Mercurial > hg > th-libs

annotate th_regex.h @ 639:8c957ad9d4c3

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Some more work on regex stuff.
author Matti Hamalainen <ccr@tnsp.org>
date Thu, 23 Jan 2020 11:38:28 +0200
parents d191ded8a790
children 9e1f9e1d1487
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
10 #ifndef TH_REGEX_H
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
11 #define TH_REGEX_H
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
12
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
13 #include "th_types.h"
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
14 #include "th_datastruct.h"
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
15
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
16
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
17 #ifdef __cplusplus
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
18 extern "C" {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
19 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
20
639
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
21
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
22 //
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
23 // Definitions
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
24 //
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
25 // Character type
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
26 typedef char th_regex_char;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
27
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
28
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
29 enum
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
30 {
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
31 TH_REF_CASEFOLD = 0x0001,
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
32 TH_REF_ANCHORED = 0x0002,
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
33 };
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
34
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
35
639
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
36 typedef struct
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
37 {
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
38 int type;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
39 th_regex_char start, end;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
40
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
41 size_t nchars;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
42 th_regex_char *chars;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
43 } th_regex_list_item;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
44
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
45
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
46 typedef struct
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
47 {
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
48 size_t nitems, itemssize;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
49 th_regex_list_item *items;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
50 } th_regex_list;
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
51
8c957ad9d4c3 Some more work on regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 635
diff changeset
52
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53 struct th_regex_ctx;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
54
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
55 typedef struct
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 int mode, type;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
58 ssize_t repeatMin, repeatMax;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
59
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
60 union
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 th_regex_char chr;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
63
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
64 struct
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
65 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
66 size_t nchars;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
67 th_regex_char *chars;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
68 } list;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
69
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70 struct th_regex_ctx *expr;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
71 } match;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
72 } th_regex_node;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
75 typedef struct th_regex_ctx
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
76 {
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
77 size_t nnodes, nodessize;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
78 th_regex_node *nodes;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 } th_regex_ctx;
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
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82 typedef struct
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
83 {
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
84 th_llist_t node;
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
85 size_t start, len;
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
86 } th_regex_match_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
635
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
89 #ifdef TH_EXPERIMENTAL_REGEX_DEBUG
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
90 enum
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
91 {
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
92 TH_DBG_RE_COMPILE = 0x0001,
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
93 TH_DBG_RE_FREE = 0x0002,
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
94 TH_DBG_RE_MATCH = 0x0004,
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
95 };
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
96
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
97 extern int th_dbg_re_flags;
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
98 #endif
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
99
d191ded8a790 Improve the experimental regex matching debugging macros.
Matti Hamalainen <ccr@tnsp.org>
parents: 610
diff changeset
100
605
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 // Functions
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
103 //
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
104 int th_regex_compile(th_regex_ctx **pexpr, const th_regex_char *pattern);
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
105 void th_regex_free(th_regex_ctx *expr);
605
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
106
610
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
107 int th_regex_match(const th_regex_ctx *expr, const th_regex_char *haystack,
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
108 size_t *pnmatches, th_regex_match_node **pmatches, const size_t maxmatches,
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
109 const int flags);
a0e8d9c6300b A bit more work on the regex stuff.
Matti Hamalainen <ccr@tnsp.org>
parents: 605
diff changeset
110 void th_regex_free_matches(th_regex_match_node *matches);
605
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 #ifdef __cplusplus
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
114 }
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
115 #endif
566e6ef41f9d Initial commit of the highly experimental and unfinished regular expression
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
116 #endif // TH_REGEX_H
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 #endif // TH_EXPERIMENTAL_REGEX