summaryrefslogtreecommitdiff
path: root/regexec.c
diff options
context:
space:
mode:
authornaruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2012-02-17 07:42:23 +0000
committernaruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2012-02-17 07:42:23 +0000
commit0424e152c684a85f4b0691f1e84aec203115333d (patch)
treeccb4af1c7fa12a09934f315edba0b9d839e57ceb /regexec.c
parent5362e7fcddf70d16fc4261383f99175c90881353 (diff)
* Merge Onigmo-5.13.1. [ruby-dev:45057] [Feature #5820]
https://github.com/k-takata/Onigmo cp reg{comp,enc,error,exec,parse,syntax}.c reg{enc,int,parse}.h cp oniguruma.h cp tool/enc-unicode.rb cp -r enc/ git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@34663 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Diffstat (limited to 'regexec.c')
-rw-r--r--regexec.c827
1 files changed, 671 insertions, 156 deletions
diff --git a/regexec.c b/regexec.c
index ee0a5fa2af..2c16dafda5 100644
--- a/regexec.c
+++ b/regexec.c
@@ -1,8 +1,9 @@
/**********************************************************************
- regexec.c - Oniguruma (regular expression library)
+ regexec.c - Onigmo (Oniguruma-mod) (regular expression library)
**********************************************************************/
/*-
* Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2011-2012 K.Takata <kentkt AT csc DOT jp>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
@@ -34,8 +35,44 @@
#ifdef USE_CRNL_AS_LINE_TERMINATOR
#define ONIGENC_IS_MBC_CRNL(enc,p,end) \
(ONIGENC_MBC_TO_CODE(enc,p,end) == 13 && \
- ONIGENC_IS_MBC_NEWLINE(enc,(p+enclen(enc,p)),end))
-#endif
+ ONIGENC_MBC_TO_CODE(enc,(p+enclen(enc,p,end)),end) == 10)
+#define ONIGENC_IS_MBC_NEWLINE_EX(enc,p,start,end,option,check_prev) \
+ 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)
+{
+ 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;
+ }
+ else
+ 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;
+ if (ONIGENC_IS_MBC_NEWLINE(enc, p, end))
+ return 1;
+ return 0;
+ }
+ }
+ else {
+ return ONIGENC_IS_MBC_NEWLINE(enc, p, end);
+ }
+}
+#else /* USE_CRNL_AS_LINE_TERMINATOR */
+#define ONIGENC_IS_MBC_NEWLINE_EX(enc,p,start,end,option,check_prev) \
+ ONIGENC_IS_MBC_NEWLINE((enc), (p), (end))
+#endif /* USE_CRNL_AS_LINE_TERMINATOR */
#ifdef USE_CAPTURE_HISTORY
static void history_tree_free(OnigCaptureTreeNode* node);
@@ -58,6 +95,8 @@ history_tree_clear(OnigCaptureTreeNode* node)
node->beg = ONIG_REGION_NOTPOS;
node->end = ONIG_REGION_NOTPOS;
node->group = -1;
+ xfree(node->childs);
+ node->childs = (OnigCaptureTreeNode** )0;
}
}
@@ -106,14 +145,20 @@ history_tree_add_child(OnigCaptureTreeNode* parent, OnigCaptureTreeNode* child)
n = HISTORY_TREE_INIT_ALLOC_SIZE;
parent->childs =
(OnigCaptureTreeNode** )xmalloc(sizeof(OnigCaptureTreeNode*) * n);
+ CHECK_NULL_RETURN_MEMERR(parent->childs);
}
else {
+ OnigCaptureTreeNode** tmp;
n = parent->allocated * 2;
- parent->childs =
+ tmp =
(OnigCaptureTreeNode** )xrealloc(parent->childs,
sizeof(OnigCaptureTreeNode*) * n);
+ if (tmp == 0) {
+ history_tree_clear(parent);
+ return ONIGERR_MEMORY;
+ }
+ parent->childs = tmp;
}
- CHECK_NULL_RETURN_MEMERR(parent->childs);
for (i = parent->allocated; i < n; i++) {
parent->childs[i] = (OnigCaptureTreeNode* )0;
}
@@ -128,7 +173,7 @@ history_tree_add_child(OnigCaptureTreeNode* parent, OnigCaptureTreeNode* child)
static OnigCaptureTreeNode*
history_tree_clone(OnigCaptureTreeNode* node)
{
- int i;
+ int i, r;
OnigCaptureTreeNode *clone, *child;
clone = history_node_new();
@@ -142,7 +187,12 @@ history_tree_clone(OnigCaptureTreeNode* node)
history_tree_free(clone);
return (OnigCaptureTreeNode* )0;
}
- history_tree_add_child(clone, child);
+ r = history_tree_add_child(clone, child);
+ if (r != 0) {
+ history_tree_free(child);
+ history_tree_free(clone);
+ return (OnigCaptureTreeNode* )0;
+ }
}
return clone;
@@ -177,11 +227,11 @@ onig_region_resize(OnigRegion* region, int n)
n = ONIG_NREGION;
if (region->allocated == 0) {
- region->beg = (int* )xmalloc(n * sizeof(int));
+ region->beg = (OnigPosition* )xmalloc(n * sizeof(OnigPosition));
if (region->beg == 0)
return ONIGERR_MEMORY;
- region->end = (int* )xmalloc(n * sizeof(int));
+ region->end = (OnigPosition* )xmalloc(n * sizeof(OnigPosition));
if (region->end == 0) {
xfree(region->beg);
return ONIGERR_MEMORY;
@@ -190,26 +240,24 @@ onig_region_resize(OnigRegion* region, int n)
region->allocated = n;
}
else if (region->allocated < n) {
- int *tmp;
+ OnigPosition *tmp;
region->allocated = 0;
- tmp = (int* )xrealloc(region->beg, n * sizeof(int));
+ tmp = (OnigPosition* )xrealloc(region->beg, n * sizeof(OnigPosition));
if (tmp == 0) {
xfree(region->beg);
xfree(region->end);
return ONIGERR_MEMORY;
}
region->beg = tmp;
- tmp = (int* )xrealloc(region->end, n * sizeof(int));
+ tmp = (OnigPosition* )xrealloc(region->end, n * sizeof(OnigPosition));
if (tmp == 0) {
xfree(region->beg);
+ xfree(region->end);
return ONIGERR_MEMORY;
}
region->end = tmp;
- if (region->beg == 0 || region->end == 0)
- return ONIGERR_MEMORY;
-
region->allocated = n;
}
@@ -247,8 +295,8 @@ onig_region_init(OnigRegion* region)
{
region->num_regs = 0;
