th-libs: th_regex.c annotate

annotate th_regex.c @ 645:b897995101b7

More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes.

author	Matti Hamalainen <ccr@tnsp.org>
date	Sat, 25 Jan 2020 05:11:40 +0200
parents	a2bf1ea05b05
children	1e7e3f96632e

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
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	13 # define DBG_RE_MATCH(...) do { if (th_dbg_re_flags) fprintf(stdout, __VA_ARGS__); } while (0)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	14 BOOL th_dbg_re_flags = FALSE;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	15 #else
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	16 # define DBG_RE_MATCH(...)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	17 #endif
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	18
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	19
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	20 #ifdef TH_EXPERIMENTAL_REGEX
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	21
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	22 enum
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	23 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	24 TH_RE_MATCH_ONCE,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	25 TH_RE_MATCH_COUNT,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	26 TH_RE_MATCH_ANCHOR_START,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	27 TH_RE_MATCH_ANCHOR_END,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	28 };
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	29
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	30
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	31 enum
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	32 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	33 TH_RE_TYPE_CHAR,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	34 TH_RE_TYPE_STR,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	35 TH_RE_TYPE_ANY_CHAR,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	36 TH_RE_TYPE_LIST,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	37 TH_RE_TYPE_LIST_REVERSE,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	38 TH_RE_TYPE_SUBEXPR,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	39 };
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	40
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	41
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	42 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	43 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	44 "ONCE",
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	45 "COUNT",
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	46 "ANCHOR START",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	47 "ANCHOR END",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	48 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	49
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	50
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	51 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	52 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	53 "CHAR",
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	54 "STR",
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	55 "ANY",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	56 "LIST",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	57 "LIST REVERSE",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	58 "SUBEXPR",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	59 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	60
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	61
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	62 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	63 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	64 int type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	65 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	66
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	67 size_t nchars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	68 th_regex_char_t *chars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	69 } th_regex_list_item_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	70
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	71
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	72 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	73 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	74 size_t nitems, itemssize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	75 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	76 } th_regex_list_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
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	79 typedef struct
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 int mode, type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	82 ssize_t repeatMin, repeatMax;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	83
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	84 struct {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	85 th_regex_char_t chr;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	86 th_regex_char_t *str;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	87 th_regex_list_t list;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	88
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	89 th_regex_t *expr;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	90 } match;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	91 } th_regex_node_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	92
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	93
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	94 struct th_regex_t
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	95 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	96 size_t nnodes, nodessize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	97 th_regex_node_t *nodes;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	98 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	99
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	100
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	101 typedef struct
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	102 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	103 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	104 size_t offs;
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 th_regex_t *data;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	107
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	108 size_t nstack, stacksize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	109 th_regex_t **stack;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	110
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	111 th_regex_char_t *buf;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	112 size_t bufSize, bufPos;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	113
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	114 } 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	115
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	116
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	117 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	118 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	119 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	120 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	121 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	122
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	123
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	124 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	125 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	126 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	127 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	128 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	129
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	130 if (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	131 return THERR_BOUNDS;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	132
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	133 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	134 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	135 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	136
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	137 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	138 (*pdst)[len] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	139
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	140 return THERR_OK;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	143
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	144 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	145 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	146 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	147 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	148 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	149 *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	150 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	151 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	153 return THERR_INVALID_DATA;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	156
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	157 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	158 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	159 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	160 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	161 ctx->stacksize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	162
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	163 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	164 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	165 return THERR_MALLOC;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	168 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	169 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	170 ctx->data = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	171
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	172 return THERR_OK;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	175
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	176 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	177 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	178 if (ctx->nstack > 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	179 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	180 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	181 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	182 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	183 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	184 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	185 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	186 return THERR_INVALID_DATA;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	189
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	190 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	191 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	192 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	193
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	194 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	195 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	196 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	197 return THERR_MALLOC;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	200 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	201 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	202 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	203 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	204 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	205 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	206 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	207
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	208 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	209 data->nnodes++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	210
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	211 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	212 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	213
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	214
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	215 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	216 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	217 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	218 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	219 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	220 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	221 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	222 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	223 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	224 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	225 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	226
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	227
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	228 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	229 ssize_t *value)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	230 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	231 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	232 BOOL neg;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	233
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	234 if (*str == '-')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	235 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	236 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	237 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	238 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	239 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	240 neg = FALSE;
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 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	243 while ((ch = *str++))
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 if (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	246 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	247 value = 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	248 *value += ch - '0';
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	249 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	250 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	251 return FALSE;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	254 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	255 value = -(value);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	256
