• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 * Copyright (C) 2012-2014, International Business Machines
6 * Corporation and others.  All Rights Reserved.
7 *******************************************************************************
8 * utf8collationiterator.cpp
9 *
10 * created on: 2012nov12 (from utf16collationiterator.cpp & uitercollationiterator.cpp)
11 * created by: Markus W. Scherer
12 */
13 
14 #include "unicode/utypes.h"
15 
16 #if !UCONFIG_NO_COLLATION
17 
18 #include "unicode/utf8.h"
19 #include "charstr.h"
20 #include "cmemory.h"
21 #include "collation.h"
22 #include "collationdata.h"
23 #include "collationfcd.h"
24 #include "collationiterator.h"
25 #include "normalizer2impl.h"
26 #include "uassert.h"
27 #include "utf8collationiterator.h"
28 
29 U_NAMESPACE_BEGIN
30 
~UTF8CollationIterator()31 UTF8CollationIterator::~UTF8CollationIterator() {}
32 
33 void
resetToOffset(int32_t newOffset)34 UTF8CollationIterator::resetToOffset(int32_t newOffset) {
35     reset();
36     pos = newOffset;
37 }
38 
39 int32_t
getOffset() const40 UTF8CollationIterator::getOffset() const {
41     return pos;
42 }
43 
44 uint32_t
handleNextCE32(UChar32 & c,UErrorCode &)45 UTF8CollationIterator::handleNextCE32(UChar32 &c, UErrorCode & /*errorCode*/) {
46     if(pos == length) {
47         c = U_SENTINEL;
48         return Collation::FALLBACK_CE32;
49     }
50     // Optimized combination of U8_NEXT_OR_FFFD() and UTRIE2_U8_NEXT32().
51     c = u8[pos++];
52     if(U8_IS_SINGLE(c)) {
53         // ASCII 00..7F
54         return trie->data32[c];
55     }
56     uint8_t t1, t2;
57     if(0xe0 <= c && c < 0xf0 &&
58             ((pos + 1) < length || length < 0) &&
59             U8_IS_VALID_LEAD3_AND_T1(c, t1 = u8[pos]) &&
60             (t2 = (u8[pos + 1] - 0x80)) <= 0x3f) {
61         // U+0800..U+FFFF except surrogates
62         c = (((c & 0xf) << 12) | ((t1 & 0x3f) << 6) | t2);
63         pos += 2;
64         return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
65     } else if(c < 0xe0 && c >= 0xc2 && pos != length && (t1 = (u8[pos] - 0x80)) <= 0x3f) {
66         // U+0080..U+07FF
67         uint32_t ce32 = trie->data32[trie->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET - 0xc0) + c] + t1];
68         c = ((c & 0x1f) << 6) | t1;
69         ++pos;
70         return ce32;
71     } else {
72         // Function call for supplementary code points and error cases.
73         // Illegal byte sequences yield U+FFFD.
74         c = utf8_nextCharSafeBody(u8, &pos, length, c, -3);
75         return data->getCE32(c);
76     }
77 }
78 
79 UBool
foundNULTerminator()80 UTF8CollationIterator::foundNULTerminator() {
81     if(length < 0) {
82         length = --pos;
83         return TRUE;
84     } else {
85         return FALSE;
86     }
87 }
88 
89 UBool
forbidSurrogateCodePoints() const90 UTF8CollationIterator::forbidSurrogateCodePoints() const {
91     return TRUE;
92 }
93 
94 UChar32
nextCodePoint(UErrorCode &)95 UTF8CollationIterator::nextCodePoint(UErrorCode & /*errorCode*/) {
96     if(pos == length) {
97         return U_SENTINEL;
98     }
99     if(u8[pos] == 0 && length < 0) {
100         length = pos;
101         return U_SENTINEL;
102     }
103     UChar32 c;
104     U8_NEXT_OR_FFFD(u8, pos, length, c);
105     return c;
106 }
107 
108 UChar32
previousCodePoint(UErrorCode &)109 UTF8CollationIterator::previousCodePoint(UErrorCode & /*errorCode*/) {
110     if(pos == 0) {
111         return U_SENTINEL;
112     }
113     UChar32 c;
114     U8_PREV_OR_FFFD(u8, 0, pos, c);
115     return c;
116 }
117 
118 void
forwardNumCodePoints(int32_t num,UErrorCode &)119 UTF8CollationIterator::forwardNumCodePoints(int32_t num, UErrorCode & /*errorCode*/) {
120     U8_FWD_N(u8, pos, length, num);
121 }
122 
123 void
backwardNumCodePoints(int32_t num,UErrorCode &)124 UTF8CollationIterator::backwardNumCodePoints(int32_t num, UErrorCode & /*errorCode*/) {
125     U8_BACK_N(u8, 0, pos, num);
126 }
127 
128 // FCDUTF8CollationIterator ------------------------------------------------ ***
129 
~FCDUTF8CollationIterator()130 FCDUTF8CollationIterator::~FCDUTF8CollationIterator() {}
131 
132 void
resetToOffset(int32_t newOffset)133 FCDUTF8CollationIterator::resetToOffset(int32_t newOffset) {
134     reset();
135     start = pos = newOffset;
136     state = CHECK_FWD;
137 }
138 
139 int32_t
getOffset() const140 FCDUTF8CollationIterator::getOffset() const {
141     if(state != IN_NORMALIZED) {
142         return pos;
143     } else if(pos == 0) {
144         return start;
145     } else {
146         return limit;
147     }
148 }
149 
150 uint32_t
handleNextCE32(UChar32 & c,UErrorCode & errorCode)151 FCDUTF8CollationIterator::handleNextCE32(UChar32 &c, UErrorCode &errorCode) {
152     for(;;) {
153         if(state == CHECK_FWD) {
154             // Combination of UTF8CollationIterator::handleNextCE32() with FCD check fastpath.
