From 38b570f99d6ce5b4f626a2cad431f27b03ec118e Mon Sep 17 00:00:00 2001 From: "Philip.Hazel" Date: Sat, 28 Jun 2014 14:23:18 +0000 Subject: [PATCH] The pcre2_match() function's basic facilities are working (though hardly tested). --- src/pcre2.h | 19 +- src/pcre2.h.in | 19 +- src/pcre2_auto_possess.c | 40 +- src/pcre2_compile.c | 751 ++--- src/pcre2_context.c | 4 +- src/pcre2_error.c | 2 - src/pcre2_internal.h | 18 +- src/pcre2_intmodedep.h | 149 +- src/pcre2_match.c | 6901 +++++++++++++++++++++++++++++++++++++- src/pcre2_newline.c | 106 +- src/pcre2_pattern_info.c | 2 - src/pcre2test.c | 8 +- 12 files changed, 7458 insertions(+), 561 deletions(-) diff --git a/src/pcre2.h b/src/pcre2.h index c810a9f..5b31de7 100644 --- a/src/pcre2.h +++ b/src/pcre2.h @@ -215,7 +215,6 @@ must all be greater than zero. */ #define PCRE2_ERROR_NULL (-50) #define PCRE2_ERROR_RECURSELOOP (-51) #define PCRE2_ERROR_RECURSIONLIMIT (-52) -#define PCRE2_ERROR_UNSET (-53) /* Request types for pcre2_pattern_info() */ @@ -309,17 +308,17 @@ typedef struct pcre2_callout_block { \ int version; /* Identifies version of block */ \ /* ------------------------ Version 0 ------------------------------- */ \ int callout_number; /* Number compiled into pattern */ \ - int *offset_vector; /* The offset vector */ \ + size_t *offset_vector; /* The offset vector */ \ PCRE2_SPTR subject; /* The subject being matched */ \ - int subject_length; /* The length of the subject */ \ - int start_match; /* Offset to start of this match attempt */ \ - int current_position; /* Where we currently are in the subject */ \ - int capture_top; /* Max current capture */ \ - int capture_last; /* Most recently closed capture */ \ + size_t subject_length; /* The length of the subject */ \ + size_t start_match; /* Offset to start of this match attempt */ \ + size_t current_position; /* Where we currently are in the subject */ \ + uint32_t capture_top; /* Max current capture */ \ + uint32_t capture_last; /* Most recently closed capture */ \ void *callout_data; /* Data passed in with the call */ \ /* ------------------- Added for Version 1 -------------------------- */ \ - int pattern_position; /* Offset to next item in the pattern */ \ - int next_item_length; /* Length of next item in the pattern */ \ + size_t pattern_position; /* Offset to next item in the pattern */ \ + size_t next_item_length; /* Length of next item in the pattern */ \ /* ------------------- Added for Version 2 -------------------------- */ \ PCRE2_SPTR mark; /* Pointer to current mark or NULL */ \ /* ------------------------------------------------------------------ */ \ @@ -381,7 +380,7 @@ PCRE2_EXP_DECL void pcre2_match_context_free(pcre2_match_context *); \ PCRE2_EXP_DECL int pcre2_set_bsr_match(pcre2_match_context *, \ uint32_t); \ PCRE2_EXP_DECL int pcre2_set_callout(pcre2_match_context *, \ - int (*)(pcre2_callout_block *, void *)); \ + int (*)(pcre2_callout_block *), void *); \ PCRE2_EXP_DECL int pcre2_set_match_limit(pcre2_match_context *, \ uint32_t); \ PCRE2_EXP_DECL int pcre2_set_newline_match(pcre2_match_context *, \ diff --git a/src/pcre2.h.in b/src/pcre2.h.in index 5d964d7..80b4a88 100644 --- a/src/pcre2.h.in +++ b/src/pcre2.h.in @@ -215,7 +215,6 @@ must all be greater than zero. */ #define PCRE2_ERROR_NULL (-50) #define PCRE2_ERROR_RECURSELOOP (-51) #define PCRE2_ERROR_RECURSIONLIMIT (-52) -#define PCRE2_ERROR_UNSET (-53) /* Request types for pcre2_pattern_info() */ @@ -309,17 +308,17 @@ typedef struct pcre2_callout_block { \ int version; /* Identifies version of block */ \ /* ------------------------ Version 0 ------------------------------- */ \ int callout_number; /* Number compiled into pattern */ \ - int *offset_vector; /* The offset vector */ \ + size_t *offset_vector; /* The offset vector */ \ PCRE2_SPTR subject; /* The subject being matched */ \ - int subject_length; /* The length of the subject */ \ - int start_match; /* Offset to start of this match attempt */ \ - int current_position; /* Where we currently are in the subject */ \ - int capture_top; /* Max current capture */ \ - int capture_last; /* Most recently closed capture */ \ + size_t subject_length; /* The length of the subject */ \ + size_t start_match; /* Offset to start of this match attempt */ \ + size_t current_position; /* Where we currently are in the subject */ \ + uint32_t capture_top; /* Max current capture */ \ + uint32_t capture_last; /* Most recently closed capture */ \ void *callout_data; /* Data passed in with the call */ \ /* ------------------- Added for Version 1 -------------------------- */ \ - int pattern_position; /* Offset to next item in the pattern */ \ - int next_item_length; /* Length of next item in the pattern */ \ + size_t pattern_position; /* Offset to next item in the pattern */ \ + size_t next_item_length; /* Length of next item in the pattern */ \ /* ------------------- Added for Version 2 -------------------------- */ \ PCRE2_SPTR mark; /* Pointer to current mark or NULL */ \ /* ------------------------------------------------------------------ */ \ @@ -381,7 +380,7 @@ PCRE2_EXP_DECL void pcre2_match_context_free(pcre2_match_context *); \ PCRE2_EXP_DECL int pcre2_set_bsr_match(pcre2_match_context *, \ uint32_t); \ PCRE2_EXP_DECL int pcre2_set_callout(pcre2_match_context *, \ - int (*)(pcre2_callout_block *, void *)); \ + int (*)(pcre2_callout_block *), void *); \ PCRE2_EXP_DECL int pcre2_set_match_limit(pcre2_match_context *, \ uint32_t); \ PCRE2_EXP_DECL int pcre2_set_newline_match(pcre2_match_context *, \ diff --git a/src/pcre2_auto_possess.c b/src/pcre2_auto_possess.c index 3dbde60..2aa113f 100644 --- a/src/pcre2_auto_possess.c +++ b/src/pcre2_auto_possess.c @@ -559,14 +559,14 @@ which case the base cannot be possessified. Arguments: code points to the byte code utf TRUE in UTF mode - cd compile data block + cb compile data block base_list the data list of the base opcode Returns: TRUE if the auto-possessification is possible */ static BOOL -compare_opcodes(PCRE2_SPTR code, BOOL utf, const compile_data *cd, +compare_opcodes(PCRE2_SPTR code, BOOL utf, const compile_block *cb, const uint32_t *base_list, PCRE2_SPTR base_end) { PCRE2_UCHAR c; @@ -654,7 +654,7 @@ for(;;) while (*next_code == OP_ALT) { - if (!compare_opcodes(code, utf, cd, base_list, base_end)) return FALSE; + if (!compare_opcodes(code, utf, cb, base_list, base_end)) return FALSE; code = next_code + 1 + LINK_SIZE; next_code += GET(next_code, 1); } @@ -674,7 +674,7 @@ for(;;) /* The bracket content will be checked by the OP_BRA/OP_CBRA case above. */ next_code += 1 + LINK_SIZE; - if (!compare_opcodes(next_code, utf, cd, base_list, base_end)) + if (!compare_opcodes(next_code, utf, cb, base_list, base_end)) return FALSE; code += PRIV(OP_lengths)[c]; @@ -686,7 +686,7 @@ for(;;) /* Check for a supported opcode, and load its properties. */ - code = get_chr_property_list(code, utf, cd->fcc, list); + code = get_chr_property_list(code, utf, cb->fcc, list); if (code == NULL) return FALSE; /* Unsupported */ /* If either opcode is a small character list, set pointers for comparing @@ -755,21 +755,21 @@ for(;;) invert_bits = TRUE; /* Fall through */ case OP_DIGIT: - set2 = (uint8_t *)(cd->cbits + cbit_digit); + set2 = (uint8_t *)(cb->cbits + cbit_digit); break; case OP_NOT_WHITESPACE: invert_bits = TRUE; /* Fall through */ case OP_WHITESPACE: - set2 = (uint8_t *)(cd->cbits + cbit_space); + set2 = (uint8_t *)(cb->cbits + cbit_space); break; case OP_NOT_WORDCHAR: invert_bits = TRUE; /* Fall through */ case OP_WORDCHAR: - set2 = (uint8_t *)(cd->cbits + cbit_word); + set2 = (uint8_t *)(cb->cbits + cbit_word); break; default: @@ -963,27 +963,27 @@ for(;;) set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */ case OP_DIGIT: - if (chr < 256 && (cd->ctypes[chr] & ctype_digit) != 0) return FALSE; + if (chr < 256 && (cb->ctypes[chr] & ctype_digit) != 0) return FALSE; break; case OP_NOT_DIGIT: - if (chr > 255 || (cd->ctypes[chr] & ctype_digit) == 0) return FALSE; + if (chr > 255 || (cb->ctypes[chr] & ctype_digit) == 0) return FALSE; break; case OP_WHITESPACE: - if (chr < 256 && (cd->ctypes[chr] & ctype_space) != 0) return FALSE; + if (chr < 256 && (cb->ctypes[chr] & ctype_space) != 0) return FALSE; break; case OP_NOT_WHITESPACE: - if (chr > 255 || (cd->ctypes[chr] & ctype_space) == 0) return FALSE; + if (chr > 255 || (cb->ctypes[chr] & ctype_space) == 0) return FALSE; break; case OP_WORDCHAR: - if (chr < 255 && (cd->ctypes[chr] & ctype_word) != 0) return FALSE; + if (chr < 255 && (cb->ctypes[chr] & ctype_word) != 0) return FALSE; break; case OP_NOT_WORDCHAR: - if (chr > 255 || (cd->ctypes[chr] & ctype_word) == 0) return FALSE; + if (chr > 255 || (cb->ctypes[chr] & ctype_word) == 0) return FALSE; break; case OP_HSPACE: @@ -1095,13 +1095,13 @@ if appropriate. This function modifies the compiled opcode! Arguments: code points to start of the byte code utf TRUE in UTF mode - cd compile data block + cb compile data block Returns: nothing */ void -PRIV(auto_possessify)(PCRE2_UCHAR *code, BOOL utf, const compile_data *cd) +PRIV(auto_possessify)(PCRE2_UCHAR *code, BOOL utf, const compile_block *cb) { register PCRE2_UCHAR c; PCRE2_SPTR end; @@ -1116,10 +1116,10 @@ for (;;) { c -= get_repeat_base(c) - OP_STAR; end = (c <= OP_MINUPTO) ? - get_chr_property_list(code, utf, cd->fcc, list) : NULL; + get_chr_property_list(code, utf, cb->fcc, list) : NULL; list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO; - if (end != NULL && compare_opcodes(end, utf, cd, list, end)) + if (end != NULL && compare_opcodes(end, utf, cb, list, end)) { switch(c) { @@ -1171,11 +1171,11 @@ for (;;) if (c >= OP_CRSTAR && c <= OP_CRMINRANGE) { /* end must not be NULL. */ - end = get_chr_property_list(code, utf, cd->fcc, list); + end = get_chr_property_list(code, utf, cb->fcc, list); list[1] = (c & 1) == 0; - if (compare_opcodes(end, utf, cd, list, end)) + if (compare_opcodes(end, utf, cb, list, end)) { switch (c) { diff --git a/src/pcre2_compile.c b/src/pcre2_compile.c index 445075c..d196935 100644 --- a/src/pcre2_compile.c +++ b/src/pcre2_compile.c @@ -43,7 +43,7 @@ POSSIBILITY OF SUCH DAMAGE. #include "config.h" #endif -#define NLBLOCK cd /* Block containing newline information */ +#define NLBLOCK cb /* Block containing newline information */ #define PSSTART start_pattern /* Field containing processed string start */ #define PSEND end_pattern /* Field containing processed string end */ @@ -72,13 +72,13 @@ by defining macros in order to minimize #if usage. */ /* Function definitions to allow mutual recursion */ static int - add_list_to_class(uint8_t *, PCRE2_UCHAR **, int, compile_data *, + add_list_to_class(uint8_t *, PCRE2_UCHAR **, int, compile_block *, const uint32_t *, unsigned int); static BOOL compile_regex(uint32_t, PCRE2_UCHAR **, PCRE2_SPTR *, int *, BOOL, BOOL, int, int, uint32_t *, int32_t *, uint32_t *, int32_t *, branch_chain *, - compile_data *, size_t *); + compile_block *, size_t *); @@ -684,17 +684,17 @@ callout points before each pattern item. Arguments: code current code pointer ptr current pattern pointer - cd general compile-time data + cb general compile-time data Returns: new code pointer */ static PCRE2_UCHAR * -auto_callout(PCRE2_UCHAR *code, PCRE2_SPTR ptr, compile_data *cd) +auto_callout(PCRE2_UCHAR *code, PCRE2_SPTR ptr, compile_block *cb) { *code++ = OP_CALLOUT; *code++ = 255; -PUT(code, 0, ptr - cd->start_pattern); /* Pattern offset */ +PUT(code, 0, ptr - cb->start_pattern); /* Pattern offset */ PUT(code, LINK_SIZE, 0); /* Default length */ return code + 2 * LINK_SIZE; } @@ -712,16 +712,16 @@ for both automatic and manual callouts. Arguments: previous_callout points to previous callout item ptr current pattern pointer - cd general compile-time data + cb general compile-time data Returns: nothing */ static void complete_callout(PCRE2_UCHAR *previous_callout, PCRE2_SPTR ptr, - compile_data *cd) + compile_block *cb) { -size_t length = ptr - cd->start_pattern - GET(previous_callout, 2); +size_t length = ptr - cb->start_pattern - GET(previous_callout, 2); PUT(previous_callout, 2 + LINK_SIZE, length); } @@ -746,7 +746,7 @@ Arguments: code points to the start of the pattern (the bracket) utf TRUE in UTF mode atend TRUE if called when the pattern is complete - cd the "compile data" structure + cb the "compile data" structure Returns: the fixed length, or -1 if there is no fixed length, @@ -756,7 +756,7 @@ Returns: the fixed length, */ static int -find_fixedlength(PCRE2_UCHAR *code, BOOL utf, BOOL atend, compile_data *cd) +find_fixedlength(PCRE2_UCHAR *code, BOOL utf, BOOL atend, compile_block *cb) { int length = -1; @@ -784,7 +784,7 @@ for (;;) case OP_ONCE: case OP_ONCE_NC: case OP_COND: - d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd); + d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cb); if (d < 0) return d; branchlength += d; do cc += GET(cc, 1); while (*cc == OP_ALT); @@ -815,10 +815,10 @@ for (;;) case OP_RECURSE: if (!atend) return -3; - cs = ce = (PCRE2_UCHAR *)cd->start_code + GET(cc, 1); /* Start subpattern */ + cs = ce = (PCRE2_UCHAR *)cb->start_code + GET(cc, 1); /* Start subpattern */ do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ if (cc > cs && cc < ce) return -1; /* Recursion */ - d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd); + d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cb); if (d < 0) return d; branchlength += d; cc += 1 + LINK_SIZE; @@ -1142,7 +1142,7 @@ Arguments: code points to start of search endcode points to where to stop utf TRUE if in UTF mode - cd compile data + cb compile data recurses chain of recurse_check to catch mutual recursion Returns: TRUE if what is matched could be empty @@ -1155,7 +1155,7 @@ typedef struct recurse_check { static BOOL could_be_empty_branch(PCRE2_SPTR code, PCRE2_SPTR endcode, BOOL utf, - compile_data *cd, recurse_check *recurses) + compile_block *cb, recurse_check *recurses) { register PCRE2_UCHAR c; recurse_check this_recurse; @@ -1187,18 +1187,18 @@ for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE); if (c == OP_RECURSE) { - PCRE2_SPTR scode = cd->start_code + GET(code, 1); + PCRE2_SPTR scode = cb->start_code + GET(code, 1); BOOL empty_branch; /* Test for forward reference or uncompleted reference. This is disabled - when called to scan a completed pattern by setting cd->start_workspace to + when called to scan a completed pattern by setting cb->start_workspace to NULL. */ - if (cd->start_workspace != NULL) + if (cb->start_workspace != NULL) { PCRE2_SPTR tcode; - for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE) - if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE; + for (tcode = cb->start_workspace; tcode < cb->hwm; tcode += LINK_SIZE) + if ((int)GET(tcode, 0) == (int)(code + 1 - cb->start_code)) return TRUE; if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ } @@ -1230,7 +1230,7 @@ for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE); do { - if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse)) + if (could_be_empty_branch(scode, endcode, utf, cb, &this_recurse)) { empty_branch = TRUE; break; @@ -1286,7 +1286,7 @@ for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE); empty_branch = FALSE; do { - if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd, NULL)) + if (!empty_branch && could_be_empty_branch(code, endcode, utf, cb, NULL)) empty_branch = TRUE; code += GET(code, 1); } @@ -1526,18 +1526,18 @@ Arguments: endcode points to where to stop (current RECURSE item) bcptr points to the chain of current (unclosed) branch starts utf TRUE if in UTF mode - cd compile data + cb compile data Returns: TRUE if what is matched could be empty */ static BOOL could_be_empty(PCRE2_SPTR code, PCRE2_SPTR endcode, branch_chain *bcptr, - BOOL utf, compile_data *cd) + BOOL utf, compile_block *cb) { while (bcptr != NULL && bcptr->current_branch >= code) { - if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd, NULL)) + if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cb, NULL)) return FALSE; bcptr = bcptr->outer; } @@ -1560,22 +1560,22 @@ Returns: 0 if all went well, else an error number */ static int -expand_workspace(compile_data *cd) +expand_workspace(compile_block *cb) { PCRE2_UCHAR *newspace; -int newsize = cd->workspace_size * 2; +int newsize = cb->workspace_size * 2; if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX; -if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX || - newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN) +if (cb->workspace_size >= COMPILE_WORK_SIZE_MAX || + newsize - cb->workspace_size < WORK_SIZE_SAFETY_MARGIN) return ERR72; -newspace = cd->cx->memctl.malloc(CU2BYTES(newsize), cd->cx->memctl.memory_data); +newspace = cb->cx->memctl.malloc(CU2BYTES(newsize), cb->cx->memctl.memory_data); if (newspace == NULL) return ERR21; -memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(PCRE2_UCHAR)); -cd->hwm = (PCRE2_UCHAR *)newspace + (cd->hwm - cd->start_workspace); -if (cd->workspace_size > COMPILE_WORK_SIZE) - cd->cx->memctl.free((void *)cd->start_workspace, cd->cx->memctl.memory_data); -cd->start_workspace = newspace; -cd->workspace_size = newsize; +memcpy(newspace, cb->start_workspace, cb->workspace_size * sizeof(PCRE2_UCHAR)); +cb->hwm = (PCRE2_UCHAR *)newspace + (cb->hwm - cb->start_workspace); +if (cb->workspace_size > COMPILE_WORK_SIZE) + cb->cx->memctl.free((void *)cb->start_workspace, cb->cx->memctl.memory_data); +cb->start_workspace = newspace; +cb->workspace_size = newsize; return 0; } @@ -1629,7 +1629,7 @@ Arguments: errorcodeptr points to the errorcode variable (containing zero) options the current options bits isclass TRUE if inside a character class - cd compile data block + cb compile data block Returns: zero => a data character positive => a special escape sequence @@ -1639,7 +1639,7 @@ Returns: zero => a data character static int check_escape(PCRE2_SPTR *ptrptr, uint32_t *chptr, int *errorcodeptr, - int options, BOOL isclass, compile_data *cd) + int options, BOOL isclass, compile_block *cb) { BOOL utf = (options & PCRE2_UTF) != 0; PCRE2_SPTR ptr = *ptrptr + 1; @@ -1652,7 +1652,7 @@ ptr--; /* Set pointer back to the last code unit */ /* If backslash is at the end of the pattern, it's an error. */ -if (c == CHAR_NULL && ptr >= cd->end_pattern) *errorcodeptr = ERR1; +if (c == CHAR_NULL && ptr >= cb->end_pattern) *errorcodeptr = ERR1; /* Non-alphanumerics are literals, so we just leave the value in c. An initial value test saves a memory lookup for code points outside the alphanumeric @@ -1798,12 +1798,12 @@ else if (negated) { - if (s > cd->bracount) + if (s > cb->bracount) { *errorcodeptr = ERR15; break; } - s = cd->bracount - (s - 1); + s = cb->bracount - (s - 1); } escape = -s; @@ -1849,7 +1849,7 @@ else *errorcodeptr = ERR61; break; } - if (s < 8 || s <= cd->bracount) /* Check for back reference */ + if (s < 8 || s <= cb->bracount) /* Check for back reference */ { escape = -s; break; @@ -2008,7 +2008,7 @@ else case CHAR_c: c = *(++ptr); - if (c == CHAR_NULL && ptr >= cd->end_pattern) + if (c == CHAR_NULL && ptr >= cb->end_pattern) { *errorcodeptr = ERR2; break; @@ -2074,14 +2074,14 @@ Arguments: ptypeptr an unsigned int that is set to the type value pdataptr an unsigned int that is set to the detailed property value errorcodeptr the error code variable - cd the compile data + cb the compile data Returns: TRUE if the type value was found, or FALSE for an invalid type */ static BOOL get_ucp(PCRE2_SPTR *ptrptr, BOOL *negptr, unsigned int *ptypeptr, - unsigned int *pdataptr, int *errorcodeptr, compile_data *cd) + unsigned int *pdataptr, int *errorcodeptr, compile_block *cb) { register PCRE2_UCHAR c; int i, bot, top; @@ -2115,7 +2115,7 @@ if (c == CHAR_LEFT_CURLY_BRACKET) /* Otherwise there is just one following character, which must be an ASCII letter. */ -else if (MAX_255(c) && (cd->ctypes[c] & ctype_letter) != 0) +else if (MAX_255(c) && (cb->ctypes[c] & ctype_letter) != 0) { name[0] = c; name[1] = 0; @@ -2527,14 +2527,14 @@ Arguments: group points to the start of the group adjust the amount by which the group is to be moved utf TRUE in UTF mode - cd compile data + cb compile data save_hwm the hwm forward reference pointer at the start of the group Returns: nothing */ static void -adjust_recurse(PCRE2_UCHAR *group, int adjust, BOOL utf, compile_data *cd, +adjust_recurse(PCRE2_UCHAR *group, int adjust, BOOL utf, compile_block *cb, PCRE2_UCHAR *save_hwm) { PCRE2_UCHAR *ptr = group; @@ -2547,10 +2547,10 @@ while ((ptr = (PCRE2_UCHAR *)find_recurse(ptr, utf)) != NULL) /* See if this recursion is on the forward reference list. If so, adjust the reference. */ - for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE) + for (hc = save_hwm; hc < cb->hwm; hc += LINK_SIZE) { offset = (int)GET(hc, 0); - if (cd->start_code + offset == ptr + 1) + if (cb->start_code + offset == ptr + 1) { PUT(hc, 0, offset + adjust); break; @@ -2560,10 +2560,10 @@ while ((ptr = (PCRE2_UCHAR *)find_recurse(ptr, utf)) != NULL) /* Otherwise, adjust the recursion offset if it's after the start of this group. */ - if (hc >= cd->hwm) + if (hc >= cb->hwm) { offset = (int)GET(ptr, 1); - if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust); + if (cb->start_code + offset >= group) PUT(ptr, 1, offset + adjust); } ptr += 1 + LINK_SIZE; @@ -2753,7 +2753,7 @@ Arguments: classbits the bit map for characters < 256 uchardptr points to the pointer for extra data options the options word - cd compile data + cb compile data start start of range character end end of range character @@ -2763,7 +2763,7 @@ Returns: the number of < 256 characters added static int add_to_class(uint8_t *classbits, PCRE2_UCHAR **uchardptr, int options, - compile_data *cd, uint32_t start, uint32_t end) + compile_block *cb, uint32_t start, uint32_t end) { uint32_t c; uint32_t classbits_end = (end <= 0xff ? end : 0xff); @@ -2789,7 +2789,7 @@ if ((options & PCRE2_CASELESS) != 0) { /* Handle a single character that has more than one other case. */ - if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd, + if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cb, PRIV(ucd_caseless_sets) + rc, oc); /* Do nothing if the other case range is within the original range. */ @@ -2802,7 +2802,7 @@ if ((options & PCRE2_CASELESS) != 0) else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */ else if (od > end && oc <= end + 1) end = od; /* Extend upwards */ - else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od); + else n8 += add_to_class(classbits, uchardptr, options, cb, oc, od); } } else @@ -2812,7 +2812,7 @@ if ((options & PCRE2_CASELESS) != 0) for (c = start; c <= classbits_end; c++) { - SETBIT(classbits, cd->fcc[c]); + SETBIT(classbits, cb->fcc[c]); n8++; } } @@ -2900,7 +2900,7 @@ Arguments: classbits the bit map for characters < 256 uchardptr points to the pointer for extra data options the options word - cd contains pointers to tables etc. + cb contains pointers to tables etc. p points to row of 32-bit values, terminated by NOTACHAR except character to omit; this is used when adding lists of case-equivalent characters to avoid including the one we @@ -2912,7 +2912,7 @@ Returns: the number of < 256 characters added static int add_list_to_class(uint8_t *classbits, PCRE2_UCHAR **uchardptr, int options, - compile_data *cd, const uint32_t *p, unsigned int except) + compile_block *cb, const uint32_t *p, unsigned int except) { int n8 = 0; while (p[0] < NOTACHAR) @@ -2921,7 +2921,7 @@ while (p[0] < NOTACHAR) if (p[0] != except) { while(p[n+1] == p[0] + n + 1) n++; - n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]); + n8 += add_to_class(classbits, uchardptr, options, cb, p[0], p[n]); } p += n + 1; } @@ -2941,7 +2941,7 @@ Arguments: classbits the bit map for characters < 256 uchardptr points to the pointer for extra data options the options word - cd contains pointers to tables etc. + cb contains pointers to tables etc. p points to row of 32-bit values, terminated by NOTACHAR Returns: the number of < 256 characters added @@ -2950,16 +2950,16 @@ Returns: the number of < 256 characters added static int add_not_list_to_class(uint8_t *classbits, PCRE2_UCHAR **uchardptr, - int options, compile_data *cd, const uint32_t *p) + int options, compile_block *cb, const uint32_t *p) { BOOL utf = (options & PCRE2_UTF) != 0; int n8 = 0; if (p[0] > 0) - n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1); + n8 += add_to_class(classbits, uchardptr, options, cb, 0, p[0] - 1); while (p[0] < NOTACHAR) { while (p[1] == p[0] + 1) p++; - n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1, + n8 += add_to_class(classbits, uchardptr, options, cb, p[0] + 1, (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1); p++; } @@ -2989,7 +2989,7 @@ Arguments: reqcuflagsptr place to put the last required code unit flags, or a negative number bcptr points to current branch chain cond_depth conditional nesting depth - cd contains pointers to tables etc. + cb contains pointers to tables etc. lengthptr NULL during the real compile phase points to length accumulator during pre-compile phase @@ -3003,7 +3003,7 @@ compile_branch(uint32_t *optionsptr, PCRE2_UCHAR **codeptr, uint32_t *firstcuptr, int32_t *firstcuflagsptr, uint32_t *reqcuptr, int32_t *reqcuflagsptr, branch_chain *bcptr, int cond_depth, - compile_data *cd, size_t *lengthptr) + compile_block *cb, size_t *lengthptr) { int repeat_type, op_type; int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ @@ -3100,7 +3100,7 @@ for (;; ptr++) BOOL xclass_has_prop; #endif int newoptions; - int recno; + uint32_t recno; int refsign; int skipbytes; uint32_t subreqcu, subfirstcu; @@ -3130,7 +3130,7 @@ for (;; ptr++) if (lengthptr != NULL) { - if (code > cd->start_workspace + cd->workspace_size - + if (code > cb->start_workspace + cb->workspace_size - WORK_SIZE_SAFETY_MARGIN) /* Check for overrun */ { *errorcodeptr = ERR52; @@ -3178,7 +3178,7 @@ for (;; ptr++) /* In the real compile phase, just check the workspace used by the forward reference list. */ - else if (cd->hwm > cd->start_workspace + cd->workspace_size - + else if (cb->hwm > cb->start_workspace + cb->workspace_size - WORK_SIZE_SAFETY_MARGIN) { *errorcodeptr = ERR52; @@ -3187,7 +3187,7 @@ for (;; ptr++) /* If in \Q...\E, check for the end; if not, we have a literal */ - if (inescq && (c != CHAR_NULL || ptr < cd->end_pattern)) + if (inescq && (c != CHAR_NULL || ptr < cb->end_pattern)) { if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E) { @@ -3200,13 +3200,13 @@ for (;; ptr++) if (previous_callout != NULL) { if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ - complete_callout(previous_callout, ptr, cd); + complete_callout(previous_callout, ptr, cb); previous_callout = NULL; } if ((options & PCRE2_AUTO_CALLOUT) != 0) { previous_callout = code; - code = auto_callout(code, ptr, cd); + code = auto_callout(code, ptr, cb); } goto NORMAL_CHAR; } @@ -3220,14 +3220,14 @@ for (;; ptr++) { for (;;) { - while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr); + while (MAX_255(c) && (cb->ctypes[c] & ctype_space) != 0) c = *(++ptr); if (c != CHAR_NUMBER_SIGN) break; ptr++; while (*ptr != CHAR_NULL) { if (IS_NEWLINE(ptr)) /* For non-fixed-length newline cases, */ - { /* IS_NEWLINE sets cd->nllen. */ - ptr += cd->nllen; + { /* IS_NEWLINE sets cb->nllen. */ + ptr += cb->nllen; break; } ptr++; @@ -3252,7 +3252,7 @@ for (;; ptr++) after_manual_callout-- <= 0) { if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ - complete_callout(previous_callout, ptr, cd); + complete_callout(previous_callout, ptr, cb); previous_callout = NULL; } @@ -3262,7 +3262,7 @@ for (;; ptr++) if ((options & PCRE2_AUTO_CALLOUT) != 0 && !is_quantifier && nestptr == NULL) { previous_callout = code; - code = auto_callout(code, ptr, cd); + code = auto_callout(code, ptr, cb); } /* Process the next pattern item. */ @@ -3273,7 +3273,7 @@ for (;; ptr++) /* The branch terminates at string end or | or ) */ case CHAR_NULL: - if (ptr < cd->end_pattern) goto NORMAL_CHAR; /* Zero data character */ + if (ptr < cb->end_pattern) goto NORMAL_CHAR; /* Zero data character */ /* Fall through */ case CHAR_VERTICAL_LINE: @@ -3412,7 +3412,7 @@ for (;; ptr++) OP_FAIL, whereas [^] must match any character, so generate OP_ALLANY. */ if (c == CHAR_RIGHT_SQUARE_BRACKET && - (cd->external_options & PCRE2_ALLOW_EMPTY_CLASS) != 0) + (cb->external_options & PCRE2_ALLOW_EMPTY_CLASS) != 0) { *code++ = negate_class? OP_ALLANY : OP_FAIL; if (firstcuflags == REQ_UNSET) firstcuflags = REQ_NONE; @@ -3464,7 +3464,7 @@ for (;; ptr++) { PCRE2_SPTR oldptr; - if (c == CHAR_NULL && ptr >= cd->end_pattern) + if (c == CHAR_NULL && ptr >= cb->end_pattern) { *errorcodeptr = ERR6; /* Missing terminating ']' */ goto FAILED; @@ -3516,7 +3516,7 @@ for (;; ptr++) { BOOL local_negate = FALSE; int posix_class, taboffset, tabopt; - register const uint8_t *cbits = cd->cbits; + register const uint8_t *cbits = cb->cbits; uint8_t pbits[32]; if (ptr[1] != CHAR_COLON) @@ -3655,7 +3655,7 @@ for (;; ptr++) if (c == CHAR_BACKSLASH) { - escape = check_escape(&ptr, &ec, errorcodeptr, options, TRUE, cd); + escape = check_escape(&ptr, &ec, errorcodeptr, options, TRUE, cb); if (*errorcodeptr != 0) goto FAILED; if (escape == 0) c = ec; /* Escaped single char */ else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */ @@ -3677,7 +3677,7 @@ for (;; ptr++) else /* Handle \d-type escapes */ { - register const uint8_t *cbits = cd->cbits; + register const uint8_t *cbits = cb->cbits; /* Every class contains at least two < 256 characters. */ class_has_8bitchar++; /* Every class contains at least two characters. */ @@ -3734,23 +3734,23 @@ for (;; ptr++) /* The rest apply in both UCP and non-UCP cases. */ case ESC_h: - (void)add_list_to_class(classbits, &class_uchardata, options, cd, + (void)add_list_to_class(classbits, &class_uchardata, options, cb, PRIV(hspace_list), NOTACHAR); break; case ESC_H: (void)add_not_list_to_class(classbits, &class_uchardata, options, - cd, PRIV(hspace_list)); + cb, PRIV(hspace_list)); break; case ESC_v: - (void)add_list_to_class(classbits, &class_uchardata, options, cd, + (void)add_list_to_class(classbits, &class_uchardata, options, cb, PRIV(vspace_list), NOTACHAR); break; case ESC_V: (void)add_not_list_to_class(classbits, &class_uchardata, options, - cd, PRIV(vspace_list)); + cb, PRIV(vspace_list)); break; #ifdef SUPPORT_UTF @@ -3759,7 +3759,7 @@ for (;; ptr++) { BOOL negated; unsigned int ptype = 0, pdata = 0; - if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr, cd)) + if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr, cb)) goto FAILED; *class_uchardata++ = ((escape == ESC_p) != negated)? XCL_PROP : XCL_NOTPROP; @@ -3803,7 +3803,7 @@ for (;; ptr++) /* Remember if \r or \n were explicitly used */ - if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE2_HASCRORLF; + if (c == CHAR_CR || c == CHAR_NL) cb->external_flags |= PCRE2_HASCRORLF; /* Check for range */ @@ -3856,7 +3856,7 @@ for (;; ptr++) if (d == CHAR_BACKSLASH) { int descape; - descape = check_escape(&ptr, &d, errorcodeptr, options, TRUE, cd); + descape = check_escape(&ptr, &d, errorcodeptr, options, TRUE, cb); if (*errorcodeptr != 0) goto FAILED; /* 0 means a character was put into d; \b is backspace; any other @@ -3902,10 +3902,10 @@ for (;; ptr++) /* Remember an explicit \r or \n, and add the range to the class. */ - if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE2_HASCRORLF; + if (d == CHAR_CR || d == CHAR_NL) cb->external_flags |= PCRE2_HASCRORLF; class_has_8bitchar += - add_to_class(classbits, &class_uchardata, options, cd, c, d); + add_to_class(classbits, &class_uchardata, options, cb, c, d); goto CONTINUE_CLASS; /* Go get the next char in the class */ } @@ -3980,7 +3980,7 @@ for (;; ptr++) has been generated. Add this character to the class. */ class_has_8bitchar += - add_to_class(classbits, &class_uchardata, options, cd, c, c); + add_to_class(classbits, &class_uchardata, options, cb, c, c); /* Continue to the next character in the class. Closing square bracket not within \Q..