th-libs: th_regex.c annotate

annotate th_regex.c @ 652:38a9302962f7

Always compile the regex code, but do not run tests on it unless enabled.

author	Matti Hamalainen <ccr@tnsp.org>
date	Sat, 25 Jan 2020 13:18:38 +0200
parents	18fe45e61b2b
children	ae601363fdad

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
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	13 th_ioctx *th_dbg_fh = NULL;
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	14
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	15 # define DBG_RE_PRINT(...) do { \
18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	16 if (th_dbg_fh != NULL) \
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	17 { \
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	18 th_regex_dump_indent(th_dbg_fh, level); \
18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	19 thfprintf(th_dbg_fh, __VA_ARGS__); \
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	20 } \
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	21 } while (0)
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 #else
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	23 # define DBG_RE_PRINT(...)
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	24 #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	25
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
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 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	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 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	30 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	31 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	32 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	33 };
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
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
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 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	37 {
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_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	39 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	40 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	41 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	42 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	43 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	44 };
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	45
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	46
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	47 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	48 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	49 "ONCE",
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	50 "COUNT",
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	51 "ANCHOR START",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	52 "ANCHOR END",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	53 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	54
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	55
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	56 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	57 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	58 "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	59 "STR",
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	60 "ANY",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	61 "LIST",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	62 "LIST REVERSE",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	63 "SUBEXPR",
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	64 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	65
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	66
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	67 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	68 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	69 int type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	70 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	71
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	72 size_t nchars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	73 th_regex_char_t *chars;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	74 } th_regex_list_item_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	75
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	76
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	77 typedef struct
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 size_t nitems, itemssize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	80 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	81 } th_regex_list_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	82
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 typedef struct
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	85 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	86 int mode, type;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	87 ssize_t repeatMin, repeatMax;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	88
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	89 struct {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	90 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	91 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	92 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	93
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	94 th_regex_t *expr;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	95 } match;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	96 } th_regex_node_t;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	97
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 struct th_regex_t
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 size_t nnodes, nodessize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	102 th_regex_node_t *nodes;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	103 };
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	104
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	105
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	106 typedef struct
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	107 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	108 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	109 size_t offs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	110
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	111 th_regex_t *data;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	112
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	113 size_t nstack, stacksize;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	114 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	115
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	116 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	117 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	118
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	119 } 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	120
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	121
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	122 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	123 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	124 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	125 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	126 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	127
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	128
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	129 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	130 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	131 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	132 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	133 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	134
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	135 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	136 return THERR_BOUNDS;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	137
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	138 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	139 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	140 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	141
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	142 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	143 (*pdst)[len] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	144
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	145 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	146 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	147
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	148
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	149 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	150 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	151 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	152 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	153 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	154 *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	155 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	156 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	157 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	158 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	159 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	160
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	161
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	162 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	163 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	164 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	165 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	166 ctx->stacksize += 16;
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 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	169 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	170 return THERR_MALLOC;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	173 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	174 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	175 ctx->data = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	176
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	177 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	178 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	179
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	180
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	181 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	182 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	183 if (ctx->nstack > 0)
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 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	186 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	187 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	188 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	189 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	190 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	191 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	192 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	193
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	194
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	195 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	196 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	197 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	198
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	199 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	200 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	201 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	202 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	203 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	204
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	205 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	206 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	207 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	208 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	209 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	210 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	211 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	212
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	213 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	214 data->nnodes++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	215
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	216 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	217 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	218
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	219
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	220 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	221 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	222 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	223 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	224 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	225 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	226 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	227 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	228 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	229 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	230 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	231
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	232
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	233 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	234 ssize_t *value)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	235 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	236 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	237 BOOL neg;
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 if (*str == '-')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	240 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	241 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	242 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	243 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	244 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	245 neg = FALSE;
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 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	248 while ((ch = *str++))
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 if (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	251 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	252 value = 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	253 *value += ch - '0';
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	254 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	255 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	256 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	257 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	258
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	259 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	260 value = -(value);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	261
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	262 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	263 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	264
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	265
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	266 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	267 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	268 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	269 }
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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	272 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	273 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	274 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	275 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	276 list->itemssize += 16;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	277
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	278 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	279 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	280 return THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	281 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	282
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	283 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	284 list->nitems++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	285
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	286 return THERR_OK;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	287 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	288
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	289
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	290 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	291 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	292 if (list != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	293 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	294 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	295 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	296 th_free(list->items[n].chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	297 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	298 th_free(list->items);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	299 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	300 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	301