155             if(pos == length) {
156                 c = U_SENTINEL;
157                 return Collation::FALLBACK_CE32;
158             }
159             c = u8[pos++];
160             if(U8_IS_SINGLE(c)) {
161                 // ASCII 00..7F
162                 return trie->data32[c];
163             }
164             uint8_t t1, t2;
165             if(0xe0 <= c && c < 0xf0 &&
166                     ((pos + 1) < length || length < 0) &&
167                     U8_IS_VALID_LEAD3_AND_T1(c, t1 = u8[pos]) &&
168                     (t2 = (u8[pos + 1] - 0x80)) <= 0x3f) {
169                 // U+0800..U+FFFF except surrogates
170                 c = (((c & 0xf) << 12) | ((t1 & 0x3f) << 6) | t2);
171                 pos += 2;
172                 if(CollationFCD::hasTccc(c) &&
173                         (CollationFCD::maybeTibetanCompositeVowel(c) ||
174                             (pos != length && nextHasLccc()))) {
175                     pos -= 3;
176                 } else {
177                     break;  // return CE32(BMP)
178                 }
179             } else if(c < 0xe0 && c >= 0xc2 && pos != length && (t1 = (u8[pos] - 0x80)) <= 0x3f) {
180                 // U+0080..U+07FF
181                 uint32_t ce32 = trie->data32[trie->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET - 0xc0) + c] + t1];
182                 c = ((c & 0x1f) << 6) | t1;
183                 ++pos;
184                 if(CollationFCD::hasTccc(c) && pos != length && nextHasLccc()) {
185                     pos -= 2;
186                 } else {
187                     return ce32;
188                 }
189             } else {
190                 // Function call for supplementary code points and error cases.
191                 // Illegal byte sequences yield U+FFFD.
192                 c = utf8_nextCharSafeBody(u8, &pos, length, c, -3);
193                 if(c == 0xfffd) {
194                     return Collation::FFFD_CE32;
195                 } else {
196                     U_ASSERT(c > 0xffff);
197                     if(CollationFCD::hasTccc(U16_LEAD(c)) && pos != length && nextHasLccc()) {
198                         pos -= 4;
199                     } else {
200                         return data->getCE32FromSupplementary(c);
201                     }
202                 }
203             }
204             if(!nextSegment(errorCode)) {
205                 c = U_SENTINEL;
206                 return Collation::FALLBACK_CE32;
207             }
208             continue;
209         } else if(state == IN_FCD_SEGMENT && pos != limit) {
210             return UTF8CollationIterator::handleNextCE32(c, errorCode);
211         } else if(state == IN_NORMALIZED && pos != normalized.length()) {
212             c = normalized[pos++];
213             break;
214         } else {
215             switchToForward();
216         }
217     }
218     return UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c);
219 }
220 
221 UBool
nextHasLccc() const222 FCDUTF8CollationIterator::nextHasLccc() const {
223     U_ASSERT(state == CHECK_FWD && pos != length);
224     // The lowest code point with ccc!=0 is U+0300 which is CC 80 in UTF-8.
225     // CJK U+4000..U+DFFF except U+Axxx are also FCD-inert. (Lead bytes E4..ED except EA.)