region->allocated = 0;
- region->beg = (int* )0;
- region->end = (int* )0;
+ region->beg = (OnigPosition* )0;
+ region->end = (OnigPosition* )0;
region->history_root = (OnigCaptureTreeNode* )0;
}
@@ -283,11 +331,13 @@ extern void
onig_region_copy(OnigRegion* to, OnigRegion* from)
{
#define RREGC_SIZE (sizeof(int) * from->num_regs)
- int i;
+ int i, r;
if (to == from) return;
- onig_region_resize(to, from->num_regs);
+ r = onig_region_resize(to, from->num_regs);
+ if (r) return;
+
for (i = 0; i < from->num_regs; i++) {
to->beg[i] = from->beg[i];
to->end[i] = from->end[i];
@@ -334,19 +384,21 @@ onig_region_copy(OnigRegion* to, OnigRegion* from)
#define STK_MASK_MEM_END_OR_MARK 0x8000 /* MEM_END or MEM_END_MARK */
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
-#define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start) do {\
+#define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start, arg_gpos) do {\
(msa).stack_p = (void* )0;\
(msa).options = (arg_option);\
(msa).region = (arg_region);\
(msa).start = (arg_start);\
+ (msa).gpos = (arg_gpos);\
(msa).best_len = ONIG_MISMATCH;\
} while(0)
#else
-#define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start) do {\
+#define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start, arg_gpos) do {\
(msa).stack_p = (void* )0;\
(msa).options = (arg_option);\
(msa).region = (arg_region);\
(msa).start = (arg_start);\
+ (msa).gpos = (arg_gpos);\
} while(0)
#endif
@@ -386,9 +438,9 @@ onig_region_copy(OnigRegion* to, OnigRegion* from)
if ((msa).state_check_buff) xfree((msa).state_check_buff);\
}\
} while(0)
-#else
+#else /* USE_COMBINATION_EXPLOSION_CHECK */
#define MATCH_ARG_FREE(msa) if ((msa).stack_p) xfree((msa).stack_p)
-#endif
+#endif /* USE_COMBINATION_EXPLOSION_CHECK */
@@ -510,13 +562,14 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
state_check_buff[x/8] |= (1<<(x%8)); \
}
-#define STACK_PUSH(stack_type,pat,s,sprev) do {\
+#define STACK_PUSH(stack_type,pat,s,sprev,keep) do {\
STACK_ENSURE(1);\
stk->type = (stack_type);\
stk->u.state.pcode = (pat);\
stk->u.state.pstr = (s);\
stk->u.state.pstr_prev = (sprev);\
stk->u.state.state_check = 0;\
+ stk->u.state.pkeep = (keep);\
STACK_INC;\
} while(0)
@@ -527,13 +580,14 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
STACK_INC;\
} while(0)
-#define STACK_PUSH_ALT_WITH_STATE_CHECK(pat,s,sprev,snum) do {\
+#define STACK_PUSH_ALT_WITH_STATE_CHECK(pat,s,sprev,snum,keep) do {\
STACK_ENSURE(1);\
stk->type = STK_ALT;\
stk->u.state.pcode = (pat);\
stk->u.state.pstr = (s);\
stk->u.state.pstr_prev = (sprev);\
stk->u.state.state_check = ((state_check_buff != NULL) ? (snum) : 0);\
+ stk->u.state.pkeep = (keep);\
STACK_INC;\
} while(0)
@@ -551,12 +605,13 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
#define ELSE_IF_STATE_CHECK_MARK(stk)
-#define STACK_PUSH(stack_type,pat,s,sprev) do {\
+#define STACK_PUSH(stack_type,pat,s,sprev,keep) do {\
STACK_ENSURE(1);\
stk->type = (stack_type);\
stk->u.state.pcode = (pat);\
stk->u.state.pstr = (s);\
stk->u.state.pstr_prev = (sprev);\
+ stk->u.state.pkeep = (keep);\
STACK_INC;\
} while(0)
@@ -567,12 +622,12 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
} while(0)
#endif /* USE_COMBINATION_EXPLOSION_CHECK */
-#define STACK_PUSH_ALT(pat,s,sprev) STACK_PUSH(STK_ALT,pat,s,sprev)
-#define STACK_PUSH_POS(s,sprev) STACK_PUSH(STK_POS,NULL_UCHARP,s,sprev)
-#define STACK_PUSH_POS_NOT(pat,s,sprev) STACK_PUSH(STK_POS_NOT,pat,s,sprev)
+#define STACK_PUSH_ALT(pat,s,sprev,keep) STACK_PUSH(STK_ALT,pat,s,sprev,keep)
+#define STACK_PUSH_POS(s,sprev,keep) STACK_PUSH(STK_POS,NULL_UCHARP,s,sprev,keep)
+#define STACK_PUSH_POS_NOT(pat,s,sprev,keep) STACK_PUSH(STK_POS_NOT,pat,s,sprev,keep)
#define STACK_PUSH_STOP_BT STACK_PUSH_TYPE(STK_STOP_BT)
-#define STACK_PUSH_LOOK_BEHIND_NOT(pat,s,sprev) \
- STACK_PUSH(STK_LOOK_BEHIND_NOT,pat,s,sprev)
+#define STACK_PUSH_LOOK_BEHIND_NOT(pat,s,sprev,keep) \
+ STACK_PUSH(STK_LOOK_BEHIND_NOT,pat,s,sprev,keep)
#define STACK_PUSH_REPEAT(id, pat) do {\
STACK_ENSURE(1);\
@@ -879,7 +934,7 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
}\
k++;\
}\
- break;\
+ break;\
}\
}\
}\
@@ -920,7 +975,7 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
}\
k++;\
}\
- break;\
+ break;\
}\
}\
else {\
@@ -983,7 +1038,7 @@ 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, int 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];
@@ -1057,20 +1112,23 @@ make_capture_history_tree(OnigCaptureTreeNode* node, OnigStackType** kp,
child = history_node_new();
CHECK_NULL_RETURN_MEMERR(child);
child->group = n;
- child->beg = (int )(k->u.mem.pstr - str);
+ child->beg = k->u.mem.pstr - str;
r = history_tree_add_child(node, child);
- if (r != 0) return r;
+ 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 = (int )(k->u.mem.pstr - str);
+ child->end = k->u.mem.pstr - str;
}
}
else if (k->type == STK_MEM_END) {
if (k->u.mem.num == node->group) {
- node->end = (int )(k->u.mem.pstr - str);
+ node->end = k->u.mem.pstr - str;
*kp = k;
return 0;
}
@@ -1080,7 +1138,7 @@ make_capture_history_tree(OnigCaptureTreeNode* node, OnigStackType** kp,
