-
Notifications
You must be signed in to change notification settings - Fork 55
/
utf_8.rs
1629 lines (1555 loc) · 61.6 KB
/
utf_8.rs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Copyright Mozilla Foundation. See the COPYRIGHT
// file at the top-level directory of this distribution.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
use super::*;
use crate::ascii::ascii_to_basic_latin;
use crate::ascii::basic_latin_to_ascii;
use crate::ascii::validate_ascii;
use crate::handles::*;
use crate::mem::convert_utf16_to_utf8_partial;
use crate::variant::*;
cfg_if! {
if #[cfg(feature = "simd-accel")] {
use ::core::intrinsics::unlikely;
use ::core::intrinsics::likely;
} else {
#[inline(always)]
fn unlikely(b: bool) -> bool {
b
}
#[inline(always)]
fn likely(b: bool) -> bool {
b
}
}
}
#[repr(align(64))] // Align to cache lines
pub struct Utf8Data {
pub table: [u8; 384],
}
// BEGIN GENERATED CODE. PLEASE DO NOT EDIT.
// Instead, please regenerate using generate-encoding-data.py
pub static UTF8_DATA: Utf8Data = Utf8Data {
table: [
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 148, 148, 148,
148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 164, 164, 164, 164, 164,
164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164,
164, 164, 164, 164, 164, 164, 164, 164, 164, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252,
252, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 8, 8, 16, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 32, 8, 8, 64, 8, 8, 8, 128, 4,
4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
],
};
// END GENERATED CODE
pub fn utf8_valid_up_to(src: &[u8]) -> usize {
let mut read = 0;
'outer: loop {
let mut byte = {
let src_remaining = &src[read..];
match validate_ascii(src_remaining) {
None => {
return src.len();
}
Some((non_ascii, consumed)) => {
read += consumed;
non_ascii
}
}
};
// Check for the longest sequence to avoid checking twice for the
// multi-byte sequences. This can't overflow with 64-bit address space,
// because full 64 bits aren't in use. In the 32-bit PAE case, for this
// to overflow would mean that the source slice would be so large that
// the address space of the process would not have space for any code.
// Therefore, the slice cannot be so long that this would overflow.
if likely(read + 4 <= src.len()) {
'inner: loop {
// At this point, `byte` is not included in `read`, because we
// don't yet know that a) the UTF-8 sequence is valid and b) that there
// is output space if it is an astral sequence.
// Inspecting the lead byte directly is faster than what the
// std lib does!
if likely(in_inclusive_range8(byte, 0xC2, 0xDF)) {
// Two-byte
let second = unsafe { *(src.get_unchecked(read + 1)) };
if !in_inclusive_range8(second, 0x80, 0xBF) {
break 'outer;
}
read += 2;
// Next lead (manually inlined)
if likely(read + 4 <= src.len()) {
byte = unsafe { *(src.get_unchecked(read)) };
if byte < 0x80 {
read += 1;
continue 'outer;
}
continue 'inner;
}
break 'inner;
}
if likely(byte < 0xF0) {
'three: loop {
// Three-byte
let second = unsafe { *(src.get_unchecked(read + 1)) };
let third = unsafe { *(src.get_unchecked(read + 2)) };
if ((UTF8_DATA.table[usize::from(second)]
& unsafe { *(UTF8_DATA.table.get_unchecked(byte as usize + 0x80)) })
| (third >> 6))
!= 2
{
break 'outer;
}
read += 3;
// Next lead (manually inlined)
if likely(read + 4 <= src.len()) {
byte = unsafe { *(src.get_unchecked(read)) };
if in_inclusive_range8(byte, 0xE0, 0xEF) {
continue 'three;
}
if likely(byte < 0x80) {
read += 1;
continue 'outer;
}
continue 'inner;
}
break 'inner;
}
}
// Four-byte
let second = unsafe { *(src.get_unchecked(read + 1)) };
let third = unsafe { *(src.get_unchecked(read + 2)) };
let fourth = unsafe { *(src.get_unchecked(read + 3)) };
if (u16::from(
UTF8_DATA.table[usize::from(second)]
& unsafe { *(UTF8_DATA.table.get_unchecked(byte as usize + 0x80)) },
) | u16::from(third >> 6)
| (u16::from(fourth & 0xC0) << 2))
!= 0x202
{
break 'outer;
}
read += 4;
// Next lead
if likely(read + 4 <= src.len()) {
byte = unsafe { *(src.get_unchecked(read)) };
if byte < 0x80 {
read += 1;
continue 'outer;
}
continue 'inner;
}
break 'inner;
}
}
// We can't have a complete 4-byte sequence, but we could still have
// one to three shorter sequences.