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	302
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	303 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	304 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	305 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	306 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	307 th_regex_list_item_t item;
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	308 int res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	309
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	310 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	311 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	312
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	313 // Handle ranges like [A-Z]
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	314 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	315 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	316 th_regex_char_t
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	317 *prev = (offs > 0) ? tmp + offs - 1 : NULL,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	318 *curr = tmp + offs,
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	319 *next = (offs + 1 < slen) ? tmp + offs + 1 : NULL;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	320
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	321 if (*curr == '-')
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	322 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	323 if (prev != NULL && next != NULL)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	324 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	325 // Range
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	326 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	327 item.type = 1;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	328 item.start = *prev;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	329 item.end = *next;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	330
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	331 if (item.start >= item.end)
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	332 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	333 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	334 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	335 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	336
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	337 curr = prev = *next = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	338
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	339 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	340 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	341 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	342 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	343 if (next != NULL)
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 res = THERR_INVALID_DATA;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	346 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	347 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	348 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	349 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	350
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	351 // Count number of remaining characters
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	352 th_regex_list_item_init(&item);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	353 item.type = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	354 item.nchars = 0;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	355
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	356 for (size_t offs = 0; offs < slen; offs++)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	357 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	358 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	359 if (curr != 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	360 item.nchars++;
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
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	363 if (item.nchars > 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	364 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	365 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	366 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	367 res = THERR_MALLOC;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	368 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	369 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	370
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	371 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	372 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	373 th_regex_char_t curr = tmp[offs];
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	374 if (curr != 0)
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 item.chars[n] = curr;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	377 n++;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	378 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	379 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	380
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	381 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	382 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	383 th_free(item.chars);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	384 goto out;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	385 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	386 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	387
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	388 out:
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	389 th_free(tmp);
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	390 return res;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	391 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	392
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	393
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	394 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	395 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	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 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	398 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	399
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 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	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 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	403 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	404 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	405 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	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 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	408 {
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 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	410 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	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
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 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	414 }
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
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 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	418 {
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 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	420
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 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	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 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	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 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	426 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	427
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 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	429 }
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 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	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 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	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 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	435 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	436
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	437 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	438 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	439
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	440
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	441 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	442 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	443 int res = THERR_OK;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	444 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	445 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	446 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	447 size_t start;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	448
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	449 // Check pointers
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	450 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	451 {
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	452 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	453 goto out;
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	454 }
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 // 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	457 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	458 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	459 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	460
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	461 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	462 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	463 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	464 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	465 }
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
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	467 // 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	468 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	469 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	470 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	471
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	472 switch (cch)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	473 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	474 case '?':
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	475 case '*':
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	476 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	477 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	478 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	479
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	480 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	481 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	482
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	483 if (cch == '?')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	484 {
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	485 // Previous token is optional (repeat 0-1 times) (non-greedy matching)
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	486 pnode->mode = TH_RE_MATCH_COUNT;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	487 pnode->repeatMin = 0;
a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	488 pnode->repeatMax = 1;
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 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	491 {
641 9a1ed82abefd Fix parsing of +? and ?. Matti Hamalainen <ccr@tnsp.org>* parents: 640 diff changeset	492 // Check if previous was a count ("*", "+", etc.)
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	493 if (pnode->mode == TH_RE_MATCH_COUNT)
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	494 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	495 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	496 goto out;
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	497 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	498
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	499 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	500
613 2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	501 if (cch == '*')
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	502 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	503 // Previous token can repeat 0 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	504 pnode->repeatMin = 0;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	505 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	506 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	507 else
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	508 {
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	509 // Previous token must repeat 1 or more times
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	510 pnode->repeatMin = 1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	511 pnode->repeatMax = -1;
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	512 }
2e3b81ae8c8a More work on regexes. Matti Hamalainen <ccr@tnsp.org> parents: 612 diff changeset	513 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	514 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	515
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	516 case '{':
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	517 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	518 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	519
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	520 // {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	521 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	522 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	523 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	524 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	525 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	526 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	527 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	528
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	529 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	530
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	531 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	532 (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	533 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	534 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	535
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	536 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	537
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	538 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	539 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	540 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	541 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	542 !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	543 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	544 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	545 goto out;
605 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 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	548 else
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 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	551 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	552 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	553 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	554 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	555 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	556 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	557
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	558 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	559 pnode->repeatMax < pnode->repeatMin)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	560 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	561 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	562 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	563 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	564 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	565 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	566
648 91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	567 /*
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	568 case '\|':
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	569 if ((res = th_regex_parse_ctx_node_commit_strchr(&ctx, FALSE)) != THERR_OK)
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	570 goto out;
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	571
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	572 // Alt pattern .. how to handle these?
