th-libs: th_regex.c annotate

annotate th_regex.c @ 642:3a35db5c1873

Cleanup.

author	Matti Hamalainen <ccr@tnsp.org>
date	Fri, 24 Jan 2020 09:39:09 +0200
parents	9a1ed82abefd
children	a2bf1ea05b05

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",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	22 "COUNT GREEDY",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	23 "COUNT NONGREEDY",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	24 "ANCHOR START",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	25 "ANCHOR END",
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
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	28 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	29 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	30 "CHAR",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	31 "ANY",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	32 "LIST",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	33 "LIST REVERSE",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	34 "SUBEXPR",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	35 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	36
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	37 #else
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	38 # define DBG_RE_COMPILE(...)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	39 # define DBG_RE_FREE(...)
d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	40 # define DBG_RE_MATCH(...)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	41 #endif
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	42
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	43
614 afcaf5e38f56 Disable regex stuff from normal builds. Matti Hamalainen <ccr@tnsp.org> parents: 613 diff changeset	44 #ifdef TH_EXPERIMENTAL_REGEX
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	45
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	46 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	47 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	48 int type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	49 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	50
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	51 size_t nchars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	52 th_regex_char_t *chars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	53 } th_regex_list_item_t;
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
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	56 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	57 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	58 size_t nitems, itemssize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	59 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	60 } th_regex_list_t;
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
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	63 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	64 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	65 int mode, type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	66 ssize_t repeatMin, repeatMax;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	67
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	68 struct {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	69 th_regex_char_t chr;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	70 th_regex_list_t list;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	71 struct th_regex_t *expr;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	72 } match;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	73 } th_regex_node_t;
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
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	76 struct th_regex_t
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	77 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	78 size_t nnodes, nodessize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	79 th_regex_node_t *nodes;
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
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	82
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	83 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	84 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	85 TH_RE_MATCH_ONCE,
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	86 TH_RE_MATCH_COUNT_GREEDY,
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	87 TH_RE_MATCH_COUNT_NONGREEDY,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	88
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	89 TH_RE_MATCH_ANCHOR_START,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	90 TH_RE_MATCH_ANCHOR_END,
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
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 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	95 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	96 TH_RE_TYPE_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	97 TH_RE_TYPE_ANY_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	98 TH_RE_TYPE_LIST,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	99 TH_RE_TYPE_LIST_REVERSE,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	100 TH_RE_TYPE_SUBEXPR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	101 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	102
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 typedef struct
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	105 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	106 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	107 size_t offs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	108
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	109 th_regex_t *data;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	110
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	111 size_t nstack, stacksize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	112 th_regex_t **stack;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	113 } 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	114
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 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	117 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	118 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	119 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	120 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	121
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	122
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	123 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	124 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	125 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	126 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	127 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	128
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	129 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	130 return THERR_BOUNDS;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	131
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	132 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	133 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	134 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	135
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	136 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	137 (*pdst)[len] = 0;
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 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	140 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	141
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	142
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	143 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	144 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	145 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	146 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	147 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	148 *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	149 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	150 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	151 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	153 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	154
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	155
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	156 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	157 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	158 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	159 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	160 ctx->stacksize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	161
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	162 if ((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	163 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	164 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	165 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	166
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	167 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	168 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	169 ctx->data = NULL;
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 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	172 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	173
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	174
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	175 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	176 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	177 if (ctx->nstack > 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	178 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	179 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	180 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	181 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	182 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	183 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	184 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	185 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	186 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	187
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	188
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	189 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	190 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	191 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	192
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	193 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	194 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	195 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	196 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	197 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	198
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	199 if (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	200 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	201 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	202 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	203 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	204 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	205 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	206
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	207 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	208 data->nnodes++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	209
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	210 DBG_RE_COMPILE(
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	211 "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	212 "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	213 data->nnodes, data->nodessize,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	214 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	215 node->repeatMin, node->repeatMax);
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 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	218 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	219
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	220
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	221 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	222 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	223 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	224 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	225 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	226 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	227 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	228 return TRUE;
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 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	231 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	232
