th-libs: th_regex.c annotate

annotate th_regex.c @ 639:8c957ad9d4c3

Some more work on regex stuff.

author	Matti Hamalainen <ccr@tnsp.org>
date	Thu, 23 Jan 2020 11:38:28 +0200
parents	c4bca120bfb0
children	9e1f9e1d1487

rev	line source
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1 /*
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	2 * Simple regular expression matching functionality
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	3 * Programmed and designed by Matti 'ccr' Hamalainen
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	4 * (C) Copyright 2020 Tecnic Software productions (TNSP)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	5 *
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	6 * Please read file 'COPYING' for information on license and distribution.
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	7 */
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	8 #include "th_util.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	9 #include "th_regex.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	10
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	11
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	12 #ifdef TH_EXPERIMENTAL_REGEX_DEBUG
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	13 int th_dbg_re_flags = 0;
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	14 # define DBG_RE_COMPILE(...) do { if (th_dbg_re_flags & TH_DBG_RE_COMPILE) fprintf(stderr, __VA_ARGS__); } while (0)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	15 # define DBG_RE_FREE(...) do { if (th_dbg_re_flags & TH_DBG_RE_FREE) fprintf(stderr, __VA_ARGS__); } while (0)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	16 # define DBG_RE_MATCH(...) do { if (th_dbg_re_flags & TH_DBG_RE_MATCH) fprintf(stderr, __VA_ARGS__); } while (0)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	17 #else
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	18 # define DBG_RE_COMPILE(...)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	19 # define DBG_RE_FREE(...)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	20 # define DBG_RE_MATCH(...)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	21 #endif
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	22
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	23
614 afcaf5e38f56 Disable regex stuff from normal builds. Matti Hamalainen <ccr@tnsp.org> parents: 613 diff changeset	24 #ifdef TH_EXPERIMENTAL_REGEX
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	25
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	26 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	27 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	28 TH_RE_MATCH_ONCE,
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	29 TH_RE_MATCH_COUNT_GREEDY,
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	30 TH_RE_MATCH_COUNT_NONGREEDY,
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	31 TH_RE_MATCH_ANCHOR_START,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	32 TH_RE_MATCH_ANCHOR_END,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	33 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	34
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	35
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	36 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	37 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	38 TH_RE_TYPE_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	39 TH_RE_TYPE_ANY_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	40 TH_RE_TYPE_LIST,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	41 TH_RE_TYPE_LIST_REVERSE,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	42 TH_RE_TYPE_SUBEXPR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	43 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	44
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	45
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	46 #ifdef TH_EXPERIMENTAL_REGEX_DEBUG
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	47 static const char *re_match_modes[] =
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	48 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	49 "ONCE",
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	50 "COUNT GREEDY",
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	51 "COUNT NONGREEDY",
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	52 "ANCHOR_START",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	53 "ANCHOR_END",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	54 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	55
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	56 static const char *re_match_types[] =
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	57 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	58 "CHAR",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	59 "ANY",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	60 "LIST",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	61 "LIST_REVERSE",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	62 "SUBEXPR",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	63 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	64 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	65
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	66
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	67 typedef struct
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	68 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	69 const th_regex_char *pattern;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	70 size_t offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	71
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	72 th_regex_ctx *data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	73
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	74 size_t nstack, stacksize;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	75 th_regex_ctx **stack;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	76 } th_regex_parse_ctx;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	77
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	78
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	79 static void th_regex_node_init(th_regex_node *node)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	80 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	81 memset(node, 0, sizeof(th_regex_node));
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	82 node->mode = TH_RE_MATCH_ONCE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	83 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	84
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	85
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	86 static int th_regex_strndup(th_regex_char **pdst,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	87 const th_regex_char *src, const size_t len)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	88 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	89 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	90 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	91
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	92 if (UINTPTR_MAX / sizeof(th_regex_char) < len + 1)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	93 return THERR_BOUNDS;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	94
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	95 if ((pdst = (th_regex_char )
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	96 th_malloc((len + 1) * sizeof(th_regex_char))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	97 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	98
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	99 memcpy(pdst, src, len sizeof(th_regex_char));
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	100 (*pdst)[len] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	101
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	102 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	103 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	104
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	105
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	106 static int th_regex_ctx_get_prev_node(
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	107 th_regex_parse_ctx ctx, th_regex_node *pnode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	108 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	109 if (ctx->data != NULL && ctx->data->nnodes > 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	110 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	111 *pnode = &ctx->data->nodes[ctx->data->nnodes - 1];
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	112 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	113 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	114 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	115 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	116 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	117
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	118
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	119 static int th_regex_ctx_push(th_regex_parse_ctx *ctx)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	120 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	121 if (ctx->stack == NULL \|\| ctx->nstack + 1 >= ctx->stacksize)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	122 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	123 ctx->stacksize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	124