\E ends the class. A NULL character terminates a @@ -4142,14 +4142,14 @@ for (;; ptr++) PCRE2_SPTR p = ptr + 1; for (;;) { - while (MAX_255(*p) && (cd->ctypes[*p] & ctype_space) != 0) p++; + while (MAX_255(*p) && (cb->ctypes[*p] & ctype_space) != 0) p++; if (*p != CHAR_NUMBER_SIGN) break; p++; while (*p != CHAR_NULL) { if (IS_NEWLINE(p)) /* For non-fixed-length newline cases, */ - { /* IS_NEWLINE sets cd->nllen. */ - p += cd->nllen; + { /* IS_NEWLINE sets cb->nllen. */ + p += cb->nllen; break; } p++; @@ -4199,11 +4199,11 @@ for (;; ptr++) /* When actually compiling, we need to check whether this was a forward reference, and if so, adjust the offset. */ - if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) + if (lengthptr == NULL && cb->hwm >= cb->start_workspace + LINK_SIZE) { - int offset = GET(cd->hwm, -LINK_SIZE); - if (offset == previous + 1 - cd->start_code) - PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); + int offset = GET(cb->hwm, -LINK_SIZE); + if (offset == previous + 1 - cb->start_code) + PUT(cb->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); } } @@ -4251,7 +4251,7 @@ for (;; ptr++) if (*previous <= OP_CHARI && repeat_min > 1) { reqcu = c; - reqcuflags = req_caseopt | cd->req_varyopt; + reqcuflags = req_caseopt | cb->req_varyopt; } } @@ -4501,7 +4501,7 @@ for (;; ptr++) if (repeat_max <= 1) /* Covers 0, 1, and unlimited */ { *code = OP_END; - adjust_recurse(previous, 1, utf, cd, save_hwm); + adjust_recurse(previous, 1, utf, cb, save_hwm); memmove(previous + 1, previous, CU2BYTES(len)); code++; if (repeat_max == 0) @@ -4525,7 +4525,7 @@ for (;; ptr++) { int offset; *code = OP_END; - adjust_recurse(previous, 2 + LINK_SIZE, utf, cd, save_hwm); + adjust_recurse(previous, 2 + LINK_SIZE, utf, cb, save_hwm); memmove(previous + 2 + LINK_SIZE, previous, CU2BYTES(len)); code += 2 + LINK_SIZE; *previous++ = OP_BRAZERO + repeat_type; @@ -4588,24 +4588,24 @@ for (;; ptr++) for (i = 1; i < repeat_min; i++) { PCRE2_UCHAR *hc; - PCRE2_UCHAR *this_hwm = cd->hwm; + PCRE2_UCHAR *this_hwm = cb->hwm; memcpy(code, previous, CU2BYTES(len)); - while (cd->hwm > cd->start_workspace + cd->workspace_size - + while (cb->hwm > cb->start_workspace + cb->workspace_size - WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) { - size_t save_offset = save_hwm - cd->start_workspace; - size_t this_offset = this_hwm - cd->start_workspace; - *errorcodeptr = expand_workspace(cd); + size_t save_offset = save_hwm - cb->start_workspace; + size_t this_offset = this_hwm - cb->start_workspace; + *errorcodeptr = expand_workspace(cb); if (*errorcodeptr != 0) goto FAILED; - save_hwm = (PCRE2_UCHAR *)cd->start_workspace + save_offset; - this_hwm = (PCRE2_UCHAR *)cd->start_workspace + this_offset; + save_hwm = (PCRE2_UCHAR *)cb->start_workspace + save_offset; + this_hwm = (PCRE2_UCHAR *)cb->start_workspace + this_offset; } for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) { - PUT(cd->hwm, 0, GET(hc, 0) + len); - cd->hwm += LINK_SIZE; + PUT(cb->hwm, 0, GET(hc, 0) + len); + cb->hwm += LINK_SIZE; } save_hwm = this_hwm; code += len; @@ -4652,7 +4652,7 @@ for (;; ptr++) else for (i = repeat_max - 1; i >= 0; i--) { PCRE2_UCHAR *hc; - PCRE2_UCHAR *this_hwm = cd->hwm; + PCRE2_UCHAR *this_hwm = cb->hwm; *code++ = OP_BRAZERO + repeat_type; @@ -4673,21 +4673,21 @@ for (;; ptr++) /* Ensure there is enough workspace for forward references before copying them. */ - while (cd->hwm > cd->start_workspace + cd->workspace_size - + while (cb->hwm > cb->start_workspace + cb->workspace_size - WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) { - size_t save_offset = save_hwm - cd->start_workspace; - size_t this_offset = this_hwm - cd->start_workspace; - *errorcodeptr = expand_workspace(cd); + size_t save_offset = save_hwm - cb->start_workspace; + size_t this_offset = this_hwm - cb->start_workspace; + *errorcodeptr = expand_workspace(cb); if (*errorcodeptr != 0) goto FAILED; - save_hwm = (PCRE2_UCHAR *)cd->start_workspace + save_offset; - this_hwm = (PCRE2_UCHAR *)cd->start_workspace + this_offset; + save_hwm = (PCRE2_UCHAR *)cb->start_workspace + save_offset; + this_hwm = (PCRE2_UCHAR *)cb->start_workspace + this_offset; } for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) { - PUT(cd->hwm, 0, GET(hc, 0) + len + ((i != 0)? 2+LINK_SIZE : 1)); - cd->hwm += LINK_SIZE; + PUT(cb->hwm, 0, GET(hc, 0) + len + ((i != 0)? 2+LINK_SIZE : 1)); + cb->hwm += LINK_SIZE; } save_hwm = this_hwm; code += len; @@ -4762,7 +4762,7 @@ for (;; ptr++) PCRE2_UCHAR *scode = bracode; do { - if (could_be_empty_branch(scode, ketcode, utf, cd, NULL)) + if (could_be_empty_branch(scode, ketcode, utf, cb, NULL)) { *bracode += OP_SBRA - OP_BRA; break; @@ -4785,7 +4785,7 @@ for (;; ptr++) { int nlen = (int)(code - bracode); *code = OP_END; - adjust_recurse(bracode, 1 + LINK_SIZE, utf, cd, save_hwm); + adjust_recurse(bracode, 1 + LINK_SIZE, utf, cb, save_hwm); memmove(bracode + 1 + LINK_SIZE, bracode, CU2BYTES(nlen)); code += 1 + LINK_SIZE; nlen += 1 + LINK_SIZE; @@ -4920,7 +4920,7 @@ for (;; ptr++) else { *code = OP_END; - adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm); + adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cb, save_hwm); memmove(tempcode + 1 + LINK_SIZE, tempcode, CU2BYTES(len)); code += 1 + LINK_SIZE; len += 1 + LINK_SIZE; @@ -4938,7 +4938,7 @@ for (;; ptr++) END_REPEAT: previous = NULL; - cd->req_varyopt |= reqvary; + cb->req_varyopt |= reqvary; break; @@ -4951,14 +4951,14 @@ for (;; ptr++) newoptions = options; skipbytes = 0; bravalue = OP_CBRA; - save_hwm = cd->hwm; + save_hwm = cb->hwm; reset_bracount = FALSE; /* First deal with various "verbs" that can be introduced by '*'. */ ptr++; if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':' - || (MAX_255(ptr[1]) && ((cd->ctypes[ptr[1]] & ctype_letter) != 0)))) + || (MAX_255(ptr[1]) && ((cb->ctypes[ptr[1]] & ctype_letter) != 0)))) { int i, namelen; int arglen = 0; @@ -4967,7 +4967,7 @@ for (;; ptr++) PCRE2_SPTR arg = NULL; previous = NULL; ptr++; - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++; + while (MAX_255(*ptr) && (cb->ctypes[*ptr] & ctype_letter) != 0) ptr++; namelen = (int)(ptr - name); /* It appears that Perl allows any characters whatsoever, other than @@ -5012,14 +5012,14 @@ for (;; ptr++) *errorcodeptr = ERR59; goto FAILED; } - cd->had_accept = TRUE; - for (oc = cd->open_caps; oc != NULL; oc = oc->next) + cb->had_accept = TRUE; + for (oc = cb->open_caps; oc != NULL; oc = oc->next) { *code++ = OP_CLOSE; PUT2INC(code, 0, oc->number); } setverb = *code++ = - (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; + (cb->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; /* Do not set firstcu after *ACCEPT */ if (firstcuflags == REQ_UNSET) firstcuflags = REQ_NONE; @@ -5055,14 +5055,14 @@ for (;; ptr++) { case OP_THEN: case OP_THEN_ARG: - cd->external_flags |= PCRE2_HASTHEN; + cb->external_flags |= PCRE2_HASTHEN; break; case OP_PRUNE: case OP_PRUNE_ARG: case OP_SKIP: case OP_SKIP_ARG: - cd->had_pruneorskip = TRUE; + cb->had_pruneorskip = TRUE; break; } @@ -5091,7 +5091,7 @@ for (;; ptr++) { case CHAR_NUMBER_SIGN: /* Comment; skip to ket */ ptr++; - while (ptr < cd->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + while (ptr < cb->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; if (*ptr != CHAR_RIGHT_PARENTHESIS) { *errorcodeptr = ERR18; @@ -5215,13 +5215,13 @@ for (;; ptr++) *errorcodeptr = ERR44; /* Group name must start with non-digit */ goto FAILED; } - if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_word) == 0) + if (!MAX_255(*ptr) || (cb->ctypes[*ptr] & ctype_word) == 0) { *errorcodeptr = ERR28; /* Assertion expected */ goto FAILED; } name = ptr++; - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) + while (MAX_255(*ptr) && (cb->ctypes[*ptr] & ctype_word) != 0) { ptr++; } @@ -5256,8 +5256,8 @@ for (;; ptr++) goto FAILED; } if (refsign != 0) recno = (refsign == CHAR_MINUS)? - cd->bracount - recno + 1 : recno + cd->bracount; - if (recno <= 0 || recno > cd->final_bracount) + cb->bracount - recno + 1 : recno + cb->bracount; + if (recno <= 0 || recno > cb->final_bracount) { *errorcodeptr = ERR15; goto FAILED; @@ -5268,11 +5268,11 @@ for (;; ptr++) /* Otherwise look for the name. */ - slot = cd->name_table; - for (i = 0; i < cd->names_found; i++) + slot = cb->name_table; + for (i = 0; i < cb->names_found; i++) { if (PRIV(strncmp)(name, slot+IMM2_SIZE, namelen) == 0) break; - slot += cd->name_entry_size; + slot += cb->name_entry_size; } /* Found the named subpattern. If the name is duplicated, add one to @@ -5280,14 +5280,14 @@ for (;; ptr++) appropriate data values. Otherwise, just insert the unique subpattern number. */ - if (i < cd->names_found) + if (i < cb->names_found) { int offset = i++; int count = 1; recno = GET2(slot, 0); /* Number from first found */ - for (; i < cd->names_found; i++) + for (; i < cb->names_found; i++) { - slot += cd->name_entry_size; + slot += cb->name_entry_size; if (PRIV(strncmp)(name, slot+IMM2_SIZE, namelen) != 0 || (slot+IMM2_SIZE)[namelen] != 0) break; count++; @@ -5360,7 +5360,7 @@ for (;; ptr++) /* ------------------------------------------------------------ */ case CHAR_EQUALS_SIGN: /* Positive lookahead */ bravalue = OP_ASSERT; - cd->assert_depth += 1; + cb->assert_depth += 1; ptr++; break; @@ -5382,7 +5382,7 @@ for (;; ptr++) continue; } bravalue = OP_ASSERT_NOT; - cd->assert_depth += 1; + cb->assert_depth += 1; break; @@ -5392,18 +5392,18 @@ for (;; ptr++) { case CHAR_EQUALS_SIGN: /* Positive lookbehind */ bravalue = OP_ASSERTBACK; - cd->assert_depth += 1; + cb->assert_depth += 1; ptr += 2; break; case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ bravalue = OP_ASSERTBACK_NOT; - cd->assert_depth += 1; + cb->assert_depth += 1; ptr += 2; break; default: /* Could be name define, else bad */ - if (MAX_255(ptr[1]) && (cd->ctypes[ptr[1]] & ctype_word) != 0) + if (MAX_255(ptr[1]) && (cb->ctypes[ptr[1]] & ctype_word) != 0) goto DEFINE_NAME; ptr++; /* Correct offset for error */ *errorcodeptr = ERR24; @@ -5440,7 +5440,7 @@ for (;; ptr++) goto FAILED; } *code++ = n; - PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */ + PUT(code, 0, (int)(ptr - cb->start_pattern + 1)); /* Pattern offset */ PUT(code, LINK_SIZE, 0); /* Default length */ code += 2 * LINK_SIZE; } @@ -5476,7 +5476,7 @@ for (;; ptr++) *errorcodeptr = ERR44; /* Group name must start with non-digit */ goto FAILED; } - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + while (MAX_255(*ptr) && (cb->ctypes[*ptr] & ctype_word) != 0) ptr++; namelen = (int)(ptr - name); /* In the pre-compile phase, do a syntax check, remember the longest @@ -5488,7 +5488,7 @@ for (;; ptr++) if (lengthptr != NULL) { named_group *ng; - uint32_t number = cd->bracount + 1; + uint32_t number = cb->bracount + 1; if (*ptr != (PCRE2_UCHAR)terminator) { @@ -5496,15 +5496,15 @@ for (;; ptr++) goto FAILED; } - if (cd->names_found >= MAX_NAME_COUNT) + if (cb->names_found >= MAX_NAME_COUNT) { *errorcodeptr = ERR49; goto FAILED; } - if (namelen + IMM2_SIZE + 1 > cd->name_entry_size) + if (namelen + IMM2_SIZE + 1 > cb->name_entry_size) { - cd->name_entry_size = namelen + IMM2_SIZE + 1; + cb->name_entry_size = namelen + IMM2_SIZE + 1; if (namelen > MAX_NAME_SIZE) { *errorcodeptr = ERR48; @@ -5519,8 +5519,8 @@ for (;; ptr++) scanning in case this is a duplicate with the same number. For non-duplicate names, give an error if the number is duplicated. */ - ng = cd->named_groups; - for (i = 0; i < cd->names_found; i++, ng++) + ng = cb->named_groups; + for (i = 0; i < cb->names_found; i++, ng++) { if (namelen == ng->length && PRIV(strncmp)(name, ng->name, namelen) == 0) @@ -5531,7 +5531,7 @@ for (;; ptr++) *errorcodeptr = ERR43; goto FAILED; } - cd->dupnames = TRUE; /* Duplicate names exist */ + cb->dupnames = TRUE; /* Duplicate names exist */ } else if (ng->number == number) { @@ -5540,35 +5540,35 @@ for (;; ptr++) } } - if (i >= cd->names_found) /* Not a duplicate with same number */ + if (i >= cb->names_found) /* Not a duplicate with same number */ { /* Increase the list size if necessary */ - if (cd->names_found >= cd->named_group_list_size) + if (cb->names_found >= cb->named_group_list_size) { - int newsize = cd->named_group_list_size * 2; + int newsize = cb->named_group_list_size * 2; named_group *newspace = - cd->cx->memctl.malloc(newsize * sizeof(named_group), - cd->cx->memctl.memory_data); + cb->cx->memctl.malloc(newsize * sizeof(named_group), + cb->cx->memctl.memory_data); if (newspace == NULL) { *errorcodeptr = ERR21; goto FAILED; } - memcpy(newspace, cd->named_groups, - cd->named_group_list_size * sizeof(named_group)); - if (cd->named_group_list_size > NAMED_GROUP_LIST_SIZE) - cd->cx->memctl.free((void *)cd->named_groups, - cd->cx->memctl.memory_data); - cd->named_groups = newspace; - cd->named_group_list_size = newsize; + memcpy(newspace, cb->named_groups, + cb->named_group_list_size * sizeof(named_group)); + if (cb->named_group_list_size > NAMED_GROUP_LIST_SIZE) + cb->cx->memctl.free((void *)cb->named_groups, + cb->cx->memctl.memory_data); + cb->named_groups = newspace; + cb->named_group_list_size = newsize; } - cd->named_groups[cd->names_found].name = name; - cd->named_groups[cd->names_found].length = namelen; - cd->named_groups[cd->names_found].number = number; - cd->names_found++; + cb->named_groups[cb->names_found].name = name; + cb->named_groups[cb->names_found].length = namelen; + cb->named_groups[cb->names_found].number = number; + cb->names_found++; } } @@ -5595,7 +5595,7 @@ for (;; ptr++) *errorcodeptr = ERR44; /* Group name must start with non-digit */ goto FAILED; } - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + while (MAX_255(*ptr) && (cb->ctypes[*ptr] & ctype_word) != 0) ptr++; namelen = (int)(ptr - name); /* In the pre-compile phase, do a syntax check. We used to just set @@ -5629,18 +5629,18 @@ for (;; ptr++) number. If the name is not found, set the value to 0 for a forward reference. */ - ng = cd->named_groups; - for (i = 0; i < cd->names_found; i++, ng++) + ng = cb->named_groups; + for (i = 0; i < cb->names_found; i++, ng++) { if (namelen == ng->length && PRIV(strncmp)(name, ng->name, namelen) == 0) break; } - recno = (i < cd->names_found)? ng->number : 0; + recno = (i < cb->names_found)? ng->number : 0; /* Count named back references. */ - if (!is_recurse) cd->namedrefcount++; + if (!is_recurse) cb->namedrefcount++; /* If duplicate names are permitted, we have to allow for a named reference to a duplicated name (this cannot be determined until the @@ -5656,16 +5656,16 @@ for (;; ptr++) else { - slot = cd->name_table; - for (i = 0; i < cd->names_found; i++) + slot = cb->name_table; + for (i = 0; i < cb->names_found; i++) { if (PRIV(strncmp)(name, slot+IMM2_SIZE, namelen) == 0 && slot[IMM2_SIZE+namelen] == 0) break; - slot += cd->name_entry_size; + slot += cb->name_entry_size; } - if (i < cd->names_found) + if (i < cb->names_found) { recno = GET2(slot, 0); } @@ -5684,17 +5684,17 @@ for (;; ptr++) /* In the second pass we must see if the name is duplicated. If so, we generate a different opcode. */ - if (lengthptr == NULL && cd->dupnames) + if (lengthptr == NULL && cb->dupnames) { int count = 1; unsigned int index = i; - PCRE2_UCHAR *cslot = slot + cd->name_entry_size; + PCRE2_UCHAR *cslot = slot + cb->name_entry_size; - for (i++; i < cd->names_found; i++) + for (i++; i < cb->names_found; i++) { if (PRIV(strcmp)(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break; count++; - cslot += cd->name_entry_size; + cslot += cb->name_entry_size; } if (count > 1) @@ -5707,18 +5707,18 @@ for (;; ptr++) /* Process each potentially referenced group. */ - for (; slot < cslot; slot += cd->name_entry_size) + for (; slot < cslot; slot += cb->name_entry_size) { open_capitem *oc; recno = GET2(slot, 0); - cd->backref_map |= (recno < 32)? (1 << recno) : 1; - if (recno > cd->top_backref) cd->top_backref = recno; + cb->backref_map |= (recno < 32)? (1 << recno) : 1; + if (recno > cb->top_backref) cb->top_backref = recno; /* Check to see if this back reference is recursive, that it, it is inside the group that it references. A flag is set so that the group can be made atomic. */ - for (oc = cd->open_caps; oc != NULL; oc = oc->next) + for (oc = cb->open_caps; oc != NULL; oc = oc->next) { if (oc->number == recno) { @@ -5792,7 +5792,7 @@ for (;; ptr++) *errorcodeptr = ERR58; goto FAILED; } - recno = cd->bracount - recno + 1; + recno = cb->bracount - recno + 1; if (recno <= 0) { *errorcodeptr = ERR15; @@ -5806,7 +5806,7 @@ for (;; ptr++) *errorcodeptr = ERR58; goto FAILED; } - recno += cd->bracount; + recno += cb->bracount; } /* Come here from code above that handles a named recursion */ @@ -5814,7 +5814,7 @@ for (;; ptr++) HANDLE_RECURSION: previous = code; - called = cd->start_code; + called = cb->start_code; /* When we are actually compiling, find the bracket that is being referenced. Temporarily end the regex in case it doesn't exist before @@ -5827,13 +5827,13 @@ for (;; ptr++) { *code = OP_END; if (recno != 0) - called = PRIV(find_bracket)(cd->start_code, utf, recno); + called = PRIV(find_bracket)(cb->start_code, utf, recno); /* Forward reference */ if (called == NULL) { - if (recno > cd->final_bracount) + if (recno > cb->final_bracount) { *errorcodeptr = ERR15; goto FAILED; @@ -5843,14 +5843,14 @@ for (;; ptr++) offset below, what it actually inserted is the reference number of the group. Then remember the forward reference. */ - called = cd->start_code + recno; - if (cd->hwm >= cd->start_workspace + cd->workspace_size - + called = cb->start_code + recno; + if (cb->hwm >= cb->start_workspace + cb->workspace_size - WORK_SIZE_SAFETY_MARGIN) { - *errorcodeptr = expand_workspace(cd); + *errorcodeptr = expand_workspace(cb); if (*errorcodeptr != 0) goto FAILED; } - PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code)); + PUTINC(cb->hwm, 0, (int)(code + 1 - cb->start_code)); } /* If not a forward reference, and the subpattern is still open, @@ -5863,7 +5863,7 @@ for (;; ptr++) conditional subpatterns will be picked up then. */ else if (GET(called, 1) == 0 && cond_depth <= 0 && - could_be_empty(called, code, bcptr, utf, cd)) + could_be_empty(called, code, bcptr, utf, cb)) { *errorcodeptr = ERR40; goto FAILED; @@ -5875,7 +5875,7 @@ for (;; ptr++) wrapped with ONCE brackets). */ *code = OP_RECURSE; - PUT(code, 1, (int)(called - cd->start_code)); + PUT(code, 1, (int)(called - cb->start_code)); code += 1 + LINK_SIZE; groupsetfirstcu = FALSE; } @@ -5900,7 +5900,7 @@ for (;; ptr++) case CHAR_J: /* Record that it changed in the external options */ *optset |= PCRE2_DUPNAMES; - cd->external_flags |= PCRE2_JCHANGED; + cb->external_flags |= PCRE2_JCHANGED; break; case CHAR_i: *optset |= PCRE2_CASELESS; break; @@ -5926,7 +5926,7 @@ for (;; ptr++) the compile phase. This can be helpful when matching -- for instance in caseless checking of required bytes. - If the code pointer is not (cd->start_code + 1 + LINK_SIZE), we are + If the code pointer is not (cb->start_code + 1 + LINK_SIZE), we are definitely *not* at the start of the pattern because something has been compiled. In the pre-compile phase, however, the code pointer can have that value after the start, because it gets reset as code is discarded @@ -5942,10 +5942,10 @@ for (;; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) { - if (code == cd->start_code + 1 + LINK_SIZE && + if (code == cb->start_code + 1 + LINK_SIZE && (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE)) { - cd->external_options = newoptions; + cb->external_options = newoptions; } else { @@ -5986,15 +5986,15 @@ for (;; ptr++) else { NUMBERED_GROUP: - cd->bracount += 1; - PUT2(code, 1+LINK_SIZE, cd->bracount); + cb->bracount += 1; + PUT2(code, 1+LINK_SIZE, cb->bracount); skipbytes = IMM2_SIZE; } /* Process nested bracketed regex. First check for parentheses nested too deeply. */ - if ((cd->parens_depth += 1) > PARENS_NEST_LIMIT) + if ((cb->parens_depth += 1) > PARENS_NEST_LIMIT) { *errorcodeptr = ERR19; goto FAILED; @@ -6008,8 +6008,8 @@ for (;; ptr++) previous = code; /* For handling repetition */ *code = bravalue; tempcode = code; - tempreqvary = cd->req_varyopt; /* Save value before bracket */ - tempbracount = cd->bracount; /* Save value before bracket */ + tempreqvary = cb->req_varyopt; /* Save value before bracket */ + tempbracount = cb->bracount; /* Save value before bracket */ length_prevgroup = 0; /* Initialize for pre-compile phase */ if (!compile_regex( @@ -6028,22 +6028,22 @@ for (;; ptr++) &subreqcu, /* For possible last char */ &subreqcuflags, bcptr, /* Current branch chain */ - cd, /* Compile data block */ + cb, /* Compile data block */ (lengthptr == NULL)? NULL : /* Actual compile phase */ &length_prevgroup /* Pre-compile phase */ )) goto FAILED; - cd->parens_depth -= 1; + cb->parens_depth -= 1; /* If this was an atomic group and there are no capturing groups within it, generate OP_ONCE_NC instead of OP_ONCE. */ - if (bravalue == OP_ONCE && cd->bracount <= tempbracount) + if (bravalue == OP_ONCE && cb->bracount <= tempbracount) *code = OP_ONCE_NC; if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) - cd->assert_depth -= 1; + cb->assert_depth -= 1; /* At the end of compiling, code is still pointing to the start of the group, while tempcode has been updated to point past the end of the group. @@ -6210,7 +6210,7 @@ for (;; ptr++) case CHAR_BACKSLASH: tempptr = ptr; - escape = check_escape(&ptr, &ec, errorcodeptr, options, FALSE, cd); + escape = check_escape(&ptr, &ec, errorcodeptr, options, FALSE, cb); if (*errorcodeptr != 0) goto FAILED; if (escape == 0) /* The escape coded a single character */ @@ -6252,7 +6252,7 @@ for (;; ptr++) PCRE2_SPTR p; uint32_t cf; - save_hwm = cd->hwm; /* Normally this is set when '(' is read */ + save_hwm = cb->hwm; /* Normally this is set when '(' is read */ terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; @@ -6322,14 +6322,14 @@ for (;; ptr++) previous = code; *code++ = ((options & PCRE2_CASELESS) != 0)? OP_REFI : OP_REF; PUT2INC(code, 0, recno); - cd->backref_map |= (recno < 32)? (1 << recno) : 1; - if (recno > cd->top_backref) cd->top_backref = recno; + cb->backref_map |= (recno < 32)? (1 << recno) : 1; + if (recno > cb->top_backref) cb->top_backref = recno; /* Check to see if this back reference is recursive, that it, it is inside the group that it references. A flag is set so that the group can be made atomic. */ - for (oc = cd->open_caps; oc != NULL; oc = oc->next) + for (oc = cb->open_caps; oc != NULL; oc = oc->next) { if (oc->number == recno) { @@ -6346,7 +6346,7 @@ for (;; ptr++) { BOOL negated; unsigned int ptype = 0, pdata = 0; - if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr, cd)) + if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr, cb)) goto FAILED; previous = code; *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP; @@ -6374,8 +6374,8 @@ for (;; ptr++) else { if ((escape == ESC_b || escape == ESC_B || escape == ESC_A) && - cd->max_lookbehind == 0) - cd->max_lookbehind = 1; + cb->max_lookbehind == 0) + cb->max_lookbehind = 1; #ifdef SUPPORT_UTF if (escape >= ESC_DU && escape <= ESC_wu) { @@ -6451,7 +6451,7 @@ for (;; ptr++) /* Remember if \r or \n were seen */ if (mcbuffer[0] == CHAR_CR || mcbuffer[0] == CHAR_NL) - cd->external_flags |= PCRE2_HASCRORLF; + cb->external_flags |= PCRE2_HASCRORLF; /* Set the first and required bytes appropriately. If no previous first byte, set it from this character, but revert to none on a zero repeat. @@ -6476,7 +6476,7 @@ for (;; ptr++) if (mclength != 1) { reqcu = code[-1]; - reqcuflags = cd->req_varyopt; + reqcuflags = cb->req_varyopt; } } else firstcuflags = reqcuflags = REQ_NONE; @@ -6494,7 +6494,7 @@ for (;; ptr++) if (mclength == 1 || req_caseopt == 0) { reqcu = code[-1]; - reqcuflags = req_caseopt | cd->req_varyopt; + reqcuflags = req_caseopt | cb->req_varyopt; } } @@ -6538,7 +6538,7 @@ Arguments: reqcuptr place to put the last required code unit reqcuflagsptr place to put the last required code unit flags, or a negative number bcptr pointer to the chain of currently open branches - cd points to the data block with tables pointers etc. + cb points to the data block with tables pointers etc. lengthptr NULL during the real compile phase points to length accumulator during pre-compile phase @@ -6551,7 +6551,7 @@ compile_regex(uint32_t options, PCRE2_UCHAR **codeptr, PCRE2_SPTR *ptrptr, int cond_depth, uint32_t *firstcuptr, int32_t *firstcuflagsptr, uint32_t *reqcuptr, int32_t *reqcuflagsptr, - branch_chain *bcptr, compile_data *cd, size_t *lengthptr) + branch_chain *bcptr, compile_block *cb, size_t *lengthptr) { PCRE2_SPTR ptr = *ptrptr; PCRE2_UCHAR *code = *codeptr; @@ -6611,9 +6611,9 @@ if (*code == OP_CBRA) { capnumber = GET2(code, 1 + LINK_SIZE); capitem.number = capnumber; - capitem.next = cd->open_caps; + capitem.next = cb->open_caps; capitem.flag = FALSE; - cd->open_caps = &capitem; + cb->open_caps = &capitem; } /* Offset is set zero to mark that this bracket is still open */ @@ -6623,14 +6623,14 @@ code += 1 + LINK_SIZE + skipunits; /* Loop for each alternative branch */ -orig_bracount = max_bracount = cd->bracount; +orig_bracount = max_bracount = cb->bracount; for (;;) { /* For a (?| group, reset the capturing bracket count so that each branch uses the same numbers. */ - if (reset_bracount) cd->bracount = orig_bracount; + if (reset_bracount) cb->bracount = orig_bracount; /* Set up dummy OP_REVERSE if lookbehind assertion */ @@ -6647,7 +6647,7 @@ for (;;) if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstcu, &branchfirstcuflags, &branchreqcu, &branchreqcuflags, &bc, - cond_depth, cd, (lengthptr == NULL)? NULL : &length)) + cond_depth, cb, (lengthptr == NULL)? NULL : &length)) { *ptrptr = ptr; return FALSE; @@ -6656,7 +6656,7 @@ for (;;) /* Keep the highest bracket count in case (?| was used and some branch has fewer than the rest. */ - if (cd->bracount > max_bracount) max_bracount = cd->bracount; + if (cb->bracount > max_bracount) max_bracount = cb->bracount; /* In the real compile phase, there is some post-processing to be done. */ @@ -6731,10 +6731,10 @@ for (;;) int fixed_length; *code = OP_END; fixed_length = find_fixedlength(last_branch, (options & PCRE2_UTF) != 0, - FALSE, cd); + FALSE, cb); if (fixed_length == -3) { - cd->check_lookbehind = TRUE; + cb->check_lookbehind = TRUE; } else if (fixed_length < 0) { @@ -6745,8 +6745,8 @@ for (;;) } else { - if (fixed_length > cd->max_lookbehind) - cd->max_lookbehind = fixed_length; + if (fixed_length > cb->max_lookbehind) + cb->max_lookbehind = fixed_length; PUT(reverse_count, 0, fixed_length); } } @@ -6786,7 +6786,7 @@ for (;;) if (capnumber > 0) { - if (cd->open_caps->flag) + if (cb->open_caps->flag) { memmove(start_bracket + 1 + LINK_SIZE, start_bracket, CU2BYTES(code - start_bracket)); @@ -6798,12 +6798,12 @@ for (;;) code += 1 + LINK_SIZE; length += 2 + 2*LINK_SIZE; } - cd->open_caps = cd->open_caps->next; + cb->open_caps = cb->open_caps->next; } /* Retain the highest bracket number, in case resetting was used. */ - cd->bracount = max_bracount; + cb->bracount = max_bracount; /* Set values to pass back */ @@ -6891,7 +6891,7 @@ Arguments: bracket_map a bitmap of which brackets we are inside while testing; this handles up to substring 31; after that we just have to take the less precise approach - cd points to the compile data block + cb points to the compile data block atomcount atomic group level Returns: TRUE or FALSE @@ -6899,7 +6899,7 @@ Returns: TRUE or FALSE static BOOL is_anchored(register PCRE2_SPTR code, unsigned int bracket_map, - compile_data *cd, int atomcount) + compile_block *cb, int atomcount) { do { PCRE2_SPTR scode = first_significant_code( @@ -6911,7 +6911,7 @@ do { if (op == OP_BRA || op == OP_BRAPOS || op == OP_SBRA || op == OP_SBRAPOS) { - if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE; + if (!is_anchored(scode, bracket_map, cb, atomcount)) return FALSE; } /* Capturing brackets */ @@ -6921,21 +6921,21 @@ do { { int n = GET2(scode, 1+LINK_SIZE); int new_map = bracket_map | ((n < 32)? (1 << n) : 1); - if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE; + if (!is_anchored(scode, new_map, cb, atomcount)) return FALSE; } /* Positive forward assertions and conditions */ else if (op == OP_ASSERT || op == OP_COND) { - if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE; + if (!is_anchored(scode, bracket_map, cb, atomcount)) return FALSE; } /* Atomic groups */ else if (op == OP_ONCE || op == OP_ONCE_NC) { - if (!is_anchored(scode, bracket_map, cd, atomcount + 1)) + if (!is_anchored(scode, bracket_map, cb, atomcount + 1)) return FALSE; } @@ -6946,8 +6946,8 @@ do { else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR)) { - if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 || - atomcount > 0 || cd->had_pruneorskip) + if (scode[1] != OP_ALLANY || (bracket_map & cb->backref_map) != 0 || + atomcount > 0 || cb->had_pruneorskip) return FALSE; } @@ -6981,14 +6981,14 @@ Arguments: bracket_map a bitmap of which brackets we are inside while testing; this handles up to substring 31; after that we just have to take the less precise approach - cd points to the compile data + cb points to the compile data atomcount atomic group level Returns: TRUE or FALSE */ static BOOL -is_startline(PCRE2_SPTR code, unsigned int bracket_map, compile_data *cd, +is_startline(PCRE2_SPTR code, unsigned int bracket_map, compile_block *cb, int atomcount) { do { @@ -7015,7 +7015,7 @@ do { return FALSE; default: /* Assertion */ - if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; + if (!is_startline(scode, bracket_map, cb, atomcount)) return FALSE; do scode += GET(scode, 1); while (*scode == OP_ALT); scode += 1 + LINK_SIZE; break; @@ -7029,7 +7029,7 @@ do { if (op == OP_BRA || op == OP_BRAPOS || op == OP_SBRA || op == OP_SBRAPOS) { - if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; + if (!is_startline(scode, bracket_map, cb, atomcount)) return FALSE; } /* Capturing brackets */ @@ -7039,21 +7039,21 @@ do { { int n = GET2(scode, 1+LINK_SIZE); int new_map = bracket_map | ((n < 32)? (1 << n) : 1); - if (!is_startline(scode, new_map, cd, atomcount)) return FALSE; + if (!is_startline(scode, new_map, cb, atomcount)) return FALSE; } /* Positive forward assertions */ else if (op == OP_ASSERT) { - if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; + if (!is_startline(scode, bracket_map, cb, atomcount)) return FALSE; } /* Atomic brackets */ else if (op == OP_ONCE || op == OP_ONCE_NC) { - if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE; + if (!is_startline(scode, bracket_map, cb, atomcount + 1)) return FALSE; } /* .