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	573 break;
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	574 */
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	575
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	576 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	577 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	578 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	579
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	580 // Start of subpattern
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	581 if ((res = th_regex_parse_ctx_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	582 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	583 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	584
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	585 case ')':
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	586 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	587 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	588
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	589 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	590 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	591 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	592
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	593 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	594 (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	595 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	596 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	597
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	598 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	599 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	600 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	601
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	602 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	603 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	604 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	605
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	606 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	607 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	608 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	609
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	610 case '$':
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	611 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	612 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	613
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	614 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	615 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	616 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	617
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	618 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	619 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	620 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	621
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	622 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	623 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	624 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	625
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	626 // 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	627 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	628 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	629 ctx.offs == start)
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	630 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	631 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	632 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	633 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	634
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	635 th_regex_node_init(&node);
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	636 if (ctx.pattern[start] == '^')
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	637 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	638 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	639 start++;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	640 }
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	641 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	642 node.type = TH_RE_TYPE_LIST;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 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_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	645 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	646 (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	647 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	648 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	649
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	650 case '.':
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	651 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	652 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	653
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	654 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	655 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	656 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	657
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	658 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	659 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	660 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	661
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	662 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	663 // Literal escape
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	664 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	665 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	666 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	667 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	668 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	669 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	670 }
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	671 // fall-through
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	672
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	673 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	674 // 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	675 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	676 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	677 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	678 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	679 goto out;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	680 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	681 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	682 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	683
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	684 // 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	685 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	686 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	687
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 // 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	689 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	690 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	691 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	692 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	693
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	694 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	695 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	696
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	697 out:
611 d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do(). Matti Hamalainen <ccr@tnsp.org> parents: 610 diff changeset	698 *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	699
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	700 // Free temporary buffers
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	701 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	702 th_free(ctx.buf);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	703 return res;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	706
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	707 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	708 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	709 if (expr != NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	710 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	711 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	712 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	713 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	714
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	715 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	716 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	717 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	718
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	719 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	720 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	721 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	722
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	723
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	724 static void th_regex_dump_indent(th_ioctx *fh, const int level)
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	725 {
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	726 for (int indent = 0; indent < level; indent++)
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	727 thfputs(" ", fh);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	728 }
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	729
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	730
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	731 static void th_regex_dump_node(th_ioctx fh, const th_regex_node_t node)
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	732 {
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	733 thfprintf(fh,
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	734 "%s %s ",
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	735 re_match_modes[node->mode],
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	736 re_match_types[node->type]);
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	737
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	738 if (node->mode == TH_RE_MATCH_COUNT)
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	739 {
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	740 thfprintf(fh, "min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T " : ",
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	741 node->repeatMin, node->repeatMax);
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	742 }
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	743
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	744 switch (node->type)
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	745 {
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	746 case TH_RE_TYPE_CHAR:
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	747 thfprintf(fh, "'%c'", node->match.chr);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	748 break;
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	749
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	750 case TH_RE_TYPE_STR:
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	751 thfprintf(fh, "\"%s\"", node->match.str);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	752 break;
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	753
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	754 case TH_RE_TYPE_ANY_CHAR:
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	755 thfprintf(fh, ".");
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	756 break;
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	757
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	758 case TH_RE_TYPE_LIST:
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	759 case TH_RE_TYPE_LIST_REVERSE:
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	760 thfputs("[ ", fh);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	761 for (size_t n = 0; n < node->match.list.nitems; n++)
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	762 {
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	763 const th_regex_list_item_t *li = &node->match.list.items[n];
