| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include <regex.h> | ||
| 2 | #include <stdlib.h> | ||
| 3 | #include "match.h" | ||
| 4 | #include "util/ascii.h" | ||
| 5 | #include "util/debug.h" | ||
| 6 | #include "util/string.h" | ||
| 7 | #include "util/xstring.h" | ||
| 8 | |||
| 9 | 58 | static size_t get_last_paired_brace_index(const char *str, size_t len) | |
| 10 | { | ||
| 11 | 58 | size_t last_paired_index = 0; | |
| 12 | 58 | size_t open_braces = 0; | |
| 13 |
2/2✓ Branch 0 (9→3) taken 973 times.
✓ Branch 1 (9→10) taken 58 times.
|
1031 | for (size_t i = 0; i < len; i++) { |
| 14 | 973 | const char ch = str[i]; | |
| 15 |
4/4✓ Branch 0 (3→4) taken 8 times.
✓ Branch 1 (3→5) taken 48 times.
✓ Branch 2 (3→6) taken 49 times.
✓ Branch 3 (3→8) taken 868 times.
|
973 | switch (ch) { |
| 16 | 8 | case '\\': | |
| 17 | 8 | i++; | |
| 18 | 8 | break; | |
| 19 | 48 | case '{': | |
| 20 | 48 | open_braces++; | |
| 21 |
1/2✓ Branch 0 (5→8) taken 48 times.
✗ Branch 1 (5→12) not taken.
|
48 | if (open_braces >= 32) { |
| 22 | // If nesting goes too deep, just return 0 and let | ||
| 23 | // ec_pattern_match() escape all braces | ||
| 24 | return 0; | ||
| 25 | } | ||
| 26 | break; | ||
| 27 | 49 | case '}': | |
| 28 |
2/2✓ Branch 0 (6→7) taken 47 times.
✓ Branch 1 (6→8) taken 2 times.
|
49 | if (open_braces != 0) { |
| 29 | 47 | last_paired_index = i; | |
| 30 | 47 | open_braces--; | |
| 31 | } | ||
| 32 | break; | ||
| 33 | } | ||
| 34 | } | ||
| 35 | |||
| 36 | // If there are unclosed braces, just return 0 | ||
| 37 |
2/2✓ Branch 0 (10→11) taken 1 times.
✓ Branch 1 (10→12) taken 57 times.
|
58 | return open_braces ? 0 : last_paired_index; |
| 38 | } | ||
| 39 | |||
| 40 | 16 | static size_t handle_bracket_expression(const char *pat, size_t len, String *buf) | |
| 41 | { | ||
| 42 | 16 | BUG_ON(len == 0); | |
| 43 |
2/2✓ Branch 0 (4→5) taken 1 times.
✓ Branch 1 (4→7) taken 15 times.
|
16 | if (len == 1) { |
| 44 | 1 | string_append_literal(buf, "\\["); | |
| 45 | 1 | return 0; | |
| 46 | } | ||
| 47 | |||
| 48 | // Skip past opening bracket | ||
| 49 | 15 | pat++; | |
| 50 | |||
| 51 | 15 | bool closed = false; | |
| 52 | 15 | size_t i = 0; | |
| 53 |
2/2✓ Branch 0 (9→8) taken 54 times.
✓ Branch 1 (9→10) taken 3 times.
|
57 | while (i < len) { |
| 54 | 54 | const char ch = pat[i++]; | |
| 55 |
2/2✓ Branch 0 (8→9) taken 42 times.
✓ Branch 1 (8→10) taken 12 times.
|
54 | if (ch == ']') { |
| 56 | closed = true; | ||
| 57 | break; | ||
| 58 | } | ||
| 59 | } | ||
| 60 | |||
| 61 |
2/2✓ Branch 0 (10→11) taken 3 times.
✓ Branch 1 (10→13) taken 12 times.
|
15 | if (!closed) { |
| 62 | 3 | string_append_literal(buf, "\\["); | |
| 63 | 3 | return 0; | |
| 64 | } | ||
| 65 | |||
| 66 | // TODO: interpret characters according to editorconfig instead | ||
| 67 | // of just copying the bracket expression to be interpreted as | ||
| 68 | // regex | ||
| 69 | |||
| 70 | 12 | char *s = string_reserve_space(buf, i + 1); | |
| 71 | 12 | s[0] = '['; | |
| 72 | 12 | memcpy(s + 1, pat, i); | |
| 73 |
2/2✓ Branch 0 (14→15) taken 10 times.
✓ Branch 1 (14→16) taken 2 times.
|
12 | s[1] = (s[1] == '!') ? '^' : s[1]; // Replace any leading ! with ^ |
| 74 | 12 | buf->len += i + 1; | |
| 75 | 12 | return i; | |
| 76 | } | ||
| 77 | |||
| 78 | // Skips past empty alternates in brace groups and returns the number | ||
| 79 | // of bytes (commas) skipped | ||
| 80 | 88 | static size_t skip_empty_alternates(const char *str, size_t len) | |
| 81 | { | ||
| 82 | 88 | size_t i = 1; | |
| 83 |
3/4✓ Branch 0 (4→5) taken 102 times.
