th-libs: th_regex.c annotate

annotate th_regex.c @ 613:2e3b81ae8c8a

More work on regexes.

author	Matti Hamalainen <ccr@tnsp.org>
date	Thu, 16 Jan 2020 16:01:27 +0200
parents	cc9ec51b4875
children	afcaf5e38f56

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 #ifdef TH_EXPERIMENTAL_REGEX
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	9 #include "th_util.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	10 #include "th_regex.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	11 //#include "th_datastruct.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	12
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	13
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	14 //#define DBG_RE_COMPILE 1
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	15 //#define DBG_RE_FREE 1
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	16 #define DBG_RE_MATCH 1
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	17
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	18 #if defined(DBG_RE_COMPILE)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	19 # define DBG_RE_PRINT_COMPILE(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	20 #else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	21 # define DBG_RE_PRINT_COMPILE(...) do { } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	22 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	23 #if defined(DBG_RE_FREE)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	24 # define DBG_RE_PRINT_FREE(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	25 #else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	26 # define DBG_RE_PRINT_FREE(...) do { } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	27 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	28 #if defined(DBG_RE_MATCH)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	29 # define DBG_RE_PRINT_MATCH(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	30 #else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	31 # define DBG_RE_PRINT_MATCH(...) do { } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	32 #endif
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 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	36 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	37 TH_RE_MATCH_ONCE,
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	38 TH_RE_MATCH_COUNT_GREEDY,
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	39 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	40 TH_RE_MATCH_ANCHOR_START,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	41 TH_RE_MATCH_ANCHOR_END,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	42 };
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 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	46 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	47 TH_RE_TYPE_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	48 TH_RE_TYPE_ANY_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	49 TH_RE_TYPE_LIST,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	50 TH_RE_TYPE_LIST_REVERSE,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	51 TH_RE_TYPE_SUBEXPR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	52 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	53
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 #if defined(DBG_RE_COMPILE) \|\| defined(DBG_RE_FREE) \|\| defined(DBG_RE_MATCH)
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_modes[] =
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 "ONCE",
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	59 "COUNT GREEDY",
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	60 "COUNT NONGREEDY",
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	61 "ANCHOR_START",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	62 "ANCHOR_END",
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	65 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	66 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	67 "CHAR",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	68 "ANY",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	69 "LIST",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	70 "LIST_REVERSE",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	71 "SUBEXPR",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	72 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	73 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	74
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	75
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	76 typedef struct
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 const th_regex_char *pattern;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	79 size_t offs;
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 th_regex_ctx *data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	82
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	83 size_t nstack, stacksize;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	84 th_regex_ctx **stack;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	85 } th_regex_parse_ctx;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	86
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	87
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	88 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	89 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	90 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	91 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	92 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	93
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 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	96 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	97 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	98 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	99 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	100
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	101 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	102 return THERR_BOUNDS;
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 if ((pdst = (th_regex_char )
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	105 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	106 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	107
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	108 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	109 (*pdst)[len] = 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 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	112 }
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	115 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	116 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	117 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	118 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	119 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	120 *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	121 return THERR_OK;
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	124 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	125 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	126
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	127
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	128 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	129 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	130 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	131 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	132 ctx->stacksize += 16;
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 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	135 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	136 return THERR_MALLOC;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	139 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	140 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	141 ctx->data = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	142
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	143 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	144 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	145
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 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	148 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	149 if (ctx->nstack > 0)
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 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	153 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	154 return THERR_OK;
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	157 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	158 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	159
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 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	162 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	163 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	164
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	165 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	166 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	167 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	168 return THERR_MALLOC;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	171 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	172 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	173 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	174 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	175 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	176 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	177 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	178
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	179 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	180 data->nnodes++;
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 DBG_RE_PRINT_COMPILE(
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	183 "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	184 "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	185 data->nnodes, data->nodessize,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	186 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	187 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	188
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	189 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	190 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	191
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 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	194 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	195 const th_regex_char delim)
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 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	198 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	199 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	200 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	201 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	202 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	203 }
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	206 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	207 ssize_t *value)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	208 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	209 th_regex_char ch;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	210 BOOL neg;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	211
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	212 if (*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 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	215 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	216 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	217 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	218 neg = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	219
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	220 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	221 while ((ch = *str++))
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 (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	224 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	225 value = 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	226 *value += ch - '0';
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	229 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	230 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	231
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	232 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	233 value = -(value);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	234
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	235 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	236 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	237
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	238
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	239 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	240 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	241 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	242 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	243 th_regex_node node, *pnode;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	244 th_regex_char *tmp = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	245 size_t start;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	246
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	247 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	248 {
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	249 res = THERR_NULLPTR;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	250 goto exit;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	251 }
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	252
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	253 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	254 ctx.pattern = pattern;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	255
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	256 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	257 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	258 th_regex_char cch = ctx.pattern[ctx.offs];
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	259 DBG_RE_PRINT_COMPILE("[%" PRIu_SIZE_T "] '%c'\n", ctx.offs, cch);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	260 switch (cch)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	261 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	262 case '?':
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	263 case '*':
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	264 case '+':
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	265 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	266 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	267