return 1; /* 1: root node ending. */
}
-#endif
+#endif /* USE_CAPTURE_HISTORY */
#ifdef USE_BACKREF_WITH_LEVEL
static int mem_is_in_memp(int mem, int num, UChar* memp)
@@ -1125,7 +1183,7 @@ static int backref_match_at_nested_level(regex_t* reg
if (ignore_case != 0) {
if (string_cmp_ic(reg->enc, case_fold_flag,
- pstart, &ss, (int )(pend - pstart), send) == 0)
+ pstart, &ss, pend - pstart, send) == 0)
return 0; /* or goto next_mem; */
}
else {
@@ -1168,14 +1226,14 @@ static struct timeval ts, te;
#define GETTIME(t) gettimeofday(&(t), (struct timezone* )0)
#define TIMEDIFF(te,ts) (((te).tv_usec - (ts).tv_usec) + \
(((te).tv_sec - (ts).tv_sec)*1000000))
-#else
+#else /* USE_TIMEOFDAY */
#ifdef HAVE_SYS_TIMES_H
#include <sys/times.h>
#endif
static struct tms ts, te;
#define GETTIME(t) times(&(t))
#define TIMEDIFF(te,ts) ((te).tms_utime - (ts).tms_utime)
-#endif
+#endif /* USE_TIMEOFDAY */
static int OpCounter[256];
static int OpPrevCounter[256];
@@ -1224,12 +1282,13 @@ onig_print_statistics(FILE* f)
MaxStackDepth = stk - stk_base;\
} while(0)
-#else
+#else /* ONIG_DEBUG_STATISTICS */
#define STACK_INC stk++
#define MOP_IN(opcode)
#define MOP_OUT
-#endif
+#endif /* ONIG_DEBUG_STATISTICS */
+
/* matching region of POSIX API */
@@ -1245,7 +1304,7 @@ void onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nex
/* match data(str - end) from position (sstart). */
/* if sstart == str then set sprev to NULL. */
-static long
+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,
@@ -1254,8 +1313,8 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
{
static const UChar FinishCode[] = { OP_FINISH };
- int i, num_mem, best_len, pop_level;
- ptrdiff_t n;
+ int i, num_mem, pop_level;
+ ptrdiff_t n, best_len;
LengthType tlen, tlen2;
MemNumType mem;
RelAddrType addr;
@@ -1264,6 +1323,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
OnigCaseFoldType case_fold_flag = reg->case_fold_flag;
UChar *s, *q, *sbegin;
UChar *p = reg->p;
+ UChar *pkeep;
char *alloca_base;
OnigStackType *stk_alloc, *stk_base, *stk, *stk_end;
OnigStackType *stkp; /* used as any purpose. */
@@ -1275,6 +1335,23 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
unsigned char* state_check_buff = msa->state_check_buff;
int num_comb_exp_check = reg->num_comb_exp_check;
#endif
+
+#ifdef USE_SUBEXP_CALL
+ /* Stack #0 is used to store the pattern itself and used for (?R), \g<0>, etc. */
+ n = reg->num_repeat + (reg->num_mem + 1) * 2;
+
+ STACK_INIT(alloca_base, n, INIT_MATCH_STACK_SIZE);
+ pop_level = reg->stack_pop_level;
+ num_mem = reg->num_mem;
+ repeat_stk = (OnigStackIndex* )alloca_base;
+
+ mem_start_stk = (OnigStackIndex* )(repeat_stk + reg->num_repeat);
+ mem_end_stk = mem_start_stk + (num_mem + 1);
+ for (i = 0; i <= num_mem; i++) {
+ mem_start_stk[i] = mem_end_stk[i] = INVALID_STACK_INDEX;
+ }
+#else /* USE_SUBEXP_CALL */
+ /* Stack #0 not is used. */
n = reg->num_repeat + reg->num_mem * 2;
STACK_INIT(alloca_base, n, INIT_MATCH_STACK_SIZE);
@@ -1291,10 +1368,11 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
for (i = 1; i <= num_mem; i++) {
mem_start_stk[i] = mem_end_stk[i] = INVALID_STACK_INDEX;
}
+#endif /* USE_SUBEXP_CALL */
#ifdef ONIG_DEBUG_MATCH
- fprintf(stderr, "match_at: str: %"PRIdPTR", end: %"PRIdPTR", start: %"PRIdPTR", sprev: %"PRIdPTR"\n",
- (intptr_t)str, (intptr_t)end, (intptr_t)sstart, (intptr_t)sprev);
+ fprintf(stderr, "match_at: str: %"PRIdPTR" (%p), end: %"PRIdPTR" (%p), start: %"PRIdPTR" (%p), sprev: %"PRIdPTR" (%p)\n",
+ (intptr_t)str, str, (intptr_t)end, end, (intptr_t)sstart, sstart, (intptr_t)sprev, sprev);
fprintf(stderr, "size: %d, start offset: %d\n",
(int )(end - str), (int )(sstart - str));
#endif
@@ -1302,6 +1380,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STACK_PUSH_ENSURED(STK_ALT, (UChar *)FinishCode); /* bottom stack */
best_len = ONIG_MISMATCH;
s = (UChar* )sstart;
+ pkeep = (UChar* )sstart;
while (1) {
#ifdef ONIG_DEBUG_MATCH
if (s) {
@@ -1309,9 +1388,11 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
int len;
fprintf(stderr, "%4d> \"", (int )(s - str));
bp = buf;
- for (i = 0, q = s; i < 7 && q < end; i++) {
- len = enclen(encode, q, end);
- while (len-- > 0) *bp++ = *q++;
+ if (*p != OP_FINISH) { /* s may not be a valid pointer if OP_FINISH. */
+ for (i = 0, q = s; i < 7 && q < end; i++) {
+ len = enclen(encode, q, end);
+ while (len-- > 0) *bp++ = *q++;
+ }
}
if (q < end) { xmemcpy(bp, "...\"", 4); bp += 4; }
else { xmemcpy(bp, "\"", 1); bp += 1; }
@@ -1332,32 +1413,32 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
if (IS_FIND_LONGEST(option)) {
if (n > msa->best_len) {
- msa->best_len = (int)n;
+ msa->best_len = n;
msa->best_s = (UChar* )sstart;
}
else
goto end_best_len;
}
#endif
- best_len = (int)n;
+ best_len = n;
region = msa->region;
if (region) {
#ifdef USE_POSIX_API_REGION_OPTION
if (IS_POSIX_REGION(msa->options)) {
posix_regmatch_t* rmt = (posix_regmatch_t* )region;
- rmt[0].rm_so = sstart - str;
- rmt[0].rm_eo = s - str;
+ rmt[0].rm_so = (regoff_t )(((pkeep > s) ? s : pkeep) - str);
+ rmt[0].rm_eo = (regoff_t )(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))