* means "start at start or after \n" if it isn't in atomic brackets or @@ -7064,8 +7064,8 @@ do { else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR) { - if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 || - atomcount > 0 || cd->had_pruneorskip) + if (scode[1] != OP_ANY || (bracket_map & cb->backref_map) != 0 || + atomcount > 0 || cb->had_pruneorskip) return FALSE; } @@ -7190,7 +7190,7 @@ name/number table, maintaining alphabetical order. Checking for permitted and forbidden duplicates has already been done. Arguments: - cd the compile data block + cb the compile data block name the name to add length the length of the name groupno the group number @@ -7199,13 +7199,13 @@ Returns: nothing */ static void -add_name_to_table(compile_data *cd, PCRE2_SPTR name, int length, +add_name_to_table(compile_block *cb, PCRE2_SPTR name, int length, unsigned int groupno) { int i; -PCRE2_UCHAR *slot = cd->name_table; +PCRE2_UCHAR *slot = cb->name_table; -for (i = 0; i < cd->names_found; i++) +for (i = 0; i < cb->names_found; i++) { int crc = memcmp(name, slot+IMM2_SIZE, CU2BYTES(length)); if (crc == 0 && slot[IMM2_SIZE+length] != 0) @@ -7218,20 +7218,20 @@ for (i = 0; i < cd->names_found; i++) if (crc < 0) { - memmove(slot + cd->name_entry_size, slot, - CU2BYTES((cd->names_found - i) * cd->name_entry_size)); + memmove(slot + cb->name_entry_size, slot, + CU2BYTES((cb->names_found - i) * cb->name_entry_size)); break; } /* Continue the loop for a later or duplicate name */ - slot += cd->name_entry_size; + slot += cb->name_entry_size; } PUT2(slot, 0, groupno); memcpy(slot + IMM2_SIZE, name, CU2BYTES(length)); slot[IMM2_SIZE + length] = 0; -cd->names_found++; +cb->names_found++; } @@ -7262,7 +7262,7 @@ pcre2_compile(PCRE2_SPTR pattern, int patlen, uint32_t options, BOOL utf; /* Set TRUE for UTF mode */ pcre2_real_code *re = NULL; /* What we will return */ pcre2_compile_context default_context; /* For use if no context given */ -compile_data cd; /* "Static" compile-time data */ +compile_block cb; /* "Static" compile-time data */ const uint8_t *tables; /* Char tables base pointer */ PCRE2_UCHAR *code; /* Current pointer in compiled code */ @@ -7276,8 +7276,8 @@ int32_t firstcuflags, reqcuflags; /* Type of first/req code unit */ uint32_t firstcu, reqcu; /* Value of first/req code unit */ uint32_t skipatstart; /* When checking (*UTF) etc */ -uint32_t limit_match = MATCH_LIMIT; /* Default match limits */ -uint32_t limit_recursion = MATCH_LIMIT_RECURSION; +uint32_t limit_match = UINT32_MAX; /* Unset match limits */ +uint32_t limit_recursion = UINT32_MAX; int newline = 0; /* Unset; can be set by the pattern */ int bsr = 0; /* Unset; can be set by the pattern */ @@ -7350,40 +7350,40 @@ if (patlen < 0) patlen = PRIV(strlen)(pattern); else tables = (ccontext->tables != NULL)? ccontext->tables : PRIV(default_tables); -cd.lcc = tables + lcc_offset; /* Individual */ -cd.fcc = tables + fcc_offset; /* character */ -cd.cbits = tables + cbits_offset; /* tables */ -cd.ctypes = tables + ctypes_offset; +cb.lcc = tables + lcc_offset; /* Individual */ +cb.fcc = tables + fcc_offset; /* character */ +cb.cbits = tables + cbits_offset; /* tables */ +cb.ctypes = tables + ctypes_offset; -cd.assert_depth = 0; -cd.bracount = cd.final_bracount = 0; -cd.cx = ccontext; -cd.dupnames = FALSE; -cd.end_pattern = pattern + patlen; -cd.external_flags = 0; -cd.external_options = options; -cd.hwm = cworkspace; -cd.max_lookbehind = 0; -cd.name_entry_size = 0; -cd.name_table = NULL; -cd.named_groups = named_groups; -cd.named_group_list_size = NAMED_GROUP_LIST_SIZE; -cd.namedrefcount = 0; -cd.names_found = 0; -cd.open_caps = NULL; -cd.parens_depth = 0; -cd.req_varyopt = 0; -cd.start_code = cworkspace; -cd.start_pattern = pattern; -cd.start_workspace = cworkspace; -cd.workspace_size = COMPILE_WORK_SIZE; +cb.assert_depth = 0; +cb.bracount = cb.final_bracount = 0; +cb.cx = ccontext; +cb.dupnames = FALSE; +cb.end_pattern = pattern + patlen; +cb.external_flags = 0; +cb.external_options = options; +cb.hwm = cworkspace; +cb.max_lookbehind = 0; +cb.name_entry_size = 0; +cb.name_table = NULL; +cb.named_groups = named_groups; +cb.named_group_list_size = NAMED_GROUP_LIST_SIZE; +cb.namedrefcount = 0; +cb.names_found = 0; +cb.open_caps = NULL; +cb.parens_depth = 0; +cb.req_varyopt = 0; +cb.start_code = cworkspace; +cb.start_pattern = pattern; +cb.start_workspace = cworkspace; +cb.workspace_size = COMPILE_WORK_SIZE; /* Maximum back reference and backref bitmap. The bitmap records up to 31 back references to help in deciding whether (.*) can be treated as anchored or not. */ -cd.top_backref = 0; -cd.backref_map = 0; +cb.top_backref = 0; +cb.backref_map = 0; /* --------------- Start looking at the pattern --------------- */ @@ -7410,7 +7410,7 @@ while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS && switch(p->type) { case PSO_OPT: - cd.external_options |= p->value; + cb.external_options |= p->value; break; case PSO_NL: @@ -7431,22 +7431,8 @@ while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS && c = c*10 + ptr[pp++] - CHAR_0; } if (ptr[pp++] != CHAR_RIGHT_PARENTHESIS) goto END_PSO; - if (p->type == PSO_LIMM) - { - if (c < limit_match) - { - limit_match = c; - cd.external_flags |= PCRE2_MLSET; - } - } - else - { - if (c < limit_recursion) - { - limit_recursion = c; - cd.external_flags |= PCRE2_RLSET; - } - } + if (p->type == PSO_LIMM) limit_match = c; + else limit_recursion = c; skipatstart += pp - skipatstart; break; } @@ -7464,7 +7450,7 @@ ptr += skipatstart; /* Can't support UTF or UCP unless PCRE2 has been compiled with UTF support. */ #ifndef SUPPORT_UTF -if ((cd->external_options & (PCRE2_UTF|PCRE2_UCP)) != 0) +if ((cb->external_options & (PCRE2_UTF|PCRE2_UCP)) != 0) { errorcode = ERR32; goto HAD_ERROR; @@ -7472,9 +7458,9 @@ if ((cd->external_options & (PCRE2_UTF|PCRE2_UCP)) != 0) #endif /* Check UTF. We have the original options in 'options', with that value as -modified by (*UTF) etc in cd->external_options. */ +modified by (*UTF) etc in cb->external_options. */ -utf = (cd.external_options & PCRE2_UTF) != 0; +utf = (cb.external_options & PCRE2_UTF) != 0; if (utf) { if ((options & PCRE2_NEVER_UTF) != 0) @@ -7494,30 +7480,31 @@ if (bsr == 0) bsr = ccontext->bsr_convention; /* Process the newline setting. */ if (newline == 0) newline = ccontext->newline_convention; +cb.nltype = NLTYPE_FIXED; switch(newline) { case PCRE2_NEWLINE_CR: - cd.nllen = 1; - cd.nl[0] = CHAR_CR; + cb.nllen = 1; + cb.nl[0] = CHAR_CR; break; case PCRE2_NEWLINE_LF: - cd.nllen = 1; - cd.nl[0] = CHAR_NL; + cb.nllen = 1; + cb.nl[0] = CHAR_NL; break; case PCRE2_NEWLINE_CRLF: - cd.nltype = NLTYPE_FIXED; - cd.nl[0] = CHAR_CR; - cd.nl[1] = CHAR_NL; + cb.nllen = 2; + cb.nl[0] = CHAR_CR; + cb.nl[1] = CHAR_NL; break; case PCRE2_NEWLINE_ANY: - cd.nltype = NLTYPE_ANY; + cb.nltype = NLTYPE_ANY; break; case PCRE2_NEWLINE_ANYCRLF: - cd.nltype = NLTYPE_ANYCRLF; + cb.nltype = NLTYPE_ANYCRLF; break; default: @@ -7533,7 +7520,7 @@ compiled code is discarded when it is no longer needed, so hopefully this workspace will never overflow, though there is a test for its doing so. On error, errorcode will be set non-zero, so we don't need to look at the -result of the function. The initial options have been put into the cd block so +result of the function. The initial options have been put into the cb block so that they can be changed if an option setting is found within the regex right at the beginning. Bringing initial option settings outside can help speed up starting point checks. We still have to pass a separate options variable (the @@ -7542,9 +7529,9 @@ first argument) because that may change as the pattern is processed. */ code = cworkspace; *code = OP_BRA; -(void)compile_regex(cd.external_options, &code, &ptr, &errorcode, FALSE, +(void)compile_regex(cb.external_options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, &firstcu, &firstcuflags, &reqcu, &reqcuflags, NULL, - &cd, &length); + &cb, &length); if (errorcode != 0) goto HAD_ERROR; if (length > MAX_PATTERN_SIZE) @@ -7557,16 +7544,16 @@ if (length > MAX_PATTERN_SIZE) name, we must allow for the possibility of named references to duplicated groups. These require an extra data item each. */ -if (cd.dupnames && cd.namedrefcount > 0) - length += cd.namedrefcount * IMM2_SIZE * sizeof(PCRE2_UCHAR); +if (cb.dupnames && cb.namedrefcount > 0) + length += cb.namedrefcount * IMM2_SIZE * sizeof(PCRE2_UCHAR); /* Compute the size of, and then get and initialize, the data block for storing the compiled pattern and names table. Integer overflow should no longer be -possible because nowadays we limit the maximum value of cd.names_found and -cd.name_entry_size. */ +possible because nowadays we limit the maximum value of cb.names_found and +cb.name_entry_size. */ re_blocksize = sizeof(pcre2_real_code) + - CU2BYTES(length + cd.names_found * cd.name_entry_size); + CU2BYTES(length + cb.names_found * cb.name_entry_size); re = (pcre2_real_code *) ccontext->memctl.malloc(re_blocksize, ccontext->memctl.memory_data); if (re == NULL) @@ -7582,8 +7569,8 @@ memset(re->start_bitmap, 0, 32 * sizeof(uint8_t)); re->blocksize = re_blocksize; re->magic_number = MAGIC_NUMBER; re->compile_options = options; -re->overall_options = cd.external_options; -re->flags = PCRE2_CODE_UNIT_WIDTH/8 | cd.external_flags; +re->overall_options = cb.external_options; +re->flags = PCRE2_CODE_UNIT_WIDTH/8 | cb.external_flags; re->limit_match = limit_match; re->limit_recursion = limit_recursion; re->first_codeunit = 0; @@ -7594,8 +7581,8 @@ re->max_lookbehind = re->minlength = 0; re->top_bracket = 0; re->top_backref = 0; -re->name_entry_size = cd.name_entry_size; -re->name_count = cd.names_found; +re->name_entry_size = cb.name_entry_size; +re->name_count = cb.names_found; /* The basic block is immediately followed by the name table, and the compiled code follows after that. */ @@ -7626,33 +7613,33 @@ from the pre-compile phase, as is the name_entry_size field. Reset the bracket count and the names_found field. Also reset the hwm field; this time it's used for remembering forward references to subpatterns. */ -cd.final_bracount = cd.bracount; /* Save for checking forward references */ -cd.parens_depth = 0; -cd.assert_depth = 0; -cd.bracount = 0; -cd.max_lookbehind = 0; -cd.name_table = (PCRE2_UCHAR *)((uint8_t *)re + sizeof(pcre2_real_code)); -cd.start_code = codestart; -cd.hwm = (PCRE2_UCHAR *)(cd.start_workspace); -cd.req_varyopt = 0; -cd.had_accept = FALSE; -cd.had_pruneorskip = FALSE; -cd.check_lookbehind = FALSE; -cd.open_caps = NULL; +cb.final_bracount = cb.bracount; /* Save for checking forward references */ +cb.parens_depth = 0; +cb.assert_depth = 0; +cb.bracount = 0; +cb.max_lookbehind = 0; +cb.name_table = (PCRE2_UCHAR *)((uint8_t *)re + sizeof(pcre2_real_code)); +cb.start_code = codestart; +cb.hwm = (PCRE2_UCHAR *)(cb.start_workspace); +cb.req_varyopt = 0; +cb.had_accept = FALSE; +cb.had_pruneorskip = FALSE; +cb.check_lookbehind = FALSE; +cb.open_caps = NULL; /* If any named groups were found, create the name/number table from the list created in the first pass. If the list was longer than the in-stack list, free the heap memory. */ -if (cd.names_found > 0) +if (cb.names_found > 0) { - int i = cd.names_found; - named_group *ng = cd.named_groups; - cd.names_found = 0; + int i = cb.names_found; + named_group *ng = cb.named_groups; + cb.names_found = 0; for (; i > 0; i--, ng++) - add_name_to_table(&cd, ng->name, ng->length, ng->number); - if (cd.named_group_list_size > NAMED_GROUP_LIST_SIZE) - ccontext->memctl.free((void *)cd.named_groups, ccontext->memctl.memory_data); + add_name_to_table(&cb, ng->name, ng->length, ng->number); + if (cb.named_group_list_size > NAMED_GROUP_LIST_SIZE) + ccontext->memctl.free((void *)cb.named_groups, ccontext->memctl.memory_data); } /* Set up a starting, non-extracting bracket, then compile the expression. On @@ -7663,17 +7650,17 @@ ptr = pattern + skipatstart; code = (PCRE2_UCHAR *)codestart; *code = OP_BRA; (void)compile_regex(re->overall_options, &code, &ptr, &errorcode, FALSE, FALSE, - 0, 0, &firstcu, &firstcuflags, &reqcu, &reqcuflags, NULL, &cd, NULL); + 0, 0, &firstcu, &firstcuflags, &reqcu, &reqcuflags, NULL, &cb, NULL); -re->top_bracket = cd.bracount; -re->top_backref = cd.top_backref; -re->max_lookbehind = cd.max_lookbehind; +re->top_bracket = cb.bracount; +re->top_backref = cb.top_backref; +re->max_lookbehind = cb.max_lookbehind; #ifdef FIXME /* Is this necessary? Set above */ -re->flags = cd.external_flags | PCRE2_MODE; +re->flags = cb.external_flags | PCRE2_MODE; #endif -if (cd.had_accept) +if (cb.had_accept) { reqcu = 0; /* Must disable after (*ACCEPT) */ reqcuflags = REQ_NONE; @@ -7683,7 +7670,7 @@ if (cd.had_accept) excess bracket. Otherwise, fill in the final opcode and check for disastrous overflow. */ -if (errorcode == 0 && ptr < cd.end_pattern) errorcode = ERR22; +if (errorcode == 0 && ptr < cb.end_pattern) errorcode = ERR22; *code++ = OP_END; if ((size_t)(code - codestart) > length) errorcode = ERR23; @@ -7697,15 +7684,15 @@ VALGRIND_MAKE_MEM_NOACCESS(code, (length - (code - codestart)) * sizeof(PCRE2_UC /* Fill in any forward references that are required. There may be repeated references; optimize for them, as searching a large regex takes time. */ -if (cd.hwm > cd.start_workspace) +if (cb.hwm > cb.start_workspace) { int prev_recno = -1; PCRE2_SPTR groupptr = NULL; - while (errorcode == 0 && cd.hwm > cd.start_workspace) + while (errorcode == 0 && cb.hwm > cb.start_workspace) { int offset, recno; - cd.hwm -= LINK_SIZE; - offset = GET(cd.hwm, 0); + cb.hwm -= LINK_SIZE; + offset = GET(cb.hwm, 0); recno = GET(codestart, offset); if (recno != prev_recno) { @@ -7720,10 +7707,10 @@ if (cd.hwm > cd.start_workspace) /* If the workspace had to be expanded, free the new memory. Set the pointer to NULL to indicate that forward references have been filled in. */ -if (cd.workspace_size > COMPILE_WORK_SIZE) - ccontext->memctl.free((void *)cd.start_workspace, +if (cb.workspace_size > COMPILE_WORK_SIZE) + ccontext->memctl.free((void *)cb.start_workspace, ccontext->memctl.memory_data); -cd.start_workspace = NULL; +cb.start_workspace = NULL; /* Give an error if there's back reference to a non-existent capturing subpattern. */ @@ -7734,7 +7721,7 @@ if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; auto-possessified. The function overwrites the appropriate opcode values. */ if ((options & PCRE2_NO_AUTO_POSSESS) == 0) - PRIV(auto_possessify)((PCRE2_UCHAR *)codestart, utf, &cd); + PRIV(auto_possessify)((PCRE2_UCHAR *)codestart, utf, &cb); /* If there were any lookbehind assertions that contained OP_RECURSE (recursions or subroutine calls), a flag is set for them to be checked here, @@ -7744,7 +7731,7 @@ OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The exceptional ones forgo this. We scan the pattern to check that they are fixed length, and set their lengths. */ -if (cd.check_lookbehind) +if (cb.check_lookbehind) { PCRE2_UCHAR *cc = (PCRE2_UCHAR *)codestart; @@ -7763,7 +7750,7 @@ if (cd.check_lookbehind) PCRE2_UCHAR *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); int end_op = *be; *be = OP_END; - fixed_length = find_fixedlength(cc, utf, TRUE, &cd); + fixed_length = find_fixedlength(cc, utf, TRUE, &cb); *be = end_op; if (fixed_length < 0) { @@ -7771,7 +7758,7 @@ if (cd.check_lookbehind) (fixed_length == -4)? ERR70 : ERR25; break; } - if (fixed_length > cd.max_lookbehind) cd.max_lookbehind = fixed_length; + if (fixed_length > cb.max_lookbehind) cb.max_lookbehind = fixed_length; PUT(cc, 1, fixed_length); } cc += 1 + LINK_SIZE; @@ -7797,7 +7784,7 @@ or anything else, such as starting with non-atomic .* when DOTALL is set and there are no occurrences of *PRUNE or *SKIP. */ if ((re->overall_options & PCRE2_ANCHORED) == 0 && - is_anchored(codestart, 0, &cd, 0)) + is_anchored(codestart, 0, &cb, 0)) re->overall_options |= PCRE2_ANCHORED; /* If the pattern is still not anchored and we do not have a first code unit, @@ -7822,7 +7809,7 @@ if ((re->overall_options & PCRE2_ANCHORED) == 0) { if (firstcu < 128 || (!utf && firstcu < 255)) { - if (cd.fcc[firstcu] != firstcu) re->flags |= PCRE2_FIRSTCASELESS; + if (cb.fcc[firstcu] != firstcu) re->flags |= PCRE2_FIRSTCASELESS; } /* The first code unit is > 128 in UTF mode, or > 255 otherwise. In @@ -7842,7 +7829,7 @@ if ((re->overall_options & PCRE2_ANCHORED) == 0) and also when all branches start with non-atomic .* for non-DOTALL matches when *PRUNE and SKIP are not present. */ - else if (is_startline(codestart, 0, &cd, 0)) re->flags |= PCRE2_STARTLINE; + else if (is_startline(codestart, 0, &cb, 0)) re->flags |= PCRE2_STARTLINE; } /* Handle the "required code unit", if one is set. In the case of an anchored @@ -7861,7 +7848,7 @@ if (reqcuflags >= 0 && { if (reqcu < 128 || (!utf && reqcu < 255)) { - if (cd.fcc[reqcu] != reqcu) re->flags |= PCRE2_LASTCASELESS; + if (cb.fcc[reqcu] != reqcu) re->flags |= PCRE2_LASTCASELESS; } #if defined SUPPORT_UTF && PCRE2_CODE_UNIT_WIDTH != 8 else if (UCD_OTHERCASE(reqcu) != reqcu) @@ -7875,7 +7862,7 @@ can be provided to applications. */ do { - if (could_be_empty_branch(codestart, code, utf, &cd, NULL)) + if (could_be_empty_branch(codestart, code, utf, &cb, NULL)) { re->flags |= PCRE2_MATCH_EMPTY; break; diff --git a/src/pcre2_context.c b/src/pcre2_context.c index 724a152..d772e8f 100644 --- a/src/pcre2_context.c +++ b/src/pcre2_context.c @@ -172,6 +172,7 @@ mcontext->stack_malloc = mcontext->malloc; mcontext->stack_free = mcontext->free; #endif mcontext->callout = NULL; +mcontext->callout_data = NULL; mcontext->newline_convention = 0; mcontext->bsr_convention = 0; mcontext->match_limit = MATCH_LIMIT; @@ -368,9 +369,10 @@ switch(newline) PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION pcre2_set_callout(pcre2_match_context *mcontext, - int (*callout)(pcre2_callout_block *, void *)) + int (*callout)(pcre2_callout_block *), void *callout_data) { mcontext->callout = callout; +mcontext->callout_data = callout_data; return 1; } diff --git a/src/pcre2_error.c b/src/pcre2_error.c index be94181..5e028a9 100644 --- a/src/pcre2_error.c +++ b/src/pcre2_error.c @@ -228,8 +228,6 @@ static const char match_error_texts[] = "NULL argument passed\0" "nested recursion at the same subject position\0" "recursion limit exceeded\0" - "unknown opcode - pattern overwritten?\0" - "value unset\0" /* Used by pcre2_pattern_info() */ ; diff --git a/src/pcre2_internal.h b/src/pcre2_internal.h index e4d3930..35e3f53 100644 --- a/src/pcre2_internal.h +++ b/src/pcre2_internal.h @@ -522,9 +522,7 @@ bytes in a code unit in that mode. */ #define PCRE2_JCHANGED 0x00000400 /* j option used in pattern */ #define PCRE2_HASCRORLF 0x00000800 /* explicit \r or \n in pattern */ #define PCRE2_HASTHEN 0x00001000 /* pattern contains (*THEN) */ -#define PCRE2_MLSET 0x00002000 /* match limit set by pattern */ -#define PCRE2_RLSET 0x00004000 /* recursion limit set by pattern */ -#define PCRE2_MATCH_EMPTY 0x00008000 /* pattern can match empty string */ +#define PCRE2_MATCH_EMPTY 0x00002000 /* pattern can match empty string */ #define PCRE2_MODE_MASK (PCRE2_MODE8 | PCRE2_MODE16 | PCRE2_MODE32) @@ -540,7 +538,7 @@ endianness. */ /* The maximum remaining length of subject we are prepared to search for a req_unit match. */ -#define REQ_UNIT_MAX 1000 +#define REQ_CU_MAX 1000 /* Bit definitions for entries in the pcre_ctypes table. */ @@ -1816,8 +1814,10 @@ compiling the library, PCRE2_CODE_UNIT_WIDTH will be defined, and we can include them at the appropriate width, after setting up suffix macros for the private structures. */ -#define compile_data PCRE2_SUFFIX(compile_data_) #define branch_chain PCRE2_SUFFIX(branch_chain_) +#define compile_block PCRE2_SUFFIX(compile_block_) +#define dfa_match_block PCRE2_SUFFIX(dfa_match_block_) +#define match_block PCRE2_SUFFIX(match_block_) #define named_group PCRE2_SUFFIX(named_group_) #include "pcre2_intmodedep.h" @@ -1845,10 +1845,11 @@ is available. */ #define _pcre2_was_newline PCRE2_SUFFIX(_pcre2_was_newline_) #define _pcre2_xclass PCRE2_SUFFIX(_pcre2_xclass_) -extern void _pcre2_auto_possessify(PCRE2_UCHAR *, BOOL, const compile_data *); +extern void _pcre2_auto_possessify(PCRE2_UCHAR *, BOOL, const compile_block *); extern void _pcre2_compile_context_init(pcre2_compile_context *, BOOL); extern PCRE2_SPTR _pcre2_find_bracket(PCRE2_SPTR, BOOL, int); -extern BOOL _pcre2_is_newline(PCRE2_SPTR, int, PCRE2_SPTR, int *, BOOL); +extern BOOL _pcre2_is_newline(PCRE2_SPTR, uint32_t, PCRE2_SPTR, uint32_t *, + BOOL); extern void _pcre2_match_context_init(pcre2_match_context *, BOOL); extern void *_pcre2_memctl_malloc(size_t, size_t, pcre2_memctl *); extern unsigned int _pcre2_ord2utf(uint32_t, PCRE2_UCHAR *); @@ -1859,7 +1860,8 @@ extern int _pcre2_strncmp(PCRE2_SPTR, PCRE2_SPTR, size_t); extern int _pcre2_strncmp_c8(PCRE2_SPTR, const char *, size_t); extern int _pcre2_study(pcre2_real_code *); extern int _pcre2_valid_utf(PCRE2_SPTR, int, size_t *); -extern BOOL _pcre2_was_newline(PCRE2_SPTR, int, PCRE2_SPTR, int *, BOOL); +extern BOOL _pcre2_was_newline(PCRE2_SPTR, uint32_t, PCRE2_SPTR, uint32_t *, + BOOL); extern BOOL _pcre2_xclass(uint32_t, PCRE2_SPTR, BOOL); #endif /* PCRE2_CODE_UNIT_WIDTH */ diff --git a/src/pcre2_intmodedep.h b/src/pcre2_intmodedep.h index 237d4d3..9ceba85 100644 --- a/src/pcre2_intmodedep.h +++ b/src/pcre2_intmodedep.h @@ -77,6 +77,7 @@ just to undefine them all. */ #undef PUT2INC #undef PUTCHAR #undef PUTINC +#undef TABLE_GET @@ -197,8 +198,12 @@ arithmetic results in a signed value. Hence the cast. */ #define PUT2(a,n,d) a[n] = d #endif -/* Other macros that are different for 8-bit mode. The maximum length of a MARK -name must fit in one code unit; currently it is set to 255 or 65535. */ +/* Other macros that are different for 8-bit mode. The MAX_255 macro checks +whether its argument is less than 256. The maximum length of a MARK name must +fit in one code unit; currently it is set to 255 or 65535. The TABLE_GET macro +is used to access elements of tables containing exactly 256 items. When code +points can be greater than 255, a check is needed before accessing these +tables. */ #if PCRE2_CODE_UNIT_WIDTH == 8 #define MAX_255(c) TRUE @@ -206,11 +211,13 @@ name must fit in one code unit; currently it is set to 255 or 65535. */ #ifdef SUPPORT_UTF #define SUPPORT_WIDE_CHARS #endif /* SUPPORT_UTF */ +#define TABLE_GET(c, table, default) ((table)[c]) #else /* Code units are 16 or 32 bits */ #define MAX_255(c) ((c) <= 255u) #define MAX_MARK ((1u << 16) - 1) #define SUPPORT_WIDE_CHARS +#define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default)) #endif @@ -557,7 +564,8 @@ typedef struct pcre2_real_match_context { void * (*stack_malloc)(size_t, void *); void (*stack_free)(void *, void *); #endif - int (*callout)(pcre2_callout_block *, void *); + int (*callout)(pcre2_callout_block *); + void *callout_data; uint16_t bsr_convention; uint16_t newline_convention; uint32_t match_limit; @@ -632,7 +640,7 @@ typedef struct named_group { /* Structure for passing "static" information around between the functions doing the compiling, so that they are thread-safe. */ -typedef struct compile_data { +typedef struct compile_block { pcre2_real_compile_context *cx; /* Points to the compile context */ const uint8_t *lcc; /* Points to lower casing table */ const uint8_t *fcc; /* Points to case-flipping table */ @@ -643,32 +651,131 @@ typedef struct compile_data { PCRE2_SPTR start_pattern; /* The start of the pattern */ PCRE2_SPTR end_pattern; /* The end of the pattern */ PCRE2_UCHAR *hwm; /* High watermark of workspace */ + PCRE2_UCHAR *name_table; /* The name/number table */ + size_t workspace_size; /* Size of workspace */ + uint16_t names_found; /* Number of entries so far */ + uint16_t name_entry_size; /* Size of each entry */ open_capitem *open_caps; /* Chain of open capture items */ named_group *named_groups; /* Points to vector in pre-compile */ - PCRE2_UCHAR *name_table; /* The name/number table */ - int names_found; /* Number of entries so far */ - int name_entry_size; /* Size of each entry */ - int named_group_list_size; /* Number of entries in the list */ - int workspace_size; /* Size of workspace */ - unsigned int bracount; /* Count of capturing parens as we compile */ - int final_bracount; /* Saved value after first pass */ - int max_lookbehind; /* Maximum lookbehind (characters) */ - int top_backref; /* Maximum back reference */ - unsigned int backref_map; /* Bitmap of low back refs */ - unsigned int namedrefcount; /* Number of backreferences by name */ - int parens_depth; /* Depth of nested parentheses */ - int assert_depth; /* Depth of nested assertions */ + uint32_t named_group_list_size; /* Number of entries in the list */ uint32_t external_options; /* External (initial) options */ uint32_t external_flags; /* External flag bits to be set */ + uint32_t bracount; /* Count of capturing parens as we compile */ + uint32_t final_bracount; /* Saved value after first pass */ + uint32_t top_backref; /* Maximum back reference */ + uint32_t backref_map; /* Bitmap of low back refs */ + uint32_t namedrefcount; /* Number of backreferences by name */ + uint32_t nltype; /* Newline type */ + uint32_t nllen; /* Newline string length */ + PCRE2_UCHAR nl[4]; /* Newline string when fixed length */ + int max_lookbehind; /* Maximum lookbehind (characters) */ + int parens_depth; /* Depth of nested parentheses */ + int assert_depth; /* Depth of nested assertions */ int req_varyopt; /* "After variable item" flag for reqbyte */ BOOL had_accept; /* (*ACCEPT) encountered */ BOOL had_pruneorskip; /* (*PRUNE) or (*SKIP) encountered */ BOOL check_lookbehind; /* Lookbehinds need later checking */ BOOL dupnames; /* Duplicate names exist */ - int nltype; /* Newline type */ - int nllen; /* Newline string length */ - PCRE2_UCHAR nl[4]; /* Newline string when fixed length */ -} compile_data; +} compile_block; + +/* Structure for items in a linked list that represents an explicit recursive +call within the pattern; used by pcre_match(). */ + +typedef struct recursion_info { + struct recursion_info *prevrec; /* Previous recursion record (or NULL) */ + unsigned int group_num; /* Number of group that was called */ + size_t *offset_save; /* Pointer to start of saved offsets */ + uint32_t saved_max; /* Number of saved offsets */ + uint32_t saved_capture_last; /* Last capture number */ + PCRE2_SPTR subject_position; /* Position at start of recursion */ +} recursion_info; + +/* A similar structure for pcre_dfa_match(). */ + +typedef struct dfa_recursion_info { + struct dfa_recursion_info *prevrec; + uint32_t group_num; + PCRE2_SPTR subject_position; +} dfa_recursion_info; + +/* Structure for building a chain of data for holding the values of the subject +pointer at the start of each subpattern, so as to detect when an empty string +has been matched by a subpattern - to break infinite loops; used by +pcre2_match(). */ + +typedef struct eptrblock { + struct eptrblock *epb_prev; + PCRE2_SPTR epb_saved_eptr; +} eptrblock; + +/* Structure for passing "static" information around between the functions +doing traditional NFA matching (pcre2_match() and friends). */ + +typedef struct match_block { + pcre2_memctl memctl; + unsigned long int match_call_count; /* As it says */ + unsigned long int match_limit; /* As it says */ + unsigned long int match_limit_recursion; /* As it says */ + BOOL hitend; /* Hit the end of the subject at some point */ + BOOL hasthen; /* Pattern contains (*THEN) */ + const uint8_t *lcc; /* Points to lower casing table */ + const uint8_t *fcc; /* Points to case-flipping table */ + const uint8_t *ctypes; /* Points to table of type maps */ + size_t *ovector; /* Pointer to the offset vector */ + size_t offset_end; /* One past the end */ + size_t offset_max; /* The maximum usable for return data */ + size_t start_offset; /* The start offset value */ + size_t end_offset_top; /* Highwater mark at end of match */ + uint16_t partial; /* PARTIAL options */ + uint16_t bsr_convention; /* \R interpretation */ + uint16_t name_count; /* Number of names in name table */ + uint16_t name_entry_size; /* Size of entry in names table */ + PCRE2_SPTR name_table; /* Table of group names */ + PCRE2_SPTR start_code; /* For use when recursing */ + PCRE2_SPTR start_subject; /* Start of the subject string */ + PCRE2_SPTR end_subject; /* End of the subject string */ + PCRE2_SPTR start_match_ptr; /* Start of matched string */ + PCRE2_SPTR end_match_ptr; /* Subject position at end match */ + PCRE2_SPTR start_used_ptr; /* Earliest consulted character */ + PCRE2_SPTR mark; /* Mark pointer to pass back on success */ + PCRE2_SPTR nomatch_mark; /* Mark pointer to pass back on failure */ + PCRE2_SPTR once_target; /* Where to back up to for atomic groups */ + uint32_t moptions; /* Match options */ + uint32_t poptions; /* Pattern options */ + uint32_t capture_last; /* Most recent capture number + overflow flag */ + uint32_t skip_arg_count; /* For counting SKIP_ARGs */ + uint32_t ignore_skip_arg; /* For re-run when SKIP arg name not found */ + uint32_t match_function_type; /* Set for certain special calls of match() */ + uint32_t nltype; /* Newline type */ + uint32_t nllen; /* Newline string length */ + PCRE2_UCHAR nl[4]; /* Newline string when fixed */ + eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */ + recursion_info *recursive; /* Linked list of recursion data */ + void *callout_data; /* To pass back to callouts */ + int (*callout)(pcre2_callout_block *); /* Callout function or NULL */ +#ifdef NO_RECURSE + void *match_frames_base; /* For remembering malloc'd frames */ +#endif +} match_block; + +/* A similar structure is used for the same purpose by the DFA matching +functions. */ + +typedef struct dfa_match_block { + PCRE2_SPTR start_code; /* Start of the compiled pattern */ + PCRE2_SPTR start_subject ; /* Start of the subject string */ + PCRE2_SPTR end_subject; /* End of subject string */ + PCRE2_SPTR start_used_ptr; /* Earliest consulted character */ + const uint8_t *tables; /* Character tables */ + int start_offset; /* The start offset value */ + uint32_t moptions; /* Match options */ + uint32_t poptions; /* Pattern options */ + int nltype; /* Newline type */ + int nllen; /* Newline string length */ + PCRE2_UCHAR nl[4]; /* Newline string when fixed */ + void *callout_data; /* To pass back to callouts */ + dfa_recursion_info *recursive; /* Linked list of recursion data */ +} dfa_match_block; #endif /* PCRE2_PCRE2TEST */ diff --git a/src/pcre2_match.c b/src/pcre2_match.c index 67f8f9e..8b7694a 100644 --- a/src/pcre2_match.c +++ b/src/pcre2_match.c @@ -43,8 +43,6131 @@ POSSIBILITY OF SUCH DAMAGE. #include "config.h" #endif +#define NLBLOCK mb /* Block containing newline information */ +#define PSSTART start_subject /* Field containing processed string start */ +#define PSEND end_subject /* Field containing processed string end */ + #include "pcre2_internal.h" +/* Masks for identifying the public options that are permitted at match time. +*/ + +#define PUBLIC_MATCH_OPTIONS \ + (PCRE2_ANCHORED|PCRE2_NOTBOL|PCRE2_NOTEOL|PCRE2_NOTEMPTY| \ + PCRE2_NOTEMPTY_ATSTART|PCRE2_NO_UTF_CHECK|PCRE2_PARTIAL_HARD| \ + PCRE2_PARTIAL_SOFT|PCRE2_NO_START_OPTIMIZE) + +/* The mb->capture_last field uses the lower 16 bits for the last captured +substring (which can never be greater than 65535) and a bit in the top half +to mean "capture vector overflowed". This odd way of doing things was +implemented when it was realized that preserving and restoring the overflow bit +whenever the last capture number was saved/restored made for a neater +interface, and doing it this way saved on (a) another variable, which would +have increased the stack frame size (a big NO-NO in PCRE) and (b) another +separate set of save/restore instructions. The following defines are used in +implementing this. */ + +#define CAPLMASK 0x0000ffff /* The bits used