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	257 return TRUE;
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	261 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	262 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	263 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	264 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	265
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	266
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	267 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	268 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	269 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	270 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	271 list->itemssize += 16;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	272
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	273 if ((list->items = th_realloc(list->items,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	274 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	275 return THERR_MALLOC;
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	278 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	279 list->nitems++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	280
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	281 return THERR_OK;
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
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	284
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	285 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	286 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	287 if (list != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	288 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	289 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	290 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	291 th_free(list->items[n].chars);
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 th_free(list->items);
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 }
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	298 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	299 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	300 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	301 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	302 th_regex_list_item_t item;
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	303 int res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	304
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	305 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	306 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	307
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	308 // Handle ranges like [A-Z]
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	309 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	310 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	311 th_regex_char_t
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	312 *prev = (offs > 0) ? tmp + offs - 1 : NULL,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	313 *curr = tmp + offs,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	314 *next = (offs + 1 < slen) ? tmp + offs + 1 : NULL;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	315
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	316 if (*curr == '-')
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	317 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	318 if (prev != NULL && next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	319 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	320 // Range
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	321 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	322 item.type = 1;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	323 item.start = *prev;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	324 item.end = *next;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	325
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	326 if (item.start >= item.end)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	327 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	328 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	329 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	330 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	331
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	332 curr = prev = *next = 0;
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 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	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 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	338 if (next != NULL)
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 res = THERR_INVALID_DATA;
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 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	344 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	345
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	346 // Count number of remaining characters
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	347 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	348 item.type = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	349 item.nchars = 0;
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 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	352 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	353 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	354 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	355 item.nchars++;
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
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	358 if (item.nchars > 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	359 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	360 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	361 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	362 res = THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	363 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	364 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	365
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	366 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	367 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	368 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	369 if (curr != 0)
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 item.chars[n] = curr;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	372 n++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	373 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	374 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	375
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	376 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	377 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	378 th_free(item.chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	379 goto out;
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
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	383 out:
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	384 th_free(tmp);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	385 return res;
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
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	389 static int th_regex_parse_ctx_node_commit_strchr_do(th_regex_parse_ctx_t *ctx,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	390 const th_regex_char_t *buf, const size_t bufLen)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	391 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	392 th_regex_node_t node;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	393 th_regex_node_init(&node);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	394
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	395 if (bufLen > 1)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	396 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	397 int res;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	398 node.type = TH_RE_TYPE_STR;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	399 if ((res = th_regex_strndup(&node.match.str, buf, bufLen)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	400 return res;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	401 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	402 else
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	403 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	404 node.type = TH_RE_TYPE_CHAR;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	405 node.match.chr = buf[0];
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	406 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	407
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	408 return th_regex_parse_ctx_node_commit(ctx, &node);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	409 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	410
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	411
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	412 static int th_regex_parse_ctx_node_commit_strchr(th_regex_parse_ctx_t *ctx, const BOOL split)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	413 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	414 int res = THERR_OK;;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	415
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	416 if (ctx->bufPos > 0)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	417 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	418 if (ctx->bufPos > 1 && split)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	419 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	420 if ((res = th_regex_parse_ctx_node_commit_strchr_do(ctx, ctx->buf, ctx->bufPos - 1)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	421 return res;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	422
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	423 res = th_regex_parse_ctx_node_commit_strchr_do(ctx, ctx->buf + ctx->bufPos - 1, 1);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	424 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	425 else
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	426 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	427 res = th_regex_parse_ctx_node_commit_strchr_do(ctx, ctx->buf, ctx->bufPos);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	428 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	429 ctx->bufPos = 0;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	430 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	431
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	432 return res;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	433 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	434
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	435
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	436 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	437 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	438 int res = THERR_OK;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	439 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	440 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	441 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	442 size_t start;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	443
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	444 // Check pointers
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	445 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	446 {
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	447 res = THERR_NULLPTR;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	448 goto out;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	449 }
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	450
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	451 // Initialize parsing context
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	452 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	453 ctx.pattern = pattern;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	454 ctx.bufSize = 256;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	455
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	456 if ((ctx.buf = th_malloc(ctx.bufSize * sizeof(th_regex_char_t))) == NULL)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	457 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	458 res = THERR_MALLOC;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	459 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	460 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	461
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	462 // Start parsing the pattern
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	463 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	464 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	465 th_regex_char_t cch = ctx.pattern[ctx.offs];
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	466
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	467 switch (cch)
605 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 '?':
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	470 case '*':
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	471 case '+':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	472 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, TRUE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	473 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	474
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	475 if ((res = th_regex_parse_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	476 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	477
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	478 if (cch == '?')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	479 {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	480 // Previous token is optional (repeat 0-1 times) (non-greedy matching)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	481 pnode->mode = TH_RE_MATCH_COUNT;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	482 pnode->repeatMin = 0;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	483 pnode->repeatMax = 1;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	484 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	485 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	486 {
641 9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	487 // Check if previous was a count ("*", "+", etc.)
