mirrors/quickjs
1
0
Fork 0
mirror of https://github.com/bellard/quickjs.git synced 2024-12-12 09:44:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

use Unicode normalization in String.prototype.localeCompare

Browse Source
This commit is contained in:
Fabrice Bellard 2023-12-22 11:05:49 +01:00
parent 7cefa7b121
commit 07ff474992
2 changed files with 100 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

131
quickjs.c
View File

@ -41109,26 +41109,6 @@ static BOOL test_final_sigma(JSString *p, int sigma_pos)
return !lre_is_cased(c1);
}
static JSValue js_string_localeCompare(JSContext *ctx, JSValueConst this_val,
int argc, JSValueConst *argv)
{
JSValue a, b;
int cmp;
a = JS_ToStringCheckObject(ctx, this_val);
if (JS_IsException(a))
return JS_EXCEPTION;
b = JS_ToString(ctx, argv[0]);
if (JS_IsException(b)) {
JS_FreeValue(ctx, a);
return JS_EXCEPTION;
}
cmp = js_string_compare(ctx, JS_VALUE_GET_STRING(a), JS_VALUE_GET_STRING(b));
JS_FreeValue(ctx, a);
JS_FreeValue(ctx, b);
return JS_NewInt32(ctx, cmp);
}
static JSValue js_string_toLowerCase(JSContext *ctx, JSValueConst this_val,
int argc, JSValueConst *argv, int to_lower)
{
@ -41214,23 +41194,38 @@ static JSValue JS_NewUTF32String(JSContext *ctx, const uint32_t *buf, int len)
return JS_EXCEPTION;
}
static int js_string_normalize1(JSContext *ctx, uint32_t **pout_buf,
JSValueConst val,
UnicodeNormalizationEnum n_type)
{
int buf_len, out_len;
uint32_t *buf, *out_buf;
buf_len = JS_ToUTF32String(ctx, &buf, val);
if (buf_len < 0)
return -1;
out_len = unicode_normalize(&out_buf, buf, buf_len, n_type,
ctx->rt, (DynBufReallocFunc *)js_realloc_rt);
js_free(ctx, buf);
if (out_len < 0)
return -1;
*pout_buf = out_buf;
return out_len;
}
static JSValue js_string_normalize(JSContext *ctx, JSValueConst this_val,
int argc, JSValueConst *argv)
{
const char *form, *p;
size_t form_len;
int is_compat, buf_len, out_len;
int is_compat, out_len;
UnicodeNormalizationEnum n_type;
JSValue val;
uint32_t *buf, *out_buf;
uint32_t *out_buf;
val = JS_ToStringCheckObject(ctx, this_val);
if (JS_IsException(val))
return val;
buf_len = JS_ToUTF32String(ctx, &buf, val);
JS_FreeValue(ctx, val);
if (buf_len < 0)
return JS_EXCEPTION;
if (argc == 0 || JS_IsUndefined(argv[0])) {
n_type = UNICODE_NFC;
@ -41256,22 +41251,96 @@ static JSValue js_string_normalize(JSContext *ctx, JSValueConst this_val,
JS_FreeCString(ctx, form);
JS_ThrowRangeError(ctx, "bad normalization form");
fail1:
js_free(ctx, buf);
JS_FreeValue(ctx, val);
return JS_EXCEPTION;
}
JS_FreeCString(ctx, form);
}
out_len = unicode_normalize(&out_buf, buf, buf_len, n_type,
ctx->rt, (DynBufReallocFunc *)js_realloc_rt);
js_free(ctx, buf);
out_len = js_string_normalize1(ctx, &out_buf, val, n_type);
JS_FreeValue(ctx, val);
if (out_len < 0)
return JS_EXCEPTION;
val = JS_NewUTF32String(ctx, out_buf, out_len);
js_free(ctx, out_buf);
return val;
}
#endif /* CONFIG_ALL_UNICODE */
/* return < 0, 0 or > 0 */
static int js_UTF32_compare(const uint32_t *buf1, int buf1_len,
const uint32_t *buf2, int buf2_len)
{
int i, len, c, res;
len = min_int(buf1_len, buf2_len);
for(i = 0; i < len; i++) {
/* Note: range is limited so a subtraction is valid */
c = buf1[i] - buf2[i];
if (c != 0)
return c;
}
if (buf1_len == buf2_len)
res = 0;
else if (buf1_len < buf2_len)
res = -1;
else
res = 1;
return res;
}
static JSValue js_string_localeCompare(JSContext *ctx, JSValueConst this_val,
int argc, JSValueConst *argv)
{
JSValue a, b;
int cmp, a_len, b_len;
uint32_t *a_buf, *b_buf;
a = JS_ToStringCheckObject(ctx, this_val);
if (JS_IsException(a))
return JS_EXCEPTION;
b = JS_ToString(ctx, argv[0]);
if (JS_IsException(b)) {
JS_FreeValue(ctx, a);
return JS_EXCEPTION;
}
a_len = js_string_normalize1(ctx, &a_buf, a, UNICODE_NFC);
JS_FreeValue(ctx, a);
if (a_len < 0) {
JS_FreeValue(ctx, b);
return JS_EXCEPTION;
}
b_len = js_string_normalize1(ctx, &b_buf, b, UNICODE_NFC);
JS_FreeValue(ctx, b);
if (b_len < 0) {
js_free(ctx, a_buf);
return JS_EXCEPTION;
}
cmp = js_UTF32_compare(a_buf, a_len, b_buf, b_len);
js_free(ctx, a_buf);
js_free(ctx, b_buf);
return JS_NewInt32(ctx, cmp);
}
#else /* CONFIG_ALL_UNICODE */
static JSValue js_string_localeCompare(JSContext *ctx, JSValueConst this_val,
int argc, JSValueConst *argv)
{
JSValue a, b;
int cmp;
a = JS_ToStringCheckObject(ctx, this_val);
if (JS_IsException(a))
return JS_EXCEPTION;
b = JS_ToString(ctx, argv[0]);
if (JS_IsException(b)) {
JS_FreeValue(ctx, a);
return JS_EXCEPTION;
}
cmp = js_string_compare(ctx, JS_VALUE_GET_STRING(a), JS_VALUE_GET_STRING(b));
JS_FreeValue(ctx, a);
JS_FreeValue(ctx, b);
return JS_NewInt32(ctx, cmp);
}
#endif /* !CONFIG_ALL_UNICODE */
/* also used for String.prototype.valueOf */
static JSValue js_string_toString(JSContext *ctx, JSValueConst this_val,