for last_capture */ +#define OVFLMASK 0xffff0000 /* The bits used for the overflow flag */ +#define OVFLBIT 0x00010000 /* The bit that is set for overflow */ + +/* Values for setting in mb->match_function_type to indicate two special types +of call to match(). We do it this way to save on using another stack variable, +as stack usage is to be discouraged. */ + +#define MATCH_CONDASSERT 1 /* Called to check a condition assertion */ +#define MATCH_CBEGROUP 2 /* Could-be-empty unlimited repeat group */ + +/* Non-error returns from the match() function. Error returns are externally +defined PCRE2_ERROR_xxx codes, which are all negative. */ + +#define MATCH_MATCH 1 +#define MATCH_NOMATCH 0 + +/* Special internal returns from the match() function. Make them sufficiently +negative to avoid the external error codes. */ + +#define MATCH_ACCEPT (-999) +#define MATCH_KETRPOS (-998) +#define MATCH_ONCE (-997) +/* The next 5 must be kept together and in sequence so that a test that checks +for any one of them can use a range. */ +#define MATCH_COMMIT (-996) +#define MATCH_PRUNE (-995) +#define MATCH_SKIP (-994) +#define MATCH_SKIP_ARG (-993) +#define MATCH_THEN (-992) +#define MATCH_BACKTRACK_MAX MATCH_THEN +#define MATCH_BACKTRACK_MIN MATCH_COMMIT + +/* Maximum number of ints of offset to save on the stack for recursive calls. +If the offset vector is bigger, malloc is used. This should be a multiple of 3, +because the offset vector is always a multiple of 3 long. */ + +#define REC_STACK_SAVE_MAX 30 + +/* Min and max values for the common repeats; for the maxima, 0 => infinity */ + +static const char rep_min[] = { 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, }; +static const char rep_max[] = { 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, }; + + + +/************************************************* +* Match a back-reference * +*************************************************/ + +/* Normally, if a back reference hasn't been set, the length that is passed is +negative, so the match always fails. However, in JavaScript compatibility mode, +the length passed is zero. Note that in caseless UTF-8 mode, the number of +subject bytes matched may be different to the number of reference bytes. + +Arguments: + offset index into the offset vector + eptr pointer into the subject + length length of reference to be matched (number of bytes) + mb points to match block + caseless TRUE if caseless + +Returns: >= 0 the number of subject bytes matched + -1 no match + -2 partial match; always given if at end subject +*/ + +static int +match_ref(int offset, register PCRE2_SPTR eptr, int length, match_block *mb, + BOOL caseless) +{ +PCRE2_SPTR eptr_start = eptr; +register PCRE2_SPTR p = mb->start_subject + mb->ovector[offset]; +#if defined SUPPORT_UTF +BOOL utf = (mb->poptions & PCRE2_UTF) != 0; +#endif + +/* Always fail if reference not set (unless PCRE2_MATCH_UNSET_BACKREF is set, +in which case the length is passed as zero). */ + +if (length < 0) return -1; + +/* Separate the caseless and UTF case for speed. */ + +if (caseless) + { +#if defined SUPPORT_UTF + if (utf) + { + /* Match characters up to the end of the reference. NOTE: the number of + code units matched may differ, because in UTF-8 there are some characters + whose upper and lower case versions code have different numbers of bytes. + For example, U+023A (2 bytes in UTF-8) is the upper case version of U+2C65 + (3 bytes in UTF-8); a sequence of 3 of the former uses 6 bytes, as does a + sequence of two of the latter. It is important, therefore, to check the + length along the reference, not along the subject (earlier code did this + wrong). */ + + PCRE2_SPTR endptr = p + length; + while (p < endptr) + { + uint32_t c, d; + const ucd_record *ur; + if (eptr >= mb->end_subject) return -2; /* Partial match */ + GETCHARINC(c, eptr); + GETCHARINC(d, p); + ur = GET_UCD(d); + if (c != d && c != d + ur->other_case) + { + const uint32_t *pp = PRIV(ucd_caseless_sets) + ur->caseset; + for (;;) + { + if (c < *pp) return -1; + if (c == *pp++) break; + } + } + } + } + else +#endif + + /* Not in UTF mode */ + + { + while (length-- > 0) + { + uint32_t cc, cp; + if (eptr >= mb->end_subject) return -2; /* Partial match */ + cc = UCHAR21TEST(eptr); + cp = UCHAR21TEST(p); + if (TABLE_GET(cp, mb->lcc, cp) != TABLE_GET(cc, mb->lcc, cc)) return -1; + p++; + eptr++; + } + } + } + +/* In the caseful case, we can just compare the bytes, whether or not we +are in UTF-8 mode. */ + +else + { + while (length-- > 0) + { + if (eptr >= mb->end_subject) return -2; /* Partial match */ + if (UCHAR21INCTEST(p) != UCHAR21INCTEST(eptr)) return -1; + } + } + +return (int)(eptr - eptr_start); +} + + + +/*************************************************************************** +**************************************************************************** + RECURSION IN THE match() FUNCTION + +The match() function is highly recursive, though not every recursive call +increases the recursive depth. Nevertheless, some regular expressions can cause +it to recurse to a great depth. I was writing for Unix, so I just let it call +itself recursively. This uses the stack for saving everything that has to be +saved for a recursive call. On Unix, the stack can be large, and this works +fine. + +It turns out that on some non-Unix-like systems there are problems with +programs that use a lot of stack. (This despite the fact that every last chip +has oodles of memory these days, and techniques for extending the stack have +been known for decades.) So.... + +There is a fudge, triggered by defining NO_RECURSE, which avoids recursive +calls by keeping local variables that need to be preserved in blocks of memory +obtained from malloc() instead instead of on the stack. Macros are used to +achieve this so that the actual code doesn't look very different to what it +always used to. + +The original heap-recursive code used longjmp(). However, it seems that this +can be very slow on some operating systems. Following a suggestion from Stan +Switzer, the use of longjmp() has been abolished, at the cost of having to +provide a unique number for each call to RMATCH. There is no way of generating +a sequence of numbers at compile time in C. I have given them names, to make +them stand out more clearly. + +Crude tests on x86 Linux show a small speedup of around 5-8%. However, on +FreeBSD, avoiding longjmp() more than halves the time taken to run the standard +tests. Furthermore, not using longjmp() means that local dynamic variables +don't have indeterminate values; this has meant that the frame size can be +reduced because the result can be "passed back" by straight setting of the +variable instead of being passed in the frame. +**************************************************************************** +***************************************************************************/ + +/* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN +below must be updated in sync. */ + +enum { RM1=1, RM2, RM3, RM4, RM5, RM6, RM7, RM8, RM9, RM10, + RM11, RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20, + RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, + RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, + RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, + RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, + RM61, RM62, RM63, RM64, RM65, RM66, RM67 }; + +/* These versions of the macros use the stack, as normal. There are debugging +versions and production versions. Note that the "rw" argument of RMATCH isn't +actually used in this definition. */ + +#ifndef NO_RECURSE +#define REGISTER register +#define RMATCH(ra,rb,rc,rd,re,rw) \ + rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1) +#define RRETURN(ra) return ra +#else + +/* These versions of the macros manage a private stack on the heap. Note that +the "rd" argument of RMATCH isn't actually used in this definition. It's the mb +argument of match(), which never changes. */ + +#define REGISTER + +#define RMATCH(ra,rb,rc,rd,re,rw)\ + {\ + heapframe *newframe = frame->Xnextframe;\ + if (newframe == NULL)\ + {\ + newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\ + if (newframe == NULL) RRETURN(PCRE2_ERROR_NOMEMORY);\ + newframe->Xnextframe = NULL;\ + frame->Xnextframe = newframe;\ + }\ + frame->Xwhere = rw;\ + newframe->Xeptr = ra;\ + newframe->Xecode = rb;\ + newframe->Xmstart = mstart;\ + newframe->Xoffset_top = rc;\ + newframe->Xeptrb = re;\ + newframe->Xrdepth = frame->Xrdepth + 1;\ + newframe->Xprevframe = frame;\ + frame = newframe;\ + goto HEAP_RECURSE;\ + L_##rw:\ + } + +#define RRETURN(ra)\ + {\ + heapframe *oldframe = frame;\ + frame = oldframe->Xprevframe;\ + if (frame != NULL)\ + {\ + rrc = ra;\ + goto HEAP_RETURN;\ + }\ + return ra;\ + } + + +/* Structure for remembering the local variables in a private frame */ + +typedef struct heapframe { + struct heapframe *Xprevframe; + struct heapframe *Xnextframe; + + /* Function arguments that may change */ + + PCRE2_SPTR Xeptr; + PCRE2_SPTR Xecode; + PCRE2_SPTR Xmstart; + int Xoffset_top; + eptrblock *Xeptrb; + unsigned int Xrdepth; + + /* Function local variables */ + + PCRE2_SPTR Xcallpat; +#ifdef SUPPORT_UTF + PCRE2_SPTR Xcharptr; +#endif + PCRE2_SPTR Xdata; + PCRE2_SPTR Xnext; + PCRE2_SPTR Xpp; + PCRE2_SPTR Xprev; + PCRE2_SPTR Xsaved_eptr; + + recursion_info Xnew_recursive; + + BOOL Xcur_is_word; + BOOL Xcondition; + BOOL Xprev_is_word; + +#ifdef SUPPORT_UTF + int Xprop_type; + unsigned int Xprop_value; + int Xprop_fail_result; + int Xoclength; + PCRE2_UCHAR Xocchars[6]; +#endif + + int Xcodelink; + int Xctype; + unsigned int Xfc; + int Xfi; + int Xlength; + int Xmax; + int Xmin; + unsigned int Xnumber; + int Xoffset; + unsigned int Xop; + uint32_t Xsave_capture_last; + int Xsave_offset1, Xsave_offset2, Xsave_offset3; + size_t Xstacksave[REC_STACK_SAVE_MAX]; + + eptrblock Xnewptrb; + + /* Where to jump back to */ + + int Xwhere; + +} heapframe; + +#endif + + +/*************************************************************************** +***************************************************************************/ + + + +/************************************************* +* Match from current position * +*************************************************/ + +/* This function is called recursively in many circumstances. Whenever it +returns a negative (error) response, the outer incarnation must also return the +same response. */ + +/* These macros pack up tests that are used for partial matching, and which +appear several times in the code. We set the "hit end" flag if the pointer is +at the end of the subject and also past the start of the subject (i.e. +something has been matched). For hard partial matching, we then return +immediately. The second one is used when we already know we are past the end of +the subject. */ + +#define CHECK_PARTIAL()\ + if (mb->partial != 0 && eptr >= mb->end_subject && \ + eptr > mb->start_used_ptr) \ + { \ + mb->hitend = TRUE; \ + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); \ + } + +#define SCHECK_PARTIAL()\ + if (mb->partial != 0 && eptr > mb->start_used_ptr) \ + { \ + mb->hitend = TRUE; \ + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); \ + } + + +/* Performance note: It might be tempting to extract commonly used fields from +the mb structure (e.g. utf, end_subject) into individual variables to improve +performance. Tests using gcc on a SPARC disproved this; in the first case, it +made performance worse. + +Arguments: + eptr pointer to current character in subject + ecode pointer to current position in compiled code + mstart pointer to the current match start position (can be modified + by encountering \K) + offset_top current top pointer + mb pointer to "static" info block for the match + eptrb pointer to chain of blocks containing eptr at start of + brackets - for testing for empty matches + rdepth the recursion depth + +Returns: MATCH_MATCH if matched ) these values are >= 0 + MATCH_NOMATCH if failed to match ) + a negative MATCH_xxx value for PRUNE, SKIP, etc + a negative PCRE2_ERROR_xxx value if aborted by an error condition + (e.g. stopped by repeated call or recursion limit) +*/ + +static int +match(REGISTER PCRE2_SPTR eptr, REGISTER PCRE2_SPTR ecode, + PCRE2_SPTR mstart, size_t offset_top, match_block *mb, eptrblock *eptrb, + unsigned int rdepth) +{ +/* These variables do not need to be preserved over recursion in this function, +so they can be ordinary variables in all cases. Mark some of them with +"register" because they are used a lot in loops. */ + +register int rrc; /* Returns from recursive calls */ +register int i; /* Used for loops not involving calls to RMATCH() */ +register uint32_t c; /* Character values not kept over RMATCH() calls */ +register BOOL utf; /* Local copy of UTF flag for speed */ + +BOOL minimize, possessive; /* Quantifier options */ +BOOL caseless; +int condcode; + +/* When recursion is not being used, all "local" variables that have to be +preserved over calls to RMATCH() are part of a "frame". We set up the top-level +frame on the stack here; subsequent instantiations are obtained from the heap +whenever RMATCH() does a "recursion". See the macro definitions above. Putting +the top-level on the stack rather than malloc-ing them all gives a performance +boost in many cases where there is not much "recursion". */ + +#ifdef NO_RECURSE +heapframe *frame = (heapframe *)mb->match_frames_base; + +/* Copy in the original argument variables */ + +frame->Xeptr = eptr; +frame->Xecode = ecode; +frame->Xmstart = mstart; +frame->Xoffset_top = offset_top; +frame->Xeptrb = eptrb; +frame->Xrdepth = rdepth; + +/* This is where control jumps back to to effect "recursion" */ + +HEAP_RECURSE: + +/* Macros make the argument variables come from the current frame */ + +#define eptr frame->Xeptr +#define ecode frame->Xecode +#define mstart frame->Xmstart +#define offset_top frame->Xoffset_top +#define eptrb frame->Xeptrb +#define rdepth frame->Xrdepth + +/* Ditto for the local variables */ + +#ifdef SUPPORT_UTF +#define charptr frame->Xcharptr +#endif +#define callpat frame->Xcallpat +#define codelink frame->Xcodelink +#define data frame->Xdata +#define next frame->Xnext +#define pp frame->Xpp +#define prev frame->Xprev +#define saved_eptr frame->Xsaved_eptr + +#define new_recursive frame->Xnew_recursive + +#define cur_is_word frame->Xcur_is_word +#define condition frame->Xcondition +#define prev_is_word frame->Xprev_is_word + +#ifdef SUPPORT_UTF +#define prop_type frame->Xprop_type +#define prop_value frame->Xprop_value +#define prop_fail_result frame->Xprop_fail_result +#define oclength frame->Xoclength +#define occhars frame->Xocchars +#endif + +#define ctype frame->Xctype +#define fc frame->Xfc +#define fi frame->Xfi +#define length frame->Xlength +#define max frame->Xmax +#define min frame->Xmin +#define number frame->Xnumber +#define offset frame->Xoffset +#define op frame->Xop +#define save_capture_last frame->Xsave_capture_last +#define save_offset1 frame->Xsave_offset1 +#define save_offset2 frame->Xsave_offset2 +#define save_offset3 frame->Xsave_offset3 +#define stacksave frame->Xstacksave + +#define newptrb frame->Xnewptrb + +/* When recursion is being used, local variables are allocated on the stack and +get preserved during recursion in the normal way. In this environment, fi and +i, and fc and c, can be the same variables. */ + +#else /* NO_RECURSE not defined */ +#define fi i +#define fc c + +/* Many of the following variables are used only in small blocks of the code. +My normal style of coding would have declared them within each of those blocks. +However, in order to accommodate the version of this code that uses an external +"stack" implemented on the heap, it is easier to declare them all here, so the +declarations can be cut out in a block. The only declarations within blocks +below are for variables that do not have to be preserved over a recursive call +to RMATCH(). */ + +#ifdef SUPPORT_UTF +PCRE2_SPTR charptr; +#endif +PCRE2_SPTR callpat; +PCRE2_SPTR data; +PCRE2_SPTR next; +PCRE2_SPTR pp; +PCRE2_SPTR prev; +PCRE2_SPTR saved_eptr; + +recursion_info new_recursive; + +BOOL cur_is_word; +BOOL condition; +BOOL prev_is_word; + +#ifdef SUPPORT_UTF +int prop_type; +unsigned int prop_value; +int prop_fail_result; +int oclength; +PCRE2_UCHAR occhars[6]; +#endif + +int codelink; +int ctype; +int length; +int max; +int min; +unsigned int number; +size_t offset; +unsigned int op; +uint32_t save_capture_last; +int save_offset1, save_offset2, save_offset3; +size_t stacksave[REC_STACK_SAVE_MAX]; + +eptrblock newptrb; + +/* There is a special fudge for calling match() in a way that causes it to +measure the size of its basic stack frame when the stack is being used for +recursion. The second argument (ecode) being NULL triggers this behaviour. It +cannot normally ever be NULL. The return is the negated value of the frame +size. */ + +if (ecode == NULL) + { + if (rdepth == 0) + return match((PCRE2_SPTR)&rdepth, NULL, NULL, 0, NULL, NULL, 1); + else + { + int len = (char *)&rdepth - (char *)eptr; + return (len > 0)? -len : len; + } + } +#endif /* NO_RECURSE */ + +/* To save space on the stack and in the heap frame, I have doubled up on some +of the local variables that are used only in localised parts of the code, but +still need to be preserved over recursive calls of match(). These macros define +the alternative names that are used. */ + +#define allow_zero cur_is_word +#define cbegroup condition +#define code_offset codelink +#define condassert condition +#define matched_once prev_is_word +#define foc number +#define save_mark data + +/* These statements are here to stop the compiler complaining about unitialized +variables. */ + +#ifdef SUPPORT_UTF +prop_value = 0; +prop_fail_result = 0; +#endif + + +/* This label is used for tail recursion, which is used in a few cases even +when NO_RECURSE is not defined, in order to reduce the amount of stack that is +used. Thanks to Ian Taylor for noticing this possibility and sending the +original patch. */ + +TAIL_RECURSE: + +/* OK, now we can get on with the real code of the function. Recursive calls +are specified by the macro RMATCH and RRETURN is used to return. When +NO_RECURSE is *not* defined, these just turn into a recursive call to match() +and a "return", respectively. However, RMATCH isn't like a function call +because it's quite a complicated macro. It has to be used in one particular +way. This shouldn't, however, impact performance when true recursion is being +used. */ + +#ifdef SUPPORT_UTF +utf = (mb->poptions & PCRE2_UTF) != 0; +#else +utf = FALSE; +#endif + +/* First check that we haven't called match() too many times, or that we +haven't exceeded the recursive call limit. */ + +if (mb->match_call_count++ >= mb->match_limit) RRETURN(PCRE2_ERROR_MATCHLIMIT); +if (rdepth >= mb->match_limit_recursion) RRETURN(PCRE2_ERROR_RECURSIONLIMIT); + +/* At the start of a group with an unlimited repeat that may match an empty +string, the variable mb->match_function_type is set to MATCH_CBEGROUP. It is +done this way to save having to use another function argument, which would take +up space on the stack. See also MATCH_CONDASSERT below. + +When MATCH_CBEGROUP is set, add the current subject pointer to the chain of +such remembered pointers, to be checked when we hit the closing ket, in order +to break infinite loops that match no characters. When match() is called in +other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must +NOT be used with tail recursion, because the memory block that is used is on +the stack, so a new one may be required for each match(). */ + +if (mb->match_function_type == MATCH_CBEGROUP) + { + newptrb.epb_saved_eptr = eptr; + newptrb.epb_prev = eptrb; + eptrb = &newptrb; + mb->match_function_type = 0; + } + +/* Now start processing the opcodes. */ + +for (;;) + { + minimize = possessive = FALSE; + op = *ecode; + + switch(op) + { + case OP_MARK: + mb->nomatch_mark = ecode + 2; + mb->mark = NULL; /* In case previously set by assertion */ + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, mb, + eptrb, RM55); + if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && + mb->mark == NULL) mb->mark = ecode + 2; + + /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an + argument, and we must check whether that argument matches this MARK's + argument. It is passed back in mb->start_match_ptr (an overloading of that + variable). If it does match, we reset that variable to the current subject + position and return MATCH_SKIP. Otherwise, pass back the return code + unaltered. */ + + else if (rrc == MATCH_SKIP_ARG && + PRIV(strcmp)(ecode + 2, mb->start_match_ptr) == 0) + { + mb->start_match_ptr = eptr; + RRETURN(MATCH_SKIP); + } + RRETURN(rrc); + + case OP_FAIL: + RRETURN(MATCH_NOMATCH); + + case OP_COMMIT: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM52); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + RRETURN(MATCH_COMMIT); + + case OP_PRUNE: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM51); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + RRETURN(MATCH_PRUNE); + + case OP_PRUNE_ARG: + mb->nomatch_mark = ecode + 2; + mb->mark = NULL; /* In case previously set by assertion */ + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, mb, + eptrb, RM56); + if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && + mb->mark == NULL) mb->mark = ecode + 2; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + RRETURN(MATCH_PRUNE); + + case OP_SKIP: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM53); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->start_match_ptr = eptr; /* Pass back current position */ + RRETURN(MATCH_SKIP); + + /* Note that, for Perl compatibility, SKIP with an argument does NOT set + nomatch_mark. When a pattern match ends with a SKIP_ARG for which there was + not a matching mark, we have to re-run the match, ignoring the SKIP_ARG + that failed and any that precede it (either they also failed, or were not + triggered). To do this, we maintain a count of executed SKIP_ARGs. If a + SKIP_ARG gets to top level, the match is re-run with mb->ignore_skip_arg + set to the count of the one that failed. */ + + case OP_SKIP_ARG: + mb->skip_arg_count++; + if (mb->skip_arg_count <= mb->ignore_skip_arg) + { + ecode += PRIV(OP_lengths)[*ecode] + ecode[1]; + break; + } + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, mb, + eptrb, RM57); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + + /* Pass back the current skip name by overloading mb->start_match_ptr and + returning the special MATCH_SKIP_ARG return code. This will either be + caught by a matching MARK, or get to the top, where it causes a rematch + with mb->ignore_skip_arg set to the value of mb->skip_arg_count. */ + + mb->start_match_ptr = ecode + 2; + RRETURN(MATCH_SKIP_ARG); + + /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that + the branch in which it occurs can be determined. Overload the start of + match pointer to do this. */ + + case OP_THEN: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM54); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->start_match_ptr = ecode; + RRETURN(MATCH_THEN); + + case OP_THEN_ARG: + mb->nomatch_mark = ecode + 2; + mb->mark = NULL; /* In case previously set by assertion */ + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, + mb, eptrb, RM58); + if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && + mb->mark == NULL) mb->mark = ecode + 2; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->start_match_ptr = ecode; + RRETURN(MATCH_THEN); + + /* Handle an atomic group that does not contain any capturing parentheses. + This can be handled like an assertion. Prior to 8.13, all atomic groups + were handled this way. In 8.13, the code was changed as below for ONCE, so + that backups pass through the group and thereby reset captured values. + However, this uses a lot more stack, so in 8.20, atomic groups that do not + contain any captures generate OP_ONCE_NC, which can be handled in the old, + less stack intensive way. + + Check the alternative branches in turn - the matching won't pass the KET + for this kind of subpattern. If any one branch matches, we carry on as at + the end of a normal bracket, leaving the subject pointer, but resetting + the start-of-match value in case it was changed by \K. */ + + case OP_ONCE_NC: + prev = ecode; + saved_eptr = eptr; + save_mark = mb->mark; + do + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, eptrb, RM64); + if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ + { + mstart = mb->start_match_ptr; + break; + } + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode,1); + mb->mark = save_mark; + } + while (*ecode == OP_ALT); + + /* If hit the end of the group (which could be repeated), fail */ + + if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH); + + /* Continue as from after the group, updating the offsets high water + mark, since extracts may have been taken. */ + + do ecode += GET(ecode, 1); while (*ecode == OP_ALT); + + offset_top = mb->end_offset_top; + eptr = mb->end_match_ptr; + + /* For a non-repeating ket, just continue at this level. This also + happens for a repeating ket if no characters were matched in the group. + This is the forcible breaking of infinite loops as implemented in Perl + 5.005. */ + + if (*ecode == OP_KET || eptr == saved_eptr) + { + ecode += 1+LINK_SIZE; + break; + } + + /* The repeating kets try the rest of the pattern or restart from the + preceding bracket, in the appropriate order. The second "call" of match() + uses tail recursion, to avoid using another stack frame. */ + + if (*ecode == OP_KETRMIN) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, eptrb, RM65); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode = prev; + goto TAIL_RECURSE; + } + else /* OP_KETRMAX */ + { + RMATCH(eptr, prev, offset_top, mb, eptrb, RM66); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += 1 + LINK_SIZE; + goto TAIL_RECURSE; + } + /* Control never gets here */ + + /* Handle a capturing bracket, other than those that are possessive with an + unlimited repeat. If there is space in the offset vector, save the current + subject position in the working slot at the top of the vector. We mustn't + change the current values of the data slot, because they may be set from a + previous iteration of this group, and be referred to by a reference inside + the group. A failure to match might occur after the group has succeeded, + if something later on doesn't match. For this reason, we need to restore + the working value and also the values of the final offsets, in case they + were set by a previous iteration of the same bracket. + + If there isn't enough space in the offset vector, treat this as if it were + a non-capturing bracket. Don't worry about setting the flag for the error + case here; that is handled in the code for KET. */ + + case OP_CBRA: + case OP_SCBRA: + number = GET2(ecode, 1+LINK_SIZE); + offset = number << 1; + + if (offset < mb->offset_max) + { + save_offset1 = mb->ovector[offset]; + save_offset2 = mb->ovector[offset+1]; + save_offset3 = mb->ovector[mb->offset_end - number]; + save_capture_last = mb->capture_last; + save_mark = mb->mark; + + mb->ovector[mb->offset_end - number] = eptr - mb->start_subject; + + for (;;) + { + if (op >= OP_SBRA) mb->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM1); + if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ + + /* If we backed up to a THEN, check whether it is within the current + branch by comparing the address of the THEN that is passed back with + the end of the branch. If it is within the current branch, and the + branch is one of two or more alternatives (it either starts or ends + with OP_ALT), we have reached the limit of THEN's action, so convert + the return code to NOMATCH, which will cause normal backtracking to + happen from now on. Otherwise, THEN is passed back to an outer + alternative. This implements Perl's treatment of parenthesized groups, + where a group not containing | does not affect the current alternative, + that is, (X) is NOT the same as (X|(*F)). */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + /* Anything other than NOMATCH is passed back. */ + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->capture_last = save_capture_last; + ecode += GET(ecode, 1); + mb->mark = save_mark; + if (*ecode != OP_ALT) break; + } + + mb->ovector[offset] = save_offset1; + mb->ovector[offset+1] = save_offset2; + mb->ovector[mb->offset_end - number] = save_offset3; + + /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */ + + RRETURN(rrc); + } + + /* FALL THROUGH ... Insufficient room for saving captured contents. Treat + as a non-capturing bracket. */ + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + + /* Non-capturing or atomic group, except for possessive with unlimited + repeat and ONCE group with no captures. Loop for all the alternatives. + + When we get to the final alternative within the brackets, we used to return + the result of a recursive call to match() whatever happened so it was + possible to reduce stack usage by turning this into a tail recursion, + except in the case of a possibly empty group. However, now that there is + the possiblity of (*THEN) occurring in the final alternative, this + optimization is no longer always possible. + + We can optimize if we know there are no (*THEN)s in the pattern; at present + this is the best that can be done. + + MATCH_ONCE is returned when the end of an atomic group is successfully + reached, but subsequent matching fails. It passes back up the tree (causing + captured values to be reset) until the original atomic group level is + reached. This is tested by comparing mb->once_target with the start of the + group. At this point, the return is converted into MATCH_NOMATCH so that + previous backup points can be taken. */ + + case OP_ONCE: + case OP_BRA: + case OP_SBRA: + + for (;;) + { + if (op >= OP_SBRA || op == OP_ONCE) + mb->match_function_type = MATCH_CBEGROUP; + + /* If this is not a possibly empty group, and there are no (*THEN)s in + the pattern, and this is the final alternative, optimize as described + above. */ + + else if (!mb->hasthen && ecode[GET(ecode, 1)] != OP_ALT) + { + ecode += PRIV(OP_lengths)[*ecode]; + goto TAIL_RECURSE; + } + + /* In all other cases, we have to make another call to match(). */ + + save_mark = mb->mark; + save_capture_last = mb->capture_last; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, eptrb, + RM2); + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) + { + if (rrc == MATCH_ONCE) + { + PCRE2_SPTR scode = ecode; + if (*scode != OP_ONCE) /* If not at start, find it */ + { + while (*scode == OP_ALT) scode += GET(scode, 1); + scode -= GET(scode, 1); + } + if (mb->once_target == scode) rrc = MATCH_NOMATCH; + } + RRETURN(rrc); + } + ecode += GET(ecode, 1); + mb->mark = save_mark; + if (*ecode != OP_ALT) break; + mb->capture_last = save_capture_last; + } + + RRETURN(MATCH_NOMATCH); + + /* Handle possessive capturing brackets with an unlimited repeat. We come + here from BRAZERO with allow_zero set TRUE. The ovector values are + handled similarly to the normal case above. However, the matching is + different. The end of these brackets will always be OP_KETRPOS, which + returns MATCH_KETRPOS without going further in the pattern. By this means + we can handle the group by iteration rather than