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	488 if (pnode->mode == TH_RE_MATCH_COUNT)
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	489 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	490 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	491 goto out;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	492 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	493
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	494 pnode->mode = TH_RE_MATCH_COUNT;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	495
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	496 if (cch == '*')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	497 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	498 // Previous token can repeat 0 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	499 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	500 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	501 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	502 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	503 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	504 // Previous token must repeat 1 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	505 pnode->repeatMin = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	506 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	507 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	508 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	509 break;
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 case '{':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	512 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, TRUE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	513 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	514
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	515 // {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	516 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	517 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	518 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	519 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	520 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	521 goto out;
605 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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	524 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	525
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	526 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	527 (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	528 ctx.offs - start)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	529 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	530
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	531 pnode->mode = TH_RE_MATCH_COUNT;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	532
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	533 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	534 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	535 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	536 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	537 !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	538 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	539 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	540 goto out;
605 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 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	543 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	544 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	545 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	546 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	547 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	548 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	549 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	550 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	551 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	552
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	553 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	554 pnode->repeatMax < pnode->repeatMin)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	555 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	556 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	557 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	558 goto out;
605 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 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	561
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	562 case '(':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	563 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	564 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	565
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	566 // Start of subpattern
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	567 if ((res = th_regex_parse_ctx_push(&ctx)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	568 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	569 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	570
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	571 case ')':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	572 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	573 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	574
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	575 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	576 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	577 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	578
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	579 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	580 (res = th_regex_parse_ctx_node_commit(&ctx, &node)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	581 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	582 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	583
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	584 case '^':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	585 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	586 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	587
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	588 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	589 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	590 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	591
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	592 if ((res = th_regex_parse_ctx_node_commit(&ctx, &node)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	593 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	594 break;
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 case '$':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	597 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	598 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	599
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	600 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	601 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	602 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	603
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	604 if ((res = th_regex_parse_ctx_node_commit(&ctx, &node)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	605 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	606 break;
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 case '[':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	609 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	610 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	611
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	612 // 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	613 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	614 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	615 ctx.offs == start)
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 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	618 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	619 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	620
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	621 th_regex_node_init(&node);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	622 if (ctx.pattern[start] == '^')
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	623 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	624 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	625 start++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	626 }
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	627 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	628 node.type = TH_RE_TYPE_LIST;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	629
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	630 if ((res = th_regex_parse_list(ctx.pattern + start,
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	631 ctx.offs - start, &node.match.list)) != THERR_OK \|\|
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	632 (res = th_regex_parse_ctx_node_commit(&ctx, &node)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	633 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	634 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	635
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	636 case '.':
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	637 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	638 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	639
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	640 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	641 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	642 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	643
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	644 if ((res = th_regex_parse_ctx_node_commit(&ctx, &node)) != THERR_OK)
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	645 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	646 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	647
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	648 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	649 // Literal escape
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	650 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	651 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	652 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	653 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	654 res = THERR_INVALID_DATA;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	655 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	656 }
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	657 // fall-through
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	658
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	659 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	660 // Given character must match
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	661 if (ctx.bufPos < ctx.bufSize)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	662 ctx.buf[ctx.bufPos++] = ctx.pattern[ctx.offs];
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	663 else
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	664 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	665 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	666 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	667 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	668 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	669
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	670 // Commit last string/char if any
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	671 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	672 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	673
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	674 // Create root node
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	675 th_regex_node_init(&node);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	676 node.type = TH_RE_TYPE_SUBEXPR;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	677 node.match.expr = ctx.data;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	678 ctx.data = NULL;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	679
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	680 if ((res = th_regex_parse_ctx_node_commit(&ctx, &node)) != THERR_OK)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	681 goto out;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	682
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	683 out:
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	684 *pexpr = ctx.data;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	685
