annotate th_regex.c @ 611:d895b0fd6ad6

Combine code from th_regex_compile() to th_regex_compile_do().
author Matti Hamalainen <ccr@tnsp.org>
date Thu, 16 Jan 2020 12:50:09 +0200
parents a0e8d9c6300b
children cc9ec51b4875
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
1 /*
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
2 * Simple regular expression matching functionality
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
3 * Programmed and designed by Matti 'ccr' Hamalainen
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
4 * (C) Copyright 2020 Tecnic Software productions (TNSP)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
5 *
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
6 * Please read file 'COPYING' for information on license and distribution.
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
7 */
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
8 #ifdef TH_EXPERIMENTAL_REGEX
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
9 #include "th_util.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
10 #include "th_regex.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
11 //#include "th_datastruct.h"
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
12
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
13
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
14 //#define DBG_RE_COMPILE 1
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
15 //#define DBG_RE_FREE 1
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
16 //#define DBG_RE_MATCH 1
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
17
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
18 #if defined(DBG_RE_COMPILE)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
19 # define DBG_RE_PRINT_COMPILE(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
20 #else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
21 # define DBG_RE_PRINT_COMPILE(...) do { } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
22 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
23 #if defined(DBG_RE_FREE)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
24 # define DBG_RE_PRINT_FREE(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
25 #else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
26 # define DBG_RE_PRINT_FREE(...) do { } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
27 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
28 #if defined(DBG_RE_MATCH)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
29 # define DBG_RE_PRINT_MATCH(...) do { fprintf(stderr, __VA_ARGS__); } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
30 #else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
31 # define DBG_RE_PRINT_MATCH(...) do { } while (0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
32 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
33
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
34
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
35 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
36 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
37 TH_RE_MATCH_ONCE,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
38 TH_RE_MATCH_COUNT,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
39 TH_RE_MATCH_ANCHOR_START,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
40 TH_RE_MATCH_ANCHOR_END,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
41 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
42
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
43
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
44 enum
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
45 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
46 TH_RE_TYPE_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
47 TH_RE_TYPE_ANY_CHAR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
48 TH_RE_TYPE_LIST,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
49 TH_RE_TYPE_LIST_REVERSE,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50 TH_RE_TYPE_SUBEXPR,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
52
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
54 #if defined(DBG_RE_COMPILE) || defined(DBG_RE_FREE) || defined(DBG_RE_MATCH)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
55 static const char *re_match_modes[] =
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
56 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
57 "ONCE",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
58 "COUNT",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
59 "ANCHOR_START",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
60 "ANCHOR_END",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
61 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
62
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
63 static const char *re_match_types[] =
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
64 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
65 "CHAR",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
66 "ANY",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
67 "LIST",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
68 "LIST_REVERSE",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
69 "SUBEXPR",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70 };
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
71 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
72
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
73
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
74 typedef struct
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
75 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
76 const th_regex_char *pattern;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
77 size_t offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
78
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 th_regex_ctx *data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
80
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
81 size_t nstack, stacksize;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82 th_regex_ctx **stack;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
83 } th_regex_parse_ctx;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
84
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
85
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
86 static void th_regex_node_init(th_regex_node *node)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
87 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
88 memset(node, 0, sizeof(th_regex_node));
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
89 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
90 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
91
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
92
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
93 static int th_regex_strndup(th_regex_char **pdst,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
94 const th_regex_char *src, const size_t len)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
95 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
96 if (pdst == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
97 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
98
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
99 if (UINTPTR_MAX / sizeof(th_regex_char) < len + 1)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
100 return THERR_BOUNDS;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
101
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
102 if ((*pdst = (th_regex_char *)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
103 th_malloc((len + 1) * sizeof(th_regex_char))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
104 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
105
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
106 memcpy(*pdst, src, len * sizeof(th_regex_char));
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
107 (*pdst)[len] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
108
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
109 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
110 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
111
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
112
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
113 static int th_regex_ctx_get_prev_node(
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
114 th_regex_parse_ctx *ctx, th_regex_node **pnode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
115 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
116 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
117 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
118 *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
119 return THERR_OK;
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
122 return THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
123 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
124
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
125
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
126 static int th_regex_ctx_push(th_regex_parse_ctx *ctx)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
127 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
128 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
129 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
130 ctx->stacksize += 16;
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 ((ctx->stack = th_realloc(ctx->stack,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
133 ctx->stacksize * sizeof(th_regex_node *))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
134 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