2
test262_errors.txt
View File

@ -5,8 +5,6 @@ test262/test/built-ins/RegExp/lookahead-quantifier-match-groups.js:27: Test262Er
test262/test/built-ins/RegExp/lookahead-quantifier-match-groups.js:27: strict mode: Test262Error: Expected [a, abc] and [a, undefined] to have the same contents. ? quantifier
test262/test/built-ins/RegExp/unicode_full_case_folding.js:20: Test262Error: \u0390 does not match \u1fd3
test262/test/built-ins/RegExp/unicode_full_case_folding.js:20: strict mode: Test262Error: \u0390 does not match \u1fd3
test262/test/built-ins/String/prototype/localeCompare/15.5.4.9_CE.js:62: Test262Error: String.prototype.localeCompare considers ö (\u006f\u0308) ≠ ö (\u00f6).
test262/test/built-ins/String/prototype/localeCompare/15.5.4.9_CE.js:62: strict mode: Test262Error: String.prototype.localeCompare considers ö (\u006f\u0308) ≠ ö (\u00f6).
test262/test/built-ins/TypedArray/prototype/sort/sort-tonumber.js:30: TypeError: ArrayBuffer is detached (Testing with Float64Array.)
test262/test/built-ins/TypedArray/prototype/sort/sort-tonumber.js:30: strict mode: TypeError: ArrayBuffer is detached (Testing with Float64Array.)
test262/test/language/expressions/assignment/target-member-computed-reference-null.js:32: Test262Error: Expected a DummyError but got a TypeError