- rmt[i].rm_so = STACK_AT(mem_start_stk[i])->u.mem.pstr - str;
+ rmt[i].rm_so = (regoff_t )(STACK_AT(mem_start_stk[i])->u.mem.pstr - str);
else
- rmt[i].rm_so = (UChar* )((void* )(mem_start_stk[i])) - str;
+ rmt[i].rm_so = (regoff_t )((UChar* )((void* )(mem_start_stk[i])) - str);
- rmt[i].rm_eo = (BIT_STATUS_AT(reg->bt_mem_end, i)
+ rmt[i].rm_eo = (regoff_t )((BIT_STATUS_AT(reg->bt_mem_end, i)
? STACK_AT(mem_end_stk[i])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[i])) - str;
+ : (UChar* )((void* )mem_end_stk[i])) - str);
}
else {
rmt[i].rm_so = rmt[i].rm_eo = ONIG_REGION_NOTPOS;
@@ -1366,16 +1447,18 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
}
else {
#endif /* USE_POSIX_API_REGION_OPTION */
- region->beg[0] = (int)(sstart - str);
- region->end[0] = (int)(s - str);
+ 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) {
- region->beg[i] = (int)((BIT_STATUS_AT(reg->bt_mem_start, i))
- ? STACK_AT(mem_start_stk[i])->u.mem.pstr - str
- : (UChar* )((void* )mem_start_stk[i]) - str);
- region->end[i] = (int)(BIT_STATUS_AT(reg->bt_mem_end, i)
- ? STACK_AT(mem_end_stk[i])->u.mem.pstr - str
- : (UChar* )((void* )mem_end_stk[i]) - str);
+ 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;
@@ -1397,8 +1480,8 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
}
node->group = 0;
- node->beg = sstart - str;
- node->end = s - str;
+ node->beg = ((pkeep > s) ? s : pkeep) - str;
+ node->end = s - str;
stkp = stk_base;
r = make_capture_history_tree(region->history_root, &stkp,
@@ -1795,7 +1878,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
DATA_ENSURE(1);
n = enclen(encode, s, end);
DATA_ENSURE(n);
- if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
+ if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
s += n;
MOP_OUT;
break;
@@ -1810,10 +1893,10 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_ANYCHAR_STAR: MOP_IN(OP_ANYCHAR_STAR);
while (DATA_ENSURE_CHECK1) {
- STACK_PUSH_ALT(p, s, sprev);
+ STACK_PUSH_ALT(p, s, sprev, pkeep);
n = enclen(encode, s, end);
DATA_ENSURE(n);
- if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
+ if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
sprev = s;
s += n;
}
@@ -1822,7 +1905,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_ANYCHAR_ML_STAR: MOP_IN(OP_ANYCHAR_ML_STAR);
while (DATA_ENSURE_CHECK1) {
- STACK_PUSH_ALT(p, s, sprev);
+ STACK_PUSH_ALT(p, s, sprev, pkeep);
n = enclen(encode, s, end);
if (n > 1) {
DATA_ENSURE(n);
@@ -1840,11 +1923,11 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_ANYCHAR_STAR_PEEK_NEXT: MOP_IN(OP_ANYCHAR_STAR_PEEK_NEXT);
while (DATA_ENSURE_CHECK1) {
if (*p == *s) {
- STACK_PUSH_ALT(p + 1, s, sprev);
+ STACK_PUSH_ALT(p + 1, s, sprev, pkeep);
}
n = enclen(encode, s, end);
DATA_ENSURE(n);
- if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
+ if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
sprev = s;
s += n;
}
@@ -1855,7 +1938,7 @@ 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) {
if (*p == *s) {
- STACK_PUSH_ALT(p + 1, s, sprev);
+ STACK_PUSH_ALT(p + 1, s, sprev, pkeep);
}
n = enclen(encode, s, end);
if (n > 1) {
@@ -1879,10 +1962,10 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STATE_CHECK_VAL(scv, mem);
if (scv) goto fail;
- STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem);
+ STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep);
n = enclen(encode, s, end);
DATA_ENSURE(n);
- if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) goto fail;
+ if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
sprev = s;
s += n;
}
@@ -1897,7 +1980,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STATE_CHECK_VAL(scv, mem);
if (scv) goto fail;
- STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem);
+ STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep);
n = enclen(encode, s, end);
if (n > 1) {
DATA_ENSURE(n);
@@ -1922,6 +2005,15 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
MOP_OUT;
break;
+ case OP_ASCII_WORD: MOP_IN(OP_ASCII_WORD);
+ DATA_ENSURE(1);
+ if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
+ goto fail;
+
+ s += enclen(encode, s, end);
+ MOP_OUT;
+ break;
+
case OP_NOT_WORD: MOP_IN(OP_NOT_WORD);
DATA_ENSURE(1);
if (ONIGENC_IS_MBC_WORD(encode, s, end))
@@ -1931,6 +2023,15 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
MOP_OUT;
break;
+ 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;
+
+ s += enclen(encode, s, end);
+ MOP_OUT;
+ break;
+
case OP_WORD_BOUND: MOP_IN(OP_WORD_BOUND);
if (ON_STR_BEGIN(s)) {
DATA_ENSURE(1);
@@ -1950,6 +2051,25 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
continue;
break;
+ 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;
+ }
+ else if (ON_STR_END(s)) {
+ 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;
+ }
+ MOP_OUT;
+ continue;
+ break;
+
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))
@@ -1968,6 +2088,24 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
continue;
break;
+ 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;
+ }
+ else if (ON_STR_END(s)) {
+ 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;
+ }
+ MOP_OUT;
+ continue;
+ break;
+