✗ Branch 1 (4→6) not taken.
✓ Branch 2 (5→3) taken 14 times.
✓ Branch 3 (5→6) taken 88 times.
|
102 | while (i < len && str[i] == ',') { |
| 84 | 14 | i++; | |
| 85 | } | ||
| 86 | 88 | return i - 1; | |
| 87 | } | ||
| 88 | |||
| 89 | 58 | bool ec_pattern_match(const char *pattern, size_t pattern_len, const char *path) | |
| 90 | { | ||
| 91 | 58 | String buf = string_new(pattern_len * 2); | |
| 92 | 58 | size_t brace_level = 0; | |
| 93 | 58 | size_t last_paired_brace_index = get_last_paired_brace_index(pattern, pattern_len); | |
| 94 | 58 | bool brace_group_has_empty_alternate[32] = {false}; | |
| 95 | |||
| 96 |
2/2✓ Branch 0 (52→5) taken 901 times.
✓ Branch 1 (52→53) taken 58 times.
|
959 | for (size_t i = 0; i < pattern_len; i++) { |
| 97 | 901 | char ch = pattern[i]; | |
| 98 |
10/10✓ Branch 0 (5→6) taken 8 times.
✓ Branch 1 (5→11) taken 4 times.
✓ Branch 2 (5→13) taken 45 times.
✓ Branch 3 (5→18) taken 16 times.
✓ Branch 4 (5→20) taken 48 times.
✓ Branch 5 (5→29) taken 48 times.
✓ Branch 6 (5→35) taken 45 times.
✓ Branch 7 (5→43) taken 61 times.
✓ Branch 8 (5→49) taken 53 times.
✓ Branch 9 (5→50) taken 573 times.
|
901 | switch (ch) { |
| 99 | 8 | case '\\': | |
| 100 |
2/2✓ Branch 0 (6→7) taken 6 times.
✓ Branch 1 (6→10) taken 2 times.
|
8 | if (i + 1 < pattern_len) { |
| 101 | 6 | ch = pattern[++i]; | |
| 102 |
1/2✓ Branch 0 (7→8) taken 6 times.
✗ Branch 1 (7→9) not taken.
|
6 | if (is_regex_special_char(ch)) { |
| 103 | 6 | string_append_byte(&buf, '\\'); | |
| 104 | } | ||
| 105 | 6 | string_append_byte(&buf, ch); | |
| 106 | } else { | ||
| 107 | 2 | string_append_literal(&buf, "\\\\"); | |
| 108 | } | ||
| 109 | break; | ||
| 110 | 4 | case '?': | |
| 111 | 4 | string_append_literal(&buf, "[^/]"); | |
| 112 | 4 | break; | |
| 113 | 45 | case '*': | |
| 114 |
4/4✓ Branch 0 (13→14) taken 42 times.
✓ Branch 1 (13→17) taken 3 times.
✓ Branch 2 (14→15) taken 10 times.
✓ Branch 3 (14→17) taken 32 times.
|
45 | if (i + 1 < pattern_len && pattern[i + 1] == '*') { |
| 115 | 10 | string_append_literal(&buf, ".*"); | |
| 116 | 10 | i++; | |
| 117 | } else { | ||
| 118 | 35 | string_append_literal(&buf, "[^/]*"); | |
| 119 | } | ||
| 120 | break; | ||
| 121 | 16 | case '[': | |
| 122 | // The entire bracket expression is handled in a separate | ||
| 123 | // loop because the POSIX regex escaping rules are different | ||
| 124 | // in that context | ||
| 125 | 16 | i += handle_bracket_expression(pattern + i, pattern_len - i, &buf); | |
| 126 | 16 | break; | |
| 127 | 48 | case '{': { | |
| 128 |
2/2✓ Branch 0 (20→21) taken 3 times.
✓ Branch 1 (20→23) taken 45 times.
|
48 | if (i >= last_paired_brace_index) { |
| 129 | 3 | string_append_literal(&buf, "\\{"); | |
| 130 | 3 | break; | |
| 131 | } | ||
| 132 | 45 | brace_level++; | |
| 133 | 45 | size_t skip = skip_empty_alternates(pattern + i, pattern_len - i); | |
| 134 |
2/2✓ Branch 0 (23→24) taken 3 times.
✓ Branch 1 (23→25) taken 42 times.
|
45 | if (skip > 0) { |
| 135 | 3 | i += skip; | |
| 136 | 3 | brace_group_has_empty_alternate[brace_level] = true; | |
| 137 | } | ||
| 138 |
3/4✓ Branch 0 (25→26) taken 45 times.