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	125 if ((ctx->stack = th_realloc(ctx->stack,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	126 ctx->stacksize * sizeof(th_regex_node *))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	127 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	128 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	129
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	130 ctx->stack[ctx->nstack] = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	131 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	132 ctx->data = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	133
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	134 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	135 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	136
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	137
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	138 static int th_regex_ctx_pop(th_regex_parse_ctx ctx, th_regex_ctx *data)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	139 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	140 if (ctx->nstack > 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	141 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	142 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	143 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	144 ctx->data = ctx->stack[ctx->nstack];
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	145 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	146 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	147 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	148 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	149 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	150
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	151
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152 static int th_regex_ctx_node_commit(th_regex_parse_ctx ctx, th_regex_node node)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	153 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	154 th_regex_ctx *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	155
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	156 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	157 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	158 if ((data = ctx->data = th_malloc0(sizeof(th_regex_ctx))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	159 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	160 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	161
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	162 if (data->nodes == NULL \|\| data->nnodes + 1 >= data->nodessize)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	163 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	164 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	165 if ((data->nodes = th_realloc(data->nodes,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	166 data->nodessize * sizeof(th_regex_node))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	167 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	168 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	169
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	170 memcpy(&data->nodes[data->nnodes], node, sizeof(th_regex_node));
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	171 data->nnodes++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	172
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	173 DBG_RE_COMPILE(
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	174 "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	175 "mode=%d, type=%d, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	176 data->nnodes, data->nodessize,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	177 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	178 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	179
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	180 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	181 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	182
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	183
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	184 static BOOL th_regex_find_next(const th_regex_char *str,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	185 const size_t start, size_t *offs,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	186 const th_regex_char delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	187 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	188 for (offs = start; str[offs] != 0; (*offs)++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	189 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	190 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	191 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	192 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	193 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	194 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	195
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	196
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	197 static BOOL th_regex_parse_ssize_t(const th_regex_char *str,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	198 ssize_t *value)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	199 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	200 th_regex_char ch;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	201 BOOL neg;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	202
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	203 if (*str == '-')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	204 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	205 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	206 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	207 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	208 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	209 neg = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	210
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	211 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	212 while ((ch = *str++))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	213 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	214 if (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	215 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	216 value = 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	217 *value += ch - '0';
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	218 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	219 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	220 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	221 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	222
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	223 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	224 value = -(value);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	225
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	226 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	227 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	228
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	229
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	230 static void th_regex_list_item_init(th_regex_list_item *item)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	231 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	232 memset(item, 0, sizeof(th_regex_list_item));
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	233 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	234
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	235
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	236 static int th_regex_list_add_item(th_regex_list list, th_regex_list_item item)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	237 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	238 if (list->items == NULL \|\| list->nitems + 1 >= list->itemssize)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	239 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	240 list->itemssize += 16;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	241
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	242 if ((list->items = th_realloc(list->items,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	243 list->itemssize * sizeof(th_regex_list_item))) == NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	244 return THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	245 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	246
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	247 memcpy(list->items + list->nitems, item, sizeof(th_regex_list_item));
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	248 list->nitems++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	249
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	250 return THERR_OK;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	251 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	252
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	253
