diff options
Diffstat (limited to 'regerror.c')
| -rw-r--r-- | regerror.c | 225 |
1 files changed, 122 insertions, 103 deletions
diff --git a/regerror.c b/regerror.c index 780ba94a0e..c61797564b 100644 --- a/regerror.c +++ b/regerror.c @@ -1,8 +1,9 @@ /********************************************************************** - regerror.c - Oniguruma (regular expression library) + regerror.c - Onigmo (Oniguruma-mod) (regular expression library) **********************************************************************/ /*- * Copyright (c) 2002-2007 K.Kosako <sndgk393 AT ybb DOT ne DOT jp> + * Copyright (c) 2011-2019 K.Takata <kentkt AT csc DOT jp> * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -30,16 +31,10 @@ #include "regint.h" #include <stdio.h> /* for vsnprintf() */ -#ifdef HAVE_STDARG_PROTOTYPES #include <stdarg.h> -#define va_init_list(a,b) va_start(a,b) -#else -#include <varargs.h> -#define va_init_list(a,b) va_start(a) -#endif extern UChar* -onig_error_code_to_format(int code) +onig_error_code_to_format(OnigPosition code) { const char *p; @@ -52,8 +47,6 @@ onig_error_code_to_format(int code) p = "no support in this configuration"; break; case ONIGERR_MEMORY: p = "failed to allocate memory"; break; - case ONIGERR_MATCH_STACK_LIMIT_OVER: - p = "match-stack limit over"; break; case ONIGERR_TYPE_BUG: p = "undefined type (bug)"; break; case ONIGERR_PARSER_BUG: @@ -64,16 +57,24 @@ onig_error_code_to_format(int code) p = "undefined bytecode (bug)"; break; case ONIGERR_UNEXPECTED_BYTECODE: p = "unexpected bytecode (bug)"; break; - case ONIGERR_DEFAULT_ENCODING_IS_NOT_SETTED: - p = "default multibyte-encoding is not setted"; break; + case ONIGERR_MATCH_STACK_LIMIT_OVER: + p = "match-stack limit over"; break; + case ONIGERR_PARSE_DEPTH_LIMIT_OVER: + p = "parse depth limit over"; break; + case ONIGERR_DEFAULT_ENCODING_IS_NOT_SET: + p = "default multibyte-encoding is not set"; break; +#if 0 case ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR: p = "can't convert to wide-char on specified multibyte-encoding"; break; +#endif case ONIGERR_INVALID_ARGUMENT: p = "invalid argument"; break; case ONIGERR_END_PATTERN_AT_LEFT_BRACE: p = "end pattern at left brace"; break; +#if 0 case ONIGERR_END_PATTERN_AT_LEFT_BRACKET: p = "end pattern at left bracket"; break; +#endif case ONIGERR_EMPTY_CHAR_CLASS: p = "empty char-class"; break; case ONIGERR_PREMATURE_END_OF_CHAR_CLASS: @@ -90,16 +91,20 @@ onig_error_code_to_format(int code) p = "invalid control-code syntax"; break; case ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE: p = "char-class value at end of range"; break; +#if 0 case ONIGERR_CHAR_CLASS_VALUE_AT_START_OF_RANGE: p = "char-class value at start of range"; break; +#endif case ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS: p = "unmatched range specifier in char-class"; break; case ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED: p = "target of repeat operator is not specified"; break; case ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID: p = "target of repeat operator is invalid"; break; +#if 0 case ONIGERR_NESTED_REPEAT_OPERATOR: p = "nested repeat operator"; break; +#endif case ONIGERR_UNMATCHED_CLOSE_PARENTHESIS: p = "unmatched close parenthesis"; break; case ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS: @@ -114,6 +119,8 @@ onig_error_code_to_format(int code) p = "invalid pattern in look-behind"; break; case ONIGERR_INVALID_REPEAT_RANGE_PATTERN: p = "invalid repeat range {lower,upper}"; break; + case ONIGERR_INVALID_CONDITION_PATTERN: + p = "invalid