Skip to content

Commit 297694a

Browse files
sliemeobnsliemeobn
committed
hardcodeASCII 9 and 10, simplified inline UTF8 decoding
1 parent 3dac1e0 commit 297694a

73 files changed

Lines changed: 3652 additions & 5696 deletions

File tree

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

‎Plugins/BridgeJS/Sources/BridgeJSLink/BridgeJSLink.swift‎

Lines changed: 52 additions & 80 deletions
Original file line numberDiff line numberDiff line change
@@ -318,8 +318,8 @@ public struct BridgeJSLink {
318318
/// The wasm32 12-byte Swift `String` layout is a standard-library
319319
/// implementation detail, not a stable ABI. Small strings (`word2 & 0x2000`)
320320
/// store up to 10 UTF-8 bytes inline. Large strings store the UTF-8 count in
321-
/// `word0` and a storage pointer in `word1` already biased by `nativeBias`
322-
/// (20), so UTF-8 starts at `word1 + 20`. Immortal large strings
321+
/// `word0` and a variant/object pointer in `word1`; UTF-8 starts at
322+
/// `word1 + nativeBias` (20 on 32-bit). Immortal large strings
323323
/// (`word2 & 0x8000`) are cached by a 53-bit `ptr + len * 2**32` key when
324324
/// `len < 2**21`; longer immortals skip the cache.
325325
private func emitDecodeString(_ printer: CodeFragmentPrinter) {
@@ -366,103 +366,74 @@ public struct BridgeJSLink {
366366
}
367367
// Match TextDecoder: consume exactly one leading UTF-8 BOM.
368368
if (start === 0 && codePoint === 0xfeff) continue;
369-
if (codePoint <= 0xffff) {
370-
switch (unitCount++) {
371-
case 0:
372-
u0 = codePoint;
373-
break;
374-
case 1:
375-
u1 = codePoint;
376-
break;
377-
case 2:
378-
u2 = codePoint;
379-
break;
380-
case 3:
381-
u3 = codePoint;
382-
break;
383-
case 4:
384-
u4 = codePoint;
385-
break;
386-
case 5:
387-
u5 = codePoint;
388-
break;
389-
case 6:
390-
u6 = codePoint;
391-
break;
392-
case 7:
393-
u7 = codePoint;
394-
break;
395-
case 8:
396-
u8 = codePoint;
397-
break;
398-
case 9:
399-
u9 = codePoint;
400-
break;
401-
}
402-
} else {
403-
codePoint -= 0x10000;
404-
switch (unitCount++) {
405-
case 0:
406-
u0 = 0xd800 + (codePoint >>> 10);
407-
break;
408-
case 1:
409-
u1 = 0xd800 + (codePoint >>> 10);
410-
break;
411-
case 2:
412-
u2 = 0xd800 + (codePoint >>> 10);
413-
break;
414-
case 3:
415-
u3 = 0xd800 + (codePoint >>> 10);
416-
break;
417-
case 4:
418-
u4 = 0xd800 + (codePoint >>> 10);
419-
break;
420-
case 5:
421-
u5 = 0xd800 + (codePoint >>> 10);
422-
break;
423-
case 6:
424-
u6 = 0xd800 + (codePoint >>> 10);
425-
break;
426-
case 7:
427-
u7 = 0xd800 + (codePoint >>> 10);
428-
break;
429-
case 8:
430-
u8 = 0xd800 + (codePoint >>> 10);
431-
break;
432-
case 9:
433-
u9 = 0xd800 + (codePoint >>> 10);
434-
break;
435-
}
369+
const supplementary = codePoint > 0xffff;
370+
const codeUnit = supplementary
371+
? 0xd800 + ((codePoint - 0x10000) >>> 10)
372+
: codePoint;
373+
switch (unitCount++) {
374+
case 0:
375+
u0 = codeUnit;
376+
break;
377+
case 1:
378+
u1 = codeUnit;
379+
break;
380+
case 2:
381+
u2 = codeUnit;
382+
break;
383+
case 3:
384+
u3 = codeUnit;
385+
break;
386+
case 4:
387+
u4 = codeUnit;
388+
break;
389+
case 5:
390+
u5 = codeUnit;
391+
break;
392+
case 6:
393+
u6 = codeUnit;
394+
break;
395+
case 7:
396+
u7 = codeUnit;
397+
break;
398+
case 8:
399+
u8 = codeUnit;
400+
break;
401+
case 9:
402+
u9 = codeUnit;
403+
break;
404+
}
405+
if (supplementary) {
406+
const lowSurrogate = 0xdc00 + ((codePoint - 0x10000) & 1023);
436407
switch (unitCount++) {
437408
case 0:
438-
u0 = 0xdc00 + (codePoint & 1023);
409+
u0 = lowSurrogate;
439410
break;
440411
case 1:
441-
u1 = 0xdc00 + (codePoint & 1023);
412+
u1 = lowSurrogate;
442413
break;
443414
case 2:
444-
u2 = 0xdc00 + (codePoint & 1023);
415+
u2 = lowSurrogate;
445416
break;
446417
case 3:
447-
u3 = 0xdc00 + (codePoint & 1023);
418+
u3 = lowSurrogate;
448419
break;
449420
case 4:
450-
u4 = 0xdc00 + (codePoint & 1023);
421+
u4 = lowSurrogate;
451422
break;
452423
case 5:
453-
u5 = 0xdc00 + (codePoint & 1023);
424+
u5 = lowSurrogate;
454425
break;
455426
case 6:
456-
u6 = 0xdc00 + (codePoint & 1023);
427+
u6 = lowSurrogate;
457428
break;
458429
case 7:
459-
u7 = 0xdc00 + (codePoint & 1023);
430+
u7 = lowSurrogate;
460431
break;
461432
case 8:
462-
u8 = 0xdc00 + (codePoint & 1023);
433+
u8 = lowSurrogate;
463434
break;
464435
case 9:
465-
u9 = 0xdc00 + (codePoint & 1023);
436+
u9 = lowSurrogate;
466437
break;
467438
}
468439
}
@@ -506,7 +477,8 @@ public struct BridgeJSLink {
506477
case 6: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff);
507478
case 7: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff);
508479
case 8: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff, word1 >>> 24);
509-
default: return decodeSmallUTF8(word0, word1, word2, count);
480+
case 9: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff, word1 >>> 24, word2 & 0xff);
481+
case 10: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff, word1 >>> 24, word2 & 0xff, (word2 >>> 16) & 0xff);
510482
}
511483
};
512484
return (word0, word1, word2) => {