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	686 // Free temporary buffers
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	687 th_free(tmp);
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	688 th_free(ctx.buf);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	689 return res;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	692
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	693 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	694 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	695 if (expr != NULL)
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 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	698 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	699 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	700
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	701 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	702 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	703 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	704
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	705 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	706 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	707 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	708
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	709
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	710 void th_regex_dump(FILE fh, const int level, const th_regex_t expr)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	711 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	712 if (expr != NULL)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	713 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	714 for (size_t nnode = 0; nnode < expr->nnodes; nnode++)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	715 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	716 th_regex_node_t *node = &expr->nodes[nnode];
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	717
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	718 for (int indent = 0; indent < level; indent++)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	719 fprintf(fh, " ");
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	720
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	721 fprintf(fh, "[%" PRIu_SIZE_T " / %" PRIu_SIZE_T "] %s %s ",
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	722 nnode + 1, expr->nnodes,
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	723 re_match_modes[node->mode],
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	724 re_match_types[node->type]);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	725
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	726 if (node->mode == TH_RE_MATCH_COUNT)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	727 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	728 fprintf(fh, "min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T " : ",
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	729 node->repeatMin, node->repeatMax);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	730 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	731
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	732 switch (node->type)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	733 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	734 case TH_RE_TYPE_CHAR:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	735 fprintf(fh, "'%c'\n", node->match.chr);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	736 break;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	737
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	738 case TH_RE_TYPE_STR:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	739 fprintf(fh, "\"%s\"\n", node->match.str);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	740 break;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	741
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	742 case TH_RE_TYPE_ANY_CHAR:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	743 fprintf(fh, ".\n");
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	744 break;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	745
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	746 case TH_RE_TYPE_LIST:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	747 case TH_RE_TYPE_LIST_REVERSE:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	748 fprintf(fh, "[ ");
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	749 for (size_t n = 0; n < node->match.list.nitems; n++)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	750 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	751 const th_regex_list_item_t *li = &node->match.list.items[n];
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	752 if (li->type)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	753 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	754 fprintf(fh, "'%c-%c' ", li->start, li->end);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	755 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	756 else
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	757 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	758 for (size_t i = 0; i < li->nchars; i++)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	759 fprintf(fh, "'%c' ", li->chars[i]);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	760 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	761 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	762 fprintf(fh, "]\n");
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	763 break;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	764
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	765 case TH_RE_TYPE_SUBEXPR:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	766 fprintf(fh, "\n");
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	767 th_regex_dump(fh, level + 1, node->match.expr);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	768 break;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	769 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	770 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	771 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	772 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	773
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	774
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	775 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	776 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	777 // 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	778 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	779 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	780 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	781
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	782 if (item->type == 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	783 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	784 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	785 if (item->chars[n] == cch)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	786 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	787 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	788 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	789 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	790 if (cch >= item->start && cch <= item->end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	791 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	792 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	793 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	794
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	795 return FALSE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	796 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	797
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	798
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	799 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	800 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	801 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	802 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	803 const size_t startnode,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	804 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	805 );
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	806
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	807
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	808 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	809 const th_regex_char_t *haystack,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	810 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	811 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	812 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	813 )
605 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 th_regex_char_t cch;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	816 BOOL res = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	817
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	818 switch (node->type)
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 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	821 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	822 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	823
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	824 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	825 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	826 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	827 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	828 else
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	829 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	830 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	831
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	832 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	833 res = !res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	834
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	835 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	836 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	837 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	838
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	839 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	840 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	841 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	842 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	843 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	844 res = TRUE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	845 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	846 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	847 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	848
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	849 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	850 if ((cch = haystack[*poffs]) == 0)
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	851 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	852 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	853 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	854 res = (cch == node->match.chr);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	855 (*poffs)++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	856 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	857 break;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	858
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	859 case TH_RE_TYPE_STR:
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	860 res = TRUE;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	861 for (th_regex_char_t str = node->match.str; str != 0; )
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	862 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	863 if (haystack[poffs] != str)
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	864 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	865 res = FALSE;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	866 break;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	867 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	868 str++;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	869 (*poffs)++;
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	870 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	871 break;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	872 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	873
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	874 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	875 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	876