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	268 if (cch == '?')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	269 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	270 // Check if previous was a count
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	271 pnode->mode = (pnode->mode == TH_RE_MATCH_COUNT_GREEDY) ?
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	272 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	273
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	274 // Previous token is optional (repeat 0-1 times)
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	275 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	276 pnode->repeatMax = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	277 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	278 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	279 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	280 // Check if previous was a count
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	281 if (pnode->mode == TH_RE_MATCH_COUNT_GREEDY \|\|
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	282 pnode->mode == TH_RE_MATCH_COUNT_NONGREEDY)
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	283 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	284 res = THERR_INVALID_DATA;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	285 goto exit;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	286 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	287
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	288 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	289
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	290 if (cch == '*')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	291 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	292 // Previous token can repeat 0 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	293 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	294 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	295 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	296 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	297 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	298 // Previous token must repeat 1 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	299 pnode->repeatMin = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	300 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	301 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	302 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	303 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	304
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	305 case '{':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	306 // {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	307 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	308 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	309 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	310 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	311 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	312 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	313 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	314
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	315 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	316
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	317 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	318 (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	319 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	320 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	321
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	322 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	323
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	324 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	325 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	326 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	327 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	328 !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	329 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	330 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	331 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	332 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	333 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	334 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	335 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	336 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	337 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	338 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	339 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	340 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	341 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	342 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	343
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	344 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	345 pnode->repeatMax < pnode->repeatMin)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	346 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	347 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	348 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	349 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	350 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	351 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	352
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	353 case '(':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	354 // 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	355 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	356 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	357 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	358
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	359 case ')':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	360 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	361 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	362 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	363
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	364 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	365 (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	366 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	367 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	368
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	369 case '^':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	370 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	371 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	372 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	373
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	374 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	375 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	376 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	377
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	378 case '$':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	379 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	380 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	381 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	382
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	383 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	384 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	385 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	386
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	387 case '[':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	388 // 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	389 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	390 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	391 ctx.offs == start)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	392 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	393 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	394 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	395 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	396
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	397 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	398 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	399 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	400 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	401 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	402 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	403 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	404
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	405 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	406 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	407 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	408
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	409 case '.':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	410 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	411 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	412 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	413
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	414 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	415 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	416 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	417
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	418 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	419 // Literal escape
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	420 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	421 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	422 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	423 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	424 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	425 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	426 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	427
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	428 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	429 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	430 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	431
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	432 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	433 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	434 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	435
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	436 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	437 // Given character must match
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	438 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	439 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	440 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	441
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	442 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	443 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	444 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	445 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	446 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	447
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	448 exit:
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	449 *pexpr = ctx.data;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	450 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	451 return res;
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
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
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 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	458 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	459 if (expr != NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	460 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	461 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	462 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	463 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	464
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	465 DBG_RE_PRINT_FREE(
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	466 "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	467 "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	468 n, expr->nnodes,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	469 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	470 node->repeatMin, node->repeatMax);
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 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	473 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	474 case TH_RE_TYPE_SUBEXPR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	475 DBG_RE_PRINT_FREE(" SUBEXPR: %p vs %p\n", expr, node->match.expr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	476 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	477 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	478
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	479 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	480 case TH_RE_TYPE_LIST_REVERSE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	481 DBG_RE_PRINT_FREE(" list='%s'\n", node->match.list.chars);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	482 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	483 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	484
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	485 #ifdef DBG_RE_FREE
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	486 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	487 DBG_RE_PRINT_FREE(" any char\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	488 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	489
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	490 case TH_RE_TYPE_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	491 DBG_RE_PRINT_FREE(" char='%c'\n", node->match.chr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	492 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	493 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	494 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	495 }
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 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	498 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	499 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	500
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	501
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	502 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	503 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	504
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 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	507 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	508 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	509 th_regex_char cch;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	510 BOOL ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	511