135 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
136
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
137 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
138 ctx->nstack++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
139 ctx->data = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
140
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
141 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
142 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
143
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 static int th_regex_ctx_pop(th_regex_parse_ctx *ctx, th_regex_ctx **data)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
146 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
147 if (ctx->nstack > 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
148 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
149 *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
150 ctx->nstack--;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
151 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
152 return THERR_OK;
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
155 return THERR_INVALID_DATA;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
158
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
159 static int th_regex_ctx_node_commit(th_regex_parse_ctx *ctx, th_regex_node *node)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
160 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
161 th_regex_ctx *data = ctx->data;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
162
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
163 if (data == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
164 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
165 if ((data = ctx->data = th_malloc0(sizeof(th_regex_ctx))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
166 return THERR_MALLOC;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
169 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
170 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
171 data->nodessize += 16;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
172 if ((data->nodes = th_realloc(data->nodes,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
173 data->nodessize * sizeof(th_regex_node))) == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
174 return THERR_MALLOC;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
175 }
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 memcpy(&data->nodes[data->nnodes], node, sizeof(th_regex_node));
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
178 data->nnodes++;
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 DBG_RE_PRINT_COMPILE(
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
181 "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
182 "mode=%d, type=%d, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
183 data->nnodes, data->nodessize,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
184 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
185 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
186
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
187 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
188 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
189
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
190
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
191 static BOOL th_regex_find_next(const th_regex_char *str,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
192 const size_t start, size_t *offs,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
193 const th_regex_char delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
194 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
195 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
196 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
197 if (str[*offs] == delim)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
198 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
199 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
200 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
201 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
202
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 static BOOL th_regex_parse_ssize_t(const th_regex_char *str,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
205 ssize_t *value)
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 th_regex_char ch;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
208 BOOL neg;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
209
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 if (*str == '-')
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 str++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
213 neg = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
214 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
215 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
216 neg = FALSE;
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 // Is the value negative?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
219 while ((ch = *str++))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
220 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
221 if (ch >= '0' && ch <= '9')
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
222 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
223 *value *= 10;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
224 *value += ch - '0';
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
227 return FALSE;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
230 if (neg)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
231 *value = -(*value);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
232
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
233 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
234 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
235
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
236
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
237 int th_regex_compile(th_regex_ctx **pexpr, const th_regex_char *pattern)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
238 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
239 int res = THERR_OK;
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
240 th_regex_parse_ctx ctx;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
241 th_regex_node node, *pnode;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
242 th_regex_char *tmp = NULL;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
243 size_t start;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
244
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
245 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
246 {
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
247 res = THERR_NULLPTR;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
248 goto exit;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
249 }
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
250
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
251 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
252 ctx.pattern = pattern;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
253
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
254 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
255 {
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
256 DBG_RE_PRINT_COMPILE("[%" PRIu_SIZE_T "] '%c'\n", ctx.offs, ctx.pattern[ctx.offs]);
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
257 switch (ctx.pattern[ctx.offs])
605
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 case '?':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
260 // Previous token is optional (repeat 0-1 times)
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
261 if ((res = th_regex_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
262 goto exit;
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 pnode->mode = TH_RE_MATCH_COUNT;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
265 pnode->repeatMin = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
266 pnode->repeatMax = 1;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
267 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
268
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
269 case '*':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
270 // Previous token can repeat 0 or more times
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
271 if ((res = th_regex_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
272 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
273
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
274 pnode->mode = TH_RE_MATCH_COUNT;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
275 pnode->repeatMin = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
276 pnode->repeatMax = -1;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
277 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
278
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
279 case '+':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
280 // Previous token must repeat 1 or more times
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
281 if ((res = th_regex_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
282 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
283
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
284 pnode->mode = TH_RE_MATCH_COUNT;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
285 pnode->repeatMin = 1;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
286 pnode->repeatMax = 1;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
287 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