conditional pattern"; break; case ONIGERR_TOO_BIG_NUMBER: p = "too big number"; break; case ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE: @@ -122,14 +129,18 @@ onig_error_code_to_format(int code) p = "upper is smaller than lower in repeat range"; break; case ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS: p = "empty range in char class"; break; +#if 0 case ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE: p = "mismatch multibyte code length in char-class range"; break; +#endif case ONIGERR_TOO_MANY_MULTI_BYTE_RANGES: p = "too many multibyte code ranges are specified"; break; case ONIGERR_TOO_SHORT_MULTI_BYTE_STRING: p = "too short multibyte code string"; break; +#if 0 case ONIGERR_TOO_BIG_BACKREF_NUMBER: p = "too big backref number"; break; +#endif case ONIGERR_INVALID_BACKREF: #ifdef USE_NAMED_GROUP p = "invalid backref number/name"; break; @@ -138,12 +149,10 @@ onig_error_code_to_format(int code) #endif case ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED: p = "numbered backref/call is not allowed. (use name)"; break; - case ONIGERR_TOO_BIG_WIDE_CHAR_VALUE: - p = "too big wide-char value"; break; + case ONIGERR_TOO_SHORT_DIGITS: + p = "too short digits"; break; case ONIGERR_TOO_LONG_WIDE_CHAR_VALUE: p = "too long wide-char value"; break; - case ONIGERR_INVALID_CODE_POINT_VALUE: - p = "invalid code point value"; break; case ONIGERR_EMPTY_GROUP_NAME: p = "group name is empty"; break; case ONIGERR_INVALID_GROUP_NAME: @@ -164,16 +173,26 @@ onig_error_code_to_format(int code) p = "multiplex definition name <%n> call"; break; case ONIGERR_NEVER_ENDING_RECURSION: p = "never ending recursion"; break; +#ifdef USE_CAPTURE_HISTORY case ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY: p = "group number is too big for capture history"; break; +#endif case ONIGERR_INVALID_CHAR_PROPERTY_NAME: p = "invalid character property name {%n}"; break; + case ONIGERR_TOO_MANY_RANGE_REPEAT: + p = "too many range repeat"; break; + case ONIGERR_TOO_MANY_NULL_CHECK: + p = "too many null check"; break; + case ONIGERR_TOO_MANY_CAPTURE_GROUPS: + p = "too many capture groups are specified"; break; + case ONIGERR_INVALID_CODE_POINT_VALUE: + p = "invalid code point value"; break; + case ONIGERR_TOO_BIG_WIDE_CHAR_VALUE: + p = "too big wide-char value"; break; case ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION: p = "not supported encoding combination"; break; case ONIGERR_INVALID_COMBINATION_OF_OPTIONS: p = "invalid combination of options"; break; - case ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT: - p = "over thread pass limit count"; break; default: p = "undefined error code"; break; @@ -184,16 +203,16 @@ onig_error_code_to_format(int code) static void sprint_byte(char* s, unsigned int v) { - sprintf(s, "%02x", (v & 0377)); + xsnprintf(s, 3, "%02x", (v & 0377)); } static void sprint_byte_with_x(char* s, unsigned int v) { - sprintf(s, "\\x%02x", (v & 0377)); + xsnprintf(s, 5, "\\x%02x", (v & 0377)); } 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; @@ -205,24 +224,24 @@ 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); @@ -232,7 +251,7 @@ static int to_ascii(OnigEncoding enc, UChar *s, UChar *end, *is_over = ((p < end) ? 1 : 0); } else { - len = MIN((end - s), buf_size); + len = (int )MIN((end - s), buf_size); xmemcpy(buf, s, (size_t )len); *is_over = ((buf_size < (end - s)) ? 