th-libs: th_regex.c annotate

annotate th_regex.c @ 643:a2bf1ea05b05

Cleanups.

author	Matti Hamalainen <ccr@tnsp.org>
date	Fri, 24 Jan 2020 12:06:43 +0200
parents	3a35db5c1873
children	b897995101b7

rev	line source
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1 /*
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	2 * Simple regular expression matching functionality
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	3 * Programmed and designed by Matti 'ccr' Hamalainen
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	4 * (C) Copyright 2020 Tecnic Software productions (TNSP)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	5 *
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	6 * Please read file 'COPYING' for information on license and distribution.
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	7 */
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	8 #include "th_util.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	9 #include "th_regex.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	10
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	11
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	12 #ifdef TH_EXPERIMENTAL_REGEX_DEBUG
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	13 # define DBG_RE_COMPILE(...) do { if (th_dbg_re_flags & TH_DBG_RE_COMPILE) fprintf(stdout, __VA_ARGS__); } while (0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	14 # define DBG_RE_FREE(...) do { if (th_dbg_re_flags & TH_DBG_RE_FREE) fprintf(stdout, __VA_ARGS__); } while (0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	15 # define DBG_RE_MATCH(...) do { if (th_dbg_re_flags & TH_DBG_RE_MATCH) fprintf(stdout, __VA_ARGS__); } while (0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	16
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	17 int th_dbg_re_flags = 0;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	18
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	19 static const char *re_match_modes[] =
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	20 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	21 "ONCE",
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	22 "COUNT",
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	23 "ANCHOR START",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	24 "ANCHOR END",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	25 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	26
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	27 static const char *re_match_types[] =
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	28 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	29 "CHAR",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	30 "ANY",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	31 "LIST",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	32 "LIST REVERSE",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	33 "SUBEXPR",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	34 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	35
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	36 #else
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	37 # define DBG_RE_COMPILE(...)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	38 # define DBG_RE_FREE(...)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	39 # define DBG_RE_MATCH(...)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	40 #endif
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	41
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	42
614 afcaf5e38f56 Disable regex stuff from normal builds. Matti Hamalainen <ccr@tnsp.org> parents: 613 diff changeset	43 #ifdef TH_EXPERIMENTAL_REGEX
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	44
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	45 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	46 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	47 int type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	48 th_regex_char_t start, end;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	49
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	50 size_t nchars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	51 th_regex_char_t *chars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	52 } th_regex_list_item_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	53
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	54
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	55 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	56 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	57 size_t nitems, itemssize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	58 th_regex_list_item_t *items;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	59 } th_regex_list_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	60
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	61
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	62 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	63 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	64 int mode, type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	65 ssize_t repeatMin, repeatMax;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	66
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	67 struct {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	68 th_regex_char_t chr;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	69 th_regex_list_t list;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	70 struct th_regex_t *expr;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	71 } match;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	72 } th_regex_node_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	73
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	74
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	75 struct th_regex_t
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	76 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	77 size_t nnodes, nodessize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	78 th_regex_node_t *nodes;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	79 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	80
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	81
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	82 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	83 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	84 TH_RE_MATCH_ONCE,
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	85 TH_RE_MATCH_COUNT,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	86
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	87 TH_RE_MATCH_ANCHOR_START,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	88 TH_RE_MATCH_ANCHOR_END,
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	91
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	92 enum
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 TH_RE_TYPE_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	95 TH_RE_TYPE_ANY_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	96 TH_RE_TYPE_LIST,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	97 TH_RE_TYPE_LIST_REVERSE,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	98 TH_RE_TYPE_SUBEXPR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	99 };
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	102 typedef struct
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	103 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	104 const th_regex_char_t *pattern;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	105 size_t offs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	106
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	107 th_regex_t *data;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	108
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	109 size_t nstack, stacksize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	110 th_regex_t **stack;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	111 } th_regex_parse_ctx_t;
605 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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	114 static void th_regex_node_init(th_regex_node_t *node)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	115 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	116 memset(node, 0, sizeof(th_regex_node_t));
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	117 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	118 }
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	121 static int th_regex_strndup(th_regex_char_t **pdst,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	122 const th_regex_char_t *src, const size_t len)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	123 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	124 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	125 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	126
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	127 if (UINTPTR_MAX / sizeof(th_regex_char_t) < len + 1)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	128 return THERR_BOUNDS;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	129
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	130 if ((pdst = (th_regex_char_t )
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	131 th_malloc((len + 1) * sizeof(th_regex_char_t))) == NULL)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	132 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	133
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	134 memcpy(pdst, src, len sizeof(th_regex_char_t));
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	135 (*pdst)[len] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	136
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	137 return THERR_OK;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	140
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	141 static int th_regex_parse_ctx_get_prev_node(
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	142 th_regex_parse_ctx_t ctx, th_regex_node_t *pnode)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	143 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	144 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	145 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	146 *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	147 return THERR_OK;
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	150 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	151 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	153
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	154 static int th_regex_parse_ctx_push(th_regex_parse_ctx_t *ctx)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	155 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	156 if (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	157 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	158 ctx->stacksize += 16;
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 if ((ctx->stack = th_realloc(ctx->stack,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	161 ctx->stacksize * sizeof(th_regex_node_t *))) == NULL)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	162 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	163 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	164
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	165 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	166 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	167 ctx->data = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	168
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	169 return THERR_OK;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	172
