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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
|
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/**
*******************************************************************************
* Copyright (C) 2006-2016, International Business Machines Corporation
* and others. All Rights Reserved.
*******************************************************************************
*/
#include "unicode/utypes.h"
#if !UCONFIG_NO_BREAK_ITERATION
#include "brkeng.h"
#include "dictbe.h"
#include "unicode/uniset.h"
#include "unicode/chariter.h"
#include "unicode/ubrk.h"
#include "uvectr32.h"
#include "uvector.h"
#include "uassert.h"
#include "unicode/normlzr.h"
#include "cmemory.h"
#include "dictionarydata.h"
U_NAMESPACE_BEGIN
/*
******************************************************************
*/
DictionaryBreakEngine::DictionaryBreakEngine() {
}
DictionaryBreakEngine::~DictionaryBreakEngine() {
}
UBool
DictionaryBreakEngine::handles(UChar32 c) const {
return fSet.contains(c);
}
int32_t
DictionaryBreakEngine::findBreaks( UText *text,
int32_t startPos,
int32_t endPos,
UVector32 &foundBreaks ) const {
(void)startPos; // TODO: remove this param?
int32_t result = 0;
// Find the span of characters included in the set.
// The span to break begins at the current position in the text, and
// extends towards the start or end of the text, depending on 'reverse'.
int32_t start = (int32_t)utext_getNativeIndex(text);
int32_t current;
int32_t rangeStart;
int32_t rangeEnd;
UChar32 c = utext_current32(text);
while((current = (int32_t)utext_getNativeIndex(text)) < endPos && fSet.contains(c)) {
utext_next32(text); // TODO: recast loop for postincrement
c = utext_current32(text);
}
rangeStart = start;
rangeEnd = current;
result = divideUpDictionaryRange(text, rangeStart, rangeEnd, foundBreaks);
utext_setNativeIndex(text, current);
return result;
}
void
DictionaryBreakEngine::setCharacters( const UnicodeSet &set ) {
fSet = set;
// Compact for caching
fSet.compact();
}
/*
******************************************************************
* PossibleWord
*/
// Helper class for improving readability of the Thai/Lao/Khmer word break
// algorithm. The implementation is completely inline.
// List size, limited by the maximum number of words in the dictionary
// that form a nested sequence.
static const int32_t POSSIBLE_WORD_LIST_MAX = 20;
class PossibleWord {
private:
// list of word candidate lengths, in increasing length order
// TODO: bytes would be sufficient for word lengths.
int32_t count; // Count of candidates
int32_t prefix; // The longest match with a dictionary word
int32_t offset; // Offset in the text of these candidates
int32_t mark; // The preferred candidate's offset
int32_t current; // The candidate we're currently looking at
int32_t cuLengths[POSSIBLE_WORD_LIST_MAX]; // Word Lengths, in code units.
int32_t cpLengths[POSSIBLE_WORD_LIST_MAX]; // Word Lengths, in code points.
public:
PossibleWord() : count(0), prefix(0), offset(-1), mark(0), current(0) {};
~PossibleWord() {};
// Fill the list of candidates if needed, select the longest, and return the number found
int32_t candidates( UText *text, DictionaryMatcher *dict, int32_t rangeEnd );
// Select the currently marked candidate, point after it in the text, and invalidate self
int32_t acceptMarked( UText *text );
// Back up from the current candidate to the next shorter one; return TRUE if that exists
// and point the text after it
UBool backUp( UText *text );
// Return the longest prefix this candidate location shares with a dictionary word
// Return value is in code points.
int32_t longestPrefix() { return prefix; };
// Mark the current candidate as the one we like
void markCurrent() { mark = current; };
// Get length in code points of the marked word.
int32_t markedCPLength() { return cpLengths[mark]; };
};
int32_t PossibleWord::candidates( UText *text, DictionaryMatcher *dict, int32_t rangeEnd ) {
// TODO: If getIndex is too slow, use offset < 0 and add discardAll()
int32_t start = (int32_t)utext_getNativeIndex(text);
if (start != offset) {
offset = start;
count = dict->matches(text, rangeEnd-start, UPRV_LENGTHOF(cuLengths), cuLengths, cpLengths, NULL, &prefix);
// Dictionary leaves text after longest prefix, not longest word. Back up.
if (count <= 0) {
utext_setNativeIndex(text, start);
}
}
if (count > 0) {
utext_setNativeIndex(text, start+cuLengths[count-1]);
}
current = count-1;
mark = current;
return count;
}
int32_t
PossibleWord::acceptMarked( UText *text ) {
utext_setNativeIndex(text, offset + cuLengths[mark]);
return cuLengths[mark];
}
UBool
PossibleWord::backUp( UText *text ) {
if (current > 0) {
utext_setNativeIndex(text, offset + cuLengths[--current]);
return TRUE;
}
return FALSE;
}
/*
******************************************************************
* ThaiBreakEngine
*/
// How many words in a row are "good enough"?
static const int32_t THAI_LOOKAHEAD = 3;
// Will not combine a non-word with a preceding dictionary word longer than this
static const int32_t THAI_ROOT_COMBINE_THRESHOLD = 3;
// Will not combine a non-word that shares at least this much prefix with a
// dictionary word, with a preceding word
static const int32_t THAI_PREFIX_COMBINE_THRESHOLD = 3;
// Ellision character
static const int32_t THAI_PAIYANNOI = 0x0E2F;
// Repeat character
static const int32_t THAI_MAIYAMOK = 0x0E46;
// Minimum word size
static const int32_t THAI_MIN_WORD = 2;
// Minimum number of characters for two words
static const int32_t THAI_MIN_WORD_SPAN = THAI_MIN_WORD * 2;
ThaiBreakEngine::ThaiBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status)
: DictionaryBreakEngine(),
fDictionary(adoptDictionary)
{
fThaiWordSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Thai:]&[:LineBreak=SA:]]"), status);
if (U_SUCCESS(status)) {
setCharacters(fThaiWordSet);
}
fMarkSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Thai:]&[:LineBreak=SA:]&[:M:]]"), status);
fMarkSet.add(0x0020);
fEndWordSet = fThaiWordSet;
fEndWordSet.remove(0x0E31); // MAI HAN-AKAT
fEndWordSet.remove(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI
fBeginWordSet.add(0x0E01, 0x0E2E); // KO KAI through HO NOKHUK
fBeginWordSet.add(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI
fSuffixSet.add(THAI_PAIYANNOI);
fSuffixSet.add(THAI_MAIYAMOK);
// Compact for caching.