recursion, thereby + reducing the amount of stack needed. */ + + case OP_CBRAPOS: + case OP_SCBRAPOS: + allow_zero = FALSE; + + POSSESSIVE_CAPTURE: + number = GET2(ecode, 1+LINK_SIZE); + offset = number << 1; + + if (offset < mb->offset_max) + { + matched_once = FALSE; + code_offset = (int)(ecode - mb->start_code); + + save_offset1 = mb->ovector[offset]; + save_offset2 = mb->ovector[offset+1]; + save_offset3 = mb->ovector[mb->offset_end - number]; + save_capture_last = mb->capture_last; + + /* Each time round the loop, save the current subject position for use + when the group matches. For MATCH_MATCH, the group has matched, so we + restart it with a new subject starting position, remembering that we had + at least one match. For MATCH_NOMATCH, carry on with the alternatives, as + usual. If we haven't matched any alternatives in any iteration, check to + see if a previous iteration matched. If so, the group has matched; + continue from afterwards. Otherwise it has failed; restore the previous + capture values before returning NOMATCH. */ + + for (;;) + { + mb->ovector[mb->offset_end - number] = eptr - mb->start_subject; + if (op >= OP_SBRA) mb->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM63); + if (rrc == MATCH_KETRPOS) + { + offset_top = mb->end_offset_top; + ecode = mb->start_code + code_offset; + save_capture_last = mb->capture_last; + matched_once = TRUE; + mstart = mb->start_match_ptr; /* In case \K changed it */ + if (eptr == mb->end_match_ptr) /* Matched an empty string */ + { + do ecode += GET(ecode, 1); while (*ecode == OP_ALT); + break; + } + eptr = mb->end_match_ptr; + continue; + } + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->capture_last = save_capture_last; + ecode += GET(ecode, 1); + if (*ecode != OP_ALT) break; + } + + if (!matched_once) + { + mb->ovector[offset] = save_offset1; + mb->ovector[offset+1] = save_offset2; + mb->ovector[mb->offset_end - number] = save_offset3; + } + + if (allow_zero || matched_once) + { + ecode += 1 + LINK_SIZE; + break; + } + + RRETURN(MATCH_NOMATCH); + } + + /* FALL THROUGH ... Insufficient room for saving captured contents. Treat + as a non-capturing bracket. */ + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + + /* Non-capturing possessive bracket with unlimited repeat. We come here + from BRAZERO with allow_zero = TRUE. The code is similar to the above, + without the capturing complication. It is written out separately for speed + and cleanliness. */ + + case OP_BRAPOS: + case OP_SBRAPOS: + allow_zero = FALSE; + + POSSESSIVE_NON_CAPTURE: + matched_once = FALSE; + code_offset = (int)(ecode - mb->start_code); + save_capture_last = mb->capture_last; + + for (;;) + { + if (op >= OP_SBRA) mb->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, mb, + eptrb, RM48); + if (rrc == MATCH_KETRPOS) + { + offset_top = mb->end_offset_top; + ecode = mb->start_code + code_offset; + matched_once = TRUE; + mstart = mb->start_match_ptr; /* In case \K reset it */ + if (eptr == mb->end_match_ptr) /* Matched an empty string */ + { + do ecode += GET(ecode, 1); while (*ecode == OP_ALT); + break; + } + eptr = mb->end_match_ptr; + continue; + } + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode, 1); + if (*ecode != OP_ALT) break; + mb->capture_last = save_capture_last; + } + + if (matched_once || allow_zero) + { + ecode += 1 + LINK_SIZE; + break; + } + RRETURN(MATCH_NOMATCH); + + /* Control never reaches here. */ + + /* Conditional group: compilation checked that there are no more than two + branches. If the condition is false, skipping the first branch takes us + past the end of the item if there is only one branch, but that's exactly + what we want. */ + + case OP_COND: + case OP_SCOND: + + /* The variable codelink will be added to ecode when the condition is + false, to get to the second branch. Setting it to the offset to the ALT + or KET, then incrementing ecode achieves this effect. We now have ecode + pointing to the condition or callout. */ + + codelink = GET(ecode, 1); /* Offset to the second branch */ + ecode += 1 + LINK_SIZE; /* From this opcode */ + + /* Because of the way auto-callout works during compile, a callout item is + inserted between OP_COND and an assertion condition. */ + + if (*ecode == OP_CALLOUT) + { + if (mb->callout != NULL) + { + pcre2_callout_block cb; + cb.version = 2; /* Version 1 of the callout block */ + cb.callout_number = ecode[1]; + cb.offset_vector = mb->ovector; + cb.subject = mb->start_subject; + cb.subject_length = (int)(mb->end_subject - mb->start_subject); + cb.start_match = (int)(mstart - mb->start_subject); + cb.current_position = (int)(eptr - mb->start_subject); + cb.pattern_position = GET(ecode, 2); + cb.next_item_length = GET(ecode, 2 + LINK_SIZE); + cb.capture_top = offset_top/2; + cb.capture_last = mb->capture_last & CAPLMASK; + cb.callout_data = mb->callout_data; + cb.mark = mb->nomatch_mark; + if ((rrc = mb->callout(&cb)) > 0) RRETURN(MATCH_NOMATCH); + if (rrc < 0) RRETURN(rrc); + } + + /* Advance ecode past the callout, so it now points to the condition. We + must adjust codelink so that the value of ecode+codelink is unchanged. */ + + ecode += PRIV(OP_lengths)[OP_CALLOUT]; + codelink -= PRIV(OP_lengths)[OP_CALLOUT]; + } + + /* Test the various possible conditions */ + + condition = FALSE; + switch(condcode = *ecode) + { + case OP_RREF: /* Numbered group recursion test */ + if (mb->recursive != NULL) /* Not recursing => FALSE */ + { + unsigned int recno = GET2(ecode, 1); /* Recursion group number*/ + condition = (recno == RREF_ANY || recno == mb->recursive->group_num); + } + break; + + case OP_DNRREF: /* Duplicate named group recursion test */ + if (mb->recursive != NULL) + { + int count = GET2(ecode, 1 + IMM2_SIZE); + PCRE2_SPTR slot = mb->name_table + GET2(ecode, 1) * mb->name_entry_size; + while (count-- > 0) + { + unsigned int recno = GET2(slot, 0); + condition = recno == mb->recursive->group_num; + if (condition) break; + slot += mb->name_entry_size; + } + } + break; + + case OP_CREF: /* Numbered group used test */ + offset = GET2(ecode, 1) << 1; /* Doubled ref number */ + condition = offset < offset_top && + mb->ovector[offset] != PCRE2_UNSET; + break; + + case OP_DNCREF: /* Duplicate named group used test */ + { + int count = GET2(ecode, 1 + IMM2_SIZE); + PCRE2_SPTR slot = mb->name_table + GET2(ecode, 1) * mb->name_entry_size; + while (count-- > 0) + { + offset = GET2(slot, 0) << 1; + condition = offset < offset_top && + mb->ovector[offset] != PCRE2_UNSET; + if (condition) break; + slot += mb->name_entry_size; + } + } + break; + + case OP_DEF: /* DEFINE - always false */ + break; + + /* The condition is an assertion. Call match() to evaluate it - setting + mb->match_function_type to MATCH_CONDASSERT causes it to stop at the end + of an assertion. */ + + default: + mb->match_function_type = MATCH_CONDASSERT; + RMATCH(eptr, ecode, offset_top, mb, NULL, RM3); + if (rrc == MATCH_MATCH) + { + if (mb->end_offset_top > offset_top) + offset_top = mb->end_offset_top; /* Captures may have happened */ + condition = TRUE; + + /* Advance ecode past the assertion to the start of the first branch, + but adjust it so that the general choosing code below works. */ + + ecode += GET(ecode, 1); + while (*ecode == OP_ALT) ecode += GET(ecode, 1); + ecode += 1 + LINK_SIZE - PRIV(OP_lengths)[condcode]; + } + + /* PCRE doesn't allow the effect of (*THEN) to escape beyond an + assertion; it is therefore treated as NOMATCH. Any other return is an + error. */ + + else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) + { + RRETURN(rrc); /* Need braces because of following else */ + } + break; + } + + /* Choose branch according to the condition */ + + ecode += condition? PRIV(OP_lengths)[condcode] : codelink; + + /* We are now at the branch that is to be obeyed. As there is only one, we + can use tail recursion to avoid using another stack frame, except when + there is unlimited repeat of a possibly empty group. In the latter case, a + recursive call to match() is always required, unless the second alternative + doesn't exist, in which case we can just plough on. Note that, for + compatibility with Perl, the | in a conditional group is NOT treated as + creating two alternatives. If a THEN is encountered in the branch, it + propagates out to the enclosing alternative (unless nested in a deeper set + of alternatives, of course). */ + + if (condition || ecode[-(1+LINK_SIZE)] == OP_ALT) + { + if (op != OP_SCOND) + { + goto TAIL_RECURSE; + } + + mb->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM49); + RRETURN(rrc); + } + + /* Condition false & no alternative; continue after the group. */ + + else + { + } + break; + + + /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, + to close any currently open capturing brackets. */ + + case OP_CLOSE: + number = GET2(ecode, 1); /* Must be less than 65536 */ + offset = number << 1; + mb->capture_last = (mb->capture_last & OVFLMASK) | number; + if (offset >= mb->offset_max) mb->capture_last |= OVFLBIT; else + { + mb->ovector[offset] = + mb->ovector[mb->offset_end - number]; + mb->ovector[offset+1] = eptr - mb->start_subject; + if (offset_top <= offset) offset_top = offset + 2; + } + ecode += 1 + IMM2_SIZE; + break; + + + /* End of the pattern, either real or forced. */ + + case OP_END: + case OP_ACCEPT: + case OP_ASSERT_ACCEPT: + + /* If we have matched an empty string, fail if not in an assertion and not + in a recursion if either PCRE2_NOTEMPTY is set, or if PCRE2_NOTEMPTY_ATSTART + is set and we have matched at the start of the subject. In both cases, + backtracking will then try other alternatives, if any. */ + + if (eptr == mstart && op != OP_ASSERT_ACCEPT && + mb->recursive == NULL && + ((mb->moptions & PCRE2_NOTEMPTY) != 0 || + ((mb->moptions & PCRE2_NOTEMPTY_ATSTART) != 0 && + mstart == mb->start_subject + mb->start_offset))) + RRETURN(MATCH_NOMATCH); + + /* Otherwise, we have a match. */ + + mb->end_match_ptr = eptr; /* Record where we ended */ + mb->end_offset_top = offset_top; /* and how many extracts were taken */ + mb->start_match_ptr = mstart; /* and the start (\K can modify) */ + + /* For some reason, the macros don't work properly if an expression is + given as the argument to RRETURN when the heap is in use. */ + + rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; + RRETURN(rrc); + + /* Assertion brackets. Check the alternative branches in turn - the + matching won't pass the KET for an assertion. If any one branch matches, + the assertion is true. Lookbehind assertions have an OP_REVERSE item at the + start of each branch to move the current point backwards, so the code at + this level is identical to the lookahead case. When the assertion is part + of a condition, we want to return immediately afterwards. The caller of + this incarnation of the match() function will have set MATCH_CONDASSERT in + mb->match_function type, and one of these opcodes will be the first opcode + that is processed. We use a local variable that is preserved over calls to + match() to remember this case. */ + + case OP_ASSERT: + case OP_ASSERTBACK: + save_mark = mb->mark; + if (mb->match_function_type == MATCH_CONDASSERT) + { + condassert = TRUE; + mb->match_function_type = 0; + } + else condassert = FALSE; + + /* Loop for each branch */ + + do + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, NULL, RM4); + + /* A match means that the assertion is true; break out of the loop + that matches its alternatives. */ + + if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) + { + mstart = mb->start_match_ptr; /* In case \K reset it */ + break; + } + + /* If not matched, restore the previous mark setting. */ + + mb->mark = save_mark; + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + /* Anything other than NOMATCH causes the entire assertion to fail, + passing back the return code. This includes COMMIT, SKIP, PRUNE and an + uncaptured THEN, which means they take their normal effect. This + consistent approach does not always have exactly the same effect as in + Perl. */ + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode, 1); + } + while (*ecode == OP_ALT); /* Continue for next alternative */ + + /* If we have tried all the alternative branches, the assertion has + failed. If not, we broke out after a match. */ + + if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); + + /* If checking an assertion for a condition, return MATCH_MATCH. */ + + if (condassert) RRETURN(MATCH_MATCH); + + /* Continue from after a successful assertion, updating the offsets high + water mark, since extracts may have been taken during the assertion. */ + + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + ecode += 1 + LINK_SIZE; + offset_top = mb->end_offset_top; + continue; + + /* Negative assertion: all branches must fail to match for the assertion to + succeed. */ + + case OP_ASSERT_NOT: + case OP_ASSERTBACK_NOT: + save_mark = mb->mark; + if (mb->match_function_type == MATCH_CONDASSERT) + { + condassert = TRUE; + mb->match_function_type = 0; + } + else condassert = FALSE; + + /* Loop for each alternative branch. */ + + do + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, NULL, RM5); + mb->mark = save_mark; /* Always restore the mark setting */ + + switch(rrc) + { + case MATCH_MATCH: /* A successful match means */ + case MATCH_ACCEPT: /* the assertion has failed. */ + RRETURN(MATCH_NOMATCH); + + case MATCH_NOMATCH: /* Carry on with next branch */ + break; + + /* See comment in the code for capturing groups above about handling + THEN. */ + + case MATCH_THEN: + next = ecode + GET(ecode,1); + if (mb->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + { + rrc = MATCH_NOMATCH; + break; + } + /* Otherwise fall through. */ + + /* COMMIT, SKIP, PRUNE, and an uncaptured THEN cause the whole + assertion to fail to match, without considering any more alternatives. + Failing to match means the assertion is true. This is a consistent + approach, but does not always have the same effect as in Perl. */ + + case MATCH_COMMIT: + case MATCH_SKIP: + case MATCH_SKIP_ARG: + case MATCH_PRUNE: + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + goto NEG_ASSERT_TRUE; /* Break out of alternation loop */ + + /* Anything else is an error */ + + default: + RRETURN(rrc); + } + + /* Continue with next branch */ + + ecode += GET(ecode,1); + } + while (*ecode == OP_ALT); + + /* All branches in the assertion failed to match. */ + + NEG_ASSERT_TRUE: + if (condassert) RRETURN(MATCH_MATCH); /* Condition assertion */ + ecode += 1 + LINK_SIZE; /* Continue with current branch */ + continue; + + /* Move the subject pointer back. This occurs only at the start of + each branch of a lookbehind assertion. If we are too close to the start to + move back, this match function fails. When working with UTF-8 we move + back a number of characters, not bytes. */ + + case OP_REVERSE: +#ifdef SUPPORT_UTF + if (utf) + { + i = GET(ecode, 1); + while (i-- > 0) + { + eptr--; + if (eptr < mb->start_subject) RRETURN(MATCH_NOMATCH); + BACKCHAR(eptr); + } + } + else +#endif + + /* No UTF-8 support, or not in UTF-8 mode: count is byte count */ + + { + eptr -= GET(ecode, 1); + if (eptr < mb->start_subject) RRETURN(MATCH_NOMATCH); + } + + /* Save the earliest consulted character, then skip to next op code */ + + if (eptr < mb->start_used_ptr) mb->start_used_ptr = eptr; + ecode += 1 + LINK_SIZE; + break; + + /* The callout item calls an external function, if one is provided, passing + details of the match so far. This is mainly for debugging, though the + function is able to force a failure. */ + + case OP_CALLOUT: + if (mb->callout != NULL) + { + pcre2_callout_block cb; + cb.version = 2; /* Version 1 of the callout block */ + cb.callout_number = ecode[1]; + cb.offset_vector = mb->ovector; + cb.subject = mb->start_subject; + cb.subject_length = (int)(mb->end_subject - mb->start_subject); + cb.start_match = (int)(mstart - mb->start_subject); + cb.current_position = (int)(eptr - mb->start_subject); + cb.pattern_position = GET(ecode, 2); + cb.next_item_length = GET(ecode, 2 + LINK_SIZE); + cb.capture_top = offset_top/2; + cb.capture_last = mb->capture_last & CAPLMASK; + cb.callout_data = mb->callout_data; + cb.mark = mb->nomatch_mark; + if ((rrc = mb->callout(&cb)) > 0) RRETURN(MATCH_NOMATCH); + if (rrc < 0) RRETURN(rrc); + } + ecode += 2 + 2*LINK_SIZE; + break; + + /* Recursion either matches the current regex, or some subexpression. The + offset data is the offset to the starting bracket from the start of the + whole pattern. (This is so that it works from duplicated subpatterns.) + + The state of the capturing groups is preserved over recursion, and + re-instated afterwards. We don't know how many are started and not yet + finished (offset_top records the completed total) so we just have to save + all the potential data. There may be up to 65535 such values, which is too + large to put on the stack, but using malloc for small numbers seems + expensive. As a compromise, the stack is used when there are no more than + REC_STACK_SAVE_MAX values to store; otherwise malloc is used. + + There are also other values that have to be saved. We use a chained + sequence of blocks that actually live on the stack. Thanks to Robin Houston + for the original version of this logic. It has, however, been hacked around + a lot, so he is not to blame for the current way it works. */ + + case OP_RECURSE: + { + recursion_info *ri; + unsigned int recno; + + callpat = mb->start_code + GET(ecode, 1); + recno = (callpat == mb->start_code)? 0 : + GET2(callpat, 1 + LINK_SIZE); + + /* Check for repeating a recursion without advancing the subject pointer. + This should catch convoluted mutual recursions. (Some simple cases are + caught at compile time.) */ + + for (ri = mb->recursive; ri != NULL; ri = ri->prevrec) + if (recno == ri->group_num && eptr == ri->subject_position) + RRETURN(PCRE2_ERROR_RECURSELOOP); + + /* Add to "recursing stack" */ + + new_recursive.group_num = recno; + new_recursive.saved_capture_last = mb->capture_last; + new_recursive.subject_position = eptr; + new_recursive.prevrec = mb->recursive; + mb->recursive = &new_recursive; + + /* Where to continue from afterwards */ + + ecode += 1 + LINK_SIZE; + + /* Now save the offset data */ + + new_recursive.saved_max = mb->offset_end; + if (new_recursive.saved_max <= REC_STACK_SAVE_MAX) + new_recursive.offset_save = stacksave; + else + { + new_recursive.offset_save = (size_t *) + (mb->memctl.malloc(new_recursive.saved_max * sizeof(size_t), + mb->memctl.memory_data)); + if (new_recursive.offset_save == NULL) RRETURN(PCRE2_ERROR_NOMEMORY); + } + memcpy(new_recursive.offset_save, mb->ovector, + new_recursive.saved_max * sizeof(size_t)); + + /* OK, now we can do the recursion. After processing each alternative, + restore the offset data and the last captured value. If there were nested + recursions, mb->recursive might be changed, so reset it before looping. + */ + + cbegroup = (*callpat >= OP_SBRA); + do + { + if (cbegroup) mb->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top, + mb, eptrb, RM6); + memcpy(mb->ovector, new_recursive.offset_save, + new_recursive.saved_max * sizeof(size_t)); + mb->capture_last = new_recursive.saved_capture_last; + mb->recursive = new_recursive.prevrec; + if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) + { + if (new_recursive.offset_save != stacksave) + mb->memctl.free(new_recursive.offset_save, mb->memctl.memory_data); + + /* Set where we got to in the subject, and reset the start in case + it was changed by \K. This *is* propagated back out of a recursion, + for Perl compatibility. */ + + eptr = mb->end_match_ptr; + mstart = mb->start_match_ptr; + goto RECURSION_MATCHED; /* Exit loop; end processing */ + } + + /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a + recursion; they cause a NOMATCH for the entire recursion. These codes + are defined in a range that can be tested for. */ + + if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX) + RRETURN(MATCH_NOMATCH); + + /* Any return code other than NOMATCH is an error. */ + + if (rrc != MATCH_NOMATCH) + { + if (new_recursive.offset_save != stacksave) + mb->memctl.free(new_recursive.offset_save, mb->memctl.memory_data); + RRETURN(rrc); + } + + mb->recursive = &new_recursive; + callpat += GET(callpat, 1); + } + while (*callpat == OP_ALT); + + mb->recursive = new_recursive.prevrec; + if (new_recursive.offset_save != stacksave) + mb->memctl.free(new_recursive.offset_save, mb->memctl.memory_data); + RRETURN(MATCH_NOMATCH); + } + + RECURSION_MATCHED: + break; + + /* An alternation is the end of a branch; scan along to find the end of the + bracketed group and go to there. */ + + case OP_ALT: + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + break; + + /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group, + indicating that it may occur zero times. It may repeat infinitely, or not + at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets + with fixed upper repeat limits are compiled as a number of copies, with the + optional ones preceded by BRAZERO or BRAMINZERO. */ + + case OP_BRAZERO: + next = ecode + 1; + RMATCH(eptr, next, offset_top, mb, eptrb, RM10); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + do next += GET(next, 1); while (*next == OP_ALT); + ecode = next + 1 + LINK_SIZE; + break; + + case OP_BRAMINZERO: + next = ecode + 1; + do next += GET(next, 1); while (*next == OP_ALT); + RMATCH(eptr, next + 1+LINK_SIZE, offset_top, mb, eptrb, RM11); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode++; + break; + + case OP_SKIPZERO: + next = ecode+1; + do next += GET(next,1); while (*next == OP_ALT); + ecode = next + 1 + LINK_SIZE; + break; + + /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything + here; just jump to the group, with allow_zero set TRUE. */ + + case OP_BRAPOSZERO: + op = *(++ecode); + allow_zero = TRUE; + if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE; + goto POSSESSIVE_NON_CAPTURE; + + /* End of a group, repeated or non-repeating. */ + + case OP_KET: + case OP_KETRMIN: + case OP_KETRMAX: + case OP_KETRPOS: + prev = ecode - GET(ecode, 1); + + /* If this was a group that remembered the subject start, in order to break + infinite repeats of empty string matches, retrieve the subject start from + the chain. Otherwise, set it NULL. */ + + if (*prev >= OP_SBRA || *prev == OP_ONCE) + { + saved_eptr = eptrb->epb_saved_eptr; /* Value at start of group */ + eptrb = eptrb->epb_prev; /* Backup to previous group */ + } + else saved_eptr = NULL; + + /* If we are at the end of an assertion group or a non-capturing atomic + group, stop matching and return MATCH_MATCH, but record the current high + water mark for use by positive assertions. We also need to record the match + start in case it was changed by \K. */ + + if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) || + *prev == OP_ONCE_NC) + { + mb->end_match_ptr = eptr; /* For ONCE_NC */ + mb->end_offset_top = offset_top; + mb->start_match_ptr = mstart; + RRETURN(MATCH_MATCH); /* Sets mb->mark */ + } + + /* For capturing groups we have to check the group number back at the start + and if necessary complete handling an extraction by setting the offsets and + bumping the high water mark. Whole-pattern recursion is coded as a recurse + into group 0, so it won't be picked up here. Instead, we catch it when the + OP_END is reached. Other recursion is handled here. We just have to record + the current subject position and start match pointer and give a MATCH + return. */ + + if (*prev == OP_CBRA || *prev == OP_SCBRA || + *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS) + { + number = GET2(prev, 1+LINK_SIZE); + offset = number << 1; + + /* Handle a recursively called group. */ + + if (mb->recursive != NULL && mb->recursive->group_num == number) + { + mb->end_match_ptr = eptr; + mb->start_match_ptr = mstart; + RRETURN(MATCH_MATCH); + } + + /* Deal with capturing */ + + mb->capture_last = (mb->capture_last & OVFLMASK) | number; + if (offset >= mb->offset_max) mb->capture_last |= OVFLBIT; else + { + /* If offset is greater than offset_top, it means that we are + "skipping" a capturing group, and that group's offsets must be marked + unset. In earlier versions of PCRE, all the offsets were unset at the + start of matching, but this doesn't work because atomic groups and + assertions can cause a value to be set that should later be unset. + Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as + part of the atomic group, but this is not on the final matching path, + so must be unset when 2 is set. (If there is no group 2, there is no + problem, because offset_top will then be 2, indicating no capture.) */ + + if (offset > offset_top) + { + register size_t *iptr = mb->ovector + offset_top; + register size_t *iend = mb->ovector + offset; + while (iptr < iend) *iptr++ = PCRE2_UNSET; + } + + /* Now make the extraction */ + + mb->ovector[offset] = mb->ovector[mb->offset_end - number]; + mb->ovector[offset+1] = eptr - mb->start_subject; + if (offset_top <= offset) offset_top = offset + 2; + } + } + + /* OP_KETRPOS is a possessive repeating ket. Remember the current position, + and return the MATCH_KETRPOS. This makes it possible to do the repeats one + at a time from the outer level, thus saving stack. This must precede the + empty string test - in this case that test is done at the outer level. */ + + if (*ecode == OP_KETRPOS) + { + mb->start_match_ptr = mstart; /* In case \K reset it */ + mb->end_match_ptr = eptr; + mb->end_offset_top = offset_top; + RRETURN(MATCH_KETRPOS); + } + + /* For an ordinary non-repeating ket, just continue at this level. This + also happens for a repeating ket if no characters were matched in the + group. This is the forcible breaking of infinite loops as implemented in + Perl 5.005. For a non-repeating atomic group that includes captures, + establish a backup point by processing the rest of the pattern at a lower + level. If this results in a NOMATCH return, pass MATCH_ONCE back to the + original OP_ONCE level, thereby bypassing intermediate backup points, but + resetting any captures that happened along the way. */ + + if (*ecode == OP_KET || eptr == saved_eptr) + { + if (*prev == OP_ONCE) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, eptrb, RM12); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ + RRETURN(MATCH_ONCE); + } + ecode += 1 + LINK_SIZE; /* Carry on at this level */ + break; + } + + /* The normal repeating kets try the rest of the pattern or restart from + the preceding bracket, in the appropriate order. In the second case, we can + use tail recursion to avoid using another stack frame, unless we have an + an atomic group or an unlimited repeat of a group that can match an empty + string. */ + + if (*ecode == OP_KETRMIN) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, eptrb, RM7); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (*prev == OP_ONCE) + { + RMATCH(eptr, prev, offset_top, mb, eptrb, RM8); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ + RRETURN(MATCH_ONCE); + } + if (*prev >= OP_SBRA) /* Could match an empty string */ + { + RMATCH(eptr, prev, offset_top, mb, eptrb, RM50); + RRETURN(rrc); + } + ecode = prev; + goto TAIL_RECURSE; + } + else /* OP_KETRMAX */ + { + RMATCH(eptr, prev, offset_top, mb, eptrb, RM13); + if (rrc == MATCH_ONCE && mb->once_target == prev) rrc = MATCH_NOMATCH; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (*prev == OP_ONCE) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, mb, eptrb, RM9); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + mb->once_target = prev; + RRETURN(MATCH_ONCE); + } + ecode += 1 + LINK_SIZE; + goto TAIL_RECURSE; + } + /* Control never gets here */ + + /* Not multiline mode: start of subject assertion, unless notbol. */ + + case OP_CIRC: + if ((mb->moptions & PCRE2_NOTBOL) != 0 && eptr == mb->start_subject) + RRETURN(MATCH_NOMATCH); + + /* Start of subject assertion */ + + case OP_SOD: + if (eptr != mb->start_subject) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Multiline mode: start of subject unless notbol, or after any newline. */ + + case OP_CIRCM: + if ((mb->moptions & PCRE2_NOTBOL) != 0 && eptr == mb->start_subject) + RRETURN(MATCH_NOMATCH); + if (eptr != mb->start_subject && + (eptr == mb->end_subject || !WAS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Start of match assertion */ + + case OP_SOM: + if (eptr != mb->start_subject + mb->start_offset) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Reset the start of match point */ + + case OP_SET_SOM: + mstart = eptr; + ecode++; + break; + + /* Multiline mode: assert before any newline, or before end of subject + unless noteol is set. */ + + case OP_DOLLM: + if (eptr < mb->end_subject) + { + if (!IS_NEWLINE(eptr)) + { + if (mb->partial != 0 && + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + UCHAR21TEST(eptr) == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + RRETURN(MATCH_NOMATCH); + } + } + else + { + if ((mb->moptions & PCRE2_NOTEOL) != 0) RRETURN(MATCH_NOMATCH); + SCHECK_PARTIAL(); + } + ecode++; + break; + + /* Not multiline mode: assert before a terminating newline or before end of + subject unless noteol is set. */ + + case OP_DOLL: + if ((mb->moptions & PCRE2_NOTEOL) != 0) RRETURN(MATCH_NOMATCH); + if ((mb->poptions & PCRE2_DOLLAR_ENDONLY) == 0) goto ASSERT_NL_OR_EOS; + + /* ... else fall through for endonly */ + + /* End of subject assertion (\z) */ + + case OP_EOD: + if (eptr < mb->end_subject) RRETURN(MATCH_NOMATCH); + SCHECK_PARTIAL(); + ecode++; + break; + + /* End of subject or ending \n assertion (\Z) */ + + case OP_EODN: + ASSERT_NL_OR_EOS: + if (eptr < mb->end_subject && + (!IS_NEWLINE(eptr) || eptr != mb->end_subject - mb->nllen)) + { + if (mb->partial != 0 && + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + UCHAR21TEST(eptr) == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + RRETURN(MATCH_NOMATCH); + } + + /* Either at end of string or \n before end. */ + + SCHECK_PARTIAL(); + ecode++; + break; + + /* Word boundary assertions */ + + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + { + + /* Find out if the previous and current characters are "word" characters. + It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to + be "non-word" characters. Remember the earliest consulted character for + partial matching. */ + +#ifdef SUPPORT_UTF + if (utf) + { + /* Get status of previous character */ + + if (eptr == mb->start_subject) prev_is_word = FALSE; else + { + PCRE2_SPTR lastptr = eptr - 1; + BACKCHAR(lastptr); + if (lastptr < mb->start_used_ptr) mb->start_used_ptr = lastptr; + GETCHAR(c, lastptr); + if ((mb->poptions & PCRE2_UCP) != 0) + { + if (c == '_') prev_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + prev_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else + prev_is_word = c < 256 && (mb->ctypes[c] & ctype_word) != 0; + } + + /* Get status of next character */ + + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + cur_is_word = FALSE; + } + else + { + GETCHAR(c, eptr); + if ((mb->poptions & PCRE2_UCP) != 0) + { + if (c == '_') cur_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + cur_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else + cur_is_word = c < 256 && (mb->ctypes[c] & ctype_word) != 0; + } + } + else +#endif /* SUPPORT UTF */ + + /* Not in UTF-8 mode, but we may still have PCRE2_UCP set, and for + consistency with the behaviour of \w we do use it in this case. */ + + { + /* Get status of