MIRROR: javascript for ๐'s, a tiny runtime with big ambitions
add codePointAt and improve charAt/charCodeAt handling
+62
-8
+11
examples/spec/strings.js
+11
examples/spec/strings.js
···
52
52
test('repeat', 'ab'.repeat(3), 'ababab');
53
53
54
54
test('charAt', 'hello'.charAt(1), 'e');
55
+
test('charAt default index', 'hello'.charAt(), 'h');
56
+
test('charAt coercion', 'hello'.charAt('1'), 'e');
55
57
test('charCodeAt', 'ABC'.charCodeAt(0), 65);
58
+
test('charCodeAt default index', 'ABC'.charCodeAt(), 65);
59
+
test('charCodeAt coercion', 'ABC'.charCodeAt('1'), 66);
60
+
test('codePointAt ascii', 'ABC'.codePointAt(0), 65);
61
+
test('codePointAt default index', 'ABC'.codePointAt(), 65);
62
+
test('codePointAt coercion', 'ABC'.codePointAt('1'), 66);
63
+
test('codePointAt out of bounds', 'ABC'.codePointAt(10), undefined);
64
+
test('codePointAt utf8 2-byte', 'รฉ'.codePointAt(0), 233);
65
+
test('codePointAt utf8 3-byte', 'ไธญ'.codePointAt(0), 20013);
66
+
test('codePointAt utf8 4-byte', '๐'.codePointAt(0), 128512);
56
67
57
68
test('replace', 'hello world'.replace('world', 'there'), 'hello there');
58
69
test('replaceAll', 'a-b-c'.replaceAll('-', '_'), 'a_b_c');
+51
-8
src/ant.c
+51
-8
src/ant.c
···
18279
18279
static jsval_t builtin_string_charCodeAt(struct js *js, jsval_t *args, int nargs) {
18280
18280
jsval_t str = to_string_val(js, js->this_val);
18281
18281
if (vtype(str) != T_STR) return js_mkerr(js, "charCodeAt called on non-string");
18282
-
if (nargs < 1 || vtype(args[0]) != T_NUM) return tov(JS_NAN);
18282
+
18283
+
double idx_d = nargs < 1 ? 0.0 : js_to_number(js, args[0]);
18284
+
if (isnan(idx_d) || isinf(idx_d)) return tov(JS_NAN);
18283
18285
18284
-
double idx_d = tod(args[0]);
18285
-
if (idx_d < 0 || idx_d != (double)(long)idx_d) return tov(JS_NAN);
18286
+
long idx_l = (long) idx_d;
18287
+
if (idx_l < 0) return tov(JS_NAN);
18286
18288
18287
-
jsoff_t idx = (jsoff_t) idx_d;
18289
+
jsoff_t idx = (jsoff_t) idx_l;
18288
18290
jsoff_t str_len = offtolen(loadoff(js, (jsoff_t) vdata(str)));
18289
18291
18290
18292
if (idx >= str_len) return tov(JS_NAN);
···
18293
18295
unsigned char ch = (unsigned char) js->mem[str_off + idx];
18294
18296
18295
18297
return tov((double) ch);
18298
+
}
18299
+
18300
+
static jsval_t builtin_string_codePointAt(struct js *js, jsval_t *args, int nargs) {
18301
+
jsval_t str = to_string_val(js, js->this_val);
18302
+
if (vtype(str) != T_STR) return js_mkerr(js, "codePointAt called on non-string");
18303
+
18304
+
double idx_d = nargs < 1 ? 0.0 : js_to_number(js, args[0]);
18305
+
if (isnan(idx_d) || isinf(idx_d)) return js_mkundef();
18306
+
18307
+
long idx_l = (long) idx_d;
18308
+
if (idx_l < 0) return js_mkundef();
18309
+
18310
+
jsoff_t idx = (jsoff_t) idx_l;
18311
+
jsoff_t str_len = offtolen(loadoff(js, (jsoff_t) vdata(str)));
18312
+
18313
+
if (idx >= str_len) return js_mkundef();
18314
+
18315
+
jsoff_t str_off = (jsoff_t) vdata(str) + sizeof(jsoff_t);
18316
+
const unsigned char *s = &js->mem[str_off + idx];