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	173 static int th_regex_parse_ctx_pop(th_regex_parse_ctx_t ctx, th_regex_t *data)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	174 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	175 if (ctx->nstack > 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	176 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	177 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	178 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	179 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	180 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	181 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	182 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	183 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	184 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	185
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	186
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	187 static int th_regex_parse_ctx_node_commit(th_regex_parse_ctx_t ctx, th_regex_node_t node)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	188 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	189 th_regex_t *data = ctx->data;
605 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 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	192 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	193 if ((data = ctx->data = th_malloc0(sizeof(th_regex_t))) == NULL)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	194 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	195 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	196
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	197 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	198 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	199 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	200 if ((data->nodes = th_realloc(data->nodes,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	201 data->nodessize * sizeof(th_regex_node_t))) == NULL)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	202 return THERR_MALLOC;
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	205 memcpy(&data->nodes[data->nnodes], node, sizeof(th_regex_node_t));
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	206 data->nnodes++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	207
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	208 DBG_RE_COMPILE(
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	209 "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	210 "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	211 data->nnodes, data->nodessize,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	212 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	213 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	214
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	215 return THERR_OK;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	218
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	219 static BOOL th_regex_find_next(const th_regex_char_t *str,
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	220 const size_t start, size_t *offs,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	221 const th_regex_char_t delim)
605 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 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	224 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	225 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	226 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	227 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	228 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	229 }
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	232 static BOOL th_regex_parse_ssize_t(const th_regex_char_t *str,
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	233 ssize_t *value)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	234 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	235 th_regex_char_t ch;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	236 BOOL neg;
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 if (*str == '-')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	239 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	240 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	241 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	242 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	243 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	244 neg = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	245
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	246 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	247 while ((ch = *str++))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	248 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	249 if (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	250 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	251 value = 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	252 *value += ch - '0';
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	253 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	254 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	255 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	256 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	257
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	258 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	259 value = -(value);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	260
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	261 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	262 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	263
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	264
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	265 static void th_regex_list_item_init(th_regex_list_item_t *item)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	266 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	267 memset(item, 0, sizeof(th_regex_list_item_t));
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	268 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	269
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	270
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	271 static int th_regex_list_add_item(th_regex_list_t list, th_regex_list_item_t item)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	272 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	273 if (list->items == NULL \|\| list->nitems + 1 >= list->itemssize)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	274 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	275 list->itemssize += 16;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	276
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	277 if ((list->items = th_realloc(list->items,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	278 list->itemssize * sizeof(th_regex_list_item_t))) == NULL)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	279 return THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	280 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	281
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	282 memcpy(&list->items[list->nitems], item, sizeof(th_regex_list_item_t));
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	283 list->nitems++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	284
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	285 return THERR_OK;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	286 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	287
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	288
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	289 static void th_regex_list_free(th_regex_list_t *list)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	290 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	291 if (list != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	292 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	293 for (size_t n = 0; n < list->nitems; n++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	294 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	295 th_free(list->items[n].chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	296 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	297 th_free(list->items);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	298 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	299 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	300
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	301
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	302 static int th_regex_parse_list(const th_regex_char_t *str,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	303 const size_t slen, th_regex_list_t *list)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	304 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	305 th_regex_char_t *tmp = NULL;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	306 th_regex_list_item_t item;
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	307 int res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	308
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	309 if ((res = th_regex_strndup(&tmp, str, slen)) != THERR_OK)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	310 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	311
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	312 // Handle ranges like [A-Z]
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	313 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	314 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	315 th_regex_char_t
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	316 *prev = (offs > 0) ? tmp + offs - 1 : NULL,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	317 *curr = tmp + offs,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	318 *next = (offs + 1 < slen) ? tmp + offs + 1 : NULL;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	319
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	320 if (*curr == '-')
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	321 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	322 if (prev != NULL && next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	323 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	324 // Range
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	325 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	326 item.type = 1;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	327 item.start = *prev;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	328 item.end = *next;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	329
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	330 if (item.start <= item.end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	331 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	332 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	333 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	334 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	335
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	336 curr = prev = *next = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	337
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	338 if ((res = th_regex_list_add_item(list, &item)) != THERR_OK)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	339 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	340 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	341 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	342 if (next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	343 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	344 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	345 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	346 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	347 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	348 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	349
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	350 // Count number of remaining characters
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	351 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	352 item.type = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	353 item.nchars = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	354