288
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
289 case '{':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
290 // {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
291 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
292 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
293 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
294 // End not found
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
295 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
296 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
297 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
298
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
299 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
300
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
301 if ((res = th_regex_ctx_get_prev_node(&ctx, &pnode)) != THERR_OK ||
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
302 (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
303 ctx.offs - start)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
304 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
305
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
306 pnode->mode = TH_RE_MATCH_COUNT;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
307
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
308 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
309 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
310 tmp[start] = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
311 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
312 !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
313 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
314 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
315 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
316 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
317 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
318 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
319 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
320 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
321 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
322 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
323 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
324 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
325 pnode->repeatMax = pnode->repeatMin;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
326 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
327
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
328 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
329 pnode->repeatMax < pnode->repeatMin)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
330 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
331 // Invalid repeat counts
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
332 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
333 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
334 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
335 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
336
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
337 case '(':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
338 // Start of subpattern
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
339 if ((res = th_regex_ctx_push(&ctx)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
340 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
341 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
342
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
343 case ')':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
344 // End of subpattern
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
345 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
346 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
347
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
348 if ((res = th_regex_ctx_pop(&ctx, &node.match.expr)) != THERR_OK ||
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
349 (res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
350 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
351 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
352
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
353 case '^':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
354 // Start of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
355 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
356 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
357
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
358 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
359 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
360 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
361
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
362 case '$':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
363 // End of line anchor
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
364 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
365 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
366
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
367 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
368 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
369 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
370
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
371 case '[':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
372 // 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
373 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
374 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
375 ctx.offs == start)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
376 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
377 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
378 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
379 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
380
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
381 th_regex_node_init(&node);
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
382 node.type = (ctx.pattern[start] == '^') ?
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
383 TH_RE_TYPE_LIST_REVERSE : TH_RE_TYPE_LIST;
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
384 node.match.list.nchars = ctx.offs - start;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
385 if ((res = th_regex_strndup(&node.match.list.chars,
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
386 ctx.pattern + start, node.match.list.nchars)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
387 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
388
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
389 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
390 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
391 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
392
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
393 case '.':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
394 // Any single character matches
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
395 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
396 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
397
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
398 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
399 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
400 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
401
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
402 case '\\':
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
403 // Literal escape
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
404 ctx.offs++;
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
405 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
406 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
407 // End of pattern, error
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
408 res = THERR_INVALID_DATA;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
409 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
410 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
411
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
412 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
413 node.type = TH_RE_TYPE_CHAR;
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
414 node.match.chr = ctx.pattern[ctx.offs];
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
415
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
416 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
417 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
418 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
419
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
420 default:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
421 // Given character must match
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
422 th_regex_node_init(&node);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
423 node.type = TH_RE_TYPE_CHAR;
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
424 node.match.chr = ctx.pattern[ctx.offs];
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
425
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
426 if ((res = th_regex_ctx_node_commit(&ctx, &node)) != THERR_OK)
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
427 goto exit;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
428 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
429 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
430 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
431
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
432 exit:
611
d895b0fd6ad6 Combine code from th_regex_compile() to th_regex_compile_do().
