Line | Branch | Exec | Source |
---|---|---|---|
1 | #ifndef REGEXP_H | ||
2 | #define REGEXP_H | ||
3 | |||
4 | #include <regex.h> | ||
5 | #include <stdbool.h> | ||
6 | #include <stddef.h> | ||
7 | #include <stdint.h> | ||
8 | #include "command/error.h" | ||
9 | #include "util/macros.h" | ||
10 | |||
11 | enum { | ||
12 | #ifdef REG_ENHANCED | ||
13 | // The REG_ENHANCED flag enables various extensions on macOS | ||
14 | // (see "enhanced features" in re_format(7)). Most of these | ||
15 | // extensions are enabled by default on Linux (in both glibc | ||
16 | // and musl) without the need for any extra flags. | ||
17 | DEFAULT_REGEX_FLAGS = REG_EXTENDED | REG_ENHANCED, | ||
18 | #else | ||
19 | // POSIX Extended Regular Expressions (ERE) are used almost | ||
20 | // everywhere in this codebase, except where Basic Regular | ||
21 | // Expressions (BRE) are explicitly called for (most notably | ||
22 | // in search_tag(), which is used for ctags patterns). | ||
23 | DEFAULT_REGEX_FLAGS = REG_EXTENDED, | ||
24 | #endif | ||
25 | }; | ||
26 | |||
27 | typedef struct { | ||
28 | const char *str; // Pattern string, interned by str_intern() | ||
29 | regex_t re; // regex(3) object, compiled with regcomp(3) | ||
30 | } InternedRegexp; | ||
31 | |||
32 | // Platform-specific patterns for matching word boundaries, as detected | ||
33 | // and initialized by regexp_init_word_boundary_tokens() | ||
34 | typedef struct { | ||
35 | char start[8]; | ||
36 | char end[8]; | ||
37 | uint8_t len; | ||
38 | } RegexpWordBoundaryTokens; | ||
39 | |||
40 | void regexp_compile_or_fatal_error(regex_t *re, const char *pattern, int flags) NONNULL_ARGS; | ||
41 | bool regexp_init_word_boundary_tokens(RegexpWordBoundaryTokens *rwbt) NONNULL_ARGS; | ||
42 | bool regexp_error_msg(ErrorBuffer *ebuf, const regex_t *re, const char *pattern, int err) NONNULL_ARG(2, 3); | ||
43 | char *regexp_escape(const char *pattern, size_t len) NONNULL_ARGS WARN_UNUSED_RESULT; | ||
44 | size_t regexp_escapeb(char *buf, size_t buflen, const char *pat, size_t plen) NONNULL_ARG(1) NONNULL_ARG_IF_NONZERO_LENGTH(3, 4); | ||
45 | |||
46 | const InternedRegexp *regexp_intern(ErrorBuffer *ebuf, const char *pattern) NONNULL_ARG(2) WARN_UNUSED_RESULT; | ||
47 | bool regexp_is_interned(const char *pattern) NONNULL_ARGS; | ||
48 | void free_interned_regexps(void); | ||
49 | |||
50 | WARN_UNUSED_RESULT NONNULL_ARGS | ||
51 | bool regexp_exec ( | ||
52 | const regex_t *re, | ||
53 | const char *buf, | ||
54 | size_t size, | ||
55 | size_t nmatch, | ||
56 | regmatch_t *pmatch, | ||
57 | int flags | ||
58 | ); | ||
59 | |||
60 | WARN_UNUSED_RESULT NONNULL_ARG(2, 3) | ||
61 | 303 | static inline bool regexp_compile(ErrorBuffer *ebuf, regex_t *re, const char *pattern, int flags) | |
62 | { | ||
63 | 303 | int err = regcomp(re, pattern, flags | DEFAULT_REGEX_FLAGS); | |
64 |
1/4✗ Branch 0 (3→4) not taken.
✓ Branch 1 (3→7) taken 303 times.
✗ Branch 2 (5→6) not taken.
✗ Branch 3 (5→7) not taken.
|
303 | return !err || regexp_error_msg(ebuf, re, pattern, err); |
65 | } | ||
66 | |||
67 | WARN_UNUSED_RESULT NONNULL_ARG(2) | ||
68 | 56 | static inline bool regexp_is_valid(ErrorBuffer *ebuf, const char *pattern, int flags) | |
69 | { | ||
70 | 56 | regex_t re; | |
71 |
1/2✓ Branch 0 (3→4) taken 56 times.
✗ Branch 1 (3→6) not taken.
|
56 | if (!regexp_compile(ebuf, &re, pattern, flags | REG_NOSUB)) { |
72 | return false; | ||
73 | } | ||
74 | 56 | regfree(&re); | |
75 | 56 | return true; | |
76 | } | ||
77 | |||
78 | #endif | ||
79 |