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	254 static void th_regex_list_free(th_regex_list *list)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	255 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	256 if (list != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	257 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	258 for (size_t n = 0; n < list->nitems; n++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	259 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	260 th_free(list->items[n].chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	261 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	262 th_free(list->items);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	263 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	264 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	265
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	266
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	267 static int th_regex_parse_list(const th_regex_char *str,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	268 const size_t slen, th_regex_list *list)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	269 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	270 th_regex_char *tmp = NULL;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	271 th_regex_list_item item;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	272 int res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	273
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	274 if ((res = th_regex_strndup(&tmp, str, slen)) != THERR_OK)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	275 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	276
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	277 // Handle ranges like [A-Z]
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	278 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	279 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	280 th_regex_char
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	281 *prev = (offs > 0) ? tmp + offs - 1 : NULL,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	282 *curr = tmp + offs,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	283 *next = (offs + 1 < slen) ? tmp + offs + 1 : NULL;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	284
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	285 if (*curr == '-')
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	286 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	287 if (prev != NULL && next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	288 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	289 // Range
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	290 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	291 item.type = 1;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	292 item.start = *prev;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	293 item.end = *next;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	294
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	295 if (item.start <= item.end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	296 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	297 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	298 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	299 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	300
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	301 curr = prev = *next = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	302
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	303 if ((res = th_regex_list_add_item(list, &item)) != THERR_OK)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	304 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	305 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	306 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	307 if (next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	308 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	309 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	310 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	311 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	312 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	313 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	314
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	315 // Count number of remaining characters
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	316 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	317 item.type = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	318 item.nchars = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	319
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	320 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	321 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	322 th_regex_char curr = tmp[offs];
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	323 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	324 item.nchars++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	325 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	326
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	327 if (item.nchars > 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	328 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	329 if ((item.chars = th_malloc(sizeof(th_regex_char) * item.nchars)) == NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	330 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	331 res = THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	332 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	333 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	334
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	335 for (size_t offs = 0, n = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	336 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	337 th_regex_char curr = tmp[offs];
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	338 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	339 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	340 item.chars[n] = curr;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	341 n++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	342 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	343 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	344
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	345 if ((res = th_regex_list_add_item(list, &item)) != THERR_OK)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	346 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	347 th_free(item.chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	348 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	349 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	350 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	351
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	352 out:
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	353 th_free(tmp);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	354 return res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	355 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	356
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	357
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	358 int th_regex_compile(th_regex_ctx *pexpr, const th_regex_char pattern)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	359 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	360 int res = THERR_OK;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	361 th_regex_parse_ctx ctx;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	362 th_regex_node node, *pnode;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	363 th_regex_char *tmp = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	364 size_t start;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	365
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	366 if (pexpr == NULL \|\| pattern == NULL)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	367 {
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	368 res = THERR_NULLPTR;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	369 goto exit;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	370 }
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	371
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	372 memset(&ctx, 0, sizeof(ctx));
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	373 ctx.pattern = pattern;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	374
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	375 for (; ctx.pattern[ctx.offs] != 0; ctx.offs++)
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	376 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	377 th_regex_char cch = ctx.pattern[ctx.offs];
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	378 DBG_RE_COMPILE("[%" PRIu_SIZE_T "] '%c'\n", ctx.offs, cch);
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	379 switch (cch)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	380 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	381 case '?':
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	382 case '*':
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	383 case '+':
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	384 if ((res = th_regex_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	385 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	386
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	387 if (cch == '?')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	388 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	389 // Check if previous was a count
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	390 pnode->mode = (pnode->mode == TH_RE_MATCH_COUNT_GREEDY) ?