fMarkSet.compact();
fEndWordSet.compact();
fBeginWordSet.compact();
fSuffixSet.compact();
}
ThaiBreakEngine::~ThaiBreakEngine() {
delete fDictionary;
}
int32_t
ThaiBreakEngine::divideUpDictionaryRange( UText *text,
int32_t rangeStart,
int32_t rangeEnd,
UVector32 &foundBreaks ) const {
utext_setNativeIndex(text, rangeStart);
utext_moveIndex32(text, THAI_MIN_WORD_SPAN);
if (utext_getNativeIndex(text) >= rangeEnd) {
return 0; // Not enough characters for two words
}
utext_setNativeIndex(text, rangeStart);
uint32_t wordsFound = 0;
int32_t cpWordLength = 0; // Word Length in Code Points.
int32_t cuWordLength = 0; // Word length in code units (UText native indexing)
int32_t current;
UErrorCode status = U_ZERO_ERROR;
PossibleWord words[THAI_LOOKAHEAD];
utext_setNativeIndex(text, rangeStart);
while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) {
cpWordLength = 0;
cuWordLength = 0;
// Look for candidate words at the current position
int32_t candidates = words[wordsFound%THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
// If we found exactly one, use that
if (candidates == 1) {
cuWordLength = words[wordsFound % THAI_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % THAI_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// If there was more than one, see which one can take us forward the most words
else if (candidates > 1) {
// If we're already at the end of the range, we're done
if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
do {
int32_t wordsMatched = 1;
if (words[(wordsFound + 1) % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) {
if (wordsMatched < 2) {
// Followed by another dictionary word; mark first word as a good candidate
words[wordsFound%THAI_LOOKAHEAD].markCurrent();
wordsMatched = 2;
}
// If we're already at the end of the range, we're done
if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
// See if any of the possible second words is followed by a third word
do {
// If we find a third word, stop right away
if (words[(wordsFound + 2) % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) {
words[wordsFound % THAI_LOOKAHEAD].markCurrent();
goto foundBest;
}
}
while (words[(wordsFound + 1) % THAI_LOOKAHEAD].backUp(text));
}
}
while (words[wordsFound % THAI_LOOKAHEAD].backUp(text));
foundBest:
// Set UText position to after the accepted word.
cuWordLength = words[wordsFound % THAI_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % THAI_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// We come here after having either found a word or not. We look ahead to the
// next word. If it's not a dictionary word, we will combine it with the word we
// just found (if there is one), but only if the preceding word does not exceed
// the threshold.
// The text iterator should now be positioned at the end of the word we found.
UChar32 uc = 0;
if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < THAI_ROOT_COMBINE_THRESHOLD) {
// if it is a dictionary word, do nothing. If it isn't, then if there is
// no preceding word, or the non-word shares less than the minimum threshold
// of characters with a dictionary word, then scan to resynchronize
if (words[wordsFound % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0
&& (cuWordLength == 0
|| words[wordsFound%THAI_LOOKAHEAD].longestPrefix() < THAI_PREFIX_COMBINE_THRESHOLD)) {
// Look for a plausible word boundary
int32_t remaining = rangeEnd - (current+cuWordLength);
UChar32 pc;
int32_t chars = 0;
for (;;) {
int32_t pcIndex = (int32_t)utext_getNativeIndex(text);
pc = utext_next32(text);
int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex;
chars += pcSize;
remaining -= pcSize;
if (remaining <= 0) {
break;
}
uc = utext_current32(text);
if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) {
// Maybe. See if it's in the dictionary.
// NOTE: In the original Apple code, checked that the next
// two characters after uc were not 0x0E4C THANTHAKHAT before
// checking the dictionary. That is just a performance filter,
// but it's not clear it's faster than checking the trie.
int32_t candidates = words[(wordsFound + 1) % THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
utext_setNativeIndex(text, current + cuWordLength + chars);
if (candidates > 0) {
break;
}
}
}
// Bump the word count if there wasn't already one
if (cuWordLength <= 0) {
wordsFound += 1;
}
// Update the length with the passed-over characters
cuWordLength += chars;
}
else {
// Back up to where we were for next iteration
utext_setNativeIndex(text, current+cuWordLength);
}
}
// Never stop before a combining mark.
int32_t currPos;
while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) {
utext_next32(text);
cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos;
}
// Look ahead for possible suffixes if a dictionary word does not follow.