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
433 *pexpr = ctx.data;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
434 th_free(tmp);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
435 return res;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
436 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
437
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
438
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
439
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
440
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
441 void th_regex_free(th_regex_ctx *expr)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
442 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
443 if (expr != NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
444 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
445 for (size_t n = 0; n < expr->nnodes; n++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
446 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
447 th_regex_node *node = &expr->nodes[n];
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
448
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
449 DBG_RE_PRINT_FREE(
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
450 "node [%" PRIu_SIZE_T " / %" PRIu_SIZE_T "]: "
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
451 "mode=%d, type=%d, rmin=%" PRId_SSIZE_T ", rmax=%" PRId_SSIZE_T "\n",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
452 n, expr->nnodes,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
453 node->mode, node->type,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
454 node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
455
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
456 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
457 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
458 case TH_RE_TYPE_SUBEXPR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
459 DBG_RE_PRINT_FREE(" SUBEXPR: %p vs %p\n", expr, node->match.expr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
460 th_regex_free(node->match.expr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
461 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
462
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
463 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
464 case TH_RE_TYPE_LIST_REVERSE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
465 DBG_RE_PRINT_FREE(" list='%s'\n", node->match.list.chars);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
466 th_free(node->match.list.chars);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
467 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
468
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
469 #ifdef DBG_RE_FREE
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
470 case TH_RE_TYPE_ANY_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
471 DBG_RE_PRINT_FREE(" any char\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
472 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
473
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
474 case TH_RE_TYPE_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
475 DBG_RE_PRINT_FREE(" char='%c'\n", node->match.chr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
476 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
477 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
478 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
479 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
480
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
481 th_free(expr->nodes);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
482 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
483 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
484
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
485
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
486 static BOOL th_regex_do_match_expr(const th_regex_ctx *expr,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
487 const th_regex_char *haystack, size_t *offs, const int flags);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
488
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
489
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
490 static BOOL th_regex_do_match_node(const th_regex_char *haystack,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
491 size_t *offs, const th_regex_node *node, const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
492 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
493 th_regex_char cch;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
494 BOOL ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
495
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
496 DBG_RE_PRINT_MATCH("node_START [%s]: '%s': ", re_match_types[node->type], haystack + *offs);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
497
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
498 switch (node->type)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
499 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
500 case TH_RE_TYPE_SUBEXPR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
501 DBG_RE_PRINT_MATCH("subexpr ..\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
502 ret = th_regex_do_match_expr(node->match.expr, haystack, offs, flags);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
503 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
504
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
505 case TH_RE_TYPE_LIST:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
506 case TH_RE_TYPE_LIST_REVERSE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
507 DBG_RE_PRINT_MATCH("[%s]\n", node->match.list.chars);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
508 ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
509 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
510 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
511
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
512 // Could be optimized, perhaps .. sort match.chars, binary search etc?
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
513 // XXX TODO Ranges and escapes are not supported yet
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
514 for (size_t n = 0; n < node->match.list.nchars; n++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
515 if (node->match.list.chars[n] == cch)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
516 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
517 ret = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
518 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
519 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
520
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
521 if (node->type == TH_RE_TYPE_LIST_REVERSE)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
522 ret = !ret;
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 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
525 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
526
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
527 case TH_RE_TYPE_ANY_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
528 DBG_RE_PRINT_MATCH("\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
529 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
530 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
531 ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
532 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
533 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
534
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
535 ret = TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
536 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
537 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
538
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
539 case TH_RE_TYPE_CHAR:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
540 DBG_RE_PRINT_MATCH("'%c'\n", node->match.chr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
541 if ((cch = haystack[*offs]) == 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
542 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
543 ret = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
544 goto out;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
545 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
546
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
547 ret = (cch == node->match.chr);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
548 (*offs)++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
549 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
550 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
551
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
552 out:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
553 DBG_RE_PRINT_MATCH("node_DONE [%s]: match %s\n", re_match_types[node->type], ret ? "YES" : "NO");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
554 return ret;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
555 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
556
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 static BOOL th_regex_do_match_expr(const th_regex_ctx *expr,
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
559 const th_regex_char *haystack, size_t *offs, const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
560 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
561 for (size_t n = 0; n < expr->nnodes; n++)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
562 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
563 const th_regex_node *node = &expr->nodes[n];
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
564 size_t soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
565
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
566 DBG_RE_PRINT_MATCH("expr %p [%" PRIu_SIZE_T "/%" PRIu_SIZE_T "]: %s ",
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