#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)) {
@@ -1979,6 +2117,16 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
goto fail;
break;
+ 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;
+ continue;
+ }
+ }
+ goto fail;
+ break;
+
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)) {
@@ -1988,6 +2136,16 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
}
goto fail;
break;
+
+ 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;
+ continue;
+ }
+ }
+ goto fail;
+ break;
#endif
case OP_BEGIN_BUF: MOP_IN(OP_BEGIN_BUF);
@@ -2005,12 +2163,18 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
break;
case OP_BEGIN_LINE: MOP_IN(OP_BEGIN_LINE);
+ op_begin_line:
if (ON_STR_BEGIN(s)) {
if (IS_NOTBOL(msa->options)) goto fail;
MOP_OUT;
continue;
}
- else if (ONIGENC_IS_MBC_NEWLINE(encode, sprev, end) && !ON_STR_END(s)) {
+ 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))
+#endif
+ && !ON_STR_END(s)) {
MOP_OUT;
continue;
}
@@ -2020,7 +2184,7 @@ 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(encode, sprev, end)) {
+ 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;
@@ -2029,23 +2193,17 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
}
#endif
}
- else if (ONIGENC_IS_MBC_NEWLINE(encode, s, end)) {
- MOP_OUT;
- continue;
- }
-#ifdef USE_CRNL_AS_LINE_TERMINATOR
- else if (ONIGENC_IS_MBC_CRNL(encode, s, end)) {
+ else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) {
MOP_OUT;
continue;
}
-#endif
goto fail;
break;
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(encode, sprev, end)) {
+ 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;
@@ -2054,32 +2212,41 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
}
#endif
}
- else if (ONIGENC_IS_MBC_NEWLINE(encode, s, end) &&
- ON_STR_END(s + enclen(encode, s, end))) {
- MOP_OUT;
- continue;
- }
+ 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;
+ continue;
+ }
#ifdef USE_CRNL_AS_LINE_TERMINATOR
- else if (ONIGENC_IS_MBC_CRNL(encode, s, end)) {
- UChar* ss = s + enclen(encode, s);
- ss += enclen(encode, ss);
- if (ON_STR_END(ss)) {
- MOP_OUT;
- continue;
- }
- }
+ else if (ss < end) {
+ ss += enclen(encode, ss, end);
+ if (ON_STR_END(ss)) {
+ MOP_OUT;
+ continue;
+ }
+ }
#endif
+ }
goto fail;
break;
case OP_BEGIN_POSITION: MOP_IN(OP_BEGIN_POSITION);
- if (s != msa->start)
+ if (s != msa->gpos)
goto fail;
MOP_OUT;
continue;
break;
+ case OP_BEGIN_POS_OR_LINE: MOP_IN(OP_BEGIN_POS_OR_LINE);
+ if (s != msa->gpos)
+ goto op_begin_line;
+
+ MOP_OUT;
+ continue;
+ break;
+
case OP_MEMORY_START_PUSH: MOP_IN(OP_MEMORY_START_PUSH);
GET_MEMNUM_INC(mem, p);
STACK_PUSH_MEM_START(mem, s);
@@ -2108,6 +2275,12 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
continue;
break;
+ case OP_KEEP: MOP_IN(OP_KEEP);
+ pkeep = s;
+ MOP_OUT;
+ continue;
+ break;
+
#ifdef USE_SUBEXP_CALL
case OP_MEMORY_END_PUSH_REC: MOP_IN(OP_MEMORY_END_PUSH_REC);
GET_MEMNUM_INC(mem, p);
@@ -2272,7 +2445,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
DATA_ENSURE(n);
sprev = s;
swork = s;
- STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, (int)n, end, is_fail);
+ 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)
@@ -2319,7 +2492,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
#if 0 /* no need: IS_DYNAMIC_OPTION() == 0 */
case OP_SET_OPTION_PUSH: MOP_IN(OP_SET_OPTION_PUSH);
GET_OPTION_INC(option, p);
- STACK_PUSH_ALT(p, s, sprev);
+ STACK_PUSH_ALT(p, s, sprev, pkeep);
p += SIZE_OP_SET_OPTION + SIZE_OP_FAIL;
MOP_OUT;
continue;
@@ -2347,8 +2520,8 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STACK_NULL_CHECK(isnull, mem, s);
if (isnull) {
#ifdef ONIG_DEBUG_MATCH
- fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%"PRIdPTR"\n",
- (int )mem, (intptr_t )s);
+ fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%"PRIdPTR" (%p)\n",
+ (int )mem, (intptr_t )s, s);
#endif
null_check_found:
/* empty loop founded, skip next instruction */
@@ -2382,8 +2555,8 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STACK_NULL_CHECK_MEMST(isnull, mem, s, reg);
if (isnull) {
#ifdef ONIG_DEBUG_MATCH
- fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%"PRIdPTR"\n",
- (int )mem, (intptr_t )s);
+ fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%"PRIdPTR" (%p)\n",
+ (int )mem, (intptr_t )s, s);
#endif
if (isnull == -1) goto fail;
goto null_check_found;
@@ -2408,8 +2581,8 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
#endif
if (isnull) {
#ifdef ONIG_DEBUG_MATCH
- fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%"PRIdPTR"\n",
- (int )mem, (intptr_t )s);
+ fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%"PRIdPTR" (%p)\n",
+ (int )mem, (intptr_t )s, s);
#endif
if (isnull == -1) goto fail;
goto null_check_found;
@@ -2433,7 +2606,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_PUSH: MOP_IN(OP_PUSH);
GET_RELADDR_INC(addr, p);
- STACK_PUSH_ALT(p + addr, s, sprev);
+ STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
MOP_OUT;
continue;
break;
@@ -2445,7 +2618,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
if (scv) goto fail;
GET_RELADDR_INC(addr, p);
- STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem);
+ STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep);
MOP_OUT;
continue;
break;
@@ -2458,7 +2631,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
p += addr;
}
else {
- STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem);
+ STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep);
}
MOP_OUT;
continue;
@@ -2485,7 +2658,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
GET_RELADDR_INC(addr, p);
if (*p == *s && DATA_ENSURE_CHECK1) {
p++;
- STACK_PUSH_ALT(p + addr, s, sprev);
+ STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
MOP_OUT;
continue;
}
@@ -2498,7 +2671,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
GET_RELADDR_INC(addr, p);
if (*p == *s) {
p++;
- STACK_PUSH_ALT(p + addr, s, sprev);
+ STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
MOP_OUT;
continue;
}
@@ -2517,7 +2690,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STACK_PUSH_REPEAT(mem, p);
if (reg->repeat_range[mem].lower == 0) {
- STACK_PUSH_ALT(p + addr, s, sprev);
+ STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
}
}
MOP_OUT;
@@ -2534,7 +2707,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
STACK_PUSH_REPEAT(mem, p);
if (reg->repeat_range[mem].lower == 0) {
- STACK_PUSH_ALT(p, s, sprev);
+ STACK_PUSH_ALT(p, s, sprev, pkeep);
p += addr;
}
}
@@ -2553,7 +2726,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
/* end of repeat. Nothing to do. */
}
else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
- STACK_PUSH_ALT(p, s, sprev);
+ STACK_PUSH_ALT(p, s, sprev, pkeep);
p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */
}
else {
@@ -2584,7 +2757,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
UChar* pcode = stkp->u.repeat.pcode;
STACK_PUSH_REPEAT_INC(si);
- STACK_PUSH_ALT(pcode, s, sprev);
+ STACK_PUSH_ALT(pcode, s, sprev, pkeep);
}
else {
p = stkp->u.repeat.pcode;
@@ -2607,7 +2780,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
break;
case OP_PUSH_POS: MOP_IN(OP_PUSH_POS);
- STACK_PUSH_POS(s, sprev);
+ STACK_PUSH_POS(s, sprev, pkeep);
MOP_OUT;
continue;
break;
@@ -2624,7 +2797,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_PUSH_POS_NOT: MOP_IN(OP_PUSH_POS_NOT);
GET_RELADDR_INC(addr, p);
- STACK_PUSH_POS_NOT(p + addr, s, sprev);
+ STACK_PUSH_POS_NOT(p + addr, s, sprev, pkeep);
MOP_OUT;
continue;
break;
@@ -2666,7 +2839,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
/* goto fail; */
}
else {
- STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev);
+ STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev, pkeep);
s = q;
sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end);
}
@@ -2696,6 +2869,18 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
break;
#endif
+ case OP_CONDITION: MOP_IN(OP_CONDITION);
+ 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;
+ }
+ MOP_OUT;
+ continue;
+ break;
+
case OP_FINISH:
goto finish;
break;
@@ -2708,6 +2893,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
p = stk->u.state.pcode;
s = stk->u.state.pstr;
sprev = stk->u.state.pstr_prev;
+ pkeep = stk->u.state.pkeep;
#ifdef USE_COMBINATION_EXPLOSION_CHECK
if (stk->u.state.state_check != 0) {
@@ -2772,7 +2958,7 @@ slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
}
s += n;
}
- return (UChar*)NULL;
+ return (UChar* )NULL;
}
while (s < end) {
if (*s == *target) {
@@ -2890,6 +3076,8 @@ slow_search_backward_ic(OnigEncoding enc, int case_fold_flag,
return (UChar* )NULL;
}
+#ifndef USE_SUNDAY_QUICK_SEARCH
+/* 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,
@@ -2900,8 +3088,8 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
ptrdiff_t skip, tlen1;
#ifdef ONIG_DEBUG_SEARCH
- fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR", text_end: %"PRIuPTR", text_range: %"PRIuPTR"\n",
- text, text_end, text_range);
+ fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
+ text, text, text_end, text_end, text_range, text_range);
#endif
tail = target_end - 1;
@@ -2946,6 +3134,7 @@ bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
return (UChar* )NULL;
}
+/* 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)
@@ -2993,10 +3182,315 @@ bm_search(regex_t* reg, const UChar* target, const UChar* target_end,
return (UChar* )NULL;
}
+/* 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 *s, *se, *t, *end;
+ const UChar *tail;
+ ptrdiff_t skip, tlen1;
+ OnigEncoding enc = reg->enc;
+ int case_fold_flag = reg->case_fold_flag;
+
+#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);
+#endif
+
+ tail = target_end - 1;
+ tlen1 = tail - target;
+ end = text_range;
+ if (end + tlen1 > text_end)
+ end = text_end - tlen1;
+
+ s = text;
+
+ if (IS_NULL(reg->int_map)) {
+ while (s < end) {
+ se = s + tlen1;
+ if (str_lower_case_match(enc, case_fold_flag, target, target_end,
+ s, se + 1))
+ return (UChar* )s;
+ skip = reg->map[*se];
+ t = s;
+ do {
+ s += enclen(reg->enc, s, end);
+ } while ((s - t) < skip && s < end);
+ }
+ }
+ else {
+ while (s < end) {
+ se = s + tlen1;
+ if (str_lower_case_match(enc, case_fold_flag, target, target_end,
+ s, se + 1))
+ return (UChar* )s;
+ skip = reg->int_map[*se];
+ t = s;
+ do {
+ s += enclen(reg->enc, s, end);