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	233
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	234 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	235 ssize_t *value)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	236 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	237 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	238 BOOL neg;
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 if (*str == '-')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	241 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	242 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	243 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	244 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	245 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	246 neg = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	247
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	248 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	249 while ((ch = *str++))
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 if (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	252 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	253 value = 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	254 *value += ch - '0';
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	255 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	256 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	257 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	258 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	259
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	260 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	261 value = -(value);
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 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	264 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	265
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	266
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	267 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	268 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	269 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	270 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	271
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	272
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	273 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	274 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	275 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	276 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	277 list->itemssize += 16;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	278
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	279 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	280 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	281 return THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	282 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	283
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	284 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	285 list->nitems++;
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 return THERR_OK;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	288 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	289
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	290
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	291 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	292 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	293 if (list != NULL)
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 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	296 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	297 th_free(list->items[n].chars);
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 th_free(list->items);
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 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	302
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	303
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	304 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	305 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	306 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	307 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	308 th_regex_list_item_t item;
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	309 int res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	310
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	311 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	312 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	313
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	314 // Handle ranges like [A-Z]
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	315 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	316 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	317 th_regex_char_t
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	318 *prev = (offs > 0) ? tmp + offs - 1 : NULL,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	319 *curr = tmp + offs,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	320 *next = (offs + 1 < slen) ? tmp + offs + 1 : NULL;
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 (*curr == '-')
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 if (prev != NULL && next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	325 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	326 // Range
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	327 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	328 item.type = 1;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	329 item.start = *prev;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	330 item.end = *next;
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 if (item.start <= item.end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	333 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	334 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	335 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	336 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	337
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	338 curr = prev = *next = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	339
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	340 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	341 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	342 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	343 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	344 if (next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	345 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	346 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	347 goto out;
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 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	351
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	352 // Count number of remaining characters
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	353 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	354 item.type = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	355 item.nchars = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	356
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	357 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	358 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	359 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	360 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	361 item.nchars++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	362 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	363
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	364 if (item.nchars > 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	365 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	366 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	367 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	368 res = THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	369 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	370 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	371
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	372 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	373 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	374 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	375 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	376 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	377 item.chars[n] = curr;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	378 n++;
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 }
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 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	383 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	384 th_free(item.chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	385 goto out;
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 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	388
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	389 out:
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	390 th_free(tmp);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	391 return res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	392 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	393
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	394
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	395 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	396 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	397 int res = THERR_OK;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	398 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	399 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	400 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	401 size_t start;
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 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	404 {
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	405 res = THERR_NULLPTR;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	406 goto exit;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	407 }
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	408
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	409 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	410 ctx.pattern = pattern;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	411
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	412 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	413 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	414 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	415 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	416 switch (cch)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	417 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	418 case '?':
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	419 case '*':
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	420 case '+':
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	421 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	422 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	423
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	424 if (cch == '?')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	425 {
642 3a35db5c1873 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 641 diff changeset	426 // Check if previous was a count
641 9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	427 if (pnode->mode == TH_RE_MATCH_COUNT_NONGREEDY)
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	428 {
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	429 res = THERR_INVALID_DATA;
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	430 goto exit;
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	431 }
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	432 else
642 3a35db5c1873 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 641 diff changeset	433 if (pnode->mode != TH_RE_MATCH_COUNT_GREEDY)
641 9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	434 {
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	435 // Previous token is optional (repeat 0-1 times)
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	436 pnode->repeatMin = 0;
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	437 pnode->repeatMax = 1;
9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	438 }
642 3a35db5c1873 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 641 diff changeset	439
3a35db5c1873 Cleanup. Matti Hamalainen <ccr@tnsp.org> parents: 641 diff changeset	440 pnode->mode = TH_RE_MATCH_COUNT_NONGREEDY;
613 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 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	443 {
641 9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	444 // Check if previous was a count ("*", "+", etc.)