'tail: loop {
// >= is better for bound check elision than ==
if read >= src.len() {
break 'outer;
}
byte = src[read];
// At this point, `byte` is not included in `read`, because we
// don't yet know that a) the UTF-8 sequence is valid and b) that there
// is output space if it is an astral sequence.
// Inspecting the lead byte directly is faster than what the
// std lib does!
if byte < 0x80 {
read += 1;
continue 'tail;
}
if in_inclusive_range8(byte, 0xC2, 0xDF) {
// Two-byte
let new_read = read + 2;
if new_read > src.len() {
break 'outer;
}
let second = src[read + 1];
if !in_inclusive_range8(second, 0x80, 0xBF) {
break 'outer;
}
read += 2;
continue 'tail;
}
// We need to exclude valid four byte lead bytes, because
// `UTF8_DATA.second_mask` covers
if byte < 0xF0 {
// Three-byte
let new_read = read + 3;
if new_read > src.len() {
break 'outer;
}
let second = src[read + 1];
let third = src[read + 2];
if ((UTF8_DATA.table[usize::from(second)]
& unsafe { *(UTF8_DATA.table.get_unchecked(byte as usize + 0x80)) })
| (third >> 6))
!= 2
{
break 'outer;
}
read += 3;
// `'tail` handles sequences shorter than 4, so
// there can't be another sequence after this one.
break 'outer;
}
break 'outer;
}
}
read
}
#[cfg_attr(feature = "cargo-clippy", allow(never_loop, cyclomatic_complexity))]
pub fn convert_utf8_to_utf16_up_to_invalid(src: &[u8], dst: &mut [u16]) -> (usize, usize) {
let mut read = 0;
let mut written = 0;
'outer: loop {
let mut byte = {
let src_remaining = &src[read..];
let dst_remaining = &mut dst[written..];
let length = ::core::cmp::min(src_remaining.len(), dst_remaining.len());
match unsafe {
ascii_to_basic_latin(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length)
} {
None => {
read += length;
written += length;
break 'outer;
}
Some((non_ascii, consumed)) => {
read += consumed;
written += consumed;
non_ascii
}
}
};
// Check for the longest sequence to avoid checking twice for the
// multi-byte sequences. This can't overflow with 64-bit address space,
// because full 64 bits aren't in use. In the 32-bit PAE case, for this
// to overflow would mean that the source slice would be so large that
// the address space of the process would not have space for any code.
// Therefore, the slice cannot be so long that this would overflow.
if likely(read + 4 <= src.len()) {
'inner: loop {
// At this point, `byte` is not included in `read`, because we
// don't yet know that a) the UTF-8 sequence is valid and b) that there
// is output space if it is an astral sequence.
// We know, thanks to `ascii_to_basic_latin` that there is output
// space for at least one UTF-16 code unit, so no need to check
// for output space in the BMP cases.