// We do this in code rather than using a rule so that the heuristic
// resynch continues to function. For example, one of the suffix characters
// could be a typo in the middle of a word.
if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cuWordLength > 0) {
if (words[wordsFound%THAI_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0
&& fSuffixSet.contains(uc = utext_current32(text))) {
if (uc == THAI_PAIYANNOI) {
if (!fSuffixSet.contains(utext_previous32(text))) {
// Skip over previous end and PAIYANNOI
utext_next32(text);
int32_t paiyannoiIndex = (int32_t)utext_getNativeIndex(text);
utext_next32(text);
cuWordLength += (int32_t)utext_getNativeIndex(text) - paiyannoiIndex; // Add PAIYANNOI to word
uc = utext_current32(text); // Fetch next character
}
else {
// Restore prior position
utext_next32(text);
}
}
if (uc == THAI_MAIYAMOK) {
if (utext_previous32(text) != THAI_MAIYAMOK) {
// Skip over previous end and MAIYAMOK
utext_next32(text);
int32_t maiyamokIndex = (int32_t)utext_getNativeIndex(text);
utext_next32(text);
cuWordLength += (int32_t)utext_getNativeIndex(text) - maiyamokIndex; // Add MAIYAMOK to word
}
else {
// Restore prior position
utext_next32(text);
}
}
}
else {
utext_setNativeIndex(text, current+cuWordLength);
}
}
// Did we find a word on this iteration? If so, push it on the break stack
if (cuWordLength > 0) {
foundBreaks.push((current+cuWordLength), status);
}
}
// Don't return a break for the end of the dictionary range if there is one there.
if (foundBreaks.peeki() >= rangeEnd) {
(void) foundBreaks.popi();
wordsFound -= 1;
}
return wordsFound;
}
/*
******************************************************************
* LaoBreakEngine
*/
// How many words in a row are "good enough"?
static const int32_t LAO_LOOKAHEAD = 3;
// Will not combine a non-word with a preceding dictionary word longer than this
static const int32_t LAO_ROOT_COMBINE_THRESHOLD = 3;
// Will not combine a non-word that shares at least this much prefix with a
// dictionary word, with a preceding word
static const int32_t LAO_PREFIX_COMBINE_THRESHOLD = 3;
// Minimum word size
static const int32_t LAO_MIN_WORD = 2;
// Minimum number of characters for two words
static const int32_t LAO_MIN_WORD_SPAN = LAO_MIN_WORD * 2;
LaoBreakEngine::LaoBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status)
: DictionaryBreakEngine(),
fDictionary(adoptDictionary)
{
fLaoWordSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Laoo:]&[:LineBreak=SA:]]"), status);
if (U_SUCCESS(status)) {
setCharacters(fLaoWordSet);
}
fMarkSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Laoo:]&[:LineBreak=SA:]&[:M:]]"), status);
fMarkSet.add(0x0020);
fEndWordSet = fLaoWordSet;
fEndWordSet.remove(0x0EC0, 0x0EC4); // prefix vowels
fBeginWordSet.add(0x0E81, 0x0EAE); // basic consonants (including holes for corresponding Thai characters)
fBeginWordSet.add(0x0EDC, 0x0EDD); // digraph consonants (no Thai equivalent)
fBeginWordSet.add(0x0EC0, 0x0EC4); // prefix vowels
// Compact for caching.
fMarkSet.compact();
fEndWordSet.compact();
fBeginWordSet.compact();
}
LaoBreakEngine::~LaoBreakEngine() {
delete fDictionary;
}
int32_t
LaoBreakEngine::divideUpDictionaryRange( UText *text,
int32_t rangeStart,
int32_t rangeEnd,
UVector32 &foundBreaks ) const {
if ((rangeEnd - rangeStart) < LAO_MIN_WORD_SPAN) {
return 0; // Not enough characters for two words
}
uint32_t wordsFound = 0;
int32_t cpWordLength = 0;
int32_t cuWordLength = 0;
int32_t current;
UErrorCode status = U_ZERO_ERROR;
PossibleWord words[LAO_LOOKAHEAD];
utext_setNativeIndex(text, rangeStart);
while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) {
cuWordLength = 0;
cpWordLength = 0;
// Look for candidate words at the current position
int32_t candidates = words[wordsFound%LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
// If we found exactly one, use that
if (candidates == 1) {
cuWordLength = words[wordsFound % LAO_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % LAO_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// If there was more than one, see which one can take us forward the most words
else if (candidates > 1) {
// If we're already at the end of the range, we're done
if (utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
do {
int32_t wordsMatched = 1;
if (words[(wordsFound + 1) % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) {
if (wordsMatched < 2) {
// Followed by another dictionary word; mark first word as a good candidate
words[wordsFound%LAO_LOOKAHEAD].markCurrent();
wordsMatched = 2;
}
// If we're already at the end of the range, we're done
if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
// See if any of the possible second words is followed by a third word
do {
// If we find a third word, stop right away
if (words[(wordsFound + 2) % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) {
words[wordsFound % LAO_LOOKAHEAD].markCurrent();
goto foundBest;
}
}
while (words[(wordsFound + 1) % LAO_LOOKAHEAD].backUp(text));
}
}
while (words[wordsFound % LAO_LOOKAHEAD].backUp(text));
foundBest:
cuWordLength = words[wordsFound % LAO_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % LAO_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// We come here after having either found a word or not. We look ahead to the
// next word. If it's not a dictionary word, we will combine it withe the word we
// just found (if there is one), but only if the preceding word does not exceed
// the threshold.
// The text iterator should now be positioned at the end of the word we found.
if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < LAO_ROOT_COMBINE_THRESHOLD) {
// if it is a dictionary word, do nothing. If it isn't, then if there is
// no preceding word, or the non-word shares less than the minimum threshold
// of characters with a dictionary word, then scan to resynchronize
if (words[wordsFound % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0
&& (cuWordLength == 0
|| words[wordsFound%LAO_LOOKAHEAD].longestPrefix() < LAO_PREFIX_COMBINE_THRESHOLD)) {
// Look for a plausible word boundary
int32_t remaining = rangeEnd - (current + cuWordLength);
UChar32 pc;
UChar32 uc;
int32_t chars = 0;
for (;;) {
int32_t pcIndex = (int32_t)utext_getNativeIndex(text);
pc = utext_next32(text);
int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex;
chars += pcSize;
remaining -= pcSize;
if (remaining <= 0) {
break;
}
uc = utext_current32(text);
if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) {
// Maybe. See if it's in the dictionary.