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	512 DBG_RE_PRINT_MATCH(" node_START [%s]: '%s': ", 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	513
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	514 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	515 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	516 case TH_RE_TYPE_SUBEXPR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	517 DBG_RE_PRINT_MATCH("subexpr ..\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	518 ret = th_regex_do_match_expr(node->match.expr, haystack, offs, flags);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	519 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	520
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	521 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	522 case TH_RE_TYPE_LIST_REVERSE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	523 DBG_RE_PRINT_MATCH("[%s]\n", node->match.list.chars);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	524 ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	525 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	526 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	527
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	528 // Could be optimized, perhaps .. sort match.chars, binary search etc?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	529 // XXX TODO Ranges and escapes are not supported yet
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	530 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	531 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	532 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	533 ret = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	534 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	535 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	536
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	537 if (node->type == TH_RE_TYPE_LIST_REVERSE)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	538 ret = !ret;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	539
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	540 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	541 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	542
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	543 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	544 DBG_RE_PRINT_MATCH("\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	545 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	546 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	547 ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	548 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	549 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	550
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	551 ret = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	552 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	553 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	554
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	555 case TH_RE_TYPE_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	556 DBG_RE_PRINT_MATCH("'%c'\n", node->match.chr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	557 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	558 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	559 ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	560 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	561 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	562
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	563 ret = (cch == node->match.chr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	564 (*offs)++;
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 out:
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	569 DBG_RE_PRINT_MATCH(" node_DONE [%s]: match %s\n", re_match_types[node->type], ret ? "YES" : "NO");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	570 return ret;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	571 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	572
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 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	575 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	576 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	577 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	578 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	579 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	580 size_t soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	581
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	582 DBG_RE_PRINT_MATCH(" expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s ",
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	583 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	584
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	585 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	586 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	587 case TH_RE_MATCH_ONCE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	588 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	589 DBG_RE_PRINT_MATCH("\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	590 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	591 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	592 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	593 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	594 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	595 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	596
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	597 case TH_RE_MATCH_COUNT_GREEDY:
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	598 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	599 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	600 BOOL done = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	601 ssize_t count = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	602
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	603 DBG_RE_PRINT_MATCH("min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n", node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	604
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	605 do
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	606 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	607 BOOL match;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	608 soffs = *offs;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	609
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	610 match = th_regex_do_match_node(haystack, &soffs, node, flags);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	611 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	612 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	613 const th_regex_node *next = &expr->nodes[qn];
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	614 do {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	615 match = th_regex_do_match_node(haystack, &soffs, next, flags);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	616 } while (haystack[soffs] != 0 && !match);
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	617 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	618
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	619 if (match)
605 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 // Node matched
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	622 count++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	623 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	624 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	625 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	626 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	627 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	628 // 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	629 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	630 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	631 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	632 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	633 } while (!done);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	634
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	635 DBG_RE_PRINT_MATCH("RESULT: %" PRId_SSIZE_T " = %s\n", count, done ? "YES" : "NO");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	636
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	637 if (!done)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	638 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	639
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	640 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	641 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	642 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	643
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	644 case TH_RE_MATCH_ANCHOR_START:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	645 DBG_RE_PRINT_MATCH("offs=%" PRIu_SIZE_T "\n", *offs);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	646 if (*offs != 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	647 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	648 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	649
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	650 case TH_RE_MATCH_ANCHOR_END:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	651 DBG_RE_PRINT_MATCH("is end=%d\n", haystack[*offs]);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	652 if (haystack[*offs] != 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	653 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	654 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	655 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	656 }
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 return TRUE;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	661
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	662 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	663 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	664 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	665 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	666 size_t nmatches = 0;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	667
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	668 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	669 *pnmatches = 0;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	670
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	671 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	672 if (expr == NULL \|\| haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	673 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	674
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	675 // Start matching
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	676 // 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	677 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	678 // - 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	679 // ..
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	680 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	681 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	682 BOOL matched;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	683 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	684
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	685 DBG_RE_PRINT_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	686 soffs, haystack + soffs);
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	687
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	688 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	689 {
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	690 // 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	691 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	692
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	693 // Deliver to caller if required
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	694 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	695 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	696
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	697 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	698 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	699 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	700 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	701 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	702
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	703 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	704 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	705
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	706 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	707 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	708
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	709 // 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	710 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	711 break;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	712
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	713 // If offset was not advanced, increase by one
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	714 // 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	715 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	716 soffs++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	717 else
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	718 soffs = coffs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	719 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	720 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	721 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	722 soffs++;
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 }
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 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	727 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	728
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	729
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	730 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	731 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	732 (void) node;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	733 // 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	734 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	735
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	736
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	737 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	738 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	739 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	740 (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	741 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	742
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	743 #endif // TH_EXPERIMENTAL_REGEX

Mercurial > hg > th-libs

annotate th_regex.c @ 613:2e3b81ae8c8a