// Inspecting the lead byte directly is faster than what the
// std lib does!
if likely(in_inclusive_range8(byte, 0xC2, 0xDF)) {
// Two-byte
let second = unsafe { *(src.get_unchecked(read + 1)) };
if !in_inclusive_range8(second, 0x80, 0xBF) {
break 'outer;
}
unsafe {
*(dst.get_unchecked_mut(written)) =
((u16::from(byte) & 0x1F) << 6) | (u16::from(second) & 0x3F)
};
read += 2;
written += 1;
// Next lead (manually inlined)
if written == dst.len() {
break 'outer;
}
if likely(read + 4 <= src.len()) {
byte = unsafe { *(src.get_unchecked(read)) };
if byte < 0x80 {
unsafe { *(dst.get_unchecked_mut(written)) = u16::from(byte) };
read += 1;
written += 1;
continue 'outer;
}
continue 'inner;
}
break 'inner;
}
if likely(byte < 0xF0) {
'three: loop {
// Three-byte
let second = unsafe { *(src.get_unchecked(read + 1)) };
let third = unsafe { *(src.get_unchecked(read + 2)) };
if ((UTF8_DATA.table[usize::from(second)]
& unsafe { *(UTF8_DATA.table.get_unchecked(byte as usize + 0x80)) })
| (third >> 6))
!= 2
{
break 'outer;
}
let point = ((u16::from(byte) & 0xF) << 12)
| ((u16::from(second) & 0x3F) << 6)
| (u16::from(third) & 0x3F);
unsafe { *(dst.get_unchecked_mut(written)) = point };
read += 3;
written += 1;
// Next lead (manually inlined)
if written == dst.len() {
break 'outer;
}
if likely(read + 4 <= src.len()) {
byte = unsafe { *(src.get_unchecked(read)) };
if in_inclusive_range8(byte, 0xE0, 0xEF) {
continue 'three;
}
if likely(byte < 0x80) {
unsafe { *(dst.get_unchecked_mut(written)) = u16::from(byte) };
read += 1;
written += 1;
continue 'outer;
}
continue 'inner;
}
break 'inner;
}
}
// Four-byte
if written + 1 == dst.len() {
break 'outer;
}
let second = unsafe { *(src.get_unchecked(read + 1)) };
let third = unsafe { *(src.get_unchecked(read + 2)) };
let fourth = unsafe { *(src.get_unchecked(read + 3)) };
if (u16::from(
UTF8_DATA.table[usize::from(second)]
& unsafe { *(UTF8_DATA.table.get_unchecked(byte as usize + 0x80)) },
) | u16::from(third >> 6)
| (u16::from(fourth & 0xC0) << 2))
!= 0x202
{
break 'outer;
}
let point = ((u32::from(byte) & 0x7) << 18)
| ((u32::from(second) & 0x3F) << 12)
| ((u32::from(third) & 0x3F) << 6)
| (u32::from(fourth) & 0x3F);
unsafe { *(dst.get_unchecked_mut(written)) = (0xD7C0 + (point >> 10)) as u16 };
unsafe {
*(dst.get_unchecked_mut(written + 1)) = (0xDC00 + (point & 0x3FF)) as u16
};
read += 4;
written += 2;
// Next lead
if written == dst.len() {
break 'outer;
}
if likely(read + 4 <= src.len()) {
byte = unsafe { *(src.get_unchecked(read)) };
if byte < 0x80 {
unsafe { *(dst.get_unchecked_mut(written)) = u16::from(byte) };
read += 1;
written += 1;
continue 'outer;
}
continue 'inner;
}
break 'inner;
}
}
// We can't have a complete 4-byte sequence, but we could still have
// one to three shorter sequences.
'tail: loop {
// >= is better for bound check elision than ==
if read >= src.len() || written >= dst.len() {
break 'outer;
}
byte = src[read];
// At this point, `byte` is not included in `read`, because we
// don't yet know that a) the UTF-8 sequence is valid and b) that there
// is output space if it is an astral sequence.