// TODO: this looks iffy; compare with old code.
int32_t candidates = words[(wordsFound + 1) % LAO_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
utext_setNativeIndex(text, current + cuWordLength + chars);
if (candidates > 0) {
break;
}
}
}
// Bump the word count if there wasn't already one
if (cuWordLength <= 0) {
wordsFound += 1;
}
// Update the length with the passed-over characters
cuWordLength += chars;
}
else {
// Back up to where we were for next iteration
utext_setNativeIndex(text, current + cuWordLength);
}
}
// Never stop before a combining mark.
int32_t currPos;
while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) {
utext_next32(text);
cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos;
}
// Look ahead for possible suffixes if a dictionary word does not follow.
// We do this in code rather than using a rule so that the heuristic
// resynch continues to function. For example, one of the suffix characters
// could be a typo in the middle of a word.
// NOT CURRENTLY APPLICABLE TO LAO
// Did we find a word on this iteration? If so, push it on the break stack
if (cuWordLength > 0) {
foundBreaks.push((current+cuWordLength), status);
}
}
// Don't return a break for the end of the dictionary range if there is one there.
if (foundBreaks.peeki() >= rangeEnd) {
(void) foundBreaks.popi();
wordsFound -= 1;
}
return wordsFound;
}
/*
******************************************************************
* BurmeseBreakEngine
*/
// How many words in a row are "good enough"?
static const int32_t BURMESE_LOOKAHEAD = 3;
// Will not combine a non-word with a preceding dictionary word longer than this
static const int32_t BURMESE_ROOT_COMBINE_THRESHOLD = 3;
// Will not combine a non-word that shares at least this much prefix with a
// dictionary word, with a preceding word
static const int32_t BURMESE_PREFIX_COMBINE_THRESHOLD = 3;
// Minimum word size
static const int32_t BURMESE_MIN_WORD = 2;
// Minimum number of characters for two words
static const int32_t BURMESE_MIN_WORD_SPAN = BURMESE_MIN_WORD * 2;
BurmeseBreakEngine::BurmeseBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status)
: DictionaryBreakEngine(),
fDictionary(adoptDictionary)
{
fBurmeseWordSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Mymr:]&[:LineBreak=SA:]]"), status);
if (U_SUCCESS(status)) {
setCharacters(fBurmeseWordSet);
}
fMarkSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Mymr:]&[:LineBreak=SA:]&[:M:]]"), status);
fMarkSet.add(0x0020);
fEndWordSet = fBurmeseWordSet;
fBeginWordSet.add(0x1000, 0x102A); // basic consonants and independent vowels
// Compact for caching.
fMarkSet.compact();
fEndWordSet.compact();
fBeginWordSet.compact();
}
BurmeseBreakEngine::~BurmeseBreakEngine() {
delete fDictionary;
}
int32_t
BurmeseBreakEngine::divideUpDictionaryRange( UText *text,
int32_t rangeStart,
int32_t rangeEnd,
UVector32 &foundBreaks ) const {
if ((rangeEnd - rangeStart) < BURMESE_MIN_WORD_SPAN) {
return 0; // Not enough characters for two words
}
uint32_t wordsFound = 0;
int32_t cpWordLength = 0;
int32_t cuWordLength = 0;
int32_t current;
UErrorCode status = U_ZERO_ERROR;
PossibleWord words[BURMESE_LOOKAHEAD];
utext_setNativeIndex(text, rangeStart);
while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) {
cuWordLength = 0;
cpWordLength = 0;
// Look for candidate words at the current position
int32_t candidates = words[wordsFound%BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
// If we found exactly one, use that
if (candidates == 1) {
cuWordLength = words[wordsFound % BURMESE_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % BURMESE_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// If there was more than one, see which one can take us forward the most words
else if (candidates > 1) {
// If we're already at the end of the range, we're done
if (utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
do {
int32_t wordsMatched = 1;
if (words[(wordsFound + 1) % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) {
if (wordsMatched < 2) {
// Followed by another dictionary word; mark first word as a good candidate
words[wordsFound%BURMESE_LOOKAHEAD].markCurrent();
wordsMatched = 2;
}
// If we're already at the end of the range, we're done
if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
// See if any of the possible second words is followed by a third word
do {
// If we find a third word, stop right away
if (words[(wordsFound + 2) % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) {
words[wordsFound % BURMESE_LOOKAHEAD].markCurrent();
goto foundBest;
}
}
while (words[(wordsFound + 1) % BURMESE_LOOKAHEAD].backUp(text));
}
}
while (words[wordsFound % BURMESE_LOOKAHEAD].backUp(text));
foundBest:
cuWordLength = words[wordsFound % BURMESE_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % BURMESE_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// We come here after having either found a word or not. We look ahead to the
// next word. If it's not a dictionary word, we will combine it withe the word we
// just found (if there is one), but only if the preceding word does not exceed
// the threshold.
// The text iterator should now be positioned at the end of the word we found.
if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < BURMESE_ROOT_COMBINE_THRESHOLD) {
// if it is a dictionary word, do nothing. If it isn't, then if there is
// no preceding word, or the non-word shares less than the minimum threshold
// of characters with a dictionary word, then scan to resynchronize
if (words[wordsFound % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0
&& (cuWordLength == 0
|| words[wordsFound%BURMESE_LOOKAHEAD].longestPrefix() < BURMESE_PREFIX_COMBINE_THRESHOLD)) {
// Look for a plausible word boundary
int32_t remaining = rangeEnd - (current + cuWordLength);
UChar32 pc;
UChar32 uc;
int32_t chars = 0;
for (;;) {
int32_t pcIndex = (int32_t)utext_getNativeIndex(text);
pc = utext_next32(text);
int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex;
chars += pcSize;
remaining -= pcSize;
if (remaining <= 0) {
break;
}
uc = utext_current32(text);
if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) {
// Maybe. See if it's in the dictionary.