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	355 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	356 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	357 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	358 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	359 item.nchars++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	360 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	361
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	362 if (item.nchars > 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	363 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	364 if ((item.chars = th_malloc(sizeof(th_regex_char_t) * item.nchars)) == NULL)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	365 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	366 res = THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	367 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	368 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	369
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	370 for (size_t offs = 0, n = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	371 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	372 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	373 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	374 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	375 item.chars[n] = curr;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	376 n++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	377 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	378 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	379
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	380 if ((res = th_regex_list_add_item(list, &item)) != THERR_OK)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	381 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	382 th_free(item.chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	383 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	384 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	385 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	386
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	387 out:
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	388 th_free(tmp);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	389 return res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	390 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	391
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	392
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	393 int th_regex_compile(th_regex_t *pexpr, const th_regex_char_t pattern)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	394 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	395 int res = THERR_OK;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	396 th_regex_parse_ctx_t ctx;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	397 th_regex_node_t node, *pnode;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	398 th_regex_char_t *tmp = NULL;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	399 size_t start;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	400
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	401 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	402 {
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	403 res = THERR_NULLPTR;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	404 goto exit;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	405 }
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	406
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	407 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	408 ctx.pattern = pattern;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	409
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	410 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	411 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	412 th_regex_char_t cch = ctx.pattern[ctx.offs];
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	413 DBG_RE_COMPILE("[%" PRIu_SIZE_T "] '%c'\n", ctx.offs, cch);
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	414 switch (cch)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	415 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	416 case '?':
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	417 case '*':
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	418 case '+':
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	419 if ((res = th_regex_parse_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	420 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	421
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	422 if (cch == '?')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	423 {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	424 // Previous token is optional (repeat 0-1 times) (non-greedy matching)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	425 pnode->mode = TH_RE_MATCH_COUNT;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	426 pnode->repeatMin = 0;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	427 pnode->repeatMax = 1;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	428 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	429 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	430 {
641 9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	431 // Check if previous was a count ("*", "+", etc.)
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	432 if (pnode->mode == TH_RE_MATCH_COUNT)
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	433 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	434 res = THERR_INVALID_DATA;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	435 goto exit;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	436 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	437
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	438 pnode->mode = TH_RE_MATCH_COUNT;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	439
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	440 if (cch == '*')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	441 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	442 // Previous token can repeat 0 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	443 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	444 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	445 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	446 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	447 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	448 // Previous token must repeat 1 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	449 pnode->repeatMin = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	450 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	451 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	452 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	453 break;
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 case '{':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	456 // {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	457 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	458 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	459 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	460 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	461 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	462 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	463 }
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 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	466
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	467 if ((res = th_regex_parse_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK \|\|
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	468 (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	469 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	470 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	471
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	472 pnode->mode = TH_RE_MATCH_COUNT;
605 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 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	475 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	476 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	477 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	478 !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	479 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	480 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	481 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	482 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	483 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	484 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	485 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	486 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	487 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	488 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	489 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	490 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	491 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	492 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	493
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	494 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	495 pnode->repeatMax < pnode->repeatMin)
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 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	498 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	499 goto exit;
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 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	502
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	503 case '(':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	504 // Start of subpattern
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	505 if ((res = th_regex_parse_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	506 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	507 break;
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 case ')':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	510 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	511 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	512 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	513
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	514 if ((res = th_regex_parse_ctx_pop(&ctx, &node.match.expr)) != THERR_OK \|\|
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	515 (res = th_regex_parse_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	516 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	517 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	518
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	519 case '^':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	520 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	521 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	522 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	523
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	524 if ((res = th_regex_parse_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	525 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	526 break;
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 case '$':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	529 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	530 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	531 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	532
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	533 if ((res = th_regex_parse_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	534 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	535 break;
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 case '[':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	538 // 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	539 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	540 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	541 ctx.offs == start)
605 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 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	544 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	545 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	546