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	391 TH_RE_MATCH_COUNT_NONGREEDY : TH_RE_MATCH_COUNT_GREEDY;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	392
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	393 // Previous token is optional (repeat 0-1 times)
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	394 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	395 pnode->repeatMax = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	396 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	397 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	398 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	399 // Check if previous was a count
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	400 if (pnode->mode == TH_RE_MATCH_COUNT_GREEDY \|\|
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	401 pnode->mode == TH_RE_MATCH_COUNT_NONGREEDY)
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	402 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	403 res = THERR_INVALID_DATA;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	404 goto exit;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	405 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	406
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	407 pnode->mode = TH_RE_MATCH_COUNT_GREEDY;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	408
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	409 if (cch == '*')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	410 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	411 // Previous token can repeat 0 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	412 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	413 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	414 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	415 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	416 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	417 // Previous token must repeat 1 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	418 pnode->repeatMin = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	419 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	420 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	421 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	422 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	423
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	424 case '{':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	425 // {n} \| {min,max}
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	426 start = ctx.offs + 1;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	427 if (!th_regex_find_next(ctx.pattern, start, &ctx.offs, '}'))
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	428 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	429 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	430 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	431 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	432 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	433
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	434 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	435
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	436 if ((res = th_regex_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK \|\|
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	437 (res = th_regex_strndup(&tmp, ctx.pattern + start,
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	438 ctx.offs - start)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	439 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	440
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	441 pnode->mode = TH_RE_MATCH_COUNT_GREEDY;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	442
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	443 if (th_regex_find_next(tmp, 0, &start, ','))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	444 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	445 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	446 if (!th_regex_parse_ssize_t(tmp, &pnode->repeatMin) \|\|
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	447 !th_regex_parse_ssize_t(tmp + start + 1, &pnode->repeatMax))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	448 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	449 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	450 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	451 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	452 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	453 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	454 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	455 if (!th_regex_parse_ssize_t(tmp, &pnode->repeatMin))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	456 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	457 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	458 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	459 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	460 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	461 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	462
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	463 if (pnode->repeatMin < 0 \|\| pnode->repeatMax < 1 \|\|
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	464 pnode->repeatMax < pnode->repeatMin)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	465 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	466 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	467 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	468 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	469 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	470 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	471
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	472 case '(':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	473 // Start of subpattern
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	474 if ((res = th_regex_ctx_push(&ctx)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	475 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	476 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	477
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	478 case ')':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	479 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	480 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	481 node.type = TH_RE_TYPE_SUBEXPR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	482
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	483 if ((res = th_regex_ctx_pop(&ctx, &node.match.expr)) != THERR_OK \|\|
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	484 (res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	485 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	486 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	487
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	488 case '^':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	489 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	490 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	491 node.mode = TH_RE_MATCH_ANCHOR_START;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	492
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	493 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	494 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	495 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	496
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	497 case '$':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	498 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	499 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	500 node.mode = TH_RE_MATCH_ANCHOR_END;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	501
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	502 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	503 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	504 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	505
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	506 case '[':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	507 // Start of char list
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	508 start = ctx.offs + 1;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	509 if (!th_regex_find_next(ctx.pattern, start, &ctx.offs, ']') \|\|
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	510 ctx.offs == start)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	511 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	512 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	513 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	514 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	515
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	516 // XXX TODO Parse/support ranges [0-9a-zA-Z_-]
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	517
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	518 th_regex_node_init(&node);
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	519 node.type = (ctx.pattern[start] == '^') ?
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	520 TH_RE_TYPE_LIST_REVERSE : TH_RE_TYPE_LIST;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	521 node.match.list.nchars = ctx.offs - start;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	522 if ((res = th_regex_strndup(&node.match.list.chars,
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	523 ctx.pattern + start, node.match.list.nchars)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	524 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	525
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	526 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	527 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	528 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	529
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	530 case '.':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	531 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	532 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	533 node.type = TH_RE_TYPE_ANY_CHAR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	534
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	535 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	536 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	537 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	538
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	539 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	540 // Literal escape
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	541 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	542 if (ctx.pattern[ctx.offs] == 0)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	543 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	544 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	545 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	546 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	547 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	548
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	549 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	550 node.type = TH_RE_TYPE_CHAR;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	551 node.match.chr = ctx.pattern[ctx.offs];
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	552
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	553 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	554 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	555 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	556
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	557 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	558 // Given character must match
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	559 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	560 node.type = TH_RE_TYPE_CHAR;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	561 node.match.chr = ctx.pattern[ctx.offs];
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	562
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	563 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	564 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	565 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	566 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	567 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	568
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	569 exit:
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	570 *pexpr = ctx.data;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	571 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	572 return res;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	573 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	574
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	575
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	576 void th_regex_free(th_regex_ctx *expr)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	577 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	578 if (expr != NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	579 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	580 for (size_t n = 0; n < expr->nnodes; n++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	581 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	582 th_regex_node *node = &expr->nodes[n];
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	583
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	584 DBG_RE_FREE(
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	585 "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	586 "mode=%d, type=%d, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	587 n, expr->nnodes,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	588 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	589 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	590
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	591 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	592 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	593 case TH_RE_TYPE_SUBEXPR:
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	594 DBG_RE_FREE(" SUBEXPR: %p vs %p\n",
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	595 (void ) expr, (void ) node->match.expr);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	596 th_regex_free(node->match.expr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	597 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	598
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	599 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	600 case TH_RE_TYPE_LIST_REVERSE:
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	601 DBG_RE_FREE(" list='%s'\n", node->match.list.chars);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	602 th_free(node->match.list.chars);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	603 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	604
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	605 #ifdef TH_EXPERIMENTAL_REGEX_DEBUG
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	606 case TH_RE_TYPE_ANY_CHAR:
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	607 DBG_RE_FREE(" any char\n");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	608 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	609
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	610 case TH_RE_TYPE_CHAR:
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	611 DBG_RE_FREE(" char='%c'\n", node->match.chr);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	612 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	613 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	614 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	615 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	616
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	617 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	618 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	619 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	620
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	621
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	622 static BOOL th_regex_do_match_list(const th_regex_list *list, const th_regex_char cch)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	623 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	624 // Could be optimized, perhaps .. sort match.chars, binary search etc?