// TODO: this looks iffy; compare with old code.
int32_t candidates = words[(wordsFound + 1) % BURMESE_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
utext_setNativeIndex(text, current + cuWordLength + chars);
if (candidates > 0) {
break;
}
}
}
// Bump the word count if there wasn't already one
if (cuWordLength <= 0) {
wordsFound += 1;
}
// Update the length with the passed-over characters
cuWordLength += chars;
}
else {
// Back up to where we were for next iteration
utext_setNativeIndex(text, current + cuWordLength);
}
}
// Never stop before a combining mark.
int32_t currPos;
while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) {
utext_next32(text);
cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos;
}
// Look ahead for possible suffixes if a dictionary word does not follow.
// We do this in code rather than using a rule so that the heuristic
// resynch continues to function. For example, one of the suffix characters
// could be a typo in the middle of a word.
// NOT CURRENTLY APPLICABLE TO BURMESE
// Did we find a word on this iteration? If so, push it on the break stack
if (cuWordLength > 0) {
foundBreaks.push((current+cuWordLength), status);
}
}
// Don't return a break for the end of the dictionary range if there is one there.
if (foundBreaks.peeki() >= rangeEnd) {
(void) foundBreaks.popi();
wordsFound -= 1;
}
return wordsFound;
}
/*
******************************************************************
* KhmerBreakEngine
*/
// How many words in a row are "good enough"?
static const int32_t KHMER_LOOKAHEAD = 3;
// Will not combine a non-word with a preceding dictionary word longer than this
static const int32_t KHMER_ROOT_COMBINE_THRESHOLD = 3;
// Will not combine a non-word that shares at least this much prefix with a
// dictionary word, with a preceding word
static const int32_t KHMER_PREFIX_COMBINE_THRESHOLD = 3;
// Minimum word size
static const int32_t KHMER_MIN_WORD = 2;
// Minimum number of characters for two words
static const int32_t KHMER_MIN_WORD_SPAN = KHMER_MIN_WORD * 2;
KhmerBreakEngine::KhmerBreakEngine(DictionaryMatcher *adoptDictionary, UErrorCode &status)
: DictionaryBreakEngine(),
fDictionary(adoptDictionary)
{
fKhmerWordSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Khmr:]&[:LineBreak=SA:]]"), status);
if (U_SUCCESS(status)) {
setCharacters(fKhmerWordSet);
}
fMarkSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Khmr:]&[:LineBreak=SA:]&[:M:]]"), status);
fMarkSet.add(0x0020);
fEndWordSet = fKhmerWordSet;
fBeginWordSet.add(0x1780, 0x17B3);
//fBeginWordSet.add(0x17A3, 0x17A4); // deprecated vowels
//fEndWordSet.remove(0x17A5, 0x17A9); // Khmer independent vowels that can't end a word
//fEndWordSet.remove(0x17B2); // Khmer independent vowel that can't end a word
fEndWordSet.remove(0x17D2); // KHMER SIGN COENG that combines some following characters
//fEndWordSet.remove(0x17B6, 0x17C5); // Remove dependent vowels
// fEndWordSet.remove(0x0E31); // MAI HAN-AKAT
// fEndWordSet.remove(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI
// fBeginWordSet.add(0x0E01, 0x0E2E); // KO KAI through HO NOKHUK
// fBeginWordSet.add(0x0E40, 0x0E44); // SARA E through SARA AI MAIMALAI
// fSuffixSet.add(THAI_PAIYANNOI);
// fSuffixSet.add(THAI_MAIYAMOK);
// Compact for caching.
fMarkSet.compact();
fEndWordSet.compact();
fBeginWordSet.compact();
// fSuffixSet.compact();
}
KhmerBreakEngine::~KhmerBreakEngine() {
delete fDictionary;
}
int32_t
KhmerBreakEngine::divideUpDictionaryRange( UText *text,
int32_t rangeStart,
int32_t rangeEnd,
UVector32 &foundBreaks ) const {
if ((rangeEnd - rangeStart) < KHMER_MIN_WORD_SPAN) {
return 0; // Not enough characters for two words
}
uint32_t wordsFound = 0;
int32_t cpWordLength = 0;
int32_t cuWordLength = 0;
int32_t current;
UErrorCode status = U_ZERO_ERROR;
PossibleWord words[KHMER_LOOKAHEAD];
utext_setNativeIndex(text, rangeStart);
while (U_SUCCESS(status) && (current = (int32_t)utext_getNativeIndex(text)) < rangeEnd) {
cuWordLength = 0;
cpWordLength = 0;
// Look for candidate words at the current position
int32_t candidates = words[wordsFound%KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
// If we found exactly one, use that
if (candidates == 1) {
cuWordLength = words[wordsFound % KHMER_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % KHMER_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// If there was more than one, see which one can take us forward the most words
else if (candidates > 1) {
// If we're already at the end of the range, we're done
if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
do {
int32_t wordsMatched = 1;
if (words[(wordsFound + 1) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) > 0) {
if (wordsMatched < 2) {
// Followed by another dictionary word; mark first word as a good candidate
words[wordsFound % KHMER_LOOKAHEAD].markCurrent();
wordsMatched = 2;
}
// If we're already at the end of the range, we're done
if ((int32_t)utext_getNativeIndex(text) >= rangeEnd) {
goto foundBest;
}
// See if any of the possible second words is followed by a third word
do {
// If we find a third word, stop right away
if (words[(wordsFound + 2) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd)) {
words[wordsFound % KHMER_LOOKAHEAD].markCurrent();
goto foundBest;
}
}
while (words[(wordsFound + 1) % KHMER_LOOKAHEAD].backUp(text));
}
}
while (words[wordsFound % KHMER_LOOKAHEAD].backUp(text));
foundBest:
cuWordLength = words[wordsFound % KHMER_LOOKAHEAD].acceptMarked(text);
cpWordLength = words[wordsFound % KHMER_LOOKAHEAD].markedCPLength();
wordsFound += 1;
}
// We come here after having either found a word or not. We look ahead to the
// next word. If it's not a dictionary word, we will combine it with the word we
// just found (if there is one), but only if the preceding word does not exceed
// the threshold.