// Inspecting the lead byte directly is faster than what the
// std lib does!
if byte < 0x80 {
dst[written] = u16::from(byte);
read += 1;
written += 1;
continue 'tail;
}
if in_inclusive_range8(byte, 0xC2, 0xDF) {
// Two-byte
let new_read = read + 2;
if new_read > src.len() {
break 'outer;
}
let second = src[read + 1];
if !in_inclusive_range8(second, 0x80, 0xBF) {
break 'outer;
}
dst[written] = ((u16::from(byte) & 0x1F) << 6) | (u16::from(second) & 0x3F);
read += 2;
written += 1;
continue 'tail;
}
// We need to exclude valid four byte lead bytes, because
// `UTF8_DATA.second_mask` covers
if byte < 0xF0 {
// Three-byte
let new_read = read + 3;
if new_read > src.len() {
break 'outer;
}
let second = src[read + 1];
let third = src[read + 2];
if ((UTF8_DATA.table[usize::from(second)]
& unsafe { *(UTF8_DATA.table.get_unchecked(byte as usize + 0x80)) })
| (third >> 6))
!= 2
{
break 'outer;
}
let point = ((u16::from(byte) & 0xF) << 12)
| ((u16::from(second) & 0x3F) << 6)
| (u16::from(third) & 0x3F);
dst[written] = point;
read += 3;
written += 1;
// `'tail` handles sequences shorter than 4, so
// there can't be another sequence after this one.
break 'outer;
}
break 'outer;
}
}
(read, written)
}
pub struct Utf8Decoder {
code_point: u32,
bytes_seen: usize, // 1, 2 or 3: counts continuations only
bytes_needed: usize, // 1, 2 or 3: counts continuations only
lower_boundary: u8,
upper_boundary: u8,
}
impl Utf8Decoder {
pub fn new_inner() -> Utf8Decoder {
Utf8Decoder {
code_point: 0,
bytes_seen: 0,
bytes_needed: 0,
lower_boundary: 0x80u8,
upper_boundary: 0xBFu8,
}
}
pub fn new() -> VariantDecoder {
VariantDecoder::Utf8(Utf8Decoder::new_inner())
}
pub fn in_neutral_state(&self) -> bool {
self.bytes_needed == 0
}
fn extra_from_state(&self) -> usize {
if self.bytes_needed == 0 {
0
} else {
self.bytes_seen + 1
}
}
pub fn max_utf16_buffer_length(&self, byte_length: usize) -> Option<usize> {
byte_length.checked_add(1 + self.extra_from_state())
}
pub fn max_utf8_buffer_length_without_replacement(&self, byte_length: usize) -> Option<usize> {
byte_length.checked_add(3 + self.extra_from_state())
}
pub fn max_utf8_buffer_length(&self, byte_length: usize) -> Option<usize> {
checked_add(
3,
checked_mul(3, byte_length.checked_add(self.extra_from_state())),
)
}
decoder_functions!(
{},
{
// This is the fast path. The rest runs only at the
// start and end for partial sequences.