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	625 for (size_t nitem = 0; nitem < list->nitems; nitem++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	626 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	627 const th_regex_list_item *item = &list->items[nitem];
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	628 if (item->type == 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	629 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	630 for (size_t n = 0; n < item->nchars; n++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	631 if (item->chars[n] == cch)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	632 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	633 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	634 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	635 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	636 if (cch >= item->start && cch <= item->end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	637 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	638 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	639 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	640
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	641 return FALSE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	642 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	643
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	644
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	645 static BOOL th_regex_do_match_expr(const th_regex_ctx *expr,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	646 const th_regex_char haystack, size_t offs, const int flags);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	647
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	648
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	649 static BOOL th_regex_do_match_node(const th_regex_char *haystack,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	650 size_t offs, const th_regex_node node, const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	651 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	652 th_regex_char cch;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	653 BOOL res = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	654
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	655 DBG_RE_MATCH(" node_START [%s]: '%s': ",
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	656 re_match_types[node->type], haystack + *offs);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	657
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	658 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	659 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	660 case TH_RE_TYPE_SUBEXPR:
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	661 DBG_RE_MATCH("subexpr ..\n");
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	662 res = th_regex_do_match_expr(node->match.expr, haystack, offs, flags);
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	663 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	664 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	665
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	666 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	667 case TH_RE_TYPE_LIST_REVERSE:
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	668 DBG_RE_MATCH("[%s]", node->match.list.chars);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	669 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	670 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	671
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	672 // Could be optimized, perhaps .. sort match.chars, binary search etc?
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	673 // XXX TODO Ranges are not supported yet
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	674 for (size_t n = 0; n < node->match.list.nchars; n++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	675 if (node->match.list.chars[n] == cch)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	676 {
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	677 res = TRUE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	678 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	679 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	680
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	681 if (node->type == TH_RE_TYPE_LIST_REVERSE)
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	682 res = !res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	683
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	684 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	685 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	686
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	687 case TH_RE_TYPE_ANY_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	688 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	689 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	690
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	691 res = TRUE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	692 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	693 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	694
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	695 case TH_RE_TYPE_CHAR:
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	696 DBG_RE_MATCH("'%c'", node->match.chr);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	697 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	698 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	699
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	700 res = (cch == node->match.chr);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	701 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	702 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	703 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	704
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	705 out:
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	706 DBG_RE_MATCH(", match=%s\n",
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	707 res ? "YES" : "NO");
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	708
c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	709 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	710 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	711
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	712
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	713 static BOOL th_regex_do_match_expr(const th_regex_ctx *expr,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	714 const th_regex_char haystack, size_t offs, const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	715 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	716 for (size_t n = 0; n < expr->nnodes; n++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	717 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	718 const th_regex_node *node = &expr->nodes[n];
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	719 size_t soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	720
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	721 DBG_RE_MATCH(" expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s ",
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	722 n, expr->nnodes, re_match_modes[node->mode]);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	723
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	724 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	725 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	726 case TH_RE_MATCH_ONCE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	727 {
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	728 DBG_RE_MATCH("\n");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	729 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	730 if (!th_regex_do_match_node(haystack, &soffs, node, flags))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	731 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	732 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	733 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	734 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	735
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	736 case TH_RE_MATCH_COUNT_GREEDY:
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	737 case TH_RE_MATCH_COUNT_NONGREEDY:
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	738 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	739 BOOL done = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	740 ssize_t count = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	741
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	742 DBG_RE_MATCH("min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n", node->repeatMin, node->repeatMax);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	743
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	744 do
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	745 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	746 BOOL match;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	747 soffs = *offs;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	748
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	749 match = th_regex_do_match_node(haystack, &soffs, node, flags);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	750 for (size_t qn = n + 1; qn < expr->nnodes && haystack[soffs] != 0; qn++)
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	751 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	752 const th_regex_node *next = &expr->nodes[qn];