// The text iterator should now be positioned at the end of the word we found.
if ((int32_t)utext_getNativeIndex(text) < rangeEnd && cpWordLength < KHMER_ROOT_COMBINE_THRESHOLD) {
// if it is a dictionary word, do nothing. If it isn't, then if there is
// no preceding word, or the non-word shares less than the minimum threshold
// of characters with a dictionary word, then scan to resynchronize
if (words[wordsFound % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0
&& (cuWordLength == 0
|| words[wordsFound % KHMER_LOOKAHEAD].longestPrefix() < KHMER_PREFIX_COMBINE_THRESHOLD)) {
// Look for a plausible word boundary
int32_t remaining = rangeEnd - (current+cuWordLength);
UChar32 pc;
UChar32 uc;
int32_t chars = 0;
for (;;) {
int32_t pcIndex = (int32_t)utext_getNativeIndex(text);
pc = utext_next32(text);
int32_t pcSize = (int32_t)utext_getNativeIndex(text) - pcIndex;
chars += pcSize;
remaining -= pcSize;
if (remaining <= 0) {
break;
}
uc = utext_current32(text);
if (fEndWordSet.contains(pc) && fBeginWordSet.contains(uc)) {
// Maybe. See if it's in the dictionary.
int32_t candidates = words[(wordsFound + 1) % KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd);
utext_setNativeIndex(text, current+cuWordLength+chars);
if (candidates > 0) {
break;
}
}
}
// Bump the word count if there wasn't already one
if (cuWordLength <= 0) {
wordsFound += 1;
}
// Update the length with the passed-over characters
cuWordLength += chars;
}
else {
// Back up to where we were for next iteration
utext_setNativeIndex(text, current+cuWordLength);
}
}
// Never stop before a combining mark.
int32_t currPos;
while ((currPos = (int32_t)utext_getNativeIndex(text)) < rangeEnd && fMarkSet.contains(utext_current32(text))) {
utext_next32(text);
cuWordLength += (int32_t)utext_getNativeIndex(text) - currPos;
}
// Look ahead for possible suffixes if a dictionary word does not follow.
// We do this in code rather than using a rule so that the heuristic
// resynch continues to function. For example, one of the suffix characters
// could be a typo in the middle of a word.
// if ((int32_t)utext_getNativeIndex(text) < rangeEnd && wordLength > 0) {
// if (words[wordsFound%KHMER_LOOKAHEAD].candidates(text, fDictionary, rangeEnd) <= 0
// && fSuffixSet.contains(uc = utext_current32(text))) {
// if (uc == KHMER_PAIYANNOI) {
// if (!fSuffixSet.contains(utext_previous32(text))) {
// // Skip over previous end and PAIYANNOI
// utext_next32(text);
// utext_next32(text);
// wordLength += 1; // Add PAIYANNOI to word
// uc = utext_current32(text); // Fetch next character
// }
// else {
// // Restore prior position
// utext_next32(text);
// }
// }
// if (uc == KHMER_MAIYAMOK) {
// if (utext_previous32(text) != KHMER_MAIYAMOK) {
// // Skip over previous end and MAIYAMOK
// utext_next32(text);
// utext_next32(text);
// wordLength += 1; // Add MAIYAMOK to word
// }
// else {
// // Restore prior position
// utext_next32(text);
// }
// }
// }
// else {
// utext_setNativeIndex(text, current+wordLength);
// }
// }
// Did we find a word on this iteration? If so, push it on the break stack
if (cuWordLength > 0) {
foundBreaks.push((current+cuWordLength), status);
}
}
// Don't return a break for the end of the dictionary range if there is one there.
if (foundBreaks.peeki() >= rangeEnd) {
(void) foundBreaks.popi();
wordsFound -= 1;
}
return wordsFound;
}
#if !UCONFIG_NO_NORMALIZATION
/*
******************************************************************
* CjkBreakEngine
*/
static const uint32_t kuint32max = 0xFFFFFFFF;
CjkBreakEngine::CjkBreakEngine(DictionaryMatcher *adoptDictionary, LanguageType type, UErrorCode &status)
: DictionaryBreakEngine(), fDictionary(adoptDictionary) {
// Korean dictionary only includes Hangul syllables
fHangulWordSet.applyPattern(UNICODE_STRING_SIMPLE("[\\uac00-\\ud7a3]"), status);
fHanWordSet.applyPattern(UNICODE_STRING_SIMPLE("[:Han:]"), status);
fKatakanaWordSet.applyPattern(UNICODE_STRING_SIMPLE("[[:Katakana:]\\uff9e\\uff9f]"), status);
fHiraganaWordSet.applyPattern(UNICODE_STRING_SIMPLE("[:Hiragana:]"), status);
nfkcNorm2 = Normalizer2::getNFKCInstance(status);
if (U_SUCCESS(status)) {
// handle Korean and Japanese/Chinese using different dictionaries
if (type == kKorean) {
setCharacters(fHangulWordSet);
} else { //Chinese and Japanese
UnicodeSet cjSet;
cjSet.addAll(fHanWordSet);
cjSet.addAll(fKatakanaWordSet);
cjSet.addAll(fHiraganaWordSet);
cjSet.add(0xFF70); // HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK
cjSet.add(0x30FC); // KATAKANA-HIRAGANA PROLONGED SOUND MARK
setCharacters(cjSet);
}
}
}
CjkBreakEngine::~CjkBreakEngine(){
delete fDictionary;
}
// The katakanaCost values below are based on the length frequencies of all
// katakana phrases in the dictionary
static const int32_t kMaxKatakanaLength = 8;
static const int32_t kMaxKatakanaGroupLength = 20;
static const uint32_t maxSnlp = 255;
static inline uint32_t getKatakanaCost(int32_t wordLength){
//TODO: fill array with actual values from dictionary!
static const uint32_t katakanaCost[kMaxKatakanaLength + 1]
= {8192, 984, 408, 240, 204, 252, 300, 372, 480};
return (wordLength > kMaxKatakanaLength) ? 8192 : katakanaCost[wordLength];
}
static inline bool isKatakana(UChar32 value) {
return (value >= 0x30A1 && value <= 0x30FE && value != 0x30FB) ||
(value >= 0xFF66 && value <= 0xFF9f);
}
// Function for accessing internal utext flags.