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	764 if (li->type)
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	765 {
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	766 thfprintf(fh, "'%c-%c' ", li->start, li->end);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	767 }
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	768 else
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	769 {
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	770 for (size_t i = 0; i < li->nchars; i++)
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	771 thfprintf(fh, "'%c' ", li->chars[i]);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	772 }
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	773 }
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	774 thfputs("]", fh);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	775 break;
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	776 }
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	777 }
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	778
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	779
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	780 void th_regex_dump(th_ioctx fh, const int level, const th_regex_t expr)
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	781 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	782 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	783 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	784 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	785 {
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	786 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	787
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	788 th_regex_dump_indent(fh, level);
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	789
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	790 thfprintf(fh,
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	791 "[%" PRIu_SIZE_T "/%" PRIu_SIZE_T "] ",
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	792 nnode + 1, expr->nnodes);
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	793
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	794 th_regex_dump_node(fh, node);
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	795 thfputs("\n", fh);
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	796
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	797 if (node->type == TH_RE_TYPE_SUBEXPR)
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	798 th_regex_dump(fh, level + 1, node->match.expr);
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	799 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	800 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	801 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	802
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	803
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	804 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	805 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	806 // 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	807 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	808 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	809 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	810
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	811 if (item->type == 0)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	812 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	813 for (size_t n = 0; n < item->nchars; n++)
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	814 {
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	815 if (item->chars[n] == cch)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	816 return TRUE;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	817 }
639 8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	818 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	819 else
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	820 {
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	821 if (cch >= item->start && cch <= item->end)
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	822 return TRUE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	823 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	824 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	825
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	826 return FALSE;
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	827 }
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	828
8c957ad9d4c3 Some more work on regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 638 diff changeset	829
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	830 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	831 const th_regex_char_t *haystack,
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	832 size_t *offs,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	833 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	834 const size_t startnode,
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	835 const int flags,
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	836 const int level
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	837 );
605 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
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	840 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	841 const th_regex_char_t *haystack,
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	842 size_t *offs,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	843 const th_regex_node_t *node,
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	844 const int flags,
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	845 const int level
640 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 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	848 th_regex_char_t cch;
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	849 BOOL res = FALSE;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	850
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	851 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	852 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	853 case TH_RE_TYPE_SUBEXPR:
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	854 res = th_regex_match_expr(haystack, offs, node->match.expr, 0, flags, level + 1);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	855 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	856
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	857 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	858 case TH_RE_TYPE_LIST_REVERSE:
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	859 if ((cch = haystack[*offs]) == 0)
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	860 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	861 else
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	862 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	863 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	864
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	865 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	866 res = !res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	867
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	868 (*offs)++;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	869 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	870 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	871
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	872 case TH_RE_TYPE_ANY_CHAR:
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	873 if ((cch = haystack[*offs]) == 0)
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	874 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	875 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	876 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	877 res = TRUE;
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	878 (*offs)++;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	879 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	880 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	881
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	882 case TH_RE_TYPE_CHAR:
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	883 if ((cch = haystack[*offs]) == 0)
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	884 res = FALSE;
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	885 else
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	886 {
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	887 res = (cch == node->match.chr);
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	888 (*offs)++;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	889 }
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	890 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	891
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	892 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	893 res = TRUE;
648 91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	894 for (th_regex_char_t *str = node->match.str;
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	895 res && *str != 0;
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	896 str++, (*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	897 {
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	898 if (haystack[offs] != str)
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	899 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	900 }
b897995101b7 More fiddling and twiddling. Add parsing to string nodes instead of separate character nodes. Matti Hamalainen <ccr@tnsp.org> parents: 643 diff changeset	901 break;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	902 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	903
638 c4bca120bfb0 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 635 diff changeset	904 return res;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	907
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	908 static BOOL th_regex_match_count(
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	909 const th_regex_char_t *haystack,
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	910 size_t *offs,
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	911 const th_regex_t *expr,
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	912 const th_regex_node_t *node,
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	913 size_t *nnode,
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	914 const int flags,
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	915 const int level
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	916 )
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	917 {
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	918 size_t toffs = *offs, noffs;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	919 BOOL res, match = FALSE;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	920 ssize_t count = 0;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	921
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	922 if (node->repeatMin > 0)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	923 do
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	924 {
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	925 noffs = toffs;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	926 match = th_regex_match_one(haystack, &toffs, node, flags, level);
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	927 if (match)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	928 {
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	929 count++;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	930 }
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	931 else
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	932 toffs = noffs;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	933
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	934 if (node->repeatMin >= 0 &&
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	935 count >= node->repeatMin &&
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	936 node->repeatMax > 0 &&
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	937 count >= node->repeatMax)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	938 break;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	939
