Implement RegExp 'v' flag, part 1 (#229)

This commit implements the flag itself and teaches the regex engine to
reject previously accepted patterns when in unicodeSets mode.

Refs: https://github.com/quickjs-ng/quickjs/issues/228

libregexp.c

@@ -63,6 +63,7 @@ typedef enum {
 
 #define TMP_BUF_SIZE 128
 
+// invariant: is_unicode ^ unicode_sets (or neither, but not both)
 typedef struct {
     DynBuf byte_code;
     const uint8_t *buf_ptr;
@@ -70,6 +71,7 @@ typedef struct {
     const uint8_t *buf_start;
     int re_flags;
     BOOL is_unicode;
+    BOOL unicode_sets;
     BOOL ignore_case;
     BOOL dotall;
     int capture_count;
@@ -853,6 +855,8 @@ static int re_emit_range(REParseState *s, const CharRange *cr)
     return 0;
 }
 
+// s->unicode turns patterns like []] into syntax errors
+// s->unicode_sets turns more patterns into errors, like [a-] or [[]
 static int re_parse_char_class(REParseState *s, const uint8_t **pp)
 {
     const uint8_t *p;
@@ -864,17 +868,43 @@ static int re_parse_char_class(REParseState *s, const uint8_t **pp)
     cr_init(cr, s->opaque, lre_realloc);
     p = *pp;
     p++;    /* skip '[' */
+
+    if (s->unicode_sets) {
+        static const char verboten[] =
+            "()[{}/-|"                              "\0"
+            "&&!!##$$%%**++,,..::;;<<==>>??@@``~~"  "\0"
+            "^^^_^^";
+        const char *s = verboten;
+        int n = 1;
+        do {
+            if (!memcmp(s, p, n))
+                if (p[n] == ']')
+                    goto invalid_class_range;
+            s += n;
+            if (!*s) {
+                s++;
+                n++;
+            }
+        } while (n < 4);
+    }
+
     invert = FALSE;
     if (*p == '^') {
         p++;
         invert = TRUE;
     }
+
     for(;;) {
         if (*p == ']')
             break;
         c1 = get_class_atom(s, cr1, &p, TRUE);
         if ((int)c1 < 0)
             goto fail;
+        if (*p == '-' && p[1] == ']' && s->unicode_sets) {
+            if (c1 >= CLASS_RANGE_BASE)
+                cr_free(cr1);
+            goto invalid_class_range;
+        }
         if (*p == '-' && p[1] != ']') {
             const uint8_t *p0 = p + 1;
             if (c1 >= CLASS_RANGE_BASE) {
@@ -1843,6 +1873,7 @@ uint8_t *lre_compile(int *plen, char *error_msg, int error_msg_size,
     is_sticky = ((re_flags & LRE_FLAG_STICKY) != 0);
     s->ignore_case = ((re_flags & LRE_FLAG_IGNORECASE) != 0);
     s->dotall = ((re_flags & LRE_FLAG_DOTALL) != 0);
+    s->unicode_sets = ((re_flags & LRE_FLAG_UNICODE_SETS) != 0);
     s->capture_count = 1;
     s->total_capture_count = -1;
     s->has_named_captures = -1;

libregexp.h

@@ -37,8 +37,8 @@
 #define LRE_FLAG_UNICODE    (1 << 4)
 #define LRE_FLAG_STICKY     (1 << 5)
 #define LRE_FLAG_INDICES    (1 << 6) /* Unused by libregexp, just recorded. */
-
 #define LRE_FLAG_NAMED_GROUPS (1 << 7) /* named groups are present in the regexp */
+#define LRE_FLAG_UNICODE_SETS (1 << 8)
 
 uint8_t *lre_compile(int *plen, char *error_msg, int error_msg_size,
                      const char *buf, size_t buf_len, int re_flags,

quickjs.c

@@ -40730,6 +40730,9 @@ static JSValue js_compile_regexp(JSContext *ctx, JSValue pattern,
             case 'u':
                 mask = LRE_FLAG_UNICODE;
                 break;
+            case 'v':
+                mask = LRE_FLAG_UNICODE_SETS;
+                break;
             case 'y':
                 mask = LRE_FLAG_STICKY;
                 break;
@@ -40746,6 +40749,10 @@ static JSValue js_compile_regexp(JSContext *ctx, JSValue pattern,
         JS_FreeCString(ctx, str);
     }
 