if self.bytes_needed == 0 {
dest.copy_utf8_up_to_invalid_from(&mut source);
}
},
{
if self.bytes_needed != 0 {
let bad_bytes = (self.bytes_seen + 1) as u8;
self.code_point = 0;
self.bytes_needed = 0;
self.bytes_seen = 0;
return (
DecoderResult::Malformed(bad_bytes, 0),
src_consumed,
dest.written(),
);
}
},
{
if self.bytes_needed == 0 {
if b < 0x80u8 {
destination_handle.write_ascii(b);
continue;
}
if b < 0xC2u8 {
return (
DecoderResult::Malformed(1, 0),
unread_handle.consumed(),
destination_handle.written(),
);
}
if b < 0xE0u8 {
self.bytes_needed = 1;
self.code_point = u32::from(b) & 0x1F;
continue;
}
if b < 0xF0u8 {
if b == 0xE0u8 {
self.lower_boundary = 0xA0u8;
} else if b == 0xEDu8 {
self.upper_boundary = 0x9Fu8;
}
self.bytes_needed = 2;
self.code_point = u32::from(b) & 0xF;
continue;
}
if b < 0xF5u8 {
if b == 0xF0u8 {
self.lower_boundary = 0x90u8;
} else if b == 0xF4u8 {
self.upper_boundary = 0x8Fu8;
}
self.bytes_needed = 3;
self.code_point = u32::from(b) & 0x7;
continue;
}
return (
DecoderResult::Malformed(1, 0),
unread_handle.consumed(),
destination_handle.written(),
);
}
// self.bytes_needed != 0
if !(b >= self.lower_boundary && b <= self.upper_boundary) {
let bad_bytes = (self.bytes_seen + 1) as u8;
self.code_point = 0;
self.bytes_needed = 0;
self.bytes_seen = 0;
self.lower_boundary = 0x80u8;
self.upper_boundary = 0xBFu8;
return (
DecoderResult::Malformed(bad_bytes, 0),
unread_handle.unread(),
destination_handle.written(),
);
}
self.lower_boundary = 0x80u8;
self.upper_boundary = 0xBFu8;
self.code_point = (self.code_point << 6) | (u32::from(b) & 0x3F);
self.bytes_seen += 1;
if self.bytes_seen != self.bytes_needed {
continue;
}
if self.bytes_needed == 3 {
destination_handle.write_astral(self.code_point);
} else {
destination_handle.write_bmp_excl_ascii(self.code_point as u16);
}
self.code_point = 0;
self.bytes_needed = 0;
self.bytes_seen = 0;
continue;
},
self,
src_consumed,
dest,
source,
b,
destination_handle,
unread_handle,
check_space_astral
);
}
#[cfg_attr(feature = "cargo-clippy", allow(never_loop))]
#[inline(never)]
pub fn convert_utf16_to_utf8_partial_inner(src: &[u16], dst: &mut [u8]) -> (usize, usize) {
let mut read = 0;
let mut written = 0;
'outer: loop {
let mut unit = {
let src_remaining = &src[read..];
let dst_remaining = &mut dst[written..];
let length = if dst_remaining.len() < src_remaining.len() {
dst_remaining.len()
} else {
src_remaining.len()
};
match unsafe {
basic_latin_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length)
} {
None => {
read += length;
written += length;
return (read, written);
}
Some((non_ascii, consumed)) => {
read += consumed;
written += consumed;
non_ascii
}
}
};
'inner: loop {
// The following loop is only broken out of as a goto forward.
loop {
// Unfortunately, this check isn't enough for the compiler to elide
// the bound checks on writes to dst, which is why they are manually
// elided, which makes a measurable difference.
if written.checked_add(4).unwrap() > dst.len() {
return (read, written);
}
read += 1;
if unit < 0x800 {
unsafe {
*(dst.get_unchecked_mut(written)) = (unit >> 6) as u8 | 0xC0u8;
written += 1;
*(dst.get_unchecked_mut(written)) = (unit & 0x3F) as u8 | 0x80u8;
written += 1;
}
break;
}
let unit_minus_surrogate_start = unit.wrapping_sub(0xD800);
if likely(unit_minus_surrogate_start > (0xDFFF - 0xD800)) {
unsafe {
*(dst.get_unchecked_mut(written)) = (unit >> 12) as u8 | 0xE0u8;
written += 1;
*(dst.get_unchecked_mut(written)) = ((unit & 0xFC0) >> 6) as u8 | 0x80u8;
written += 1;
*(dst.get_unchecked_mut(written)) = (unit & 0x3F) as u8 | 0x80u8;
written += 1;
}
break;
}
if likely(unit_minus_surrogate_start <= (0xDBFF - 0xD800)) {
// high surrogate
// read > src.len() is impossible, but using
// >= instead of == allows the compiler to elide a bound check.