// Replicates an internal UText function.
static inline int32_t utext_i32_flag(int32_t bitIndex) {
return (int32_t)1 << bitIndex;
}
/*
* @param text A UText representing the text
* @param rangeStart The start of the range of dictionary characters
* @param rangeEnd The end of the range of dictionary characters
* @param foundBreaks vector<int32> to receive the break positions
* @return The number of breaks found
*/
int32_t
CjkBreakEngine::divideUpDictionaryRange( UText *inText,
int32_t rangeStart,
int32_t rangeEnd,
UVector32 &foundBreaks ) const {
if (rangeStart >= rangeEnd) {
return 0;
}
// UnicodeString version of input UText, NFKC normalized if necessary.
UnicodeString inString;
// inputMap[inStringIndex] = corresponding native index from UText inText.
// If NULL then mapping is 1:1
LocalPointer<UVector32> inputMap;
UErrorCode status = U_ZERO_ERROR;
// if UText has the input string as one contiguous UTF-16 chunk
if ((inText->providerProperties & utext_i32_flag(UTEXT_PROVIDER_STABLE_CHUNKS)) &&
inText->chunkNativeStart <= rangeStart &&
inText->chunkNativeLimit >= rangeEnd &&
inText->nativeIndexingLimit >= rangeEnd - inText->chunkNativeStart) {
// Input UText is in one contiguous UTF-16 chunk.
// Use Read-only aliasing UnicodeString.
inString.setTo(FALSE,
inText->chunkContents + rangeStart - inText->chunkNativeStart,
rangeEnd - rangeStart);
} else {
// Copy the text from the original inText (UText) to inString (UnicodeString).
// Create a map from UnicodeString indices -> UText offsets.
utext_setNativeIndex(inText, rangeStart);
int32_t limit = rangeEnd;
U_ASSERT(limit <= utext_nativeLength(inText));
if (limit > utext_nativeLength(inText)) {
limit = (int32_t)utext_nativeLength(inText);
}
inputMap.adoptInsteadAndCheckErrorCode(new UVector32(status), status);
if (U_FAILURE(status)) {
return 0;
}
while (utext_getNativeIndex(inText) < limit) {
int32_t nativePosition = (int32_t)utext_getNativeIndex(inText);
UChar32 c = utext_next32(inText);
U_ASSERT(c != U_SENTINEL);
inString.append(c);
while (inputMap->size() < inString.length()) {
inputMap->addElement(nativePosition, status);
}
}
inputMap->addElement(limit, status);
}
if (!nfkcNorm2->isNormalized(inString, status)) {
UnicodeString normalizedInput;
// normalizedMap[normalizedInput position] == original UText position.
LocalPointer<UVector32> normalizedMap(new UVector32(status), status);
if (U_FAILURE(status)) {
return 0;
}
UnicodeString fragment;
UnicodeString normalizedFragment;
for (int32_t srcI = 0; srcI < inString.length();) { // Once per normalization chunk
fragment.remove();
int32_t fragmentStartI = srcI;
UChar32 c = inString.char32At(srcI);
for (;;) {
fragment.append(c);
srcI = inString.moveIndex32(srcI, 1);
if (srcI == inString.length()) {
break;
}
c = inString.char32At(srcI);
if (nfkcNorm2->hasBoundaryBefore(c)) {
break;
}
}
nfkcNorm2->normalize(fragment, normalizedFragment, status);
normalizedInput.append(normalizedFragment);
// Map every position in the normalized chunk to the start of the chunk
// in the original input.
int32_t fragmentOriginalStart = inputMap.isValid() ?
inputMap->elementAti(fragmentStartI) : fragmentStartI+rangeStart;
while (normalizedMap->size() < normalizedInput.length()) {
normalizedMap->addElement(fragmentOriginalStart, status);
if (U_FAILURE(status)) {
break;
}
}
}
U_ASSERT(normalizedMap->size() == normalizedInput.length());
int32_t nativeEnd = inputMap.isValid() ?
inputMap->elementAti(inString.length()) : inString.length()+rangeStart;
normalizedMap->addElement(nativeEnd, status);
inputMap.moveFrom(normalizedMap);
inString.moveFrom(normalizedInput);
}
int32_t numCodePts = inString.countChar32();
if (numCodePts != inString.length()) {
// There are supplementary characters in the input.
// The dictionary will produce boundary positions in terms of code point indexes,
// not in terms of code unit string indexes.
// Use the inputMap mechanism to take care of this in addition to indexing differences
// from normalization and/or UTF-8 input.
UBool hadExistingMap = inputMap.isValid();
if (!hadExistingMap) {
inputMap.adoptInsteadAndCheckErrorCode(new UVector32(status), status);
if (U_FAILURE(status)) {
return 0;
}
}
int32_t cpIdx = 0;
for (int32_t cuIdx = 0; ; cuIdx = inString.moveIndex32(cuIdx, 1)) {
U_ASSERT(cuIdx >= cpIdx);
if (hadExistingMap) {
inputMap->setElementAt(inputMap->elementAti(cuIdx), cpIdx);
} else {
inputMap->addElement(cuIdx+rangeStart, status);
}
cpIdx++;
if (cuIdx == inString.length()) {
break;
}
}
}
// bestSnlp[i] is the snlp of the best segmentation of the first i
// code points in the range to be matched.