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	940 } while (match && toffs > noffs);
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	941
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	942 if (count > 0 \|\| node->repeatMin == 0)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	943 {
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	944 DBG_RE_PRINT("count=%" PRId_SSIZE_T " \"%s\"\n",
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	945 count, haystack + toffs);
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	946
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	947 match = th_regex_match_expr(haystack, &toffs, expr, *nnode + 1, flags, level + 1);
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	948
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	949 DBG_RE_PRINT("rest expr match=%s \"%s\"\n",
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	950 match ? "YES" : "NO", haystack + toffs);
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	951 }
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	952
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	953 if (match)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	954 {
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	955 *offs = toffs;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	956 *nnode = expr->nnodes;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	957 }
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	958
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	959 res = match &&
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	960 (
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	961 (node->repeatMax > 0 && count >= node->repeatMax) \|\|
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	962 (node->repeatMin >= 0 && count >= node->repeatMin)
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	963 );
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	964
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	965 DBG_RE_PRINT("RESULT: match=%s, res=%s\n",
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	966 match ? "YES" : "NO", res ? "YES" : "NO");
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	967
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	968 return res;
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	969 }
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	970
2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	971
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	972 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	973 const th_regex_char_t *haystack,
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	974 size_t *offs,
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	975 const th_regex_t *expr,
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	976 const size_t startnode,
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	977 const int flags,
1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	978 const int level
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	979 )
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	980 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	981 BOOL res = TRUE;
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	982 size_t soffs = *offs;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	983
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	984 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	985 {
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	986 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	987
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	988 #ifdef TH_EXPERIMENTAL_REGEX_DEBUG
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	989 if (th_dbg_fh != NULL)
648 91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	990 {
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	991 th_regex_dump_indent(th_dbg_fh, level);
18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	992
18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	993 thfprintf(th_dbg_fh,
648 91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	994 "[%" PRIu_SIZE_T "/%" PRIu_SIZE_T "] ",
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	995 nnode + 1, expr->nnodes);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	996
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	997 th_regex_dump_node(th_dbg_fh, node);
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	998
651 18fe45e61b2b Moar re-work. Matti Hamalainen <ccr@tnsp.org> parents: 649 diff changeset	999 thfprintf(th_dbg_fh, " <-> \"%s\"\n",
648 91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	1000 haystack + soffs);
91c43398c6fc Twiddle. Matti Hamalainen <ccr@tnsp.org> parents: 647 diff changeset	1001 }
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	1002 #endif
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1003
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1004 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1005 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1006 case TH_RE_MATCH_ONCE:
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	1007 res = th_regex_match_one(haystack, &soffs, node, flags, level);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1008 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1009
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	1010 case TH_RE_MATCH_COUNT:
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	1011 res = th_regex_match_count(haystack, &soffs, expr, node, &nnode, flags, level);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1012 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1013
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1014 case TH_RE_MATCH_ANCHOR_START:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	1015 res = (soffs == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1016 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1017
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1018 case TH_RE_MATCH_ANCHOR_END:
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	1019 res = (haystack[soffs] == 0);
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1020 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1021 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1022 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1023
643 a2bf1ea05b05 Cleanups. Matti Hamalainen <ccr@tnsp.org> parents: 642 diff changeset	1024 if (res)
649 2c9260f5cf44 Tweedle. Matti Hamalainen <ccr@tnsp.org> parents: 648 diff changeset	1025 *offs = soffs;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1026
9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1027 return res;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1028 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1029
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1030
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1031 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	1032 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	1033 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1034 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1035 size_t nmatches = 0;
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	1036 int level = 0;
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1037 (void) flags;
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1038
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1039 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1040 *pnmatches = 0;
605 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 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1043 if (expr == NULL \|\| haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1044 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1045
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1046 // Start matching
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1047 // 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	1048 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1049 // - 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	1050 // ..
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1051 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	1052 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1053 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1054
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	1055 if (th_regex_match_expr(haystack, &coffs, expr, 0, flags, level))
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1056 {
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1057 // 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	1058 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1059
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1060 // Deliver to caller if required
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1061 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1062 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1063
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1064 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1065 {
647 1e7e3f96632e And some more work. Matti Hamalainen <ccr@tnsp.org> parents: 645 diff changeset	1066 // Add the match region to the list
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1067 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	1068 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1069 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1070
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1071 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1072 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1073
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1074 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	1075 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1076
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1077 // 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	1078 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1079 break;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1080
612 cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1081 // If offset was not advanced, increase by one
cc9ec51b4875 Add some comments and debug messages. Matti Hamalainen <ccr@tnsp.org> parents: 611 diff changeset	1082 // 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	1083 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1084 soffs++;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1085 else
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1086 soffs = coffs;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1087 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1088 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1089 {
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1090 soffs++;
605 566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1091 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1092 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1093
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1094 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1095 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression Matti Hamalainen <ccr@tnsp.org> parents: diff changeset	1096
610 a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1097
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1098 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	1099 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1100 (void) node;
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1101 // 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	1102 }
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1103
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1104
640 9e1f9e1d1487 Aaand some more work. Still just a broken concept. Matti Hamalainen <ccr@tnsp.org> parents: 639 diff changeset	1105 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	1106 {
a0e8d9c6300b A bit more work on the regex stuff. Matti Hamalainen <ccr@tnsp.org> parents: 609 diff changeset	1107 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	1108 (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	1109 }

Mercurial > hg > th-libs

annotate th_regex.c @ 652:38a9302962f7