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	445 if (pnode->mode == TH_RE_MATCH_COUNT_GREEDY \|\|
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	446 pnode->mode == TH_RE_MATCH_COUNT_NONGREEDY)
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 res = THERR_INVALID_DATA;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	449 goto exit;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	450 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	451
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	452 pnode->mode = TH_RE_MATCH_COUNT_GREEDY;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	453
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	454 if (cch == '*')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	455 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	456 // Previous token can repeat 0 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	457 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	458 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	459 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	460 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	461 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	462 // Previous token must repeat 1 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	463 pnode->repeatMin = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	464 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	465 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	466 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	467 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	468
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	469 case '{':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	470 // {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	471 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	472 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	473 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	474 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	475 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	476 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	477 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	478
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	479 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	480
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	481 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	482 (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	483 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	484 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	485
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	486 pnode->mode = TH_RE_MATCH_COUNT_GREEDY;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	487
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	488 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	489 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	490 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	491 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	492 !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	493 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	494 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	495 goto exit;
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 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	498 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	499 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	500 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	501 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	502 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	503 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	504 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	505 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	506 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	507
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	508 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	509 pnode->repeatMax < pnode->repeatMin)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	510 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	511 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	512 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	513 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	514 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	515 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	516
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	517 case '(':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	518 // Start of subpattern
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	519 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	520 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	521 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	522
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	523 case ')':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	524 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	525 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	526 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	527
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	528 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	529 (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	530 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	531 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	532
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	533 case '^':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	534 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	535 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	536 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	537
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	538 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	539 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	540 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	541
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	542 case '$':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	543 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	544 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	545 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	546
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	547 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	548 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	549 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	550
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	551 case '[':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	552 // 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	553 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	554 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	555 ctx.offs == start)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	556 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	557 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	558 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	559 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	560
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	561 th_regex_node_init(&node);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	562 if (ctx.pattern[start] == '^')
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	563 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	564 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	565 start++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	566 }
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	567 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	568 node.type = TH_RE_TYPE_LIST;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	569
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	570 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	571 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	572 (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	573 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	574 break;
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 case '.':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	577 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	578 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	579 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	580
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	581 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	582 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	583 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	584
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	585 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	586 // Literal escape
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	587 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	588 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	589 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	590 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	591 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	592 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	593 }
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	594 // fall-through
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	595
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	596 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	597 // Given character must match
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	598 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	599 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	600 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	601
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	602 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	603 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	604 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	605 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	606 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	607
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	608 exit:
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	609 *pexpr = ctx.data;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	610 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	611 return res;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	612 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	613
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	614
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	615 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	616 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	617 if (expr != NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	618 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	619 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	620 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	621 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	622
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	623 DBG_RE_FREE(
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	624 "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	625 "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	626 nnode, expr->nnodes,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	627 re_match_modes[node->mode],
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	628 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	629 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	630
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	631 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	632 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	633 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	634
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	635 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	636 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	637 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	638
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	639
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	640 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	641 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	642 // 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	643 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	644 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	645 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	646
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	647 if (item->type == 0)
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 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	650 if (item->chars[n] == cch)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	651 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	652 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	653 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	654 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	655 if (cch >= item->start && cch <= item->end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	656 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	657 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	658 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	659
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	660 return FALSE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	661 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	662
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	663
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	664 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	665 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	666 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	667 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	668 const size_t startnode,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	669 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	670 );
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	671
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	672
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	673 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	674 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	675 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	676 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	677 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	678 )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	679 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	680 th_regex_char_t cch;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	681 BOOL res = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	682
