, gnum); CHECK_NULL_RETURN_MEMERR(*np); @@ -6541,9 +6542,9 @@ parse_exp(Node** np, OnigToken* tok, int term, case TK_INTERVAL: if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS)) { if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS)) - return ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED; + return ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED; else - *np = node_new_empty(); + *np = node_new_empty(); } else { goto tk_byte; @@ -6565,49 +6566,49 @@ parse_exp(Node** np, OnigToken* tok, int term, repeat: if (r == TK_OP_REPEAT || r == TK_INTERVAL) { if (is_invalid_quantifier_target(*targetp)) - return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID; + return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID; qn = node_new_quantifier(tok->u.repeat.lower, tok->u.repeat.upper, - (r == TK_INTERVAL ? 1 : 0)); + (r == TK_INTERVAL ? 1 : 0)); CHECK_NULL_RETURN_MEMERR(qn); NQTFR(qn)->greedy = tok->u.repeat.greedy; r = set_quantifier(qn, *targetp, group, env); if (r < 0) { - onig_node_free(qn); - return r; + onig_node_free(qn); + return r; } if (tok->u.repeat.possessive != 0) { - Node* en; - en = node_new_enclose(ENCLOSE_STOP_BACKTRACK); - if (IS_NULL(en)) { - onig_node_free(qn); - return ONIGERR_MEMORY; - } - NENCLOSE(en)->target = qn; - qn = en; + Node* en; + en = node_new_enclose(ENCLOSE_STOP_BACKTRACK); + if (IS_NULL(en)) { + onig_node_free(qn); + return ONIGERR_MEMORY; + } + NENCLOSE(en)->target = qn; + qn = en; } if (r == 0) { - *targetp = qn; + *targetp = qn; } else if (r == 1) { - onig_node_free(qn); + onig_node_free(qn); } else if (r == 2) { /* split case: /abc+/ */ - Node *tmp; + Node *tmp; - *targetp = node_new_list(*targetp, NULL); - if (IS_NULL(*targetp)) { - onig_node_free(qn); - return ONIGERR_MEMORY; - } - tmp = NCDR(*targetp) = node_new_list(qn, NULL); - if (IS_NULL(tmp)) { - onig_node_free(qn); - return ONIGERR_MEMORY; - } - targetp = &(NCAR(tmp)); + *targetp = node_new_list(*targetp, NULL); + if (IS_NULL(*targetp)) { + onig_node_free(qn); + return ONIGERR_MEMORY; + } + tmp = NCDR(*targetp) = node_new_list(qn, NULL); + if (IS_NULL(tmp)) { + onig_node_free(qn); + return ONIGERR_MEMORY; + } + targetp = &(NCAR(tmp)); } goto re_entry; } @@ -6618,7 +6619,7 @@ parse_exp(Node** np, OnigToken* tok, int term, static int parse_branch(Node** top, OnigToken* tok, int term, - UChar** src, UChar* end, ScanEnv* env) + UChar** src, UChar* end, ScanEnv* env) { int r; Node *node, **headp; @@ -6639,18 +6640,18 @@ parse_branch(Node** top, OnigToken* tok, int term, while (r != TK_EOT && r != term && r != TK_ALT) { r = parse_exp(&node, tok, term, src, end, env); if (r < 0) { - onig_node_free(node); - return r; + onig_node_free(node); + return r; } if (NTYPE(node) == NT_LIST) { - *headp = node; - while (IS_NOT_NULL(NCDR(node))) node = NCDR(node); - headp = &(NCDR(node)); + *headp = node; + while (IS_NOT_NULL(NCDR(node))) node = NCDR(node); + headp = &(NCDR(node)); } else { - *headp = node_new_list(node, NULL); - headp = &(NCDR(*headp)); + *headp = node_new_list(node, NULL); + headp = &(NCDR(*headp)); } } } @@ -6661,7 +6662,7 @@ parse_branch(Node** top, OnigToken* tok, int term, /* term_tok: TK_EOT or TK_SUBEXP_CLOSE */ static int parse_subexp(Node** top, OnigToken* tok, int term, - UChar** src, UChar* end, ScanEnv* env) + UChar** src, UChar* end, ScanEnv* env) { int r; Node *node, **headp; @@ -6684,11 +6685,14 @@ parse_subexp(Node** top, OnigToken* tok, int term, headp = &(NCDR(*top)); while (r == TK_ALT) { r = fetch_token(tok, src, end, env); - if (r < 0) return r; + if (r < 0) { + onig_node_free(node); + return r; + } r = parse_branch(&node, tok, term, src, end, env); if (r < 0) { - onig_node_free(node); - return r; + onig_node_free(node); + return r; } *headp = onig_node_new_alt(node, NULL); @@ -6733,8 +6737,8 @@ parse_regexp(Node** top, UChar** src, UChar* end, ScanEnv* env) NENCLOSE(np)->target = *top; r = scan_env_set_mem_node(env, num, np); if (r != 0) { - onig_node_free(np); - return r; + onig_node_free(np); + return r; } *top = np; } @@ -6744,7 +6748,7 @@ parse_regexp(Node** top, UChar** src, UChar* end, ScanEnv* env) extern int onig_parse_make_tree(Node** root, const UChar* pattern, const UChar* end, - regex_t* reg, ScanEnv* env) + regex_t* reg, ScanEnv* env) { int r; UChar* p; @@ -6771,7 +6775,7 @@ onig_parse_make_tree(Node** root, const UChar* pattern, const UChar* end, extern void onig_scan_env_set_error_string(ScanEnv* env, int ecode ARG_UNUSED, - UChar* arg, UChar* arg_end) + UChar* arg, UChar* arg_end) { env->error = arg; env->error_end = arg_end; diff --git a/regparse.h b/regparse.h index de980d0ac8..dd35d48525 100644 --- a/regparse.h +++ b/regparse.h @@ -69,8 +69,8 @@ RUBY_SYMBOL_EXPORT_BEGIN #define NTYPE(node) ((node)->u.base.type) #define SET_NTYPE(node, ntype) \ do { \ - int value = ntype; \ - memcpy(&((node)->u.base.type), &value, sizeof(int)); \ + int value = ntype; \ + memcpy(&((node)->u.base.type), &value, sizeof(int)); \ } while (0) #define NSTR(node) (&((node)->u.str)) |