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	547 th_regex_node_init(&node);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	548 if (ctx.pattern[start] == '^')
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	549 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	550 node.type = TH_RE_TYPE_LIST_REVERSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	551 start++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	552 }
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	553 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	554 node.type = TH_RE_TYPE_LIST;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	555
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	556 if ((res = th_regex_parse_list(ctx.pattern + start,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	557 ctx.offs - start, &node.match.list) != THERR_OK) \|\|
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	558 (res = th_regex_parse_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	559 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	560 break;
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 case '.':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	563 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	564 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	565 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	566
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	567 if ((res = th_regex_parse_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	568 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	569 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	570
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	571 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	572 // Literal escape
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	573 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	574 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	575 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	576 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	577 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	578 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	579 }
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	580 // fall-through
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	581
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	582 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	583 // Given character must match
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	584 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	585 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	586 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	587
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	588 if ((res = th_regex_parse_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	589 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	590 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	591 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	592 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	593
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	594 exit:
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	595 *pexpr = ctx.data;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	596 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	597 return res;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	598 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	599
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	600
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	601 void th_regex_free(th_regex_t *expr)
605 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 if (expr != NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	604 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	605 for (size_t nnode = 0; nnode < expr->nnodes; nnode++)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	606 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	607 th_regex_node_t *node = &expr->nodes[nnode];
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	608
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	609 DBG_RE_FREE(
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	610 "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	611 "mode=%s, type=%s, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	612 nnode, expr->nnodes,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	613 re_match_modes[node->mode],
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	614 re_match_types[node->type],
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	615 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	616
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	617 th_regex_free(node->match.expr);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	618 th_regex_list_free(&node->match.list);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	619 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	620
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	621 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	622 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	623 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	624
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	625
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	626 static BOOL th_regex_match_list(const th_regex_list_t *list, const th_regex_char_t cch)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	627 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	628 // Could be optimized, perhaps .. sort match.chars, binary search etc?
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	629 for (size_t nitem = 0; nitem < list->nitems; nitem++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	630 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	631 const th_regex_list_item_t *item = &list->items[nitem];
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	632
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	633 if (item->type == 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	634 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	635 for (size_t n = 0; n < item->nchars; n++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	636 if (item->chars[n] == cch)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	637 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	638 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	639 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	640 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	641 if (cch >= item->start && cch <= item->end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	642 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	643 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	644 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	645
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	646 return FALSE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	647 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	648
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	649
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	650 static BOOL th_regex_match_expr(
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	651 const th_regex_char_t *haystack,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	652 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	653 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	654 const size_t startnode,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	655 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	656 );
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	657
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	658
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	659 static BOOL th_regex_match_one(
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	660 const th_regex_char_t *haystack,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	661 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	662 const th_regex_node_t *node,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	663 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	664 )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	665 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	666 th_regex_char_t cch;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	667 BOOL res = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	668
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	669 switch (node->type)
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 case TH_RE_TYPE_SUBEXPR:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	672 // push current context in stack
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	673 // new context for subexpr
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	674 res = th_regex_match_expr(haystack, poffs, node->match.expr, 0, flags);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	675 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	676
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	677 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	678 case TH_RE_TYPE_LIST_REVERSE:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	679 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	680 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	681 else
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	682 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	683 res = th_regex_match_list(&node->match.list, cch);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	684
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	685 if (node->type == TH_RE_TYPE_LIST_REVERSE)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	686 res = !res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	687
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	688 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	689 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	690 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	691
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	692 case TH_RE_TYPE_ANY_CHAR:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	693 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	694 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	695 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	696 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	697 res = TRUE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	698 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	699 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	700 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	701
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	702 case TH_RE_TYPE_CHAR:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	703 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	704 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	705 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	706 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	707 res = (cch == node->match.chr);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	708 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	709 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	710 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	711 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	712
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	713 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	714 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	715
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	716
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	717 static BOOL th_regex_match_expr(
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	718 const th_regex_char_t *haystack,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	719 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	720 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	721 const size_t startnode,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	722 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	723 )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	724 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	725 BOOL res = TRUE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	726 size_t soffs = *poffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	727