UVector32 bestSnlp(numCodePts + 1, status);
bestSnlp.addElement(0, status);
for(int32_t i = 1; i <= numCodePts; i++) {
bestSnlp.addElement(kuint32max, status);
}
// prev[i] is the index of the last CJK code point in the previous word in
// the best segmentation of the first i characters.
UVector32 prev(numCodePts + 1, status);
for(int32_t i = 0; i <= numCodePts; i++){
prev.addElement(-1, status);
}
const int32_t maxWordSize = 20;
UVector32 values(numCodePts, status);
values.setSize(numCodePts);
UVector32 lengths(numCodePts, status);
lengths.setSize(numCodePts);
UText fu = UTEXT_INITIALIZER;
utext_openUnicodeString(&fu, &inString, &status);
// Dynamic programming to find the best segmentation.
// In outer loop, i is the code point index,
// ix is the corresponding string (code unit) index.
// They differ when the string contains supplementary characters.
int32_t ix = 0;
bool is_prev_katakana = false;
for (int32_t i = 0; i < numCodePts; ++i, ix = inString.moveIndex32(ix, 1)) {
if ((uint32_t)bestSnlp.elementAti(i) == kuint32max) {
continue;
}
int32_t count;
utext_setNativeIndex(&fu, ix);
count = fDictionary->matches(&fu, maxWordSize, numCodePts,
NULL, lengths.getBuffer(), values.getBuffer(), NULL);
// Note: lengths is filled with code point lengths
// The NULL parameter is the ignored code unit lengths.
// if there are no single character matches found in the dictionary
// starting with this character, treat character as a 1-character word
// with the highest value possible, i.e. the least likely to occur.
// Exclude Korean characters from this treatment, as they should be left
// together by default.
if ((count == 0 || lengths.elementAti(0) != 1) &&
!fHangulWordSet.contains(inString.char32At(ix))) {
values.setElementAt(maxSnlp, count); // 255
lengths.setElementAt(1, count++);
}
for (int32_t j = 0; j < count; j++) {
uint32_t newSnlp = (uint32_t)bestSnlp.elementAti(i) + (uint32_t)values.elementAti(j);
int32_t ln_j_i = lengths.elementAti(j) + i;
if (newSnlp < (uint32_t)bestSnlp.elementAti(ln_j_i)) {
bestSnlp.setElementAt(newSnlp, ln_j_i);
prev.setElementAt(i, ln_j_i);
}
}
// In Japanese,
// Katakana word in single character is pretty rare. So we apply
// the following heuristic to Katakana: any continuous run of Katakana
// characters is considered a candidate word with a default cost
// specified in the katakanaCost table according to its length.
bool is_katakana = isKatakana(inString.char32At(ix));
int32_t katakanaRunLength = 1;
if (!is_prev_katakana && is_katakana) {
int32_t j = inString.moveIndex32(ix, 1);
// Find the end of the continuous run of Katakana characters
while (j < inString.length() && katakanaRunLength < kMaxKatakanaGroupLength &&
isKatakana(inString.char32At(j))) {
j = inString.moveIndex32(j, 1);
katakanaRunLength++;
}
if (katakanaRunLength < kMaxKatakanaGroupLength) {
uint32_t newSnlp = bestSnlp.elementAti(i) + getKatakanaCost(katakanaRunLength);
if (newSnlp < (uint32_t)bestSnlp.elementAti(i+katakanaRunLength)) {
bestSnlp.setElementAt(newSnlp, i+katakanaRunLength);
prev.setElementAt(i, i+katakanaRunLength); // prev[j] = i;
}
}
}
is_prev_katakana = is_katakana;
}
utext_close(&fu);
// Start pushing the optimal offset index into t_boundary (t for tentative).
// prev[numCodePts] is guaranteed to be meaningful.
// We'll first push in the reverse order, i.e.,
// t_boundary[0] = numCodePts, and afterwards do a swap.
UVector32 t_boundary(numCodePts+1, status);
int32_t numBreaks = 0;
// No segmentation found, set boundary to end of range
if ((uint32_t)bestSnlp.elementAti(numCodePts) == kuint32max) {
t_boundary.addElement(numCodePts, status);
numBreaks++;
} else {
for (int32_t i = numCodePts; i > 0; i = prev.elementAti(i)) {
t_boundary.addElement(i, status);
numBreaks++;
}
U_ASSERT(prev.elementAti(t_boundary.elementAti(numBreaks - 1)) == 0);
}
// Add a break for the start of the dictionary range if there is not one
// there already.
if (foundBreaks.size() == 0 || foundBreaks.peeki() < rangeStart) {
t_boundary.addElement(0, status);
numBreaks++;
}
// Now that we're done, convert positions in t_boundary[] (indices in
// the normalized input string) back to indices in the original input UText
// while reversing t_boundary and pushing values to foundBreaks.
int32_t prevCPPos = -1;
int32_t prevUTextPos = -1;
for (int32_t i = numBreaks-1; i >= 0; i--) {
int32_t cpPos = t_boundary.elementAti(i);
U_ASSERT(cpPos > prevCPPos);
int32_t utextPos = inputMap.isValid() ? inputMap->elementAti(cpPos) : cpPos + rangeStart;
U_ASSERT(utextPos >= prevUTextPos);
if (utextPos > prevUTextPos) {
// Boundaries are added to foundBreaks output in ascending order.
U_ASSERT(foundBreaks.size() == 0 || foundBreaks.peeki() < utextPos);
foundBreaks.push(utextPos, status);
} else {
// Normalization expanded the input text, the dictionary found a boundary
// within the expansion, giving two boundaries with the same index in the
// original text. Ignore the second. See ticket #12918.
--numBreaks;
}
prevCPPos = cpPos;
prevUTextPos = utextPos;
}
(void)prevCPPos; // suppress compiler warnings about unused variable
// inString goes out of scope
// inputMap goes out of scope
return numBreaks;
}
#endif
U_NAMESPACE_END
#endif /* #if !UCONFIG_NO_BREAK_ITERATION */
|