| 1 | #include <locale.h> // setlocale()
|
| 2 | #include <regex.h> // regcomp()
|
| 3 |
|
| 4 | #include "mycpp/runtime.h"
|
| 5 | #include "vendor/greatest.h"
|
| 6 |
|
| 7 | void FindAll(const char* p, const char* s) {
|
| 8 | regex_t pat;
|
| 9 |
|
| 10 | int cflags = REG_EXTENDED;
|
| 11 | if (regcomp(&pat, p, cflags) != 0) {
|
| 12 | FAIL();
|
| 13 | }
|
| 14 | int outlen = pat.re_nsub + 1; // number of captures
|
| 15 |
|
| 16 | // TODO: Could statically allocate 99, and assert that re_nsub is less than
|
| 17 | // 99. Would speed up loops.
|
| 18 | regmatch_t* pmatch =
|
| 19 | static_cast<regmatch_t*>(malloc(sizeof(regmatch_t) * outlen));
|
| 20 |
|
| 21 | int cur_pos = 0;
|
| 22 | // int n = strlen(s);
|
| 23 | while (true) {
|
| 24 | // Necessary so ^ doesn't match in the middle!
|
| 25 | int eflags = cur_pos == 0 ? 0 : REG_NOTBOL;
|
| 26 | bool match = regexec(&pat, s + cur_pos, outlen, pmatch, eflags) == 0;
|
| 27 |
|
| 28 | if (!match) {
|
| 29 | break;
|
| 30 | }
|
| 31 | int i;
|
| 32 | for (i = 0; i < outlen; i++) {
|
| 33 | int start = pmatch[i].rm_so;
|
| 34 | int end = pmatch[i].rm_eo;
|
| 35 | int len = end - start;
|
| 36 | BigStr* m = StrFromC(s + cur_pos + start, len);
|
| 37 | log("%d GROUP %d (%d .. %d) = [%s]", cur_pos, i, start, end, m->data_);
|
| 38 | }
|
| 39 | log("");
|
| 40 | int match_len = pmatch[0].rm_eo;
|
| 41 | if (match_len == 0) {
|
| 42 | break;
|
| 43 | }
|
| 44 | cur_pos += match_len;
|
| 45 | }
|
| 46 |
|
| 47 | free(pmatch);
|
| 48 | regfree(&pat);
|
| 49 | }
|
| 50 |
|
| 51 | // adjacent matches
|
| 52 | const char* s = "a345y-axy- there b789y- cy-";
|
| 53 |
|
| 54 | TEST regex_unanchored() {
|
| 55 | const char* unanchored = "[abc]([0-9]*)(x?)(y)-";
|
| 56 | FindAll(unanchored, s);
|
| 57 |
|
| 58 | PASS();
|
| 59 | }
|
| 60 |
|
| 61 | TEST regex_caret() {
|
| 62 | const char* anchored = "^[abc]([0-9]*)(x?)(y)-";
|
| 63 | FindAll(anchored, s);
|
| 64 |
|
| 65 | PASS();
|
| 66 | }
|
| 67 |
|
| 68 | TEST regex_lexer() {
|
| 69 | // like the Yaks / Make-a-Lisp pattern
|
| 70 | const char* lexer = "([a-z]+)|([0-9]+)|([ ]+)|([+-])";
|
| 71 | FindAll(lexer, s);
|
| 72 |
|
| 73 | PASS();
|
| 74 | }
|
| 75 |
|
| 76 | TEST regex_repeat_with_capture() {
|
| 77 | const char* lexer = "(([a-z]+)([0-9]+)-)*((A+)|(Z+))*";
|
| 78 | FindAll(lexer, "a0-b1-c2-AAZZZA");
|
| 79 | // Groups are weird
|
| 80 | // whole match 0: a0-b1-c2-
|
| 81 | // 1: c2- # last repetition
|
| 82 | // 2: c # last one
|
| 83 | // 3: 2 # last one
|
| 84 | //
|
| 85 | // And then there's an empty match
|
| 86 | //
|
| 87 | // Ideas:
|
| 88 | // - disallow nested groups in Eggex?
