Line | Branch | Exec | Source |
---|---|---|---|
1 | #include <errno.h> | ||
2 | #include <stdlib.h> | ||
3 | #include "regexp.h" | ||
4 | #include "util/arith.h" | ||
5 | #include "util/ascii.h" | ||
6 | #include "util/debug.h" | ||
7 | #include "util/hashmap.h" | ||
8 | #include "util/intern.h" | ||
9 | #include "util/xmalloc.h" | ||
10 | #include "util/xsnprintf.h" | ||
11 | #include "util/xstring.h" | ||
12 | |||
13 | // NOLINTNEXTLINE(*-avoid-non-const-global-variables) | ||
14 | static HashMap interned_regexps = {.flags = HMAP_BORROWED_KEYS}; | ||
15 | |||
16 | ✗ | bool regexp_error_msg(ErrorBuffer *ebuf, const regex_t *re, const char *pattern, int err) | |
17 | { | ||
18 | ✗ | if (!ebuf) { | |
19 | return false; | ||
20 | } | ||
21 | ✗ | char msg[1024]; | |
22 | ✗ | regerror(err, re, msg, sizeof(msg)); | |
23 | ✗ | return error_msg(ebuf, "%s: %s", msg, pattern); | |
24 | } | ||
25 | |||
26 | 3 | void regexp_compile_or_fatal_error(regex_t *re, const char *pattern, int flags) | |
27 | { | ||
28 | // Note: DEFAULT_REGEX_FLAGS isn't used here because this function | ||
29 | // is only used for compiling built-in patterns, where we explicitly | ||
30 | // avoid using "enhanced" features | ||
31 | 3 | int err = regcomp(re, pattern, flags | REG_EXTENDED); | |
32 |
1/2✗ Branch 0 (3→4) not taken.
✓ Branch 1 (3→6) taken 3 times.
|
3 | if (unlikely(err)) { |
33 | ✗ | char msg[1024]; | |
34 | ✗ | regerror(err, re, msg, sizeof(msg)); | |
35 | − | fatal_error(msg, EINVAL); | |
36 | } | ||
37 | 3 | } | |
38 | |||
39 | 143 | bool regexp_exec ( | |
40 | const regex_t *re, | ||
41 | const char *buf, | ||
42 | size_t size, | ||
43 | size_t nmatch, | ||
44 | regmatch_t *pmatch, | ||
45 | int flags | ||
46 | ) { | ||
47 | // ASan's __interceptor_regexec() doesn't support REG_STARTEND | ||
48 | #if defined(REG_STARTEND) && ASAN_ENABLED == 0 && MSAN_ENABLED == 0 | ||
49 | // "If REG_STARTEND is specified, pmatch must point to at least | ||
50 | // one regmatch_t (even if nmatch is 0 or REG_NOSUB was specified), | ||
51 | // to hold the input offsets for REG_STARTEND." | ||
52 | // -- https://man.openbsd.org/regex.3 | ||
53 | 143 | pmatch[0].rm_so = 0; | |
54 | 143 | pmatch[0].rm_eo = size; | |
55 | 143 | return !regexec(re, buf, nmatch, pmatch, flags | REG_STARTEND); | |
56 | #else | ||
57 | // Buffer must be null-terminated if REG_STARTEND isn't supported | ||
58 | char *tmp = xstrcut(buf, size); | ||
59 | int ret = !regexec(re, tmp, nmatch, pmatch, flags); | ||
60 | free(tmp); | ||
61 | return ret; | ||
62 | #endif | ||
63 | } | ||
64 | |||
65 | // Check which word boundary tokens are supported by regcomp(3) | ||
66 | // (if any) and initialize `rwbt` with them for later use | ||
67 | 11 | bool regexp_init_word_boundary_tokens(RegexpWordBoundaryTokens *rwbt) | |
68 | { | ||
69 | 11 | static const char text[] = "SSfooEE SSfoo fooEE foo SSfooEE"; | |
70 | 11 | const regoff_t match_start = 20, match_end = 23; | |
71 | 11 | static const RegexpWordBoundaryTokens pairs[] = { | |
72 | {"\\<", "\\>", 2}, | ||
73 | {"[[:<:]]", "[[:>:]]", 7}, | ||
74 | {"\\b", "\\b", 2}, | ||
75 | }; | ||
76 | |||
77 | 11 | BUG_ON(ARRAYLEN(text) <= match_end); | |
78 | 11 | BUG_ON(!mem_equal(text + match_start - 1, " foo ", 5)); | |
79 | |||
80 |
1/2✓ Branch 0 (17→5) taken 11 times.
✗ Branch 1 (17→18) not taken.
|
11 | for (size_t i = 0; i < ARRAYLEN(pairs); i++) { |
81 | 11 | const RegexpWordBoundaryTokens *p = &pairs[i]; | |
82 | 11 | char patt[32]; | |
83 | 11 | xmempcpy3(patt, p->start, p->len, STRN("(foo)"), p->end, p->len + 1); | |
84 | 11 | regex_t re; | |
85 |
1/2✗ Branch 0 (7→8) not taken.