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	683 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	684 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	685 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	686 // push current context in stack
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	687 // new context for subexpr
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	688 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	689 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	690
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	691 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	692 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	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
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	696 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	697 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	698
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	699 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	700 res = !res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	701
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	702 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	703 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	704 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	705
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	706 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	707 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	708 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	709 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	710 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	711 res = TRUE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	712 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	713 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	714 break;
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 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	717 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	718 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	719 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	720 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	721 res = (cch == node->match.chr);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	722 (*poffs)++;
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 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	725 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	726
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	727 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	728 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	729
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	730
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	731 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	732 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	733 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	734 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	735 const size_t startnode,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	736 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	737 )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	738 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	739 BOOL res = TRUE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	740 size_t soffs = *poffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	741
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	742 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	743 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	744 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	745
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	746 DBG_RE_MATCH(" expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s %s '%s'\n",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	747 nnode, expr->nnodes,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	748 re_match_modes[node->mode], re_match_types[node->type],
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	749 haystack + *poffs);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	750
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	751 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	752 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	753 case TH_RE_MATCH_ONCE:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	754 res = th_regex_match_one(haystack, poffs, node, flags);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	755 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	756
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	757 case TH_RE_MATCH_COUNT_GREEDY:
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	758 case TH_RE_MATCH_COUNT_NONGREEDY:
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	759 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	760 ssize_t count = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	761
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	762 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	763 node->repeatMin, node->repeatMax);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	764
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	765 do
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	766 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	767 BOOL match;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	768 size_t toffs = *poffs, tnode = nnode;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	769
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	770 do {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	771 match = th_regex_match_one(haystack, &toffs, node, flags);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	772 if (match && haystack[toffs] != 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	773 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	774 for (tnode = nnode + 1; match && tnode < expr->nnodes && haystack[toffs] != 0; tnode++)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	775 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	776 size_t noffs = toffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	777 match = th_regex_match_one(haystack, &toffs, &expr->nodes[tnode], flags);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	778 DBG_RE_MATCH(" '%s': %d\n", haystack + noffs, match);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	779 }
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	780 }
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	781 } while (!match && haystack[toffs] != 0);
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	782
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	783 if (match)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	784 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	785 // Node matched
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	786 count++;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	787 *poffs = soffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	788 nnode = tnode;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	789 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	790 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	791 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	792 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	793 // 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	794 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	795 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	796
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	797 } while (!res);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	798
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	799 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	800 count, res ? "YES" : "NO");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	801 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	802 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	803
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	804
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	805 case TH_RE_MATCH_ANCHOR_START:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	806 res = (*poffs == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	807 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	808
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	809 case TH_RE_MATCH_ANCHOR_END:
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	810 res = (haystack[*poffs] == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	811 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	812 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	813 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	814
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	815 if (!res)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	816 *poffs = soffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	817
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	818 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	819 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	820
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	821
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	822 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	823 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	824 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	825 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	826 size_t nmatches = 0;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	827 (void) flags;
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	828
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	829 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	830 *pnmatches = 0;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	831
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	832 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	833 if (expr == NULL \|\| haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	834 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	835
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	836 // Start matching
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	837 // 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	838 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	839 // - 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	840 // ..
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	841 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	842 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	843 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	844
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	845 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	846 soffs, haystack + soffs);
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	847
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	848 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	849 {
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	850 // 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	851 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	852
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	853 // Deliver to caller if required
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	854 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	855 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	856
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	857 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	858 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	859 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	860 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	861 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	862
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	863 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	864 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	865
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	866 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	867 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	868
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	869 // 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	870 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	871 break;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	872
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	873 // If offset was not advanced, increase by one
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	874 // 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	875 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	876 soffs++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	877 else
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	878 soffs = coffs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	879 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	880 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	881 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	882 soffs++;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	883 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	884 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	885
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	886 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	887 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	888
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	889
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	890 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	891 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	892 (void) node;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	893 // 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	894 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	895
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	896
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	897 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	898 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	899 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	900 (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	901 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	902
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	903 #endif // TH_EXPERIMENTAL_REGEX

Mercurial > hg > th-libs

annotate th_regex.c @ 642:3a35db5c1873