previous character */ + + if (eptr == mb->start_subject) prev_is_word = FALSE; else + { + if (eptr <= mb->start_used_ptr) mb->start_used_ptr = eptr - 1; +#ifdef SUPPORT_UTF + if ((mb->poptions & PCRE2_UCP) != 0) + { + c = eptr[-1]; + if (c == '_') prev_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + prev_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else +#endif + prev_is_word = MAX_255(eptr[-1]) + && ((mb->ctypes[eptr[-1]] & ctype_word) != 0); + } + + /* Get status of next character */ + + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + cur_is_word = FALSE; + } + else +#ifdef SUPPORT_UTF + if ((mb->poptions & PCRE2_UCP) != 0) + { + c = *eptr; + if (c == '_') cur_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + cur_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else +#endif + cur_is_word = MAX_255(*eptr) + && ((mb->ctypes[*eptr] & ctype_word) != 0); + } + + /* Now see if the situation is what we want */ + + if ((*ecode++ == OP_WORD_BOUNDARY)? + cur_is_word == prev_is_word : cur_is_word != prev_is_word) + RRETURN(MATCH_NOMATCH); + } + break; + + /* Match any single character type except newline; have to take care with + CRLF newlines and partial matching. */ + + case OP_ANY: + if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); + if (mb->partial != 0 && + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + UCHAR21TEST(eptr) == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + + /* Fall through */ + + /* Match any single character whatsoever. */ + + case OP_ALLANY: + if (eptr >= mb->end_subject) /* DO NOT merge the eptr++ here; it must */ + { /* not be updated before SCHECK_PARTIAL. */ + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr++; +#ifdef SUPPORT_UTF + if (utf) ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); +#endif + ecode++; + break; + + /* Match a single byte, even in UTF-8 mode. This opcode really does match + any byte, even newline, independent of the setting of PCRE2_DOTALL. */ + + case OP_ANYBYTE: + if (eptr >= mb->end_subject) /* DO NOT merge the eptr++ here; it must */ + { /* not be updated before SCHECK_PARTIAL. */ + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr++; + ecode++; + break; + + case OP_NOT_DIGIT: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_WIDE_CHARS + c < 256 && +#endif + (mb->ctypes[c] & ctype_digit) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_DIGIT: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_WIDE_CHARS + c > 255 || +#endif + (mb->ctypes[c] & ctype_digit) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_WHITESPACE: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_WIDE_CHARS + c < 256 && +#endif + (mb->ctypes[c] & ctype_space) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_WHITESPACE: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_WIDE_CHARS + c > 255 || +#endif + (mb->ctypes[c] & ctype_space) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_WORDCHAR: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_WIDE_CHARS + c < 256 && +#endif + (mb->ctypes[c] & ctype_word) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_WORDCHAR: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#ifdef SUPPORT_WIDE_CHARS + c > 255 || +#endif + (mb->ctypes[c] & ctype_word) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_ANYNL: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + + case CHAR_CR: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + } + else if (UCHAR21TEST(eptr) == CHAR_LF) eptr++; + break; + + case CHAR_LF: + break; + + case CHAR_VT: + case CHAR_FF: + case CHAR_NEL: +#ifndef EBCDIC + case 0x2028: + case 0x2029: +#endif /* Not EBCDIC */ + if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) RRETURN(MATCH_NOMATCH); + break; + } + ecode++; + break; + + case OP_NOT_HSPACE: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + HSPACE_CASES: RRETURN(MATCH_NOMATCH); /* Byte and multibyte cases */ + default: break; + } + ecode++; + break; + + case OP_HSPACE: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + HSPACE_CASES: break; /* Byte and multibyte cases */ + default: RRETURN(MATCH_NOMATCH); + } + ecode++; + break; + + case OP_NOT_VSPACE: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + VSPACE_CASES: RRETURN(MATCH_NOMATCH); + default: break; + } + ecode++; + break; + + case OP_VSPACE: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + VSPACE_CASES: break; + default: RRETURN(MATCH_NOMATCH); + } + ecode++; + break; + +#ifdef SUPPORT_UTF + /* Check the next character by Unicode property. We will get here only + if the support is in the binary; otherwise a compile-time error occurs. */ + + case OP_PROP: + case OP_NOTPROP: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + { + const uint32_t *cp; + const ucd_record *prop = GET_UCD(c); + + switch(ecode[1]) + { + case PT_ANY: + if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); + break; + + case PT_LAMP: + if ((prop->chartype == ucp_Lu || + prop->chartype == ucp_Ll || + prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_GC: + if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_PC: + if ((ecode[2] != prop->chartype) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_SC: + if ((ecode[2] != prop->script) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + /* These are specials */ + + case PT_ALNUM: + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + /* Perl space used to exclude VT, but from Perl 5.18 it is included, + which means that Perl space and POSIX space are now identical. PCRE + was changed at release 8.34. */ + + case PT_SPACE: /* Perl space */ + case PT_PXSPACE: /* POSIX space */ + switch(c) + { + HSPACE_CASES: + VSPACE_CASES: + if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); + break; + + default: + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == + (op == OP_NOTPROP)) RRETURN(MATCH_NOMATCH); + break; + } + break; + + case PT_WORD: + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N || + c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_CLIST: + cp = PRIV(ucd_caseless_sets) + ecode[2]; + for (;;) + { + if (c < *cp) + { if (op == OP_PROP) { RRETURN(MATCH_NOMATCH); } else break; } + if (c == *cp++) + { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } } + } + break; + + case PT_UCNC: + if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || + c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || + c >= 0xe000) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + /* This should never occur */ + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + + ecode += 3; + } + break; + + /* Match an extended Unicode sequence. We will get here only if the support + is in the binary; otherwise a compile-time error occurs. */ + + case OP_EXTUNI: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + else + { + int lgb, rgb; + GETCHARINCTEST(c, eptr); + lgb = UCD_GRAPHBREAK(c); + while (eptr < mb->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + rgb = UCD_GRAPHBREAK(c); + if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + lgb = rgb; + eptr += len; + } + } + CHECK_PARTIAL(); + ecode++; + break; +#endif /* SUPPORT_UTF */ + + + /* Match a back reference, possibly repeatedly. Look past the end of the + item to see if there is repeat information following. The code is similar + to that for character classes, but repeated for efficiency. Then obey + similar code to character type repeats - written out again for speed. + However, if the referenced string is the empty string, always treat + it as matched, any number of times (otherwise there could be infinite + loops). If the reference is unset, there are two possibilities: + + (a) In the default, Perl-compatible state, set the length negative; + this ensures that every attempt at a match fails. We can't just fail + here, because of the possibility of quantifiers with zero minima. + + (b) If the JavaScript compatibility flag is set, set the length to zero + so that the back reference matches an empty string. + + Otherwise, set the length to the length of what was matched by the + referenced subpattern. + + The OP_REF and OP_REFI opcodes are used for a reference to a numbered group + or to a non-duplicated named group. For a duplicated named group, OP_DNREF + and OP_DNREFI are used. In this case we must scan the list of groups to + which the name refers, and use the first one that is set. */ + + case OP_DNREF: + case OP_DNREFI: + caseless = op == OP_DNREFI; + { + int count = GET2(ecode, 1+IMM2_SIZE); + PCRE2_SPTR slot = mb->name_table + GET2(ecode, 1) * mb->name_entry_size; + ecode += 1 + 2*IMM2_SIZE; + + /* Setting the default length first and initializing 'offset' avoids + compiler warnings in the REF_REPEAT code. */ + + length = ((mb->poptions & PCRE2_MATCH_UNSET_BACKREF) != 0)? 0 : -1; + offset = 0; + + while (count-- > 0) + { + offset = GET2(slot, 0) << 1; + if (offset < offset_top && mb->ovector[offset] != PCRE2_UNSET) + { + length = mb->ovector[offset+1] - mb->ovector[offset]; + break; + } + slot += mb->name_entry_size; + } + } + goto REF_REPEAT; + + case OP_REF: + case OP_REFI: + caseless = op == OP_REFI; + offset = GET2(ecode, 1) << 1; /* Doubled ref number */ + ecode += 1 + IMM2_SIZE; + if (offset >= offset_top || mb->ovector[offset] == PCRE2_UNSET) + length = ((mb->poptions & PCRE2_MATCH_UNSET_BACKREF) != 0)? 0 : -1; + else + length = mb->ovector[offset+1] - mb->ovector[offset]; + + /* Set up for repetition, or handle the non-repeated case */ + + REF_REPEAT: + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 1 + IMM2_SIZE); + if (max == 0) max = INT_MAX; + ecode += 1 + 2 * IMM2_SIZE; + break; + + default: /* No repeat follows */ + if ((length = match_ref(offset, eptr, length, mb, caseless)) < 0) + { + if (length == -2) eptr = mb->end_subject; /* Partial match */ + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += length; + continue; /* With the main loop */ + } + + /* Handle repeated back references. If the length of the reference is + zero, just continue with the main loop. If the length is negative, it + means the reference is unset in non-Java-compatible mode. If the minimum is + zero, we can continue at the same level without recursion. For any other + minimum, carrying on will result in NOMATCH. */ + + if (length == 0) continue; + if (length < 0 && min == 0) continue; + + /* First, ensure the minimum number of matches are present. We get back + the length of the reference string explicitly rather than passing the + address of eptr, so that eptr can be a register variable. */ + + for (i = 1; i <= min; i++) + { + int slength; + if ((slength = match_ref(offset, eptr, length, mb, caseless)) < 0) + { + if (slength == -2) eptr = mb->end_subject; /* Partial match */ + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += slength; + } + + /* If min = max, continue at the same level without recursion. + They are not both allowed to be zero. */ + + if (min == max) continue; + + /* If minimizing, keep trying and advancing the pointer */ + + if (minimize) + { + for (fi = min;; fi++) + { + int slength; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM14); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if ((slength = match_ref(offset, eptr, length, mb, caseless)) < 0) + { + if (slength == -2) eptr = mb->end_subject; /* Partial match */ + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += slength; + } + /* Control never gets here */ + } + + /* If maximizing, find the longest string and work backwards */ + + else + { + pp = eptr; + for (i = min; i < max; i++) + { + int slength; + if ((slength = match_ref(offset, eptr, length, mb, caseless)) < 0) + { + /* Can't use CHECK_PARTIAL because we don't want to update eptr in + the soft partial matching case. */ + + if (slength == -2 && mb->partial != 0 && + mb->end_subject > mb->start_used_ptr) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + break; + } + eptr += slength; + } + + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM15); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr -= length; + } + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* Match a bit-mapped character class, possibly repeatedly. This op code is + used when all the characters in the class have values in the range 0-255, + and either the matching is caseful, or the characters are in the range + 0-127 when UTF-8 processing is enabled. The only difference between + OP_CLASS and OP_NCLASS occurs when a data character outside the range is + encountered. + + First, look past the end of the item to see if there is repeat information + following. Then obey similar code to character type repeats - written out + again for speed. */ + + case OP_NCLASS: + case OP_CLASS: + { + /* The data variable is saved across frames, so the byte map needs to + be stored there. */ +#define BYTE_MAP ((uint8_t *)data) + data = ecode + 1; /* Save for matching */ + ecode += 1 + (32 / sizeof(PCRE2_UCHAR)); /* Advance past the item */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + case OP_CRPOSSTAR: + case OP_CRPOSPLUS: + case OP_CRPOSQUERY: + c = *ecode++ - OP_CRSTAR; + if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0; + else possessive = TRUE; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + case OP_CRPOSRANGE: + minimize = (*ecode == OP_CRMINRANGE); + possessive = (*ecode == OP_CRPOSRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 1 + IMM2_SIZE); + if (max == 0) max = INT_MAX; + ecode += 1 + 2 * IMM2_SIZE; + break; + + default: /* No repeat follows */ + min = max = 1; + break; + } + + /* First, ensure the minimum number of matches are present. */ + +#ifdef SUPPORT_UTF + if (utf) + { + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + c = *eptr++; +#if PCRE2_CODE_UNIT_WIDTH != 8 + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else +#endif + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + + /* If max == min we can continue with the main loop without the + need to recurse. */ + + if (min == max) continue; + + /* If minimizing, keep testing the rest of the expression and advancing + the pointer while it matches the class. */ + + if (minimize) + { +#ifdef SUPPORT_UTF + if (utf) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM16); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM17); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + c = *eptr++; +#if PCRE2_CODE_UNIT_WIDTH != 8 + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else +#endif + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* If maximizing, find the longest possible run, then work backwards. */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF + if (utf) + { + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c > 255) + { + if (op == OP_CLASS) break; + } + else + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; + eptr += len; + } + + if (possessive) continue; /* No backtracking */ + + for (;;) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM18); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; +#if PCRE2_CODE_UNIT_WIDTH != 8 + if (c > 255) + { + if (op == OP_CLASS) break; + } + else +#endif + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; + eptr++; + } + + if (possessive) continue; /* No backtracking */ + + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM19); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } +#undef BYTE_MAP + } + /* Control never gets here */ + + + /* Match an extended character class. In the 8-bit library, this opcode is + encountered only when UTF-8 mode mode is supported. In the 16-bit and + 32-bit libraries, codepoints greater than 255 may be encountered even when + UTF is not supported. */ + +#ifdef SUPPORT_WIDE_CHARS + case OP_XCLASS: + { + data = ecode + 1 + LINK_SIZE; /* Save for matching */ + ecode += GET(ecode, 1); /* Advance past the item */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + case OP_CRPOSSTAR: + case OP_CRPOSPLUS: + case OP_CRPOSQUERY: + c = *ecode++ - OP_CRSTAR; + if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0; + else possessive = TRUE; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + case OP_CRPOSRANGE: + minimize = (*ecode == OP_CRMINRANGE); + possessive = (*ecode == OP_CRPOSRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 1 + IMM2_SIZE); + if (max == 0) max = INT_MAX; + ecode += 1 + 2 * IMM2_SIZE; + break; + + default: /* No repeat follows */ + min = max = 1; + break; + } + + /* First, ensure the minimum number of matches are present. */ + + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); + } + + /* If max == min we can continue with the main loop without the + need to recurse. */ + + if (min == max) continue; + + /* If minimizing, keep testing the rest of the expression and advancing + the pointer while it matches the class. */ + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM20); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* If maximizing, find the longest possible run, then work backwards. */ + + else + { + pp = eptr; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } +#ifdef SUPPORT_UTF + GETCHARLENTEST(c, eptr, len); +#else + c = *eptr; +#endif + if (!PRIV(xclass)(c, data, utf)) break; + eptr += len; + } + + if (possessive) continue; /* No backtracking */ + + for(;;) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM21); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ +#ifdef SUPPORT_UTF + if (utf) BACKCHAR(eptr); +#endif + } + RRETURN(MATCH_NOMATCH); + } + + /* Control never gets here */ + } +#endif /* End of XCLASS */ + + /* Match a single character, casefully */ + + case OP_CHAR: +#ifdef SUPPORT_UTF + if (utf) + { + length = 1; + ecode++; + GETCHARLEN(fc, ecode, length); + if (length > mb->end_subject - eptr) + { + CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ + RRETURN(MATCH_NOMATCH); + } + while (length-- > 0) if (*ecode++ != UCHAR21INC(eptr)) RRETURN(MATCH_NOMATCH); + } + else +#endif + /* Not UTF mode */ + { + if (mb->end_subject - eptr < 1) + { + SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ + RRETURN(MATCH_NOMATCH); + } + if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); + ecode += 2; + } + break; + + /* Match a single character, caselessly. If we are at the end of the + subject, give up immediately. */ + + case OP_CHARI: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + +#ifdef SUPPORT_UTF + if (utf) + { + length = 1; + ecode++; + GETCHARLEN(fc, ecode, length); + + /* If the pattern character's value is < 128, we have only one byte, and + we know that its other case must also be one byte long, so we can use the + fast lookup table. We know that there is at least one byte left in the + subject. */ + + if (fc < 128) + { + uint32_t cc = UCHAR21(eptr); + if (mb->lcc[fc] != TABLE_GET(cc, mb->lcc, cc)) RRETURN(MATCH_NOMATCH); + ecode++; + eptr++; + } + + /* Otherwise we must pick up the subject character. Note that we cannot + use the value of "length" to check for sufficient bytes left, because the + other case of the character may have more or fewer bytes. */ + + else + { + uint32_t dc; + GETCHARINC(dc, eptr); + ecode += length; + + /* If we have Unicode property support, we can use it to test the other + case of the character, if there is one. */ + + if (fc != dc) + { +#ifdef SUPPORT_UTF + if (dc != UCD_OTHERCASE(fc)) +#endif + RRETURN(MATCH_NOMATCH); + } + } + } + else +#endif /* SUPPORT_UTF */ + + /* Not UTF mode */ + { + if (TABLE_GET(ecode[1], mb->lcc, ecode[1]) + != TABLE_GET(*eptr, mb->lcc, *eptr)) RRETURN(MATCH_NOMATCH); + eptr++; + ecode += 2; + } + break; + + /* Match a single character repeatedly. */ + + case OP_EXACT: + case OP_EXACTI: + min = max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATCHAR; + + case OP_POSUPTO: + case OP_POSUPTOI: + possessive = TRUE; + /* Fall through */ + + case OP_UPTO: + case OP_UPTOI: + case OP_MINUPTO: + case OP_MINUPTOI: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; + ecode += 1 + IMM2_SIZE; + goto REPEATCHAR; + + case OP_POSSTAR: + case OP_POSSTARI: + possessive = TRUE; + min = 0; + max = INT_MAX; + ecode++; + goto REPEATCHAR; + + case OP_POSPLUS: + case OP_POSPLUSI: + possessive = TRUE; + min = 1; + max = INT_MAX; + ecode++; + goto REPEATCHAR; + + case OP_POSQUERY: + case OP_POSQUERYI: + possessive = TRUE; + min = 0; + max = 1; + ecode++; + goto REPEATCHAR; + + case OP_STAR: + case OP_STARI: + case OP_MINSTAR: + case OP_MINSTARI: + case OP_PLUS: + case OP_PLUSI: + case OP_MINPLUS: + case OP_MINPLUSI: + case OP_QUERY: + case OP_QUERYI: + case OP_MINQUERY: + case OP_MINQUERYI: + c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI); + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single-character matches. We first check + for the minimum number of characters. If the minimum equals the maximum, we + are done. Otherwise, if minimizing, check the rest of the pattern for a + match; if there isn't one, advance up to the maximum, one character at a + time. + + If maximizing, advance up to the maximum number of matching characters, + until eptr is past the end of the maximum run. If possessive, we are + then done (no backing up). Otherwise, match at this position; anything + other than no match is immediately returned. For nomatch, back up one + character, unless we are matching \R and the last thing matched was + \r\n, in which case, back up two bytes. When we reach the first optional + character position, we can save stack by doing a tail recurse. + + The various UTF/non-UTF and caseful/caseless cases are handled separately, + for speed. */ + + REPEATCHAR: +#ifdef SUPPORT_UTF + if (utf) + { + length = 1; + charptr = ecode; + GETCHARLEN(fc, ecode, length); + ecode += length; + + /* Handle multibyte character matching specially here. There is + support for caseless matching if UCP support is present. */ + + if (length > 1) + { + uint32_t othercase; + if (op >= OP_STARI && /* Caseless */ + (othercase = UCD_OTHERCASE(fc)) != fc) + oclength = PRIV(ord2utf)(othercase, occhars); + else oclength = 0; + + for (i = 1; i <= min; i++) + { + if (eptr <= mb->end_subject - length && + memcmp(eptr, charptr, CU2BYTES(length)) == 0) eptr += length; + else if (oclength > 0 && + eptr <= mb->end_subject - oclength && + memcmp(eptr, occhars, CU2BYTES(oclength)) == 0) eptr += oclength; + else + { + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + } + + if (min == max) continue; + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM22); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr <= mb->end_subject - length && + memcmp(eptr, charptr, CU2BYTES(length)) == 0) eptr += length; + else if (oclength > 0 && + eptr <= mb->end_subject - oclength && + memcmp(eptr, occhars, CU2BYTES(oclength)) == 0) eptr += oclength; + else + { + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + else /* Maximize */ + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr <= mb->end_subject - length && + memcmp(eptr, charptr, CU2BYTES(length)) == 0) eptr += length; + else if (oclength > 0 && + eptr <= mb->end_subject - oclength && + memcmp(eptr, occhars, CU2BYTES(oclength)) == 0) eptr += oclength; + else + { + CHECK_PARTIAL(); + break; + } + } + + if (possessive) continue; /* No backtracking */ + for(;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM23); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + BACKCHAR(eptr); + } + } + /* Control never gets here */ + } + + /* If the length of a UTF-8 character is 1, we fall through here, and + obey the code as for non-UTF-8 characters below, though in this case the + value of fc will always be < 128. */ + } + else +#endif /* SUPPORT_UTF */ + + /* When not in UTF-8 mode, load a single-byte character. */ + fc = *ecode++; + + /* The value of fc at this point is always one character, though we may + or may not be in UTF mode. The code is duplicated for the caseless and + caseful cases, for speed, since matching characters is likely to be quite + common. First, ensure the minimum number of matches are present. If min = + max, continue at the same level without recursing. Otherwise, if + minimizing, keep trying the rest of the expression and advancing one + matching character if failing, up to the maximum. Alternatively, if + maximizing, find the maximum number of characters and work backwards. */ + + if (op >= OP_STARI) /* Caseless */ + { +#if PCRE2_CODE_UNIT_WIDTH == 8 + /* fc must be < 128 if UTF is enabled. */ + foc = mb->fcc[fc]; +#else +#ifdef SUPPORT_UTF + if (utf && fc > 127) + foc = UCD_OTHERCASE(fc); + else +#endif /* SUPPORT_UTF */ + foc = TABLE_GET(fc, mb->fcc, fc); +#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */ + + for (i = 1; i <= min; i++) + { + uint32_t cc; /* Faster than PCRE2_UCHAR */ + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21TEST(eptr); + if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH); + eptr++; + } + if (min == max) continue; + if (minimize) + { + for (fi = min;; fi++) + { + uint32_t cc; /* Faster than PCRE2_UCHAR */ + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM24); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21TEST(eptr); + if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH); + eptr++; + } + /* Control never gets here */ + } + else /* Maximize */ + { + pp = eptr; + for (i = min; i < max; i++) + { + uint32_t cc; /* Faster than PCRE2_UCHAR */ + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + cc = UCHAR21TEST(eptr); + if (fc != cc && foc != cc) break; + eptr++; + } + if (possessive) continue; /* No backtracking */ + for (;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM25); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + /* Control never gets here */ + } + } + + /* Caseful comparisons (includes all multi-byte characters) */ + + else + { + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc != UCHAR21INCTEST(eptr)) RRETURN(MATCH_NOMATCH); + } + + if (min == max) continue; + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM26); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc != UCHAR21INCTEST(eptr)) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + else /* Maximize */ + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc != UCHAR21TEST(eptr)) break; + eptr++; + } + if (possessive) continue; /* No backtracking */ + for (;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM27); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + /* Control never gets here */ + } + } + /* Control never gets here */ + + /* Match a negated single one-byte character. The character we are + checking can be multibyte. */ + + case OP_NOT: + case OP_NOTI: + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t ch, och; + + ecode++; + GETCHARINC(ch, ecode); + GETCHARINC(c, eptr); + + if (op == OP_NOT) + { + if (ch == c) RRETURN(MATCH_NOMATCH); + } + else + { + if (ch > 127) + och = UCD_OTHERCASE(ch); + else + och = TABLE_GET(ch, mb->fcc, ch); + if (ch == c || och == c) RRETURN(MATCH_NOMATCH); + } + } + else +#endif /* SUPPORT_UTF */ + { + register uint32_t ch = ecode[1]; + c = *eptr++; + if (ch == c || (op == OP_NOTI && TABLE_GET(ch, mb->fcc, ch) == c)) + RRETURN(MATCH_NOMATCH); + ecode += 2; + } + break; + + /* Match a negated single one-byte character repeatedly. This is almost a + repeat of the code for a repeated single character, but I haven't found a + nice way of commoning these up that doesn't require a test of the + positive/negative option for each character match. Maybe that wouldn't add + very much to the time taken, but character matching *is* what this is all + about... */ + + case OP_NOTEXACT: + case OP_NOTEXACTI: + min = max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATNOTCHAR; + + case OP_NOTUPTO: + case OP_NOTUPTOI: + case OP_NOTMINUPTO: + case OP_NOTMINUPTOI: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; + ecode += 1 + IMM2_SIZE; + goto REPEATNOTCHAR; + + case OP_NOTPOSSTAR: + case OP_NOTPOSSTARI: + possessive = TRUE; + min = 0; + max = INT_MAX; + ecode++; + goto REPEATNOTCHAR; + + case OP_NOTPOSPLUS: + case OP_NOTPOSPLUSI: + possessive = TRUE; + min = 1; + max = INT_MAX; + ecode++; + goto REPEATNOTCHAR; + + case OP_NOTPOSQUERY: + case OP_NOTPOSQUERYI: + possessive = TRUE; + min = 0; + max = 1; + ecode++; + goto REPEATNOTCHAR; + + case OP_NOTPOSUPTO: + case OP_NOTPOSUPTOI: + possessive = TRUE; + min = 0; + max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATNOTCHAR; + + case OP_NOTSTAR: + case OP_NOTSTARI: + case OP_NOTMINSTAR: + case OP_NOTMINSTARI: + case OP_NOTPLUS: + case OP_NOTPLUSI: + case OP_NOTMINPLUS: + case OP_NOTMINPLUSI: + case OP_NOTQUERY: + case OP_NOTQUERYI: + case OP_NOTMINQUERY: + case OP_NOTMINQUERYI: + c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR); + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single-byte matches. */ + + REPEATNOTCHAR: + GETCHARINCTEST(fc, ecode); + + /* The code is duplicated for the caseless and caseful cases, for speed, + since matching characters is likely to be quite common. First, ensure the + minimum number of matches are present. If min = max, continue at the same + level without recursing. Otherwise, if minimizing, keep trying the rest of + the expression and advancing one matching character if failing, up to the + maximum. Alternatively, if maximizing, find the maximum number of + characters and work backwards. */ + + if (op >= OP_NOTSTARI) /* Caseless */ + { +#ifdef SUPPORT_UTF + if (utf && fc > 127) + foc = UCD_OTHERCASE(fc); + else +#endif /* SUPPORT_UTF */ + foc = TABLE_GET(fc, mb->fcc, fc); + +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t d; + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif /* SUPPORT_UTF */ + /* Not UTF mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); + eptr++; + } + } + + if (min == max) continue; + + if (minimize) + { +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t d; + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM28); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif /*SUPPORT_UTF */ + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM29); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); + eptr++; + } + } + /* Control never gets here */ + } + + /* Maximize case */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t d; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(d, eptr, len); + if (fc == d || (unsigned int)foc == d) break; + eptr += len; + } + if (possessive) continue; /* No backtracking */ + for(;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM30); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + BACKCHAR(eptr); + } + } + else +#endif /* SUPPORT_UTF */ + /* Not UTF mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc == *eptr || foc == *eptr) break; + eptr++; + } + if (possessive) continue; /* No backtracking */ + for (;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM31); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + /* Control never gets here */ + } + } + + /* Caseful comparisons */ + + else + { +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t d; + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr++) RRETURN(MATCH_NOMATCH); + } + } + + if (min == max) continue; + + if (minimize) + { +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t d; + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM32); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM33); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr++) RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* Maximize case */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF + if (utf) + { + register uint32_t d; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(d, eptr, len); + if (fc == d) break; + eptr += len; + } + if (possessive) continue; /* No backtracking */ + for(;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM34); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc == *eptr) break; + eptr++; + } + if (possessive) continue; /* No backtracking */ + for (;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM35); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + /* Control never gets here */ + } + } + /* Control never gets here */ + + /* Match a single character type repeatedly; several different opcodes + share code. This is very similar to the code for single characters, but we + repeat it in the interests of efficiency. */ + + case OP_TYPEEXACT: + min = max = GET2(ecode, 1); + minimize = TRUE; + ecode += 1 + IMM2_SIZE; + goto REPEATTYPE; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_TYPEMINUPTO; + ecode += 1 + IMM2_SIZE; + goto REPEATTYPE; + + case OP_TYPEPOSSTAR: + possessive = TRUE; + min = 0; + max = INT_MAX; + ecode++; + goto REPEATTYPE; + + case OP_TYPEPOSPLUS: + possessive = TRUE; + min = 1; + max = INT_MAX; + ecode++; + goto REPEATTYPE; + + case OP_TYPEPOSQUERY: + possessive = TRUE; + min = 0; + max = 1; + ecode++; + goto REPEATTYPE; + + case OP_TYPEPOSUPTO: + possessive = TRUE; + min = 0; + max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATTYPE; + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + c = *ecode++ - OP_TYPESTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single character type matches. Note that + in UTF-8 mode, '.' matches a character of any length, but for the other + character types, the valid characters are all one-byte long. */ + + REPEATTYPE: + ctype = *ecode++; /* Code for the character type */ + +#ifdef SUPPORT_UTF + if (ctype == OP_PROP || ctype == OP_NOTPROP) + { + prop_fail_result = ctype == OP_NOTPROP; + prop_type = *ecode++; + prop_value = *ecode++; + } + else prop_type = -1; +#endif + + /* First, ensure the minimum number of matches are present. Use inline + code for maximizing the speed, and do the type test once at the start + (i.e. keep it out of the loop). Separate the UTF-8 code completely as that + is tidier. Also separate the UCP code, which can be the same for both UTF-8 + and single-bytes. */ + + if (min > 0) + { +#ifdef SUPPORT_UTF + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + } + break; + + case PT_LAMP: + for (i = 1; i <= min; i++) + { + int chartype; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + chartype = UCD_CHARTYPE(c); + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_GC: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_PC: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_SC: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_ALNUM: + for (i = 1; i <= min; i++) + { + int category; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + /* Perl space used to exclude VT, but from Perl 5.18 it is included, + which means that Perl space and POSIX space are now identical. PCRE + was changed at release 8.34. */ + + case PT_SPACE: /* Perl space */ + case PT_PXSPACE: /* POSIX space */ + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + HSPACE_CASES: + VSPACE_CASES: + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + break; + + default: + if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + break; + } + } + break; + + case PT_WORD: + for (i = 1; i <= min; i++) + { + int category; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_CLIST: + for (i = 1; i <= min; i++) + { + const uint32_t *cp; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + cp = PRIV(ucd_caseless_sets) + prop_value; + for (;;) + { + if (c < *cp) + { if (prop_fail_result) break; else { RRETURN(MATCH_NOMATCH); } } + if (c == *cp++) + { if (prop_fail_result) { RRETURN(MATCH_NOMATCH); } else break; } + } + } + break; + + case PT_UCNC: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || + c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || + c >= 0xe000) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + /* This should not occur */ + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + else + { + int lgb, rgb; + GETCHARINCTEST(c, eptr); + lgb = UCD_GRAPHBREAK(c); + while (eptr < mb->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + rgb = UCD_GRAPHBREAK(c); + if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + lgb = rgb; + eptr += len; + } + } + CHECK_PARTIAL(); + } + } + + else +#endif /* SUPPORT_UTF */ + +/* Handle all other cases when the coding is UTF-8 */ + +#ifdef SUPPORT_UTF + if (utf) switch(ctype) + { + case OP_ANY: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); + if (mb->partial != 0 && + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + UCHAR21(eptr) == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + eptr++; + ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); + } + break; + + case OP_ALLANY: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr++; + ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); + } + break; + + case OP_ANYBYTE: + if (eptr > mb->end_subject - min) RRETURN(MATCH_NOMATCH); + eptr += min; + break; + + case OP_ANYNL: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + + case CHAR_CR: + if (eptr < mb->end_subject && UCHAR21(eptr) == CHAR_LF) eptr++; + break; + + case CHAR_LF: + break; + + case CHAR_VT: + case CHAR_FF: + case CHAR_NEL: +#ifndef EBCDIC + case 0x2028: + case 0x2029: +#endif /* Not EBCDIC */ + if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) RRETURN(MATCH_NOMATCH); + break; + } + } + break; + + case OP_NOT_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + HSPACE_CASES: RRETURN(MATCH_NOMATCH); /* Byte and multibyte cases */ + default: break; + } + } + break; + + case OP_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + HSPACE_CASES: break; /* Byte and multibyte cases */ + default: RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_NOT_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + VSPACE_CASES: RRETURN(MATCH_NOMATCH); + default: break; + } + } + break; + + case OP_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + VSPACE_CASES: break; + default: RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_NOT_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + if (c < 128 && (mb->ctypes[c] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_DIGIT: + for (i = 1; i <= min; i++) + { + uint32_t cc; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21(eptr); + if (cc >= 128 || (mb->ctypes[cc] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + case OP_NOT_WHITESPACE: + for (i = 1; i <= min; i++) + { + uint32_t cc; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21(eptr); + if (cc < 128 && (mb->ctypes[cc] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); + } + break; + + case OP_WHITESPACE: + for (i = 1; i <= min; i++) + { + uint32_t cc; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21(eptr); + if (cc >= 128 || (mb->ctypes[cc] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + case OP_NOT_WORDCHAR: + for (i = 1; i <= min; i++) + { + uint32_t cc; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21(eptr); + if (cc < 128 && (mb->ctypes[cc] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); + } + break; + + case OP_WORDCHAR: + for (i = 1; i <= min; i++) + { + uint32_t cc; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + cc = UCHAR21(eptr); + if (cc >= 128 || (mb->ctypes[cc] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } /* End switch(ctype) */ + + else +#endif /* SUPPORT_UTF */ + + /* Code for the non-UTF-8 case for minimum matching of operators other + than OP_PROP and OP_NOTPROP. */ + + switch(ctype) + { + case OP_ANY: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); + if (mb->partial != 0 && + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + eptr++; + } + break; + + case OP_ALLANY: + if (eptr > mb->end_subject - min) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += min; + break; + + case OP_ANYBYTE: + if (eptr > mb->end_subject - min) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += min; + break; + + case OP_ANYNL: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: RRETURN(MATCH_NOMATCH); + + case CHAR_CR: + if (eptr < mb->end_subject && *eptr == CHAR_LF) eptr++; + break; + + case CHAR_LF: + break; + + case CHAR_VT: + case CHAR_FF: + case CHAR_NEL: +#if PCRE2_CODE_UNIT_WIDTH != 8 + case 0x2028: + case 0x2029: +#endif + if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) RRETURN(MATCH_NOMATCH); + break; + } + } + break; + + case OP_NOT_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: break; + HSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + HSPACE_MULTIBYTE_CASES: +#endif + RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: RRETURN(MATCH_NOMATCH); + HSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + HSPACE_MULTIBYTE_CASES: +#endif + break; + } + } + break; + + case OP_NOT_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + VSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + VSPACE_MULTIBYTE_CASES: +#endif + RRETURN(MATCH_NOMATCH); + default: break; + } + } + break; + + case OP_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: RRETURN(MATCH_NOMATCH); + VSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + VSPACE_MULTIBYTE_CASES: +#endif + break; + } + } + break; + + case OP_NOT_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (MAX_255(*eptr) && (mb->ctypes[*eptr] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (!MAX_255(*eptr) || (mb->ctypes[*eptr] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_NOT_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (MAX_255(*eptr) && (mb->ctypes[*eptr] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (!MAX_255(*eptr) || (mb->ctypes[*eptr] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_NOT_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (MAX_255(*eptr) && (mb->ctypes[*eptr] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (!MAX_255(*eptr) || (mb->ctypes[*eptr] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + } + + /* If min = max, continue at the same level without recursing */ + + if (min == max) continue; + + /* If minimizing, we have to test the rest of the pattern before each + subsequent match. Again, separate the UTF-8 case for speed, and also + separate the UCP cases. */ + + if (minimize) + { +#ifdef SUPPORT_UTF + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM36); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_LAMP: + for (fi = min;; fi++) + { + int chartype; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM37); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + chartype = UCD_CHARTYPE(c); + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_GC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM38); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_PC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM39); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_SC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM40); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_ALNUM: + for (fi = min;; fi++) + { + int category; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM59); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* Perl space used to exclude VT, but from Perl 5.18 it is included, + which means that Perl space and POSIX space are now identical. PCRE + was changed at release 8.34. */ + + case PT_SPACE: /* Perl space */ + case PT_PXSPACE: /* POSIX space */ + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM61); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + HSPACE_CASES: + VSPACE_CASES: + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + break; + + default: + if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + break; + } + } + /* Control never gets here */ + + case PT_WORD: + for (fi = min;; fi++) + { + int category; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM62); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || + category == ucp_N || + c == CHAR_UNDERSCORE) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_CLIST: + for (fi = min;; fi++) + { + const uint32_t *cp; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM67); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + cp = PRIV(ucd_caseless_sets) + prop_value; + for (;;) + { + if (c < *cp) + { if (prop_fail_result) break; else { RRETURN(MATCH_NOMATCH); } } + if (c == *cp++) + { if (prop_fail_result) { RRETURN(MATCH_NOMATCH); } else break; } + } + } + /* Control never gets here */ + + case PT_UCNC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM60); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || + c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || + c >= 0xe000) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* This should never occur */ + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM41); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + else + { + int lgb, rgb; + GETCHARINCTEST(c, eptr); + lgb = UCD_GRAPHBREAK(c); + while (eptr < mb->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + rgb = UCD_GRAPHBREAK(c); + if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + lgb = rgb; + eptr += len; + } + } + CHECK_PARTIAL(); + } + } + else +#endif /* SUPPORT_UTF */ + +#ifdef SUPPORT_UTF + if (utf) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM42); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (ctype == OP_ANY && IS_NEWLINE(eptr)) + RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + switch(ctype) + { + case OP_ANY: /* This is the non-NL case */ + if (mb->partial != 0 && /* Take care with CRLF partial */ + eptr >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + c == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + break; + + case OP_ALLANY: + case OP_ANYBYTE: + break; + + case OP_ANYNL: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case CHAR_CR: + if (eptr < mb->end_subject && UCHAR21(eptr) == CHAR_LF) eptr++; + break; + + case CHAR_LF: + break; + + case CHAR_VT: + case CHAR_FF: + case CHAR_NEL: +#ifndef EBCDIC + case 0x2028: + case 0x2029: +#endif /* Not EBCDIC */ + if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) RRETURN(MATCH_NOMATCH); + break; + } + break; + + case OP_NOT_HSPACE: + switch(c) + { + HSPACE_CASES: RRETURN(MATCH_NOMATCH); + default: break; + } + break; + + case OP_HSPACE: + switch(c) + { + HSPACE_CASES: break; + default: RRETURN(MATCH_NOMATCH); + } + break; + + case OP_NOT_VSPACE: + switch(c) + { + VSPACE_CASES: RRETURN(MATCH_NOMATCH); + default: break; + } + break; + + case OP_VSPACE: + switch(c) + { + VSPACE_CASES: break; + default: RRETURN(MATCH_NOMATCH); + } + break; + + case OP_NOT_DIGIT: + if (c < 256 && (mb->ctypes[c] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + if (c >= 256 || (mb->ctypes[c] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + if (c < 256 && (mb->ctypes[c] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + if (c >= 256 || (mb->ctypes[c] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + if (c < 256 && (mb->ctypes[c] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + if (c >= 256 || (mb->ctypes[c] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM43); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (ctype == OP_ANY && IS_NEWLINE(eptr)) + RRETURN(MATCH_NOMATCH); + c = *eptr++; + switch(ctype) + { + case OP_ANY: /* This is the non-NL case */ + if (mb->partial != 0 && /* Take care with CRLF partial */ + eptr >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + c == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + break; + + case OP_ALLANY: + case OP_ANYBYTE: + break; + + case OP_ANYNL: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case CHAR_CR: + if (eptr < mb->end_subject && *eptr == CHAR_LF) eptr++; + break; + + case CHAR_LF: + break; + + case CHAR_VT: + case CHAR_FF: + case CHAR_NEL: +#if PCRE2_CODE_UNIT_WIDTH != 8 + case 0x2028: + case 0x2029: +#endif + if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) RRETURN(MATCH_NOMATCH); + break; + } + break; + + case OP_NOT_HSPACE: + switch(c) + { + default: break; + HSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + HSPACE_MULTIBYTE_CASES: +#endif + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_HSPACE: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + HSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + HSPACE_MULTIBYTE_CASES: +#endif + break; + } + break; + + case OP_NOT_VSPACE: + switch(c) + { + default: break; + VSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + VSPACE_MULTIBYTE_CASES: +#endif + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_VSPACE: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + VSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + VSPACE_MULTIBYTE_CASES: +#endif + break; + } + break; + + case OP_NOT_DIGIT: + if (MAX_255(c) && (mb->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + if (!MAX_255(c) || (mb->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + if (MAX_255(c) && (mb->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + if (!MAX_255(c) || (mb->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + if (MAX_255(c) && (mb->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + if (!MAX_255(c) || (mb->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + } + } + /* Control never gets here */ + } + + /* If maximizing, it is worth using inline code for speed, doing the type + test once at the start (i.e. keep it out of the loop). Again, keep the + UTF-8 and UCP stuff separate. */ + + else + { + pp = eptr; /* Remember where we started */ + +#ifdef SUPPORT_UTF + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if (prop_fail_result) break; + eptr+= len; + } + break; + + case PT_LAMP: + for (i = min; i < max; i++) + { + int chartype; + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + chartype = UCD_CHARTYPE(c); + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_GC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break; + eptr+= len; + } + break; + + case PT_PC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break; + eptr+= len; + } + break; + + case PT_SC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break; + eptr+= len; + } + break; + + case PT_ALNUM: + for (i = min; i < max; i++) + { + int category; + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N) == prop_fail_result) + break; + eptr+= len; + } + break; + + /* Perl space used to exclude VT, but from Perl 5.18 it is included, + which means that Perl space and POSIX space are now identical. PCRE + was changed at release 8.34. */ + + case PT_SPACE: /* Perl space */ + case PT_PXSPACE: /* POSIX space */ + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + switch(c) + { + HSPACE_CASES: + VSPACE_CASES: + if (prop_fail_result) goto ENDLOOP99; /* Break the loop */ + break; + + default: + if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result) + goto ENDLOOP99; /* Break the loop */ + break; + } + eptr+= len; + } + ENDLOOP99: + break; + + case PT_WORD: + for (i = min; i < max; i++) + { + int category; + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N || + c == CHAR_UNDERSCORE) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_CLIST: + for (i = min; i < max; i++) + { + const uint32_t *cp; + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + cp = PRIV(ucd_caseless_sets) + prop_value; + for (;;) + { + if (c < *cp) + { if (prop_fail_result) break; else goto GOT_MAX; } + if (c == *cp++) + { if (prop_fail_result) goto GOT_MAX; else break; } + } + eptr += len; + } + GOT_MAX: + break; + + case PT_UCNC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || + c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || + c >= 0xe000) == prop_fail_result) + break; + eptr += len; + } + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + + /* eptr is now past the end of the maximum run */ + + if (possessive) continue; /* No backtracking */ + for(;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM44); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + if (utf) BACKCHAR(eptr); + } + } + + /* Match extended Unicode grapheme clusters. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + else + { + int lgb, rgb; + GETCHARINCTEST(c, eptr); + lgb = UCD_GRAPHBREAK(c); + while (eptr < mb->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + rgb = UCD_GRAPHBREAK(c); + if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + lgb = rgb; + eptr += len; + } + } + CHECK_PARTIAL(); + } + + /* eptr is now past the end of the maximum run */ + + if (possessive) continue; /* No backtracking */ + + for(;;) + { + int lgb, rgb; + PCRE2_SPTR fptr; + + if (eptr == pp) goto TAIL_RECURSE; /* At start of char run */ + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM45); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + + /* Backtracking over an extended grapheme cluster involves inspecting + the previous two characters (if present) to see if a break is + permitted between them. */ + + eptr--; + if (!utf) c = *eptr; else + { + BACKCHAR(eptr); + GETCHAR(c, eptr); + } + rgb = UCD_GRAPHBREAK(c); + + for (;;) + { + if (eptr == pp) goto TAIL_RECURSE; /* At start of char run */ + fptr = eptr - 1; + if (!utf) c = *fptr; else + { + BACKCHAR(fptr); + GETCHAR(c, fptr); + } + lgb = UCD_GRAPHBREAK(c); + if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + eptr = fptr; + rgb = lgb; + } + } + } + + else +#endif /* SUPPORT_UTF */ + +#ifdef SUPPORT_UTF + if (utf) + { + switch(ctype) + { + case OP_ANY: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (IS_NEWLINE(eptr)) break; + if (mb->partial != 0 && /* Take care with CRLF partial */ + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + UCHAR21(eptr) == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + eptr++; + ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); + } + break; + + case OP_ALLANY: + if (max < INT_MAX) + { + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + eptr++; + ACROSSCHAR(eptr < mb->end_subject, *eptr, eptr++); + } + } + else + { + eptr = mb->end_subject; /* Unlimited UTF-8 repeat */ + SCHECK_PARTIAL(); + } + break; + + /* The byte case is the same as non-UTF8 */ + + case OP_ANYBYTE: + c = max - min; + if (c > (unsigned int)(mb->end_subject - eptr)) + { + eptr = mb->end_subject; + SCHECK_PARTIAL(); + } + else eptr += c; + break; + + case OP_ANYNL: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c == CHAR_CR) + { + if (++eptr >= mb->end_subject) break; + if (UCHAR21(eptr) == CHAR_LF) eptr++; + } + else + { + if (c != CHAR_LF && + (mb->bsr_convention == PCRE2_BSR_ANYCRLF || + (c != CHAR_VT && c != CHAR_FF && c != CHAR_NEL +#ifndef EBCDIC + && c != 0x2028 && c != 0x2029 +#endif /* Not EBCDIC */ + ))) + break; + eptr += len; + } + } + break; + + case OP_NOT_HSPACE: + case OP_HSPACE: + for (i = min; i < max; i++) + { + BOOL gotspace; + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + switch(c) + { + HSPACE_CASES: gotspace = TRUE; break; + default: gotspace = FALSE; break; + } + if (gotspace == (ctype == OP_NOT_HSPACE)) break; + eptr += len; + } + break; + + case OP_NOT_VSPACE: + case OP_VSPACE: + for (i = min; i < max; i++) + { + BOOL gotspace; + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + switch(c) + { + VSPACE_CASES: gotspace = TRUE; break; + default: gotspace = FALSE; break; + } + if (gotspace == (ctype == OP_NOT_VSPACE)) break; + eptr += len; + } + break; + + case OP_NOT_DIGIT: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c < 256 && (mb->ctypes[c] & ctype_digit) != 0) break; + eptr+= len; + } + break; + + case OP_DIGIT: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c >= 256 ||(mb->ctypes[c] & ctype_digit) == 0) break; + eptr+= len; + } + break; + + case OP_NOT_WHITESPACE: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c < 256 && (mb->ctypes[c] & ctype_space) != 0) break; + eptr+= len; + } + break; + + case OP_WHITESPACE: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c >= 256 ||(mb->ctypes[c] & ctype_space) == 0) break; + eptr+= len; + } + break; + + case OP_NOT_WORDCHAR: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c < 256 && (mb->ctypes[c] & ctype_word) != 0) break; + eptr+= len; + } + break; + + case OP_WORDCHAR: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c >= 256 || (mb->ctypes[c] & ctype_word) == 0) break; + eptr+= len; + } + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + + if (possessive) continue; /* No backtracking */ + for(;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM46); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + BACKCHAR(eptr); + if (ctype == OP_ANYNL && eptr > pp && UCHAR21(eptr) == CHAR_NL && + UCHAR21(eptr - 1) == CHAR_CR) eptr--; + } + } + else +#endif /* SUPPORT_UTF */ + /* Not UTF mode */ + { + switch(ctype) + { + case OP_ANY: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (IS_NEWLINE(eptr)) break; + if (mb->partial != 0 && /* Take care with CRLF partial */ + eptr + 1 >= mb->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + mb->hitend = TRUE; + if (mb->partial > 1) RRETURN(PCRE2_ERROR_PARTIAL); + } + eptr++; + } + break; + + case OP_ALLANY: + case OP_ANYBYTE: + c = max - min; + if (c > (unsigned int)(mb->end_subject - eptr)) + { + eptr = mb->end_subject; + SCHECK_PARTIAL(); + } + else eptr += c; + break; + + case OP_ANYNL: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; + if (c == CHAR_CR) + { + if (++eptr >= mb->end_subject) break; + if (*eptr == CHAR_LF) eptr++; + } + else + { + if (c != CHAR_LF && (mb->bsr_convention == PCRE2_BSR_ANYCRLF || + (c != CHAR_VT && c != CHAR_FF && c != CHAR_NEL +#if PCRE2_CODE_UNIT_WIDTH != 8 + && c != 0x2028 && c != 0x2029 +#endif + ))) break; + eptr++; + } + } + break; + + case OP_NOT_HSPACE: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + switch(*eptr) + { + default: eptr++; break; + HSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + HSPACE_MULTIBYTE_CASES: +#endif + goto ENDLOOP00; + } + } + ENDLOOP00: + break; + + case OP_HSPACE: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + switch(*eptr) + { + default: goto ENDLOOP01; + HSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + HSPACE_MULTIBYTE_CASES: +#endif + eptr++; break; + } + } + ENDLOOP01: + break; + + case OP_NOT_VSPACE: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + switch(*eptr) + { + default: eptr++; break; + VSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + VSPACE_MULTIBYTE_CASES: +#endif + goto ENDLOOP02; + } + } + ENDLOOP02: + break; + + case OP_VSPACE: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + switch(*eptr) + { + default: goto ENDLOOP03; + VSPACE_BYTE_CASES: +#if PCRE2_CODE_UNIT_WIDTH != 8 + VSPACE_MULTIBYTE_CASES: +#endif + eptr++; break; + } + } + ENDLOOP03: + break; + + case OP_NOT_DIGIT: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (MAX_255(*eptr) && (mb->ctypes[*eptr] & ctype_digit) != 0) break; + eptr++; + } + break; + + case OP_DIGIT: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!MAX_255(*eptr) || (mb->ctypes[*eptr] & ctype_digit) == 0) break; + eptr++; + } + break; + + case OP_NOT_WHITESPACE: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (MAX_255(*eptr) && (mb->ctypes[*eptr] & ctype_space) != 0) break; + eptr++; + } + break; + + case OP_WHITESPACE: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!MAX_255(*eptr) || (mb->ctypes[*eptr] & ctype_space) == 0) break; + eptr++; + } + break; + + case OP_NOT_WORDCHAR: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (MAX_255(*eptr) && (mb->ctypes[*eptr] & ctype_word) != 0) break; + eptr++; + } + break; + + case OP_WORDCHAR: + for (i = min; i < max; i++) + { + if (eptr >= mb->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!MAX_255(*eptr) || (mb->ctypes[*eptr] & ctype_word) == 0) break; + eptr++; + } + break; + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + + if (possessive) continue; /* No backtracking */ + for (;;) + { + if (eptr == pp) goto TAIL_RECURSE; + RMATCH(eptr, ecode, offset_top, mb, eptrb, RM47); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + if (ctype == OP_ANYNL && eptr > pp && *eptr == CHAR_LF && + eptr[-1] == CHAR_CR) eptr--; + } + } + + /* Control never gets here */ + } + + /* There's been some horrible disaster. Arrival here can only mean there is + something seriously wrong in the code above or the OP_xxx definitions. */ + + default: + RRETURN(PCRE2_ERROR_INTERNAL); + } + + /* Do not stick any code in here without much thought; it is assumed + that "continue" in the code above comes out to here to repeat the main + loop. */ + + } /* End of main loop */ +/* Control never reaches here */ + + +/* When compiling to use the heap rather than the stack for recursive calls to +match(), the RRETURN() macro jumps here. The number that is saved in +frame->Xwhere indicates which label we actually want to return to. */ + +#ifdef NO_RECURSE +#define LBL(val) case val: goto L_RM##val; +HEAP_RETURN: +switch (frame->Xwhere) + { + LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8) + LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17) + LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33) + LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52) + LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64) + LBL(65) LBL(66) +#ifdef SUPPORT_WIDE_CHARS + LBL(20) LBL(21) +#endif +#ifdef SUPPORT_UTF + LBL(16) LBL(18) + LBL(22) LBL(23) LBL(28) LBL(30) + LBL(32) LBL(34) LBL(42) LBL(46) + LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45) + LBL(59) LBL(60) LBL(61) LBL(62) LBL(67) +#endif /* SUPPORT_UTF */ + default: + return PCRE2_ERROR_INTERNAL; + } +#undef LBL +#endif /* NO_RECURSE */ +} + + +/*************************************************************************** +**************************************************************************** + RECURSION IN THE match() FUNCTION + +Undefine all the macros that were defined above to handle this. */ + +#ifdef NO_RECURSE +#undef eptr +#undef ecode +#undef mstart +#undef offset_top +#undef eptrb +#undef flags + +#undef callpat +#undef charptr +#undef data +#undef next +#undef pp +#undef prev +#undef saved_eptr + +#undef new_recursive + +#undef cur_is_word +#undef condition +#undef prev_is_word + +#undef ctype +#undef length +#undef max +#undef min +#undef number +#undef offset +#undef op +#undef save_capture_last +#undef save_offset1 +#undef save_offset2 +#undef save_offset3 +#undef stacksave + +#undef newptrb + +#endif + +/* These two are defined as macros in both cases */ + +#undef fc +#undef fi + +/*************************************************************************** +***************************************************************************/ + + +#ifdef NO_RECURSE +/************************************************* +* Release allocated heap frames * +*************************************************/ + +/* This function releases all the allocated frames. The base frame is on the +machine stack, and so must not be freed. + +Argument: the address of the base frame +Returns: nothing +*/ + +static void +release_match_heapframes (heapframe *frame_base) +{ +heapframe *nextframe = frame_base->Xnextframe; +while (nextframe != NULL) + { + heapframe *oldframe = nextframe; + nextframe = nextframe->Xnextframe; + (PUBL(stack_free))(oldframe); + } +} +#endif + + + +#ifdef FIXME + +/* FIXME: This must be in a new function..... */ + +/* Check for the special magic call that measures the size of the stack used +per recursive call of match(). Without the funny casting for sizeof, a Windows +compiler gave this error: "unary minus operator applied to unsigned type, +result still unsigned". Hopefully the cast fixes that. */ + +if (re == NULL && extra_data == NULL && subject == NULL && length == -999 && + start_offset == -999) +#ifdef NO_RECURSE + return -((int)sizeof(heapframe)); +#else + return match(NULL, NULL, NULL, 0, NULL, NULL, 0); +#endif + + +#endif /* FIXME */ + /************************************************* * Match a Regular Expression * @@ -61,87 +6184,739 @@ Arguments: length length of subject string (may contain binary zeros) start_offset where to start in the subject string options option bits - match_data points to a match_data block + match_data points to a match_data block Returns: > 0 => success; value is the number of ovector pairs filled = 0 => success, but ovector is not big enough - -1 => failed to match (PCRE_ERROR_NOMATCH) - < -1 => some kind of unexpected problem + -1 => failed to match (PCRE2_ERROR_NOMATCH) + -2 => partial match (PCRE2_ERROR_PARTIAL) + < -2 => some kind of unexpected problem */ -/* FIXME: this is currently a placeholder function */ - PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION -pcre2_match(const pcre2_code *code, PCRE2_SPTR subject, int length, - size_t start_offset, uint32_t options, pcre2_match_data *match_data, +pcre2_match(const pcre2_code *code, PCRE2_SPTR subject, int length, + size_t start_offset, uint32_t options, pcre2_match_data *match_data, pcre2_match_context *mcontext) { -int rc = PCRE2_ERROR_NOMATCH; +int rc; +int newline; +int ocount; -mcontext=mcontext;length=length; -options=options; +const uint8_t *start_bits = NULL; -/* Fudges for testing pcre2test */ +const pcre2_real_code *re = (const pcre2_real_code *)code; +pcre2_match_context default_context; /* For use if no context given */ -if (subject[0] == 'Y') - { - rc = 0; - match_data->code = code; - match_data->subject = subject; - match_data->leftchar = 0; - match_data->rightchar = 3; - match_data->startchar = 0; - match_data->mark = NULL; + +BOOL anchored; +BOOL firstline; +BOOL has_first_cu = FALSE; +BOOL has_req_cu = FALSE; +BOOL startline; +BOOL using_temporary_offsets = FALSE; +BOOL utf; + +PCRE2_UCHAR first_cu = 0; +PCRE2_UCHAR first_cu2 = 0; +PCRE2_UCHAR req_cu = 0; +PCRE2_UCHAR req_cu2 = 0; + +PCRE2_SPTR end_subject; +PCRE2_SPTR start_match = subject + start_offset; +PCRE2_SPTR req_cu_ptr = start_match - 1; +PCRE2_SPTR start_partial = NULL; +PCRE2_SPTR match_partial = NULL; + +/* We need to have mb pointing to a match block, because the IS_NEWLINE macro +is used below, and it expects NLBLOCK to be defined as a pointer. */ + +match_block actual_match_block; +match_block *mb = &actual_match_block; + +#ifdef NO_RECURSE +heapframe frame_zero; +frame_zero.Xprevframe = NULL; /* Marks the top level */ +frame_zero.Xnextframe = NULL; /* None are allocated yet */ +mb->match_frames_base = &frame_zero; +#endif + +/* A negative length implies a zero-terminated subject string. */ + +if (length < 0) length = PRIV(strlen)(subject); + +/* Plausibility checks */ + +if ((options & ~PUBLIC_MATCH_OPTIONS) != 0) return PCRE2_ERROR_BADOPTION; +if (code == NULL || subject == NULL || match_data == NULL) + return PCRE2_ERROR_NULL; +if ((int)start_offset > length) return PCRE2_ERROR_BADOFFSET; + +/* Check that the first field in the block is the magic number. If it is not, +return with PCRE2_ERROR_BADMAGIC. However, if the magic number is equal to +REVERSED_MAGIC_NUMBER we return with PCRE2_ERROR_BADENDIANNESS, which +means that the pattern is likely compiled with different endianness. */ + +if (re->magic_number != MAGIC_NUMBER) + return re->magic_number == REVERSED_MAGIC_NUMBER? + PCRE2_ERROR_BADENDIANNESS:PCRE2_ERROR_BADMAGIC; + +/* Check the code unit width. */ + +if ((re->flags & PCRE2_MODE_MASK) != PCRE2_CODE_UNIT_WIDTH/8) + return PCRE2_ERROR_BADMODE; - switch (match_data->oveccount) +/* A NULL match context means "use a default context" */ + +if (mcontext == NULL) + { + PRIV(match_context_init)(&default_context, TRUE); + mcontext = &default_context; + } + +/* These two settings are used in the code for checking a UTF string that +follows immediately afterwards. Other values in the mb block are used only +during interpretive pcre_match() processing, not when the JIT support is in +use, so they are set up later. */ + +utf = (re->overall_options & PCRE2_UTF) != 0; +mb->partial = ((options & PCRE2_PARTIAL_HARD) != 0)? 