✓ Branch 1 (7→9) taken 11 times.
|
11 | if (regcomp(&re, patt, DEFAULT_REGEX_FLAGS) != 0) { |
86 | ✗ | continue; | |
87 | } | ||
88 | 11 | regmatch_t m[2]; | |
89 | 11 | bool match = !regexec(&re, text, ARRAYLEN(m), m, 0); | |
90 | 11 | regfree(&re); | |
91 |
3/6✓ Branch 0 (11→12) taken 11 times.
✗ Branch 1 (11→15) not taken.
✓ Branch 2 (12→13) taken 11 times.
✗ Branch 3 (12→15) not taken.
✓ Branch 4 (13→14) taken 11 times.
✗ Branch 5 (13→15) not taken.
|
11 | if (match && m[0].rm_so == match_start && m[0].rm_eo == match_end) { |
92 | 11 | *rwbt = pairs[i]; | |
93 | 11 | return true; | |
94 | } | ||
95 | } | ||
96 | |||
97 | return false; | ||
98 | } | ||
99 | |||
100 | 1 | size_t regexp_escapeb(char *buf, size_t buflen, const char *pat, size_t plen) | |
101 | { | ||
102 | 1 | BUG_ON(buflen < (2 * plen) + 1); | |
103 | size_t o = 0; | ||
104 |
2/2✓ Branch 0 (7→4) taken 27 times.
✓ Branch 1 (7→8) taken 1 times.
|
28 | for (size_t i = 0; i < plen; i++) { |
105 | 27 | char ch = pat[i]; | |
106 |
2/2✓ Branch 0 (4→5) taken 15 times.
✓ Branch 1 (4→6) taken 12 times.
|
27 | if (is_regex_special_char(ch)) { |
107 | 15 | buf[o++] = '\\'; | |
108 | } | ||
109 | 27 | buf[o++] = ch; | |
110 | } | ||
111 | 1 | buf[o] = '\0'; | |
112 | 1 | return o; | |
113 | } | ||
114 | |||
115 | 1 | char *regexp_escape(const char *pattern, size_t len) | |
116 | { | ||
117 | 1 | size_t buflen = xmul(2, len) + 1; | |
118 | 1 | char *buf = xmalloc(buflen); | |
119 | 1 | regexp_escapeb(buf, buflen, pattern, len); | |
120 | 1 | return buf; | |
121 | } | ||
122 | |||
123 | 33 | const InternedRegexp *regexp_intern(ErrorBuffer *ebuf, const char *pattern) | |
124 | { | ||
125 |
1/2✓ Branch 0 (2→3) taken 33 times.
✗ Branch 1 (2→14) not taken.
|
33 | if (pattern[0] == '\0') { |
126 | return NULL; | ||
127 | } | ||
128 | |||
129 | 33 | InternedRegexp *ir = hashmap_get(&interned_regexps, pattern); | |
130 |
2/2✓ Branch 0 (4→5) taken 32 times.
✓ Branch 1 (4→14) taken 1 times.
|
33 | if (ir) { |
131 | return ir; | ||
132 | } | ||
133 | |||
134 | 32 | ir = xmalloc(sizeof(*ir)); | |
135 | 32 | int err = regcomp(&ir->re, pattern, DEFAULT_REGEX_FLAGS | REG_NEWLINE | REG_NOSUB); | |
136 |
1/2✗ Branch 0 (7→8) not taken.
✓ Branch 1 (7→10) taken 32 times.
|
32 | if (unlikely(err)) { |
137 | ✗ | regexp_error_msg(ebuf, &ir->re, pattern, err); | |
138 | ✗ | free(ir); | |
139 | ✗ | return NULL; | |
140 | } | ||
141 | |||
142 | 32 | BUG_ON(!(interned_regexps.flags & HMAP_BORROWED_KEYS)); | |
143 | 32 | const char *str = str_intern(pattern); | |
144 | 32 | ir->str = str; | |
145 | 32 | return hashmap_insert(&interned_regexps, (char*)str, ir); | |
146 | } | ||
147 | |||
148 | 58 | bool regexp_is_interned(const char *pattern) | |
149 | { | ||
150 | 58 | return !!hashmap_find(&interned_regexps, pattern); | |
151 | } | ||
152 | |||
153 | 32 | static void free_interned_regexp(InternedRegexp *ir) | |
154 | { | ||
155 | 32 | regfree(&ir->re); | |
156 | 32 | free(ir); | |
157 | 32 | } | |
158 | |||
159 | 11 | void free_interned_regexps(void) | |
160 | { | ||
161 | 11 | BUG_ON(!(interned_regexps.flags & HMAP_BORROWED_KEYS)); | |
162 | 11 | hashmap_free(&interned_regexps, FREE_FUNC(free_interned_regexp)); | |
163 | 11 | } | |
164 |