567 (void *) expr, n, expr->nnodes, re_match_modes[node->mode]);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
568
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
569 switch (node->mode)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
570 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
571 case TH_RE_MATCH_ONCE:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
572 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
573 DBG_RE_PRINT_MATCH("\n");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
574 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
575 if (!th_regex_do_match_node(haystack, &soffs, node, flags))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
576 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
577 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
578 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
579 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
580
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
581 case TH_RE_MATCH_COUNT:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
582 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
583 BOOL done = FALSE, match = FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
584 ssize_t count = 0;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
585
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
586 DBG_RE_PRINT_MATCH("min=%" PRId_SSIZE_T ", max=%" PRId_SSIZE_T "\n", node->repeatMin, node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
587
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
588 do
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
589 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
590 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
591 if ((match = th_regex_do_match_node(haystack, &soffs, node, flags)))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
592 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
593 // Node matched
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
594 count++;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
595 done = (node->repeatMax > 0 && count >= node->repeatMax);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
596 *offs = soffs;
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 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
599 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
600 // Node did not match, check if we got the minimum if set
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
601 done = (node->repeatMin >= 0 && count >= node->repeatMin);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
602 soffs = *offs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
603 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
604 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
605 } while (!done);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
606
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
607 DBG_RE_PRINT_MATCH("RESULT: %" PRId_SSIZE_T " = %s\n", count, done ? "YES" : "NO");
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
608
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
609 if (!done)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
610 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
611
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
612 *offs = soffs;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
613 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
614 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
615
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
616 case TH_RE_MATCH_ANCHOR_START:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
617 DBG_RE_PRINT_MATCH("offs=%" PRIu_SIZE_T "\n", *offs);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
618 if (*offs != 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
619 return FALSE;
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 TH_RE_MATCH_ANCHOR_END:
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
623 DBG_RE_PRINT_MATCH("is end=%d\n", haystack[*offs]);
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
624 if (haystack[*offs] != 0)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
625 return FALSE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
626 break;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
627 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
628 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
629
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
630 return TRUE;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
631 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
632
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
633
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
634 int th_regex_match(const th_regex_ctx *expr, const th_regex_char *haystack,
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
635 size_t *pnmatches, th_regex_match_node **pmatches, const size_t maxmatches,
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
636 const int flags)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
637 {
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
638 size_t nmatches = 0;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
639
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
640 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
641 *pnmatches = 0;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
642
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
643 // Check given pattern and string
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
644 if (expr == NULL || haystack == NULL)
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
645 return THERR_NULLPTR;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
646
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
647 // Start matching
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
648 // 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
649 // - anchored and unanchored expressions
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
650 // - 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
651 // ..
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
652 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
653 {
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
654 BOOL matched;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
655 size_t coffs = soffs;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
656
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
657 if ((matched = th_regex_do_match_expr(expr, haystack, &coffs, flags)))
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
658 {
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
659 nmatches++;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
660
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
661 if (pnmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
662 *pnmatches = nmatches;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
663
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
664 if (pmatches != NULL)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
665 {
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
666 th_regex_match_node *match = th_malloc0(sizeof(th_regex_match_node));
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
667 if (match == NULL)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
668 return THERR_MALLOC;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
669
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
670 match->start = soffs;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
671 match->len = coffs - soffs;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
672
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
673 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
674 }
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
675
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
676 if (maxmatches > 0 && nmatches >= maxmatches)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
677 break;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
678
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
679 if (soffs == coffs)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
680 soffs++;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
681 else
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
682 soffs = coffs;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
683 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
684 else
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
685 {
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
686 soffs++;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
687 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
688 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
689
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
690 return THERR_OK;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
691 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
692
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
693
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
694 static void th_regex_free_match(th_regex_match_node *node)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
695 {
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
696 (void) node;
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
697 // 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
698 }
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
699
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
700
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
701 void th_regex_free_matches(th_regex_match_node *matches)
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
702 {
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
703 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
704 (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
705 }
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 609
diff changeset
706
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
707 #endif // TH_EXPERIMENTAL_REGEX