fix a bug when matching with empty strings

related to kokke/tiny-regex-c/issues/56
matyalatte · Jul 5, 2024 · 2e64692 · 2e64692
1 parent 61a0d57
commit 2e64692
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 4 deletions.
diff --git a/src/re.c b/src/re.c
@@ -84,8 +84,9 @@ int re_matchp(re_t pattern, const char* text, int* matchlength) {
 
             do {
                 if (matchpattern(pattern, text, rune_size, matchlength)) {
-                    if (text[0] == '\0')
-                        return -1;
+                    // Maybe we don't need this
+                    // if (text[0] == '\0')
+                    //     return -1;
 
                     return (int)(text - prepoint);
                 }

diff --git a/tests/re_test.h b/tests/re_test.h
@@ -140,7 +140,7 @@ INSTANTIATE_TEST_SUITE_P(RegexTestInstantiation_Escaped,
 // https://github.com/python/cpython/blob/main/Lib/test/re_tests.py
 const RegexCase regex_cases_python[] = {
     // { ")", "", TSM_SYNTAX_ERROR },  // () operator does not supported yet.
-    // { "", "", TSM_OK },  // this fails somehow
+    { "", "", TSM_OK },  // this fails somehow
     { "abc", "abc", TSM_OK },
     { "abc", "xbc", TSM_FAIL },
     { "abc", "axc", TSM_FAIL },
@@ -197,7 +197,7 @@ const RegexCase regex_cases_python[] = {
     { "a+b+c", "aabbabc", TSM_OK },
     { "[^ab]*", "cde", TSM_OK },
     { "abc", "", TSM_FAIL },
-    // { "a*", "", TSM_OK },  // This fails somehow
+    { "a*", "", TSM_OK },  // This fails somehow
     { "abcd*efg", "abcdefg", TSM_OK },
     { "ab*", "xabyabbbz", TSM_OK },
     { "ab*", "xayabbbz", TSM_OK },
@@ -222,6 +222,21 @@ INSTANTIATE_TEST_SUITE_P(RegexTestInstantiation_Python,
     RegexTest,
     ::testing::ValuesIn(regex_cases_python));
 
+// Original test cases.
+const RegexCase regex_cases_tsm[] = {
+    { "^[a-zA-Z_][z-zA-Z0-9_]*", "a", TSM_OK },
+    { "^[a-zA-Z_][z-zA-Z0-9_]*", "abcd1234_", TSM_OK },
+    { "^[a-zA-Z_][z-zA-Z0-9_]*", "1", TSM_FAIL },
+    { "^[a-zA-Z_][z-zA-Z0-9_]*", "1bcd1234_", TSM_FAIL },
+    { ".", "", TSM_FAIL },
+    { "^.$", "", TSM_FAIL },
+    { ".*", "", TSM_OK },
+};
+
+INSTANTIATE_TEST_SUITE_P(RegexTestInstantiation_Tsm,
+    RegexTest,
+    ::testing::ValuesIn(regex_cases_tsm));
+
 TEST_P(RegexTest, tsm_regex_match) {
     const RegexCase test_case = GetParam();
     int actual = tsm_regex_match(test_case.pattern, test_case.str);