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	753 do {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	754 match = th_regex_do_match_node(haystack, &soffs, next, flags);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	755 } while (haystack[soffs] != 0 && !match);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	756 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	757
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	758 if (match)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	759 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	760 // Node matched
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	761 count++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	762 done = (node->repeatMax > 0 && count >= node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	763 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	764 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	765 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	766 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	767 // Node did not match, check if we got the minimum if set
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	768 done = (node->repeatMin >= 0 && count >= node->repeatMin);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	769 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	770 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	771 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	772 } while (!done);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	773
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	774 DBG_RE_MATCH("RESULT: %" PRId_SSIZE_T " = %s\n", count, done ? "YES" : "NO");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	775
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	776 if (!done)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	777 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	778
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	779 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	780 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	781 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	782
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	783 case TH_RE_MATCH_ANCHOR_START:
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	784 DBG_RE_MATCH("offs=%" PRIu_SIZE_T "\n", *offs);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	785 if (*offs != 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	786 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	787 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	788
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	789 case TH_RE_MATCH_ANCHOR_END:
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	790 DBG_RE_MATCH("is end=%d\n", haystack[*offs]);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	791 if (haystack[*offs] != 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	792 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	793 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	794 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	795 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	796
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	797 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	798 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	799
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	800
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	801 int th_regex_match(const th_regex_ctx expr, const th_regex_char haystack,
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	802 size_t pnmatches, th_regex_match_node *pmatches, const size_t maxmatches,
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	803 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	804 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	805 size_t nmatches = 0;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	806
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	807 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	808 *pnmatches = 0;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	809
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	810 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	811 if (expr == NULL \|\| haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	812 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	813
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	814 // Start matching
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	815 // XXX NOTE .. lots to think about and to take into account:
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	816 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	817 // - how to check if the expression has consumed all possibilities?
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	818 // ..
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	819 for (size_t soffs = 0; haystack[soffs] != 0; )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	820 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	821 BOOL matched;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	822 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	823
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	824 DBG_RE_MATCH("\nTRY_MATCH @ startoffs=%" PRIu_SIZE_T ": '%s'\n",
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	825 soffs, haystack + soffs);
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	826
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	827 if ((matched = th_regex_do_match_expr(expr, haystack, &coffs, flags)))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	828 {
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	829 // A match was found, increase count
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	830 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	831
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	832 // Deliver to caller if required
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	833 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	834 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	835
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	836 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	837 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	838 th_regex_match_node *match = th_malloc0(sizeof(th_regex_match_node));
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	839 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	840 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	841
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	842 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	843 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	844
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	845 th_llist_append_node((th_llist_t *) pmatches, (th_llist_t ) match);
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	846 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	847
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	848 // Check match count limit, if set
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	849 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	850 break;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	851
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	852 // If offset was not advanced, increase by one
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	853 // otherwise use end of match offset as new start
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	854 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	855 soffs++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	856 else
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	857 soffs = coffs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	858 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	859 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	860 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	861 soffs++;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	862 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	863 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	864
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	865 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	866 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	867
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	868
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	869 static void th_regex_free_match(th_regex_match_node *node)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	870 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	871 (void) node;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	872 // Nothing to do here at the moment
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	873 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	874
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	875
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	876 void th_regex_free_matches(th_regex_match_node *matches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	877 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	878 th_llist_free_func_node((th_llist_t *) matches,
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	879 (void ()(th_llist_t )) th_regex_free_match);
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	880 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	881
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	882 #endif // TH_EXPERIMENTAL_REGEX

Mercurial > hg > th-libs

annotate th_regex.c @ 639:8c957ad9d4c3