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