if read >= src.len() {
debug_assert_eq!(read, src.len());
// Unpaired surrogate at the end of the buffer.
unsafe {
*(dst.get_unchecked_mut(written)) = 0xEFu8;
written += 1;
*(dst.get_unchecked_mut(written)) = 0xBFu8;
written += 1;
*(dst.get_unchecked_mut(written)) = 0xBDu8;
written += 1;
}
return (read, written);
}
let second = src[read];
let second_minus_low_surrogate_start = second.wrapping_sub(0xDC00);
if likely(second_minus_low_surrogate_start <= (0xDFFF - 0xDC00)) {
// The next code unit is a low surrogate. Advance position.
read += 1;
let astral = (u32::from(unit) << 10) + u32::from(second)
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32);
unsafe {
*(dst.get_unchecked_mut(written)) = (astral >> 18) as u8 | 0xF0u8;
written += 1;
*(dst.get_unchecked_mut(written)) =
((astral & 0x3F000u32) >> 12) as u8 | 0x80u8;
written += 1;
*(dst.get_unchecked_mut(written)) =
((astral & 0xFC0u32) >> 6) as u8 | 0x80u8;
written += 1;
*(dst.get_unchecked_mut(written)) = (astral & 0x3F) as u8 | 0x80u8;
written += 1;
}
break;
}
// The next code unit is not a low surrogate. Don't advance
// position and treat the high surrogate as unpaired.
// Fall through
}
// Unpaired low surrogate
unsafe {
*(dst.get_unchecked_mut(written)) = 0xEFu8;
written += 1;
*(dst.get_unchecked_mut(written)) = 0xBFu8;
written += 1;
*(dst.get_unchecked_mut(written)) = 0xBDu8;
written += 1;
}
break;
}
// Now see if the next unit is Basic Latin
// read > src.len() is impossible, but using
// >= instead of == allows the compiler to elide a bound check.
if read >= src.len() {
debug_assert_eq!(read, src.len());
return (read, written);
}
unit = src[read];
if unlikely(unit < 0x80) {
// written > dst.len() is impossible, but using
// >= instead of == allows the compiler to elide a bound check.
if written >= dst.len() {
debug_assert_eq!(written, dst.len());
return (read, written);
}
dst[written] = unit as u8;
read += 1;
written += 1;
// Mysteriously, adding a punctuation check here makes
// the expected benificiary cases *slower*!
continue 'outer;
}
continue 'inner;
}
}
}
#[inline(never)]
pub fn convert_utf16_to_utf8_partial_tail(src: &[u16], dst: &mut [u8]) -> (usize, usize) {
// Everything below is cold code!
let mut read = 0;
let mut written = 0;
let mut unit = src[read];
// We now have up to 3 output slots, so an astral character
// will not fit.
if unit < 0x800 {
loop {
if unit < 0x80 {
if written >= dst.len() {
return (read, written);
}
read += 1;
dst[written] = unit as u8;
written += 1;
} else if unit < 0x800 {
if written + 2 > dst.len() {
return (read, written);
}
read += 1;
dst[written] = (unit >> 6) as u8 | 0xC0u8;
written += 1;
dst[written] = (unit & 0x3F) as u8 | 0x80u8;
written += 1;
} else {
return (read, written);
}
// read > src.len() is impossible, but using
// >= instead of == allows the compiler to elide a bound check.
if read >= src.len() {
debug_assert_eq!(read, src.len());
return (read, written);
}
unit = src[read];
}
}
// Could be an unpaired surrogate, but we'll need 3 output
// slots in any case.
if written + 3 > dst.len() {
return (read, written);
}
read += 1;
let unit_minus_surrogate_start = unit.wrapping_sub(0xD800);
if unit_minus_surrogate_start <= (0xDFFF - 0xD800) {
// Got surrogate
if unit_minus_surrogate_start <= (0xDBFF - 0xD800) {
// Got high surrogate
if read >= src.len() {
// Unpaired high surrogate
unit = 0xFFFD;
} else {
let second = src[read];
if in_inclusive_range16(second, 0xDC00, 0xDFFF) {
// Valid surrogate pair, but we know it won't fit.