‎Plugins/BridgeJS/Tests/BridgeJSToolTests/__Snapshots__/BridgeJSLinkTests/Alias.js‎

Lines changed: 50 additions & 78 deletions
Original file line numberDiff line numberDiff line change
@@ -726,103 +726,74 @@ export async function createInstantiator(options, swift) {
726726
}
727727
// Match TextDecoder: consume exactly one leading UTF-8 BOM.
728728
if (start === 0 && codePoint === 0xfeff) continue;
729-
if (codePoint <= 0xffff) {
730-
switch (unitCount++) {
731-
case 0:
732-
u0 = codePoint;
733-
break;
734-
case 1:
735-
u1 = codePoint;
736-
break;
737-
case 2:
738-
u2 = codePoint;
739-
break;
740-
case 3:
741-
u3 = codePoint;
742-
break;
743-
case 4:
744-
u4 = codePoint;
745-
break;
746-
case 5:
747-
u5 = codePoint;
748-
break;
749-
case 6:
750-
u6 = codePoint;
751-
break;
752-
case 7:
753-
u7 = codePoint;
754-
break;
755-
case 8:
756-
u8 = codePoint;
757-
break;
758-
case 9:
759-
u9 = codePoint;
760-
break;
761-
}
762-
} else {
763-
codePoint -= 0x10000;
764-
switch (unitCount++) {
765-
case 0:
766-
u0 = 0xd800 + (codePoint >>> 10);
767-
break;
768-
case 1:
769-
u1 = 0xd800 + (codePoint >>> 10);
770-
break;
771-
case 2:
772-
u2 = 0xd800 + (codePoint >>> 10);
773-
break;
774-
case 3:
775-
u3 = 0xd800 + (codePoint >>> 10);
776-
break;
777-
case 4:
778-
u4 = 0xd800 + (codePoint >>> 10);
779-
break;
780-
case 5:
781-
u5 = 0xd800 + (codePoint >>> 10);
782-
break;
783-
case 6:
784-
u6 = 0xd800 + (codePoint >>> 10);
785-
break;
786-
case 7:
787-
u7 = 0xd800 + (codePoint >>> 10);
788-
break;
789-
case 8:
790-
u8 = 0xd800 + (codePoint >>> 10);
791-
break;
792-
case 9:
793-
u9 = 0xd800 + (codePoint >>> 10);
794-
break;
795-
}
729+
const supplementary = codePoint > 0xffff;
730+
const codeUnit = supplementary
731+
? 0xd800 + ((codePoint - 0x10000) >>> 10)
732+
: codePoint;
733+
switch (unitCount++) {
734+
case 0:
735+
u0 = codeUnit;
736+
break;
737+
case 1:
738+
u1 = codeUnit;
739+
break;
740+
case 2:
741+
u2 = codeUnit;
742+
break;
743+
case 3:
744+
u3 = codeUnit;
745+
break;
746+
case 4:
747+
u4 = codeUnit;
748+
break;
749+
case 5:
750+
u5 = codeUnit;
751+
break;
752+
case 6:
753+
u6 = codeUnit;
754+
break;
755+
case 7:
756+
u7 = codeUnit;
757+
break;
758+
case 8:
759+
u8 = codeUnit;
760+
break;
761+
case 9:
762+
u9 = codeUnit;
763+
break;
764+
}
765+
if (supplementary) {
766+
const lowSurrogate = 0xdc00 + ((codePoint - 0x10000) & 1023);
796767
switch (unitCount++) {
797768
case 0:
798-
u0 = 0xdc00 + (codePoint & 1023);
769+
u0 = lowSurrogate;
799770
break;
800771
case 1:
801-
u1 = 0xdc00 + (codePoint & 1023);
772+
u1 = lowSurrogate;
802773
break;
803774
case 2:
804-
u2 = 0xdc00 + (codePoint & 1023);
775+
u2 = lowSurrogate;
805776
break;
806777
case 3:
807-
u3 = 0xdc00 + (codePoint & 1023);
778+
u3 = lowSurrogate;
808779
break;
809780
case 4:
810-
u4 = 0xdc00 + (codePoint & 1023);
781+
u4 = lowSurrogate;
811782
break;
812783
case 5:
813-
u5 = 0xdc00 + (codePoint & 1023);
784+
u5 = lowSurrogate;
814785
break;
815786
case 6:
816-
u6 = 0xdc00 + (codePoint & 1023);
787+
u6 = lowSurrogate;
817788
break;
818789
case 7:
819-
u7 = 0xdc00 + (codePoint & 1023);
790+
u7 = lowSurrogate;
820791
break;
821792
case 8:
822-
u8 = 0xdc00 + (codePoint & 1023);
793+
u8 = lowSurrogate;
823794
break;
824795
case 9:
825-
u9 = 0xdc00 + (codePoint & 1023);
796+
u9 = lowSurrogate;
826797
break;
827798
}
828799
}
@@ -866,7 +837,8 @@ export async function createInstantiator(options, swift) {
866837
case 6: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff);
867838
case 7: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff);
868839
case 8: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff, word1 >>> 24);
869-
default: return decodeSmallUTF8(word0, word1, word2, count);
840+
case 9: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff, word1 >>> 24, word2 & 0xff);
841+
case 10: return char(word0 & 0xff, (word0 >>> 8) & 0xff, (word0 >>> 16) & 0xff, word0 >>> 24, word1 & 0xff, (word1 >>> 8) & 0xff, (word1 >>> 16) & 0xff, word1 >>> 24, word2 & 0xff, (word2 >>> 16) & 0xff);
870842
}
871843
};
872844
return (word0, word1, word2) => {

0 commit comments

Comments
 (0)