1 : 0); } @@ -241,26 +260,20 @@ static int to_ascii(OnigEncoding enc, UChar *s, UChar *end, } -/* for ONIG_MAX_ERROR_MESSAGE_LEN */ -#define MAX_ERROR_PAR_LEN 30 +/* < ONIG_MAX_ERROR_MESSAGE_LEN - max length of messages with %n */ +#define MAX_ERROR_PAR_LEN 50 extern int -#ifdef HAVE_STDARG_PROTOTYPES -onig_error_code_to_str(UChar* s, int code, ...) -#else -onig_error_code_to_str(s, code, va_alist) - UChar* s; - int code; - va_dcl -#endif +onig_error_code_to_str(UChar* s, OnigPosition code, ...) { UChar *p, *q; OnigErrorInfo* einfo; - int len, is_over; + size_t len; + int is_over; UChar parbuf[MAX_ERROR_PAR_LEN]; va_list vargs; - va_init_list(vargs, code); + va_start(vargs, code); switch (code) { case ONIGERR_UNDEFINED_NAME_REFERENCE: @@ -272,27 +285,27 @@ onig_error_code_to_str(s, code, va_alist) 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'; @@ -301,64 +314,47 @@ onig_error_code_to_str(s, code, va_alist) default: q = onig_error_code_to_format(code); - len = onigenc_str_bytelen_null(ONIG_ENCODING_ASCII, q); - xmemcpy(s, q, len); + if (q) { + len = onigenc_str_bytelen_null(ONIG_ENCODING_ASCII, q); + xmemcpy(s, q, len); + } + else { + len = 0; + } s[len] = '\0'; break; } va_end(vargs); - return len; + return (int )len; } - void -#ifdef HAVE_STDARG_PROTOTYPES -onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, - UChar* pat, UChar* pat_end, const UChar *fmt, ...) -#else -onig_snprintf_with_pattern(buf, bufsize, enc, pat, pat_end, fmt, va_alist) - UChar buf[]; - int bufsize; - OnigEncoding enc; - UChar* pat; - UChar* pat_end; - const UChar *fmt; - va_dcl -#endif +onig_vsnprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, + UChar* pat, UChar* pat_end, const char *fmt, va_list args) { - int n, need, len; + size_t need; + int n, len; UChar *p, *s, *bp; UChar bs[6]; - va_list args; - va_init_list(args, fmt); n = xvsnprintf((char* )buf, bufsize, (const char* )fmt, args); - va_end(args); need = (pat_end - pat) * 4 + 4; - if (n + need < bufsize) { - strcat((char* )buf, ": /"); + if (n + need < (size_t )bufsize) { + static const char sep[] = ": /"; + memcpy((char* )buf + n, sep, sizeof(sep)); s = buf + onigenc_str_bytelen_null(ONIG_ENCODING_ASCII, buf); p = pat; while (p < pat_end) { - if (*p == '\\') { - *s++ = *p++; - len = enclen(enc, p, pat_end); - while (len-- > 0) *s++ = *p++; - } - else if (*p == '/') { - *s++ = (unsigned char )'\\'; - *s++ = *p++; - } - else if (ONIGENC_IS_MBC_HEAD(enc, p, pat_end)) { + if (ONIGENC_IS_MBC_HEAD(enc, p, pat_end)) { len = enclen(enc, p, pat_end); if (ONIGENC_MBC_MINLEN(enc) == 1) { while (len-- > 0) *s++ = *p++; } - else { /* for UTF16 */ + else { /* for UTF16/32 */ int blen; while (len-- > 0) { @@ -369,15 +365,25 @@ onig_snprintf_with_pattern(buf, bufsize, enc, pat, pat_end, fmt, va_alist) } } } + else if (*p == '\\') { + *s++ = *p++; + len = enclen(enc, p, pat_end); + while (len-- > 0) *s++ = *p++; + } + else if (*p == '/') { + *s++ = (unsigned char )'\\'; + *s++ = *p++; + } else if (!ONIGENC_IS_CODE_PRINT(enc, *p) && - !ONIGENC_IS_CODE_SPACE(enc, *p)) { - sprint_byte_with_x((char* )bs, (unsigned int )(*p++)); - len = onigenc_str_bytelen_null(ONIG_ENCODING_ASCII, bs); + (!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); bp = bs; - while (len-- > 0) *s++ = *bp++; + while (len-- > 0) *s++ = *bp++; } else { - *s++ = *p++; + *s++ = *p++; } } @@ -385,3 +391,16 @@ onig_snprintf_with_pattern(buf, bufsize, enc, pat, pat_end, fmt, va_alist) *s = '\0'; } } + +#if 0 /* unused */ +void +onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, + UChar* pat, UChar* pat_end, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + onig_vsnprintf_with_pattern(buf, bufsize, enc, + pat, pat_end, fmt, args); + va_end(args); +} +#endif |