18317
+
jsoff_t remaining = str_len - idx;
18318
+
18319
+
unsigned char c0 = s[0];
18320
+
if (c0 < 0x80) return tov((double) c0);
18321
+
18322
+
if ((c0 & 0xE0) == 0xC0 && remaining >= 2 && (s[1] & 0xC0) == 0x80) {
18323
+
uint32_t cp = ((c0 & 0x1F) << 6) | (s[1] & 0x3F);
18324
+
return tov((double) cp);
18325
+
}
18326
+
18327
+
if ((c0 & 0xF0) == 0xE0 && remaining >= 3 && (s[1] & 0xC0) == 0x80 && (s[2] & 0xC0) == 0x80) {
18328
+
uint32_t cp = ((c0 & 0x0F) << 12) | ((s[1] & 0x3F) << 6) | (s[2] & 0x3F);
18329
+
return tov((double) cp);
18330
+
}
18331
+
18332
+
if ((c0 & 0xF8) == 0xF0 && remaining >= 4 && (s[1] & 0xC0) == 0x80 && (s[2] & 0xC0) == 0x80 && (s[3] & 0xC0) == 0x80) {
18333
+
uint32_t cp = ((c0 & 0x07) << 18) | ((s[1] & 0x3F) << 12) | ((s[2] & 0x3F) << 6) | (s[3] & 0x3F);
18334
+
return tov((double) cp);
18335
+
}
18336
+
18337
+
return tov((double) c0);
18296
18338
}
18297
18339
18298
18340
static jsval_t builtin_string_toLowerCase(struct js *js, jsval_t *args, int nargs) {
···
18487
18529
static jsval_t builtin_string_charAt(struct js *js, jsval_t *args, int nargs) {
18488
18530
jsval_t str = to_string_val(js, js->this_val);
18489
18531
if (vtype(str) != T_STR) return js_mkerr(js, "charAt called on non-string");
18490
-
if (nargs < 1 || vtype(args[0]) != T_NUM) return js_mkstr(js, "", 0);
18491
18532
18492
-
double idx_d = tod(args[0]);
18533
+
double idx_d = nargs < 1 ? 0.0 : js_to_number(js, args[0]);
18493
18534
if (isnan(idx_d)) idx_d = 0;
18494
18535
else if (idx_d < 0) idx_d = -floor(-idx_d);
18495
18536
else idx_d = floor(idx_d);
···
18509
18550
static jsval_t builtin_string_at(struct js *js, jsval_t *args, int nargs) {
18510
18551
jsval_t str = to_string_val(js, js->this_val);
18511
18552
if (vtype(str) != T_STR) return js_mkerr(js, "at called on non-string");
18512
-
if (nargs < 1 || vtype(args[0]) != T_NUM) return js_mkundef();
18553
+
18554
+
double idx_d = nargs < 1 ? 0.0 : js_to_number(js, args[0]);
18555
+
if (isnan(idx_d) || isinf(idx_d)) return js_mkundef();
18513
18556
18514
18557
jsoff_t str_len = offtolen(loadoff(js, (jsoff_t) vdata(str)));
18515
-
double idx_d = tod(args[0]);
18516
18558
long idx = (long) idx_d;
18517
18559
18518
18560
if (idx < 0) idx += (long) str_len;
···
22041
22083
setprop(js, string_proto, js_mkstr(js, "match", 5), js_mkfun(builtin_string_match));
22042
22084
setprop(js, string_proto, js_mkstr(js, "template", 8), js_mkfun(builtin_string_template));
22043
22085
setprop(js, string_proto, js_mkstr(js, "charCodeAt", 10), js_mkfun(builtin_string_charCodeAt));
22086
+
setprop(js, string_proto, js_mkstr(js, "codePointAt", 11), js_mkfun(builtin_string_codePointAt));
22044
22087
setprop(js, string_proto, js_mkstr(js, "toLowerCase", 11), js_mkfun(builtin_string_toLowerCase));
22045
22088
setprop(js, string_proto, js_mkstr(js, "toUpperCase", 11), js_mkfun(builtin_string_toUpperCase));
22046
22089
setprop(js, string_proto, js_mkstr(js, "toLocaleLowerCase", 17), js_mkfun(builtin_string_toLowerCase));