|
| 89 | // - I really care about the inner ones -- groups 2 and 3
|
| 90 | // - I want flat groups
|
| 91 |
|
| 92 | PASS();
|
| 93 | }
|
| 94 |
|
| 95 | // Disallow this in eggex, as well as the above
|
| 96 | TEST regex_nested_capture() {
|
| 97 | const char* lexer = "(([a-z]+)([0-9]+))";
|
| 98 | FindAll(lexer, "a0");
|
| 99 | PASS();
|
| 100 | }
|
| 101 |
|
| 102 | // I think we allow this in eggex
|
| 103 | TEST regex_alt_with_capture() {
|
| 104 | const char* lexer = "([a-z]+)|([0-9]+)(-)";
|
| 105 | FindAll(lexer, "x-");
|
| 106 | FindAll(lexer, "7-");
|
| 107 | PASS();
|
| 108 | }
|
| 109 |
|
| 110 | bool RegexMatch(const char* s, const char* regex_str) {
|
| 111 | regex_t pat;
|
| 112 | int status = regcomp(&pat, regex_str, REG_EXTENDED);
|
| 113 | if (status != 0) {
|
| 114 | assert(false);
|
| 115 | }
|
| 116 | log("*** Matching string %s against regex %s", s, regex_str);
|
| 117 |
|
| 118 | int num_groups = pat.re_nsub + 1; // number of captures
|
| 119 |
|
| 120 | regmatch_t* pmatch =
|
| 121 | static_cast<regmatch_t*>(malloc(sizeof(regmatch_t) * num_groups));
|
| 122 | int eflags = 0;
|
| 123 | bool match = regexec(&pat, s, num_groups, pmatch, eflags) == 0;
|
| 124 | if (match) {
|
| 125 | printf("match\n");
|
| 126 | for (int i = 0; i < num_groups; i++) {
|
| 127 | int start = pmatch[i].rm_so;
|
| 128 | int end = pmatch[i].rm_eo;
|
| 129 | printf("start %d - %d\n", start, end);
|
| 130 | }
|
| 131 | }
|
| 132 | free(pmatch);
|
| 133 | regfree(&pat);
|
| 134 |
|
| 135 | return match;
|
| 136 | }
|
| 137 |
|
| 138 | TEST regex_unicode() {
|
| 139 | regex_t pat;
|
| 140 |
|
| 141 | const char* p = "_._"; // 1 byte, not code point?
|
| 142 |
|
| 143 | bool matched;
|
| 144 |
|
| 145 | matched = RegexMatch("_xyz_", p);
|
| 146 | ASSERT(not matched);
|
| 147 |
|
| 148 | matched = RegexMatch("_x_", p);
|
| 149 | ASSERT(matched);
|
| 150 |
|
| 151 | matched = RegexMatch("_\x01_", p);
|
| 152 | ASSERT(matched);
|
| 153 |
|
| 154 | const char* u1 = "_μ_";
|
| 155 | const char* u2 = "_\u03bc_";
|
| 156 | const char* u3 = "_\xce\xbc_"; // utf-8 encoding
|
| 157 |
|
| 158 | // Doesn't match without UTF-8 setting
|
| 159 | matched = RegexMatch(u1, p);
|
| 160 | // log("u1 %d", matched);
|
| 161 | ASSERT(not matched);
|
| 162 |
|
| 163 | matched = RegexMatch(u2, p);
|
| 164 | // log("u2 %d", matched);
|
| 165 | ASSERT(not matched);
|
| 166 |
|
| 167 | matched = RegexMatch(u3, p);
|
| 168 | // log("u3 %d", matched);
|
| 169 | ASSERT(not matched);
|
| 170 |
|
| 171 | // SETS GLOBAL
|
| 172 | char* saved_locale = setlocale(LC_ALL, "");
|
| 173 | log("saved_locale %s", saved_locale);
|
| 174 | if (saved_locale == nullptr) {
|
| 175 | FAIL();
|
| 176 | }
|
| 177 |
|
| 178 | // Now it matches
|
| 179 | matched = RegexMatch(u1, p);
|
| 180 | ASSERT(matched);
|
| 181 | matched = RegexMatch(u2, p);
|
| 182 | ASSERT(matched);
|
| 183 | matched = RegexMatch(u3, p);
|
| 184 | ASSERT(matched);
|
| 185 |
|
| 186 | // [^a] can match a code point
|
| 187 | matched = RegexMatch(u3, "_[^a]_");
|
| 188 | ASSERT(matched);
|
| 189 |
|
| 190 | const char* unicode_char_class = "[a\xce\xbc]";
|
| 191 | const char* s = "\xce\xbc";
|
| 192 | matched = RegexMatch(s, unicode_char_class);
|
| 193 | ASSERT(matched);
|
| 194 |
|
| 195 | PASS();
|
| 196 | }
|
| 197 |
|
| 198 | GREATEST_MAIN_DEFS();
|
| 199 |
|
| 200 | int main(int argc, char** argv) {
|
| 201 | gHeap.Init();
|
| 202 |
|
| 203 | GREATEST_MAIN_BEGIN();
|
| 204 |
|
| 205 | RUN_TEST(regex_unanchored);
|
| 206 | RUN_TEST(regex_caret);
|
| 207 | RUN_TEST(regex_lexer);
|
| 208 | RUN_TEST(regex_repeat_with_capture);
|
| 209 | RUN_TEST(regex_alt_with_capture);
|
| 210 | RUN_TEST(regex_nested_capture);
|
| 211 | RUN_TEST(regex_unicode);
|
| 212 |
|
| 213 | gHeap.CleanProcessExit();
|
| 214 |
|
| 215 | GREATEST_MAIN_END();
|
| 216 | return 0;
|
| 217 | }
|