2 : + ((options & PCRE2_PARTIAL_SOFT) != 0)? 1 : 0; + +/* Check a UTF string for validity if required. For 8-bit and 16-bit strings, +we must also check that a starting offset does not point into the middle of a +multiunit character. */ + +#ifdef SUPPORT_UTF +if (utf && (options & PCRE2_NO_UTF_CHECK) == 0) + { + match_data->rc = PRIV(valid_utf)(subject, length, &(match_data->rightchar)); + if (match_data->rc != 0) { - case 0: break; - - case 1: match_data->ovector[0] = start_offset; - match_data->ovector[1] = start_offset + 4; - break; - - default: - case 6: match_data->ovector[10] = PCRE2_UNSET; - match_data->ovector[11] = PCRE2_UNSET; - - case 5: match_data->ovector[8] = PCRE2_UNSET; - match_data->ovector[9] = PCRE2_UNSET; - - case 4: match_data->ovector[6] = start_offset + 3; - match_data->ovector[7] = start_offset + 4; - rc += 2; - - case 3: match_data->ovector[4] = PCRE2_UNSET; - match_data->ovector[5] = PCRE2_UNSET; - - case 2: match_data->ovector[0] = start_offset; - match_data->ovector[1] = start_offset + 4; - match_data->ovector[2] = start_offset + 1; - match_data->ovector[3] = start_offset + 3; - match_data->mark = subject; - rc += 2; - break; - } - - } + match_data->leftchar = 0; + return match_data->rc; + } +#if PCRE2_CODE_UNIT_WIDTH != 32 + if (start_offset > 0 && (int)start_offset < length && + NOT_FIRSTCHAR(subject[start_offset])) + return PCRE2_ERROR_BADUTFOFFSET; +#endif /* PCRE2_CODE_UNIT_WIDTH != 32 */ + } +#endif /* SUPPORT_UTF */ -else if (subject[0] == 'P') +/* If the pattern was successfully studied with JIT support, run the JIT +executable instead of the rest of this function. Most options must be set at +compile time for the JIT code to be usable. Fallback to the normal code path if +an unsupported option is set or if JIT returns BADOPTION (which means that the +selected normal or partial matching mode was not compiled). */ + +#ifdef SUPPORT_JIT +if (re->executable_jit != NULL && (options & ~PUBLIC_JIT_EXEC_OPTIONS) == 0 && + mcontext->bsr_convention == 0 && mcontext->newline_convention == 0) { - rc = PCRE2_ERROR_PARTIAL; - match_data->code = code; - match_data->subject = subject; - match_data->leftchar = 0; - match_data->rightchar = length; - match_data->startchar = 1; - match_data->mark = NULL; + rc = PRIV(jit_exec)(subject, length, start_offset, options, match_data, + mcontext); + if (rc != PCRE2_ERROR_JIT_BADOPTION) return rc; + } +#endif + +/* Carry on with non-JIT matching. */ + +anchored = ((re->overall_options | options) & PCRE2_ANCHORED) != 0; +firstline = (re->overall_options & PCRE2_FIRSTLINE) != 0; +startline = (re->flags & PCRE2_STARTLINE) != 0; + +/* Fill in the fields in the match block. */ + +if (mcontext == NULL) + { + mb->memctl = re->memctl; + mb->callout = NULL; + } +else + { + mb->memctl = mcontext->memctl; + mb->callout = mcontext->callout; + mb->callout_data = NULL; /* FIXME */ } + +mb->start_subject = subject; +mb->start_offset = start_offset; +mb->end_subject = end_subject = mb->start_subject + length; +mb->hasthen = (re->flags & PCRE2_HASTHEN) != 0; + +mb->moptions = options; /* Match options */ +mb->poptions = re->overall_options; /* Pattern options */ + +mb->ignore_skip_arg = 0; +mb->mark = mb->nomatch_mark = NULL; /* In case never set */ +mb->recursive = NULL; /* No recursion at top level */ +mb->hitend = FALSE; + +/* The name table is needed for finding all the numbers associated with a +given name, for condition testing. The code follows the name table. */ + +mb->name_table = (PCRE2_UCHAR *)((uint8_t *)re + sizeof(pcre2_real_code)); +mb->name_count = re->name_count; +mb->name_entry_size = re->name_entry_size; +mb->start_code = mb->name_table + re->name_count * re->name_entry_size; + +/* Limits set in the pattern override the match context only if they are +smaller. */ + +mb->match_limit = (mcontext->match_limit < re->limit_match)? + mcontext->match_limit : re->limit_match; +mb->match_limit_recursion = (mcontext->recursion_limit > re->limit_recursion)? + mcontext->recursion_limit : re->limit_recursion; + +/* Pointers to the individual character tables */ + +mb->lcc = re->tables + lcc_offset; +mb->fcc = re->tables + fcc_offset; +mb->ctypes = re->tables + ctypes_offset; + +/* The match context /R convention, if set, overrides. */ + +mb->bsr_convention = (mcontext->bsr_convention != 0)? + mcontext->bsr_convention : re->bsr_convention; + +/* Process the newline setting. */ + +newline = (mcontext->newline_convention == 0)? + re->newline_convention : mcontext->newline_convention; + +mb->nltype = NLTYPE_FIXED; +switch(newline) + { + case PCRE2_NEWLINE_CR: + mb->nllen = 1; + mb->nl[0] = CHAR_CR; + break; + + case PCRE2_NEWLINE_LF: + mb->nllen = 1; + mb->nl[0] = CHAR_NL; + break; + + case PCRE2_NEWLINE_CRLF: + mb->nllen = 2; + mb->nl[0] = CHAR_CR; + mb->nl[1] = CHAR_NL; + break; + + case PCRE2_NEWLINE_ANY: + mb->nltype = NLTYPE_ANY; + break; + + case PCRE2_NEWLINE_ANYCRLF: + mb->nltype = NLTYPE_ANYCRLF; + break; + + default: return PCRE2_ERROR_INTERNAL; + } -else if (subject[0] == 'E') rc = PCRE2_ERROR_MATCHLIMIT; +/* If the expression has got more back references than the offsets supplied can +hold, we get a temporary chunk of memory to use during the matching. Otherwise, +we can use the vector supplied. The size of the ovector is three times the +value in the oveccount field. Two-thirds of it is pairs for storing matching +offsets, and the top third is working space. */ + +if (re->top_backref >= match_data->oveccount) + { + ocount = re->top_backref * 3 + 3; + mb->ovector = (size_t *)(mb->memctl.malloc(ocount * sizeof(size_t), + mb->memctl.memory_data)); + if (mb->ovector == NULL) return PCRE2_ERROR_NOMEMORY; + using_temporary_offsets = TRUE; + } +else + { + ocount = 3 * match_data->oveccount; + mb->ovector = match_data->ovector; + } + +mb->offset_end = ocount; +mb->offset_max = (2*ocount)/3; +mb->capture_last = 0; + +/* Reset the working variable associated with each extraction. These should +never be used unless previously set, but they get saved and restored, and so we +initialize them to avoid reading uninitialized locations. Also, unset the +offsets for the matched string. This is really just for tidiness with callouts, +in case they inspect these fields. */ + +if (ocount > 0) + { + register size_t *iptr = mb->ovector + ocount; + register size_t *iend = iptr - re->top_bracket; + if (iend < mb->ovector + 2) iend = mb->ovector + 2; + while (--iptr >= iend) *iptr = PCRE2_UNSET; + mb->ovector[0] = mb->ovector[1] = PCRE2_UNSET; + } + +/* Set up the first code unit to match, if available. The first_codeunit value +is never set for an anchored regular expression, but the anchoring may be +forced at run time, so we have to test for anchoring. The first code unit may +be unset for an unanchored pattern, of course. If there's no first code unit +there may be a bitmap of possible first characters. */ + +if (!anchored) + { + if ((re->flags & PCRE2_FIRSTSET) != 0) + { + has_first_cu = TRUE; + first_cu = first_cu2 = (PCRE2_UCHAR)(re->first_codeunit); + if ((re->flags & PCRE2_FIRSTCASELESS) != 0) + { + first_cu2 = TABLE_GET(first_cu, mb->fcc, first_cu); +#if defined SUPPORT_UTF && PCRE2_CODE_UNIT_WIDTH != 8 + if (utf && first_cu > 127) first_cu2 = UCD_OTHERCASE(first_cu); +#endif + } + } + else + if (!startline && (re->flags & PCRE2_FIRSTMAPSET) != 0) + start_bits = re->start_bitmap; + } + +/* For anchored or unanchored matches, there may be a "last known required +character" set. */ + +if ((re->flags & PCRE2_LASTSET) != 0) + { + has_req_cu = TRUE; + req_cu = req_cu2 = (PCRE2_UCHAR)(re->last_codeunit); + if ((re->flags & PCRE2_LASTCASELESS) != 0) + { + req_cu2 = TABLE_GET(req_cu, mb->fcc, req_cu); +#if defined SUPPORT_UTF && PCRE2_CODE_UNIT_WIDTH != 8 + if (utf && req_cu > 127) req_cu2 = UCD_OTHERCASE(req_cu); +#endif + } + } -match_data->rc = rc; -return rc; -} +/* ==========================================================================*/ + +/* Loop for handling unanchored repeated matching attempts; for anchored regexs +the loop runs just once. */ + +for(;;) + { + PCRE2_SPTR new_start_match; + + /* ----------------- Start of match optimizations ---------------- */ + + /* There are some optimizations that avoid running the match if a known + starting point is not found, or if a known later code unit is not present. + However, there is an option (settable at compile or match time) that disables + these, for testing and for ensuring that all callouts do actually occur. */ + + if (((options | re->overall_options) & PCRE2_NO_START_OPTIMIZE) == 0) + { + PCRE2_SPTR save_end_subject = end_subject; + + /* If firstline is TRUE, the start of the match is constrained to the first + line of a multiline string. That is, the match must be before or at the + first newline. Implement this by temporarily adjusting end_subject so that + we stop the optimization scans at a newline. If the match fails at the + newline, later code breaks this loop. */ + + if (firstline) + { + PCRE2_SPTR t = start_match; +#ifdef SUPPORT_UTF + if (utf) + { + while (t < mb->end_subject && !IS_NEWLINE(t)) + { + t++; + ACROSSCHAR(t < end_subject, *t, t++); + } + } + else +#endif + while (t < mb->end_subject && !IS_NEWLINE(t)) t++; + end_subject = t; + } + + /* Advance to a unique first code unit if there is one. */ + + if (has_first_cu) + { + PCRE2_UCHAR smc; + if (first_cu != first_cu2) + while (start_match < end_subject && + (smc = UCHAR21TEST(start_match)) != first_cu && smc != first_cu2) + start_match++; + else + while (start_match < end_subject && UCHAR21TEST(start_match) != first_cu) + start_match++; + } + + /* Or to just after a linebreak for a multiline match */ + + else if (startline) + { + if (start_match > mb->start_subject + start_offset) + { +#ifdef SUPPORT_UTF + if (utf) + { + while (start_match < end_subject && !WAS_NEWLINE(start_match)) + { + start_match++; + ACROSSCHAR(start_match < end_subject, *start_match, + start_match++); + } + } + else +#endif + while (start_match < end_subject && !WAS_NEWLINE(start_match)) + start_match++; + + /* If we have just passed a CR and the newline option is ANY or + ANYCRLF, and we are now at a LF, advance the match position by one more + code unit. */ + + if (start_match[-1] == CHAR_CR && + (mb->nltype == NLTYPE_ANY || mb->nltype == NLTYPE_ANYCRLF) && + start_match < end_subject && + UCHAR21TEST(start_match) == CHAR_NL) + start_match++; + } + } + + /* Or to a non-unique first code unit if any have been identified. The + bitmap contains only 256 bits. When code units are 16 or 32 bits wide, all + code units greater than 254 set the 255 bit. */ + + else if (start_bits != NULL) + { + while (start_match < end_subject) + { + register uint32_t c = UCHAR21TEST(start_match); +#if PCRE2_CODE_UNIT_WIDTH != 8 + if (c > 255) c = 255; +#endif + if ((start_bits[c/8] & (1 << (c&7))) != 0) break; + start_match++; + } + } + + /* Restore fudged end_subject */ + + end_subject = save_end_subject; + + /* The following two optimizations are disabled for partial matching. */ + + if (!mb->partial) + { + /* The minimum matching length is a lower bound; no actual string of that + length may actually match the pattern. Although the value is, strictly, + in characters, we treat it as code units to avoid spending too much time + in this optimization. */ + + if (end_subject - start_match < re->minlength) + { + rc = MATCH_NOMATCH; + break; + } + + /* If req_cu is set, we know that that code unit must appear in the + subject for the match to succeed. If the first code unit is set, req_cu + must be later in the subject; otherwise the test starts at the match + point. This optimization can save a huge amount of backtracking in + patterns with nested unlimited repeats that aren't going to match. + Writing separate code for cased/caseless versions makes it go faster, as + does using an autoincrement and backing off on a match. + + HOWEVER: when the subject string is very, very long, searching to its end + can take a long time, and give bad performance on quite ordinary + patterns. This showed up when somebody was matching something like + /^\d+C/ on a 32-megabyte string... so we don't do this when the string is + sufficiently long. */ + + if (has_req_cu && end_subject - start_match < REQ_CU_MAX) + { + register PCRE2_SPTR p = start_match + (has_first_cu? 1:0); + + /* We don't need to repeat the search if we haven't yet reached the + place we found it at last time. */ + + if (p > req_cu_ptr) + { + if (req_cu != req_cu2) + { + while (p < end_subject) + { + register uint32_t pp = UCHAR21INCTEST(p); + if (pp == req_cu || pp == req_cu2) { p--; break; } + } + } + else + { + while (p < end_subject) + { + if (UCHAR21INCTEST(p) == req_cu) { p--; break; } + } + } + + /* If we can't find the required code unit, break the matching loop, + forcing a match failure. */ + + if (p >= end_subject) + { + rc = MATCH_NOMATCH; + break; + } + + /* If we have found the required code unit, save the point where we + found it, so that we don't search again next time round the loop if + the start hasn't passed this code unit yet. */ + + req_cu_ptr = p; + } + } + } + } + + /* ------------ End of start of match optimizations ------------ */ + + /* OK, we can now run the match. If "hitend" is set afterwards, remember the + first starting point for which a partial match was found. */ + + mb->start_match_ptr = start_match; + mb->start_used_ptr = start_match; + mb->match_call_count = 0; + mb->match_function_type = 0; + mb->end_offset_top = 0; + mb->skip_arg_count = 0; + rc = match(start_match, mb->start_code, start_match, 2, mb, NULL, 0); + + if (mb->hitend && start_partial == NULL) + { + start_partial = mb->start_used_ptr; + match_partial = start_match; + } + + switch(rc) + { + /* If MATCH_SKIP_ARG reaches this level it means that a MARK that matched + the SKIP's arg was not found. In this circumstance, Perl ignores the SKIP + entirely. The only way we can do that is to re-do the match at the same + point, with a flag to force SKIP with an argument to be ignored. Just + treating this case as NOMATCH does not work because it does not check other + alternatives in patterns such as A(*SKIP:A)B|AC when the subject is AC. */ + + case MATCH_SKIP_ARG: + new_start_match = start_match; + mb->ignore_skip_arg = mb->skip_arg_count; + break; + + /* SKIP passes back the next starting point explicitly, but if it is no + greater than the match we have just done, treat it as NOMATCH. */ + + case MATCH_SKIP: + if (mb->start_match_ptr > start_match) + { + new_start_match = mb->start_match_ptr; + break; + } + /* Fall through */ + + /* NOMATCH and PRUNE advance by one character. THEN at this level acts + exactly like PRUNE. Unset ignore SKIP-with-argument. */ + + case MATCH_NOMATCH: + case MATCH_PRUNE: + case MATCH_THEN: + mb->ignore_skip_arg = 0; + new_start_match = start_match + 1; +#ifdef SUPPORT_UTF + if (utf) + ACROSSCHAR(new_start_match < end_subject, *new_start_match, + new_start_match++); +#endif + break; + + /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */ + + case MATCH_COMMIT: + rc = MATCH_NOMATCH; + goto ENDLOOP; + + /* Any other return is either a match, or some kind of error. */ + + default: + goto ENDLOOP; + } + + /* Control reaches here for the various types of "no match at this point" + result. Reset the code to MATCH_NOMATCH for subsequent checking. */ + + rc = MATCH_NOMATCH; + + /* If PCRE2_FIRSTLINE is set, the match must happen before or at the first + newline in the subject (though it may continue over the newline). Therefore, + if we have just failed to match, starting at a newline, do not continue. */ + + if (firstline && IS_NEWLINE(start_match)) break; + + /* Advance to new matching position */ + + start_match = new_start_match; + + /* Break the loop if the pattern is anchored or if we have passed the end of + the subject. */ + + if (anchored || start_match > end_subject) break; + + /* If we have just passed a CR and we are now at a LF, and the pattern does + not contain any explicit matches for \r or \n, and the newline option is CRLF + or ANY or ANYCRLF, advance the match position by one more code unit. In + normal matching start_match will aways be greater than the first position at + this stage, but a failed *SKIP can cause a return at the same point, which is + why the first test exists. */ + + if (start_match > subject + start_offset && + start_match[-1] == CHAR_CR && + start_match < end_subject && + *start_match == CHAR_NL && + (re->flags & PCRE2_HASCRORLF) == 0 && + (mb->nltype == NLTYPE_ANY || + mb->nltype == NLTYPE_ANYCRLF || + mb->nllen == 2)) + start_match++; + + mb->mark = NULL; /* Reset for start of next match attempt */ + } /* End of for(;;) "bumpalong" loop */ + +/* ==========================================================================*/ + +/* When we reach here, one of the stopping conditions is true: + +(1) The match succeeded, either completely, or partially; + +(2) The pattern is anchored or the match was failed by (*COMMIT); + +(3) We are past the end of the subject; + +(4) PCRE2_FIRSTLINE is set and we have failed to match at a newline, because + this option requests that a match occur at or before the first newline in + the subject. + +(5) Some kind of error occurred. + +*/ + +ENDLOOP: + +#ifdef NO_RECURSE +release_match_heapframes(&frame_zero); +#endif + +/* Handle a fully successful match. */ + +if (rc == MATCH_MATCH || rc == MATCH_ACCEPT) + { + uint32_t arg_offset_max = 2 * match_data->oveccount; + + /* When the offset vector is big enough to deal with any backreferences, + captured substring offsets will already be set up. In the case where we had + to get some local memory to hold offsets for backreference processing, copy + those that we can. In this case there need not be overflow if certain parts + of the pattern were not used, even though there are more capturing + parentheses than vector slots. */ + + if (using_temporary_offsets) + { + if (arg_offset_max >= 4) + { + memcpy(match_data->ovector + 2, mb->ovector + 2, + (arg_offset_max - 2) * sizeof(size_t)); + } + if (mb->end_offset_top > arg_offset_max) mb->capture_last |= OVFLBIT; + mb->memctl.free(mb->ovector, mb->memctl.memory_data); + } + + /* Set the return code to the number of captured strings, or 0 if there were + too many to fit into the ovector. */ + + match_data->rc = ((mb->capture_last & OVFLBIT) != 0 && + mb->end_offset_top >= arg_offset_max)? + 0 : mb->end_offset_top/2; + + /* If there is space in the offset vector, set any unused pairs at the end to + PCRE2_UNSET for backwards compatibility. It is documented that this happens. + In earlier versions, the whole set of potential capturing offsets was + initialized each time round the loop, but this is handled differently now. + "Gaps" are set to PCRE2_UNSET dynamically instead (this fixes a bug). Thus, + it is only those at the end that need setting here. We can't just set them + all at the start of the whole thing because they may get set in one branch + that is not the final matching branch. */ + + if (mb->end_offset_top/2 <= re->top_bracket) + { + register size_t *iptr, *iend; + int resetcount = re->top_bracket + 1; + if (resetcount > match_data->oveccount) resetcount = match_data->oveccount; + iptr = match_data->ovector + mb->end_offset_top; + iend = match_data->ovector + 2 * resetcount; + while (iptr < iend) *iptr++ = PCRE2_UNSET; + } + + /* If there is space, set up the whole thing as substring 0. The value of + mb->start_match_ptr might be modified if \K was encountered on the success + matching path. */ + + if (match_data->oveccount < 1) rc = 0; else + { + match_data->ovector[0] = mb->start_match_ptr - mb->start_subject; + match_data->ovector[1] = mb->end_match_ptr - mb->start_subject; + } + + /* Fill in the remaining fields that are returned in the match data. */ + + match_data->code = re; + match_data->subject = subject; + match_data->leftchar = mb->start_used_ptr - subject; + match_data->rightchar = 0; /* FIXME */ + match_data->startchar = start_match - subject; + match_data->mark = mb->mark; + return match_data->rc; + } + +/* Control gets here if there has been a partial match, an error, or if the +overall match attempt has failed at all permitted starting positions. For +anything other than nomatch or partial match, just return the code. */ + +if (rc != MATCH_NOMATCH && rc != PCRE2_ERROR_PARTIAL) + match_data->rc = rc; + +/* Handle a partial match. */ + +else if (match_partial != NULL) + { + if (match_data->oveccount > 0) + { + match_data->ovector[0] = match_partial - subject; + match_data->ovector[1] = end_subject - subject; + } + match_data->leftchar = start_partial - subject; + match_data->rc = PCRE2_ERROR_PARTIAL; + } + +/* This is the classic nomatch case. */ + +else + { + match_data->rc = PCRE2_ERROR_NOMATCH; + match_data->mark = mb->nomatch_mark; + } + +if (using_temporary_offsets) + mb->memctl.free(mb->ovector, mb->memctl.memory_data); +return match_data->rc; +} /* End of pcre2_match.c */ diff --git a/src/pcre2_newline.c b/src/pcre2_newline.c index 1110db7..0b08c99 100644 --- a/src/pcre2_newline.c +++ b/src/pcre2_newline.c @@ -60,8 +60,10 @@ http://unicode.org/unicode/reports/tr18/. */ * Check for newline at given position * *************************************************/ -/* It is guaranteed that the initial value of ptr is less than the end of the -string that is being processed. +/* This function is called only via the IS_NEWLINE macro, which does so only +when the newline type is NLTYPE_ANY or NLTYPE_ANYCRLF. The case of a fixed +newline (NLTYPE_FIXED) is handled inline. It is guaranteed that the code unit +pointed to by ptr is less than the end of the string. Arguments: ptr pointer to possible newline @@ -74,27 +76,30 @@ Returns: TRUE or FALSE */ BOOL -PRIV(is_newline)(PCRE2_SPTR ptr, int type, PCRE2_SPTR endptr, int *lenptr, - BOOL utf) +PRIV(is_newline)(PCRE2_SPTR ptr, uint32_t type, PCRE2_SPTR endptr, + uint32_t *lenptr, BOOL utf) { uint32_t c; #ifdef SUPPORT_UTF -if (utf) { GETCHAR(c, ptr); } else +if (utf) { GETCHAR(c, ptr); } else c = *ptr; #else (void)utf; +c = *ptr; #endif /* SUPPORT_UTF */ - c = *ptr; - -/* Note that this function is called only for ANY or ANYCRLF. */ - if (type == NLTYPE_ANYCRLF) switch(c) { - case CHAR_LF: *lenptr = 1; return TRUE; - case CHAR_CR: *lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1; - return TRUE; - default: return FALSE; + case CHAR_LF: + *lenptr = 1; + return TRUE; + + case CHAR_CR: + *lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1; + return TRUE; + + default: + return FALSE; } /* NLTYPE_ANY */ @@ -106,7 +111,9 @@ else switch(c) #endif case CHAR_LF: case CHAR_VT: - case CHAR_FF: *lenptr = 1; return TRUE; + case CHAR_FF: + *lenptr = 1; + return TRUE; case CHAR_CR: *lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1; @@ -114,17 +121,26 @@ else switch(c) #ifndef EBCDIC #if PCRE2_CODE_UNIT_WIDTH == 8 - case CHAR_NEL: *lenptr = utf? 2 : 1; return TRUE; - case 0x2028: /* LS */ - case 0x2029: *lenptr = 3; return TRUE; /* PS */ + case CHAR_NEL: + *lenptr = utf? 2 : 1; + return TRUE; + + case 0x2028: /* LS */ + case 0x2029: /* PS */ + *lenptr = 3; + return TRUE; + #else /* 16-bit or 32-bit code units */ case CHAR_NEL: - case 0x2028: /* LS */ - case 0x2029: *lenptr = 1; return TRUE; /* PS */ + case 0x2028: /* LS */ + case 0x2029: /* PS */ + *lenptr = 1; + return TRUE; #endif #endif /* Not EBCDIC */ - default: return FALSE; + default: + return FALSE; } } @@ -134,8 +150,10 @@ else switch(c) * Check for newline at previous position * *************************************************/ -/* It is guaranteed that the initial value of ptr is greater than the start of -the string that is being processed. +/* This function is called only via the WAS_NEWLINE macro, which does so only +when the newline type is NLTYPE_ANY or NLTYPE_ANYCRLF. The case of a fixed +newline (NLTYPE_FIXED) is handled inline. It is guaranteed that the initial +value of ptr is greater than the start of the string that is being processed. Arguments: ptr pointer to possible newline @@ -148,8 +166,8 @@ Returns: TRUE or FALSE */ BOOL -PRIV(was_newline)(PCRE2_SPTR ptr, int type, PCRE2_SPTR startptr, int *lenptr, - BOOL utf) +PRIV(was_newline)(PCRE2_SPTR ptr, uint32_t type, PCRE2_SPTR startptr, + uint32_t *lenptr, BOOL utf) { uint32_t c; ptr--; @@ -160,23 +178,24 @@ if (utf) BACKCHAR(ptr); GETCHAR(c, ptr); } -else +else c = *ptr; #else (void)utf; +c = *ptr; #endif /* SUPPORT_UTF */ - c = *ptr; - -/* Note that this function is called only for ANY or ANYCRLF. */ - if (type == NLTYPE_ANYCRLF) switch(c) { case CHAR_LF: *lenptr = (ptr > startptr && ptr[-1] == CHAR_CR)? 2 : 1; return TRUE; - case CHAR_CR: *lenptr = 1; return TRUE; - default: return FALSE; + case CHAR_CR: + *lenptr = 1; + return TRUE; + + default: + return FALSE; } /* NLTYPE_ANY */ @@ -192,21 +211,32 @@ else switch(c) #endif case CHAR_VT: case CHAR_FF: - case CHAR_CR: *lenptr = 1; return TRUE; + case CHAR_CR: + *lenptr = 1; + return TRUE; #ifndef EBCDIC #if PCRE2_CODE_UNIT_WIDTH == 8 - case CHAR_NEL: *lenptr = utf? 2 : 1; return TRUE; - case 0x2028: /* LS */ - case 0x2029: *lenptr = 3; return TRUE; /* PS */ + case CHAR_NEL: + *lenptr = utf? 2 : 1; + return TRUE; + + case 0x2028: /* LS */ + case 0x2029: /* PS */ + *lenptr = 3; + return TRUE; + #else /* 16-bit or 32-bit code units */ case CHAR_NEL: - case 0x2028: /* LS */ - case 0x2029: *lenptr = 1; return TRUE; /* PS */ + case 0x2028: /* LS */ + case 0x2029: /* PS */ + *lenptr = 1; + return TRUE; #endif #endif /* Not EBCDIC */ - default: return FALSE; + default: + return FALSE; } } diff --git a/src/pcre2_pattern_info.c b/src/pcre2_pattern_info.c index c04c6ac..940dd1f 100644 --- a/src/pcre2_pattern_info.c +++ b/src/pcre2_pattern_info.c @@ -150,7 +150,6 @@ switch(what) break; case PCRE2_INFO_MATCHLIMIT: - if ((re->flags & PCRE2_MLSET) == 0) return PCRE2_ERROR_UNSET; *((uint32_t *)where) = re->limit_match; break; @@ -179,7 +178,6 @@ switch(what) break; case PCRE2_INFO_RECURSIONLIMIT: - if ((re->flags & PCRE2_RLSET) == 0) return PCRE2_ERROR_UNSET; *((uint32_t *)where) = re->limit_recursion; break; diff --git a/src/pcre2test.c b/src/pcre2test.c index 7e813e4..1d67c87 100644 --- a/src/pcre2test.c +++ b/src/pcre2test.c @@ -2632,7 +2632,7 @@ pattern_info(int what, void *where) { int rc; PCRE2_PATTERN_INFO(rc, compiled_code, what, where); -if (rc >= 0 || rc == PCRE2_ERROR_UNSET) return 0; +if (rc >= 0) return 0; fprintf(outfile, "Error %d from pcre2_pattern_info_%d(%d)\n", rc, test_mode, what); if (rc == PCRE2_ERROR_BADMODE) @@ -2831,7 +2831,7 @@ if ((pat_patctl.control & CTL_INFO) != 0) int nameentrysize, namecount; uint32_t bsr_convention, newline_convention; uint32_t first_cunit, last_cunit; - uint32_t match_limit = 0, recursion_limit = 0; + uint32_t match_limit, recursion_limit; /* These info requests should always succeed. */ @@ -2865,10 +2865,10 @@ if ((pat_patctl.control & CTL_INFO) != 0) if (maxlookbehind > 0) fprintf(outfile, "Max lookbehind = %d\n", maxlookbehind); - if (match_limit > 0) + if (match_limit != UINT32_MAX) fprintf(outfile, "Match limit = %u\n", match_limit); - if (recursion_limit > 0) + if (recursion_limit != UINT32_MAX) fprintf(outfile, "Recursion limit = %u\n", recursion_limit); if (namecount > 0)