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	728 for (size_t nnode = startnode; res && nnode < expr->nnodes; nnode++)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	729 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	730 const th_regex_node_t *node = &expr->nodes[nnode];
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	731
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	732 DBG_RE_MATCH(" expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s %s '%s'\n",
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	733 nnode + 1, expr->nnodes,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	734 re_match_modes[node->mode], re_match_types[node->type],
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	735 haystack + soffs);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	736
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	737 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	738 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	739 case TH_RE_MATCH_ONCE:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	740 res = th_regex_match_one(haystack, &soffs, node, flags);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	741 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	742
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	743 case TH_RE_MATCH_COUNT:
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	744 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	745 ssize_t count = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	746
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	747 DBG_RE_MATCH(" COUNT min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	748 node->repeatMin, node->repeatMax);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	749
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	750 do
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	751 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	752 BOOL match;
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	753 size_t toffs = soffs, tnode;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	754
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	755 DBG_RE_MATCH(" START '%s'\n", haystack + toffs);
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	756 do {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	757 match = TRUE;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	758 for (tnode = nnode; match && tnode < expr->nnodes && haystack[toffs] != 0; )
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	759 {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	760 match = th_regex_match_one(haystack, &toffs, &expr->nodes[tnode], flags);
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	761 if (match)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	762 tnode++;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	763 }
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	764 DBG_RE_MATCH(" '%s': %d (tnode=%" PRIu_SIZE_T ")\n", haystack + toffs, match, tnode);
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	765 if (node->repeatMin >= 0 && match)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	766 break;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	767 } while (!match && haystack[toffs] != 0);
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	768
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	769 DBG_RE_MATCH(" END '%s': %d (tnode=%" PRIu_SIZE_T ")\n", haystack + toffs, match, tnode);
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	770
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	771 if (match)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	772 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	773 // Node matched
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	774 count++;
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	775 soffs = toffs;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	776 nnode = tnode;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	777 res = (node->repeatMax > 0 && count >= node->repeatMax);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	778 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	779 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	780 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	781 // Node did not match, check if we got the minimum if set
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	782 res = (node->repeatMin >= 0 && count >= node->repeatMin);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	783 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	784
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	785 } while (!res);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	786
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	787 DBG_RE_MATCH(" RESULT: count=%" PRId_SSIZE_T ", done=%s\n",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	788 count, res ? "YES" : "NO");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	789 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	790 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	791
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	792 case TH_RE_MATCH_ANCHOR_START:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	793 res = (soffs == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	794 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	795
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	796 case TH_RE_MATCH_ANCHOR_END:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	797 res = (haystack[soffs] == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	798 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	799 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	800 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	801
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	802 if (res)
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	803 *poffs = soffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	804
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	805 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	806 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	807
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	808
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	809 int th_regex_match(const th_regex_t expr, const th_regex_char_t haystack,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	810 size_t pnmatches, th_regex_match_t *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	811 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	812 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	813 size_t nmatches = 0;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	814 (void) flags;
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	815
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	816 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	817 *pnmatches = 0;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	818
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	819 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	820 if (expr == NULL \|\| haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	821 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	822
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	823 // Start matching
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	824 // 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	825 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	826 // - 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	827 // ..
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	828 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	829 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	830 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	831
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	832 DBG_RE_MATCH("\nTRY_MATCH @ startoffs=%" PRIu_SIZE_T ": '%s'\n",
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	833 soffs, haystack + soffs);
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	834
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	835 if (th_regex_match_expr(haystack, &coffs, expr, 0, flags))
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	836 {
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	837 // 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	838 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	839
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	840 // Deliver to caller if required
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	841 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	842 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	843
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	844 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	845 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	846 th_regex_match_t *match = th_malloc0(sizeof(th_regex_match_t));
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	847 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	848 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	849
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	850 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	851 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	852
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	853 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	854 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	855
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	856 // 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	857 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	858 break;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	859
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	860 // If offset was not advanced, increase by one
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	861 // 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	862 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	863 soffs++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	864 else
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	865 soffs = coffs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	866 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	867 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	868 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	869 soffs++;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	870 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	871 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	872
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	873 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	874 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	875
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	876
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	877 static void th_regex_free_match(th_regex_match_t *node)
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	878 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	879 (void) node;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	880 // 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	881 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	882
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	883
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	884 void th_regex_free_matches(th_regex_match_t *matches)
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	885 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	886 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	887 (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	888 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	889
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	890 #endif // TH_EXPERIMENTAL_REGEX

Mercurial > hg > th-libs

annotate th_regex.c @ 643:a2bf1ea05b05