✗ Branch 1 (25→28) not taken.
✓ Branch 2 (26→27) taken 1 times.
✓ Branch 3 (26→28) taken 44 times.
|
45 | if (i + 1 < pattern_len && pattern[i + 1] == '}') { |
| 139 | // If brace group contains only empty alternates, emit nothing | ||
| 140 | 1 | brace_group_has_empty_alternate[brace_level] = false; | |
| 141 | 1 | i++; | |
| 142 | 1 | brace_level--; | |
| 143 | } else { | ||
| 144 | 44 | string_append_byte(&buf, '('); | |
| 145 | } | ||
| 146 | break; | ||
| 147 | } | ||
| 148 | 48 | case '}': | |
| 149 |
2/2✓ Branch 0 (29→30) taken 4 times.
✓ Branch 1 (29→31) taken 44 times.
|
48 | if (i > last_paired_brace_index || brace_level == 0) { |
| 150 | 4 | goto append_byte; | |
| 151 | } | ||
| 152 | 44 | string_append_byte(&buf, ')'); | |
| 153 |
2/2✓ Branch 0 (32→33) taken 9 times.
✓ Branch 1 (32→34) taken 35 times.
|
44 | if (brace_group_has_empty_alternate[brace_level]) { |
| 154 | 9 | string_append_byte(&buf, '?'); | |
| 155 | } | ||
| 156 | 44 | brace_group_has_empty_alternate[brace_level] = false; | |
| 157 | 44 | brace_level--; | |
| 158 | 44 | break; | |
| 159 | 45 | case ',': { | |
| 160 |
2/2✓ Branch 0 (35→36) taken 2 times.
✓ Branch 1 (35→37) taken 43 times.
|
45 | if (i >= last_paired_brace_index || brace_level == 0) { |
| 161 | 2 | goto append_byte; | |
| 162 | } | ||
| 163 | 43 | size_t skip = skip_empty_alternates(pattern + i, pattern_len - i); | |
| 164 |
2/2✓ Branch 0 (37→38) taken 8 times.
✓ Branch 1 (37→39) taken 35 times.
|
43 | if (skip > 0) { |
| 165 | 8 | i += skip; | |
| 166 | 8 | brace_group_has_empty_alternate[brace_level] = true; | |
| 167 | } | ||
| 168 |
3/4✓ Branch 0 (39→40) taken 43 times.
✗ Branch 1 (39→42) not taken.
✓ Branch 2 (40→41) taken 7 times.
✓ Branch 3 (40→42) taken 36 times.
|
43 | if (i + 1 < pattern_len && pattern[i + 1] == '}') { |
| 169 | 7 | brace_group_has_empty_alternate[brace_level] = true; | |
| 170 | } else { | ||
| 171 | 36 | string_append_byte(&buf, '|'); | |
| 172 | } | ||
| 173 | break; | ||
| 174 | } | ||
| 175 | 61 | case '/': | |
| 176 |
4/4✓ Branch 0 (43→44) taken 59 times.
✓ Branch 1 (43→48) taken 2 times.
✓ Branch 2 (45→46) taken 1 times.
✓ Branch 3 (45→48) taken 58 times.
|
61 | if (i + 3 < pattern_len && mem_equal(pattern + i, "/**/", 4)) { |
| 177 | 1 | string_append_literal(&buf, "(/|/.*/)"); | |
| 178 | 1 | i += 3; | |
| 179 | 1 | break; | |
| 180 | } | ||
| 181 | 60 | goto append_byte; | |
| 182 | 53 | case '.': | |
| 183 | case '(': | ||
| 184 | case ')': | ||
| 185 | case '|': | ||
| 186 | case '+': | ||
| 187 | 53 | string_append_byte(&buf, '\\'); | |
| 188 | // Fallthrough | ||
| 189 | default: | ||
| 190 | 692 | append_byte: | |
| 191 | 692 | string_append_byte(&buf, ch); | |
| 192 | } | ||
| 193 | } | ||
| 194 | |||
| 195 | 58 | string_append_byte(&buf, '$'); | |
| 196 | 58 | char *regex_pattern = string_steal_cstring(&buf); | |
| 197 | |||
| 198 | 58 | regex_t re; | |
| 199 | 58 | bool compiled = !regcomp(&re, regex_pattern, REG_EXTENDED | REG_NOSUB); | |
| 200 | 58 | free(regex_pattern); | |
| 201 |
1/2✓ Branch 0 (56→57) taken 58 times.
✗ Branch 1 (56→60) not taken.
|
58 | if (!compiled) { |
| 202 | return false; | ||
| 203 | } | ||
| 204 | |||
| 205 | 58 | int res = regexec(&re, path, 0, NULL, 0); | |
| 206 | 58 | regfree(&re); | |
| 207 | 58 | return res == 0; | |
| 208 | } | ||
| 209 |