+ } while ((s - t) < skip && s < end);
+ }
+ }
+
+ return (UChar* )NULL;
+}
+
+/* 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 *s, *p, *end;
+ const UChar *tail;
+ OnigEncoding enc = reg->enc;
+ int case_fold_flag = reg->case_fold_flag;
+
+#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);
+#endif
+
+ end = text_range + (target_end - target) - 1;
+ if (end > text_end)
+ end = text_end;
+
+ tail = target_end - 1;
+ s = text + (target_end - target) - 1;
+ if (IS_NULL(reg->int_map)) {
+ 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;
+ s += reg->map[*s];
+ }
+ }
+ else { /* see int_map[] */
+ 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;
+ s += reg->int_map[*s];
+ }
+ }
+ return (UChar* )NULL;
+}
+
+#else /* USE_SUNDAY_QUICK_SEARCH */
+
+/* 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 *s, *se, *t, *p, *end;
+ const UChar *tail;
+ ptrdiff_t skip, tlen1;
+ OnigEncoding enc = reg->enc;
+
+#ifdef ONIG_DEBUG_SEARCH
+ fprintf(stderr, "bm_search_notrev: 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);
+#endif
+
+ tail = target_end - 1;
+ tlen1 = tail - target;
+ end = text_range;
+ if (end + tlen1 > text_end)
+ end = text_end - tlen1;
+
+ s = text;
+
+ if (IS_NULL(reg->int_map)) {
+ while (s < end) {
+ p = se = s + tlen1;
+ t = tail;
+ while (*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);
+ } while ((s - t) < skip && s < end);
+ }
+ }
+ else {
+ while (s < end) {
+ p = se = s + tlen1;
+ t = tail;
+ while (*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);
+ } while ((s - t) < skip && s < end);
+ }
+ }
+
+ return (UChar* )NULL;
+}
+
+/* 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 *s, *t, *p, *end;
+ const UChar *tail;
+ ptrdiff_t tlen1;
+
+ tail = target_end - 1;
+ tlen1 = tail - target;
+ end = text_range + tlen1;
+ if (end > text_end)
+ end = text_end;
+
+ s = text + tlen1;
+ if (IS_NULL(reg->int_map)) {
+ while (s < end) {
+ p = s;
+ t = tail;
+ while (*p == *t) {
+ if (t == target) return (UChar* )p;
+ p--; t--;
+ }
+ if (s + 1 >= end) break;
+ s += reg->map[s[1]];
+ }
+ }
+ else { /* see int_map[] */
+ while (s < end) {
+ p = s;
+ t = tail;
+ while (*p == *t) {
+ if (t == target) return (UChar* )p;
+ p--; t--;
+ }
+ if (s + 1 >= end) break;
+ s += reg->int_map[s[1]];
+ }
+ }
+ return (UChar* )NULL;
+}
+
+/* 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 *s, *se, *t, *end;
+ const UChar *tail;
+ ptrdiff_t skip, tlen1;
+ OnigEncoding enc = reg->enc;
+ int case_fold_flag = reg->case_fold_flag;
+
+#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);
+#endif
+
+ tail = target_end - 1;
+ tlen1 = tail - target;
+ end = text_range;
+ if (end + tlen1 > text_end)
+ end = text_end - tlen1;
+
+ s = text;
+
+ if (IS_NULL(reg->int_map)) {
+ while (s < end) {
+ se = s + tlen1;
+ if (str_lower_case_match(enc, case_fold_flag, target, target_end,
+ s, se + 1))
+ return (UChar* )s;
+ if (s + 1 >= end) break;
+ skip = reg->map[se[1]];
+ t = s;
+ do {
+ s += enclen(enc, s, end);
+ } while ((s - t) < skip && s < end);
+ }
+ }
+ else {
+ while (s < end) {
+ se = s + tlen1;
+ if (str_lower_case_match(enc, case_fold_flag, target, target_end,
+ 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);
+ } while ((s - t) < skip && s < end);
+ }
+ }
+
+ return (UChar* )NULL;
+}
+
+/* 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 *s, *p, *end;
+ const UChar *tail;
+ ptrdiff_t tlen1;
+ OnigEncoding enc = reg->enc;
+ int case_fold_flag = reg->case_fold_flag;
+
+#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);
+#endif
+
+ tail = target_end - 1;
+ tlen1 = tail - target;
+ end = text_range + tlen1;
+ if (end > text_end)
+ end = text_end;
+
+ s = text + tlen1;
+ if (IS_NULL(reg->int_map)) {
+ while (s < end) {
+ p = s - tlen1;
+ if (str_lower_case_match(enc, case_fold_flag, target, target_end,
+ p, s + 1))
+ return (UChar* )p;
+ if (s + 1 >= end) break;
+ s += reg->map[s[1]];
+ }
+ }
+ else { /* see int_map[] */
+ while (s < end) {
+ p = s - tlen1;
+ if (str_lower_case_match(enc, case_fold_flag, target, target_end,
+ p, s + 1))
+ return (UChar* )p;
+ if (s + 1 >= end) break;
+ s += reg->int_map[s[1]];
+ }
+ }
+ return (UChar* )NULL;
+}
+#endif /* USE_SUNDAY_QUICK_SEARCH */
+
static int
set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,
int** skip)
-
{
int i, len;
@@ -3005,7 +3499,7 @@ set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,
if (IS_NULL(*skip)) return ONIGERR_MEMORY;
}
- len = (int)(end - s);
+ len = (int )(end - s);
for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++)
(*skip)[i] = len;
@@ -3073,11 +3567,11 @@ map_search_backward(OnigEncoding enc, UChar map[],
return (UChar* )NULL;
}
-extern long
+extern OnigPosition
onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, OnigRegion* region,
OnigOptionType option)
{
- long r;
+ ptrdiff_t r;
UChar *prev;
OnigMatchArg msa;
@@ -3106,7 +3600,7 @@ onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, On
THREAD_ATOMIC_END;
#endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */
- MATCH_ARG_INIT(msa, option, region, at);
+ MATCH_ARG_INIT(msa, option, region, at, at);