+    if (re_flags & LRE_FLAG_UNICODE)
+        if (re_flags & LRE_FLAG_UNICODE_SETS)
+            return JS_ThrowSyntaxError(ctx, "invalid regular expression flags");
+
     str = JS_ToCStringLen2(ctx, &len, pattern, !(re_flags & LRE_FLAG_UNICODE));
     if (!str)
         return JS_EXCEPTION;
@@ -41067,6 +41074,11 @@ static JSValue js_regexp_get_flags(JSContext *ctx, JSValue this_val)
         goto exception;
     if (res)
         *p++ = 'u';
+    res = JS_ToBoolFree(ctx, JS_GetPropertyStr(ctx, this_val, "unicodeSets"));
+    if (res < 0)
+        goto exception;
+    if (res)
+        *p++ = 'v';
     res = JS_ToBoolFree(ctx, JS_GetPropertyStr(ctx, this_val, "sticky"));
     if (res < 0)
         goto exception;
@@ -42152,6 +42164,7 @@ static const JSCFunctionListEntry js_regexp_proto_funcs[] = {
     JS_CGETSET_MAGIC_DEF("multiline", js_regexp_get_flag, NULL, LRE_FLAG_MULTILINE ),
     JS_CGETSET_MAGIC_DEF("dotAll", js_regexp_get_flag, NULL, LRE_FLAG_DOTALL ),
     JS_CGETSET_MAGIC_DEF("unicode", js_regexp_get_flag, NULL, LRE_FLAG_UNICODE ),
+    JS_CGETSET_MAGIC_DEF("unicodeSets", js_regexp_get_flag, NULL, LRE_FLAG_UNICODE_SETS ),
     JS_CGETSET_MAGIC_DEF("sticky", js_regexp_get_flag, NULL, LRE_FLAG_STICKY ),
     JS_CGETSET_MAGIC_DEF("hasIndices", js_regexp_get_flag, NULL, LRE_FLAG_INDICES ),
     JS_CFUNC_DEF("exec", 1, js_regexp_exec ),

test262.conf

@@ -157,7 +157,7 @@ regexp-lookbehind
 regexp-match-indices
 regexp-named-groups
 regexp-unicode-property-escapes
-regexp-v-flag=skip
+regexp-v-flag
 resizable-arraybuffer=skip
 rest-parameters
 Set
@@ -223,5 +223,147 @@ test262/test/built-ins/ThrowTypeError/unique-per-realm-function-proto.js
 #test262/test/built-ins/RegExp/CharacterClassEscapes/
 #test262/test/built-ins/RegExp/property-escapes/
 
+# in progress regexp-v-flag support, see https://github.com/quickjs-ng/quickjs/issues/228
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Basic_Emoji-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Basic_Emoji-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Basic_Emoji-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Basic_Emoji.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Emoji_Keycap_Sequence-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Emoji_Keycap_Sequence-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Emoji_Keycap_Sequence-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/Emoji_Keycap_Sequence.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Flag_Sequence-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Flag_Sequence-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Flag_Sequence-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Flag_Sequence.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Modifier_Sequence-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Modifier_Sequence-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Modifier_Sequence-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Modifier_Sequence.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Tag_Sequence-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Tag_Sequence-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Tag_Sequence-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_Tag_Sequence.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_ZWJ_Sequence-negative-CharacterClass.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_ZWJ_Sequence-negative-P.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_ZWJ_Sequence-negative-u.js
+test262/test/built-ins/RegExp/property-escapes/generated/strings/RGI_Emoji_ZWJ_Sequence.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-difference-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-difference-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-difference-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-difference-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-difference-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-difference-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-difference-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-difference-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-difference-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-difference-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-difference-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-difference-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-intersection-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-intersection-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-intersection-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-intersection-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-intersection-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-intersection-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-union-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-union-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-union-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-union-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-union-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-escape-union-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-intersection-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-intersection-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-intersection-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-intersection-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-intersection-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-intersection-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-union-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-union-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-union-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-union-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-union-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-class-union-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-difference-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-difference-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-difference-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-difference-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-difference-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-difference-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-intersection-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-intersection-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-intersection-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-intersection-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-intersection-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-intersection-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-difference-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-difference-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-difference-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-difference-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-difference-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-difference-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-intersection-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-intersection-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-intersection-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-intersection-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-intersection-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-intersection-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-union-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-union-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-union-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-union-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-union-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-property-escape-union-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-union-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-union-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-union-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-union-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-union-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/character-union-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-difference-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-difference-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-difference-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-difference-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-difference-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-difference-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-intersection-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-intersection-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-intersection-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-intersection-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-intersection-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-intersection-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-union-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-union-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-union-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-union-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-union-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/property-of-strings-escape-union-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/rgi-emoji-13.1.js
+test262/test/built-ins/RegExp/unicodeSets/generated/rgi-emoji-14.0.js
+test262/test/built-ins/RegExp/unicodeSets/generated/rgi-emoji-15.0.js
+test262/test/built-ins/RegExp/unicodeSets/generated/rgi-emoji-15.1.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-difference-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-difference-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-difference-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-difference-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-difference-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-difference-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-intersection-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-intersection-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-intersection-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-intersection-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-intersection-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-intersection-string-literal.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-union-character-class-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-union-character-class.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-union-character-property-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-union-character.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-union-property-of-strings-escape.js
+test262/test/built-ins/RegExp/unicodeSets/generated/string-literal-union-string-literal.js
+
 [tests]
 # list test files or use config.testdir

tests/test_builtin.js

@@ -646,6 +646,17 @@ function test_regexp()
     assert(/{1a}/.toString(), "/{1a}/");
     a = /a{1+/.exec("a{11");
     assert(a, ["a{11"] );
+
+    eval("/[a-]/");  // accepted with no flag
+    eval("/[a-]/u"); // accepted with 'u' flag
+
+    let ex;
+    try {
+        eval("/[a-]/v"); // rejected with 'v' flag
+    } catch (_ex) {
+        ex = _ex;
+    }
+    assert(ex?.message, "invalid class range");
 }
 
 function test_symbol()