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	877
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	878 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	879 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	880 size_t *poffs,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	881 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	882 const size_t startnode,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	883 const int flags
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	884 )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	885 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	886 BOOL res = TRUE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	887 size_t soffs = *poffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	888
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	889 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	890 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	891 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	892
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	893 DBG_RE_MATCH(" expr [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s %s '%s'\n",
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	894 nnode + 1, expr->nnodes,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	895 re_match_modes[node->mode], re_match_types[node->type],
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	896 haystack + soffs);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	897
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	898 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	899 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	900 case TH_RE_MATCH_ONCE:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	901 res = th_regex_match_one(haystack, &soffs, node, flags);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	902 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	903
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	904 case TH_RE_MATCH_COUNT:
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	905 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	906 ssize_t count = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	907
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	908 DBG_RE_MATCH(" min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n",
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	909 node->repeatMin, node->repeatMax);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	910
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	911 do
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	912 {
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	913 BOOL match;
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	914 size_t toffs = soffs, tnode;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	915
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	916 DBG_RE_MATCH(" ROUND #%" PRIu_SIZE_T ": '%s'\n",
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	917 count, haystack + toffs);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	918
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	919 do
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	920 {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	921 match = TRUE;
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	922 size_t noffs = toffs;
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	923 for (tnode = nnode; match && tnode < expr->nnodes && haystack[toffs] != 0; )
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	924 {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	925 match = th_regex_match_one(haystack, &toffs, &expr->nodes[tnode], flags);
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	926 if (match)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	927 tnode++;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	928 }
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	929
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	930 DBG_RE_MATCH(" '%s': %d (tnode=%" PRIu_SIZE_T ")\n",
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	931 haystack + noffs, match, tnode);
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	932
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	933 if (node->repeatMin >= 0 && match)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	934 break;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	935 } while (!match && haystack[toffs] != 0);
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	936
645 b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	937 DBG_RE_MATCH(" ROUND #%" PRIu_SIZE_T " END: match=%s, tnode=%" PRIu_SIZE_T "\n",
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	938 count, match ? "YES" : "NO", tnode);
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	939
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	940 if (match)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	941 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	942 // Node matched
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	943 count++;
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	944 soffs = toffs;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	945 nnode = tnode;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	946 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	947 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	948 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	949 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	950 // 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	951 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	952 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	953
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	954 } while (!res);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	955
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	956 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	957 count, res ? "YES" : "NO");
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	958 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	959 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	960
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	961 case TH_RE_MATCH_ANCHOR_START:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	962 res = (soffs == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	963 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	964
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	965 case TH_RE_MATCH_ANCHOR_END:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	966 res = (haystack[soffs] == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	967 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	968 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	969 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	970
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	971 if (res)
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	972 *poffs = soffs;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	973
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	974 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	975 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	976
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	977
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	978 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	979 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	980 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	981 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	982 size_t nmatches = 0;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	983 (void) flags;
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	984
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	985 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	986 *pnmatches = 0;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	987
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	988 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	989 if (expr == NULL \|\| haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	990 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	991
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	992 // Start matching
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	993 // 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	994 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	995 // - 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	996 // ..
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	997 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	998 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	999 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1000
635 d191ded8a790 Improve the experimental regex matching debugging macros. Matti Hamalainen <ccr@tnsp.org> parents: 614 diff changeset	1001 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	1002 soffs, haystack + soffs);
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1003
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1004 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	1005 {
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1006 // 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	1007 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1008
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1009 // Deliver to caller if required
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1010 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1011 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1012
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1013 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1014 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1015 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	1016 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1017 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1018
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1019 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1020 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1021
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1022 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	1023 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1024
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1025 // 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	1026 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1027 break;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1028
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1029 // If offset was not advanced, increase by one
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1030 // 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	1031 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1032 soffs++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1033 else
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1034 soffs = coffs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1035 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1036 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1037 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1038 soffs++;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1039 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1040 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1041
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1042 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1043 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1044
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1045
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1046 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	1047 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1048 (void) node;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1049 // 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	1050 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1051
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1052
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1053 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	1054 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1055 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	1056 (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	1057 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1058
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1059 #endif // TH_EXPERIMENTAL_REGEX

Mercurial > hg > th-libs

annotate th_regex.c @ 645:b897995101b7