#ifdef USE_COMBINATION_EXPLOSION_CHECK
{
int offset = at - str;
@@ -3145,8 +3639,8 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
UChar *p, *pprev = (UChar* )NULL;
#ifdef ONIG_DEBUG_SEARCH
- fprintf(stderr, "forward_search_range: str: %"PRIuPTR", end: %"PRIuPTR", s: %"PRIuPTR", range: %"PRIuPTR"\n",
- str, end, s, range);
+ fprintf(stderr, "forward_search_range: str: %"PRIuPTR" (%p), end: %"PRIuPTR" (%p), s: %"PRIuPTR" (%p), range: %"PRIuPTR" (%p)\n",
+ str, str, end, end, s, s, range, range);
#endif
p = s;
@@ -3178,6 +3672,14 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
p = bm_search_notrev(reg, reg->exact, reg->exact_end, p, end, range);
break;
+ case ONIG_OPTIMIZE_EXACT_BM_IC:
+ p = bm_search_ic(reg, reg->exact, reg->exact_end, p, end, range);
+ break;
+
+ case ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC:
+ p = bm_search_notrev_ic(reg, reg->exact, reg->exact_end, p, end, range);
+ break;
+
case ONIG_OPTIMIZE_MAP:
p = map_search(reg->enc, reg->map, p, range, end);
break;
@@ -3199,7 +3701,7 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
if (!ON_STR_BEGIN(p)) {
prev = onigenc_get_prev_char_head(reg->enc,
(pprev ? pprev : str), p, end);
- if (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end))
+ if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0))
goto retry_gate;
}
break;
@@ -3209,15 +3711,11 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
#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(reg->enc, prev, end))
+ 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(reg->enc, p, end)
-#ifdef USE_CRNL_AS_LINE_TERMINATOR
- && ! ONIGENC_IS_MBC_CRNL(reg->enc, p, end)
-#endif
- )
+ else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1))
goto retry_gate;
break;
}
@@ -3269,7 +3767,7 @@ static int set_bm_backward_skip P_((UChar* s, UChar* end, OnigEncoding enc,
#define BM_BACKWARD_SEARCH_LENGTH_THRESHOLD 100
-static long
+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)
@@ -3289,6 +3787,8 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
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);
@@ -3321,7 +3821,7 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
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(reg->enc, prev, end)) {
+ if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) {
p = prev;
goto retry;
}
@@ -3333,17 +3833,13 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
#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(reg->enc, prev, end)) {
+ 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(reg->enc, p, end)
-#ifdef USE_CRNL_AS_LINE_TERMINATOR
- && ! ONIGENC_IS_MBC_CRNL(reg->enc, p, end)
-#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;
@@ -3374,11 +3870,19 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
}
-extern long
+extern OnigPosition
onig_search(regex_t* reg, const UChar* str, const UChar* end,
const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option)
{
- long r;
+ 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)
+{
+ ptrdiff_t r;
UChar *s, *prev;
OnigMatchArg msa;
const UChar *orig_start = start;
@@ -3413,8 +3917,8 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
#ifdef ONIG_DEBUG_SEARCH
fprintf(stderr,
- "onig_search (entry point): str: %"PRIuPTR", end: %"PRIuPTR", start: %"PRIuPTR", range: %"PRIuPTR"\n",
- str, end - str, start - str, range - str);
+ "onig_search (entry point): str: %"PRIuPTR" (%p), end: %"PRIuPTR", start: %"PRIuPTR", range: %"PRIuPTR"\n",
+ str, str, end - str, start - str, range - str);
#endif
if (region
@@ -3546,6 +4050,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* 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;
}
@@ -3575,7 +4080,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
s = (UChar* )start;
prev = (UChar* )NULL;
- MATCH_ARG_INIT(msa, option, region, start);
+ MATCH_ARG_INIT(msa, option, region, start, start);
#ifdef USE_COMBINATION_EXPLOSION_CHECK
msa.state_check_buff = (void* )0;
msa.state_check_buff_size = 0; /* NO NEED, for valgrind */
@@ -3591,7 +4096,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
(int )(end - str), (int )(start - str), (int )(range - str));
#endif
- MATCH_ARG_INIT(msa, option, region, orig_start);
+ MATCH_ARG_INIT(msa, option, region, start, global_pos);
#ifdef USE_COMBINATION_EXPLOSION_CHECK
{
int offset = (MIN(start, range) - str);
@@ -3644,9 +4149,19 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
if ((reg->anchor & ANCHOR_ANYCHAR_STAR) != 0) {
do {
+ if ((reg->anchor & ANCHOR_BEGIN_POSITION) == 0)
+ msa.gpos = s; /* move \G position */
MATCH_AND_RETURN_CHECK(orig_range);
prev = s;
s += enclen(reg->enc, s, end);
+
+ if ((reg->anchor & ANCHOR_LOOK_BEHIND) == 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;
}
@@ -3666,7 +4181,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
else { /* backward search */
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
if (orig_start < end)
- orig_start += enclen(reg->enc, orig_start, end); /* is upper range */
+ orig_start += enclen(reg->enc, orig_start, end); /* is upper range */
#endif
if (reg->optimize != ONIG_OPTIMIZE_NONE) {