read -= 1;
return (read, written);
}
// Unpaired high
unit = 0xFFFD;
}
} else {
// Unpaired low
unit = 0xFFFD;
}
}
dst[written] = (unit >> 12) as u8 | 0xE0u8;
written += 1;
dst[written] = ((unit & 0xFC0) >> 6) as u8 | 0x80u8;
written += 1;
dst[written] = (unit & 0x3F) as u8 | 0x80u8;
written += 1;
debug_assert_eq!(written, dst.len());
(read, written)
}
pub struct Utf8Encoder;
impl Utf8Encoder {
pub fn new(encoding: &'static Encoding) -> Encoder {
Encoder::new(encoding, VariantEncoder::Utf8(Utf8Encoder))
}
pub fn max_buffer_length_from_utf16_without_replacement(
&self,
u16_length: usize,
) -> Option<usize> {
u16_length.checked_mul(3)
}
pub fn max_buffer_length_from_utf8_without_replacement(
&self,
byte_length: usize,
) -> Option<usize> {
Some(byte_length)
}
pub fn encode_from_utf16_raw(
&mut self,
src: &[u16],
dst: &mut [u8],
_last: bool,
) -> (EncoderResult, usize, usize) {
let (read, written) = convert_utf16_to_utf8_partial(src, dst);
(
if read == src.len() {
EncoderResult::InputEmpty
} else {
EncoderResult::OutputFull
},
read,
written,
)
}
pub fn encode_from_utf8_raw(
&mut self,
src: &str,
dst: &mut [u8],
_last: bool,
) -> (EncoderResult, usize, usize) {
let bytes = src.as_bytes();
let mut to_write = bytes.len();
if to_write <= dst.len() {
(&mut dst[..to_write]).copy_from_slice(bytes);
return (EncoderResult::InputEmpty, to_write, to_write);
}
to_write = dst.len();
// Move back until we find a UTF-8 sequence boundary.
while (bytes[to_write] & 0xC0) == 0x80 {
to_write -= 1;
}
(&mut dst[..to_write]).copy_from_slice(&bytes[..to_write]);
(EncoderResult::OutputFull, to_write, to_write)
}
}
// Any copyright to the test code below this comment is dedicated to the
// Public Domain. http://creativecommons.org/publicdomain/zero/1.0/
#[cfg(all(test, feature = "alloc"))]
mod tests {
use super::super::testing::*;
use super::super::*;
// fn decode_utf8_to_utf16(bytes: &[u8], expect: &[u16]) {
// decode_to_utf16_without_replacement(UTF_8, bytes, expect);
// }
fn decode_utf8_to_utf8(bytes: &[u8], expect: &str) {
decode_to_utf8(UTF_8, bytes, expect);
}
fn decode_valid_utf8(string: &str) {
decode_utf8_to_utf8(string.as_bytes(), string);
}
fn encode_utf8_from_utf16(string: &[u16], expect: &[u8]) {
encode_from_utf16(UTF_8, string, expect);
}
fn encode_utf8_from_utf8(string: &str, expect: &[u8]) {
encode_from_utf8(UTF_8, string, expect);
}
fn encode_utf8_from_utf16_with_output_limit(
string: &[u16],
expect: &str,
limit: usize,
expect_result: EncoderResult,
) {
let mut dst = Vec::new();
{
dst.resize(limit, 0u8);
let mut encoder = UTF_8.new_encoder();
let (result, read, written) =
encoder.encode_from_utf16_without_replacement(string, &mut dst, false);
assert_eq!(result, expect_result);
if expect_result == EncoderResult::InputEmpty {
assert_eq!(read, string.len());
}
assert_eq!(&dst[..written], expect.as_bytes());
}
{
dst.resize(64, 0u8);
for (i, elem) in dst.iter_mut().enumerate() {
*elem = i as u8;
}
let mut encoder = UTF_8.new_encoder();
let (_, _, mut j) =
encoder.encode_from_utf16_without_replacement(string, &mut dst, false);
while j < dst.len() {
assert_eq!(usize::from(dst[j]), j);
j += 1;
}
}
}
#[test]
fn test_utf8_decode() {
// Empty
decode_valid_utf8("");
// ASCII
decode_valid_utf8("ab");
// Low BMP
decode_valid_utf8("a\u{E4}Z");
// High BMP
decode_valid_utf8("a\u{2603}Z");
// Astral
decode_valid_utf8("a\u{1F4A9}Z");
// Low BMP with last byte missing
decode_utf8_to_utf8(b"a\xC3Z", "a\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xC3", "a\u{FFFD}");
// High BMP with last byte missing
decode_utf8_to_utf8(b"a\xE2\x98Z", "a\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xE2\x98", "a\u{FFFD}");
// Astral with last byte missing
decode_utf8_to_utf8(b"a\xF0\x9F\x92Z", "a\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xF0\x9F\x92", "a\u{FFFD}");
// Lone highest continuation
decode_utf8_to_utf8(b"a\xBFZ", "a\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xBF", "a\u{FFFD}");
// Two lone highest continuations
decode_utf8_to_utf8(b"a\xBF\xBFZ", "a\u{FFFD}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xBF\xBF", "a\u{FFFD}\u{FFFD}");
// Low BMP followed by lowest lone continuation
decode_utf8_to_utf8(b"a\xC3\xA4\x80Z", "a\u{E4}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xC3\xA4\x80", "a\u{E4}\u{FFFD}");
// Low BMP followed by highest lone continuation
decode_utf8_to_utf8(b"a\xC3\xA4\xBFZ", "a\u{E4}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xC3\xA4\xBF", "a\u{E4}\u{FFFD}");
// High BMP followed by lowest lone continuation
decode_utf8_to_utf8(b"a\xE2\x98\x83\x80Z", "a\u{2603}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xE2\x98\x83\x80", "a\u{2603}\u{FFFD}");
// High BMP followed by highest lone continuation
decode_utf8_to_utf8(b"a\xE2\x98\x83\xBFZ", "a\u{2603}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xE2\x98\x83\xBF", "a\u{2603}\u{FFFD}");
// Astral followed by lowest lone continuation
decode_utf8_to_utf8(b"a\xF0\x9F\x92\xA9\x80Z", "a\u{1F4A9}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xF0\x9F\x92\xA9\x80", "a\u{1F4A9}\u{FFFD}");
// Astral followed by highest lone continuation
decode_utf8_to_utf8(b"a\xF0\x9F\x92\xA9\xBFZ", "a\u{1F4A9}\u{FFFD}Z");
decode_utf8_to_utf8(b"a\xF0\x9F\x92\xA9\xBF", "a\u{1F4A9}\u{FFFD}");
// Boundary conditions
// Lowest single-byte
decode_valid_utf8("Z\x00");
decode_valid_utf8("Z\x00Z");
// Lowest single-byte as two-byte overlong sequence
decode_utf8_to_utf8(b"a\xC0\x80", "a\u{FFFD}\u{FFFD}");
decode_utf8_to_utf8(b"a\xC0\x80Z", "a\u{FFFD}\u{FFFD}Z");
// Lowest single-byte as three-byte overlong sequence
decode_utf8_to_utf8(b"a\xE0\x80\x80", "a\u{FFFD}\u{FFFD}\u{FFFD}");
decode_utf8_to_utf8(b"a\xE0\x80\x80Z", "a\u{FFFD}\u{FFFD}\u{FFFD}Z");