226     UChar32 c = u8[pos];
227     if(c < 0xcc || (0xe4 <= c && c <= 0xed && c != 0xea)) { return FALSE; }
228     int32_t i = pos;
229     U8_NEXT_OR_FFFD(u8, i, length, c);
230     if(c > 0xffff) { c = U16_LEAD(c); }
231     return CollationFCD::hasLccc(c);
232 }
233 
234 UBool
previousHasTccc() const235 FCDUTF8CollationIterator::previousHasTccc() const {
236     U_ASSERT(state == CHECK_BWD && pos != 0);
237     UChar32 c = u8[pos - 1];
238     if(U8_IS_SINGLE(c)) { return FALSE; }
239     int32_t i = pos;
240     U8_PREV_OR_FFFD(u8, 0, i, c);
241     if(c > 0xffff) { c = U16_LEAD(c); }
242     return CollationFCD::hasTccc(c);
243 }
244 
245 UChar
handleGetTrailSurrogate()246 FCDUTF8CollationIterator::handleGetTrailSurrogate() {
247     if(state != IN_NORMALIZED) { return 0; }
248     U_ASSERT(pos < normalized.length());
249     UChar trail;
250     if(U16_IS_TRAIL(trail = normalized[pos])) { ++pos; }
251     return trail;
252 }
253 
254 UBool
foundNULTerminator()255 FCDUTF8CollationIterator::foundNULTerminator() {
256     if(state == CHECK_FWD && length < 0) {
257         length = --pos;
258         return TRUE;
259     } else {
260         return FALSE;
261     }
262 }
263 
264 UChar32
nextCodePoint(UErrorCode & errorCode)265 FCDUTF8CollationIterator::nextCodePoint(UErrorCode &errorCode) {
266     UChar32 c;
267     for(;;) {
268         if(state == CHECK_FWD) {
269             if(pos == length || ((c = u8[pos]) == 0 && length < 0)) {
270                 return U_SENTINEL;
271             }
272             if(U8_IS_SINGLE(c)) {
273                 ++pos;
274                 return c;
275             }
276             U8_NEXT_OR_FFFD(u8, pos, length, c);
277             if(CollationFCD::hasTccc(c <= 0xffff ? c : U16_LEAD(c)) &&
278                     (CollationFCD::maybeTibetanCompositeVowel(c) ||
279                         (pos != length && nextHasLccc()))) {
280                 // c is not FCD-inert, therefore it is not U+FFFD and it has a valid byte sequence
281                 // and we can use U8_LENGTH() rather than a previous-position variable.
282                 pos -= U8_LENGTH(c);
283                 if(!nextSegment(errorCode)) {
284                     return U_SENTINEL;
285                 }
286                 continue;
287             }
288             return c;
289         } else if(state == IN_FCD_SEGMENT && pos != limit) {
290             U8_NEXT_OR_FFFD(u8, pos, length, c);
291             return c;
292         } else if(state == IN_NORMALIZED && pos != normalized.length()) {
293             c = normalized.char32At(pos);
294             pos += U16_LENGTH(c);
295             return c;
296         } else {
297             switchToForward();
298         }
299     }
300 }
301 
302 UChar32
previousCodePoint(UErrorCode & errorCode)303 FCDUTF8CollationIterator::previousCodePoint(UErrorCode &errorCode) {
304     UChar32 c;
305     for(;;) {
306         if(state == CHECK_BWD) {
307             if(pos == 0) {
308                 return U_SENTINEL;
309             }
310             if(U8_IS_SINGLE(c = u8[pos - 1])) {
311                 --pos;
312                 return c;
313             }
314             U8_PREV_OR_FFFD(u8, 0, pos, c);
315             if(CollationFCD::hasLccc(c <= 0xffff ? c : U16_LEAD(c)) &&
316                     (CollationFCD::maybeTibetanCompositeVowel(c) ||
317                         (pos != 0 && previousHasTccc()))) {
318                 // c is not FCD-inert, therefore it is not U+FFFD and it has a valid byte sequence
319                 // and we can use U8_LENGTH() rather than a previous-position variable.
320                 pos += U8_LENGTH(c);
321                 if(!previousSegment(errorCode)) {
322                     return U_SENTINEL;
323                 }
324                 continue;
325             }
326             return c;
327         } else if(state == IN_FCD_SEGMENT && pos != start) {
328             U8_PREV_OR_FFFD(u8, 0, pos, c);
329             return c;
330         } else if(state >= IN_NORMALIZED && pos != 0) {
331             c = normalized.char32At(pos - 1);
332             pos -= U16_LENGTH(c);
333             return c;
334         } else {
335             switchToBackward();
336         }
337     }
338 }
339 
340 void
forwardNumCodePoints(int32_t num,UErrorCode & errorCode)341 FCDUTF8CollationIterator::forwardNumCodePoints(int32_t num, UErrorCode &errorCode) {
342     // Specify the class to avoid a virtual-function indirection.
343     // In Java, we would declare this class final.
344     while(num > 0 && FCDUTF8CollationIterator::nextCodePoint(errorCode) >= 0) {
345         --num;
346     }
347 }
348 
349 void
backwardNumCodePoints(int32_t num,UErrorCode & errorCode)350 FCDUTF8CollationIterator::backwardNumCodePoints(int32_t num, UErrorCode &errorCode) {
351     // Specify the class to avoid a virtual-function indirection.
352     // In Java, we would declare this class final.
353     while(num > 0 && FCDUTF8CollationIterator::previousCodePoint(errorCode) >= 0) {
354         --num;
355     }
356 }
357 
358 void
switchToForward()359 FCDUTF8CollationIterator::switchToForward() {
360     U_ASSERT(state == CHECK_BWD ||
361              (state == IN_FCD_SEGMENT && pos == limit) ||
362              (state == IN_NORMALIZED && pos == normalized.length()));
363     if(state == CHECK_BWD) {
364         // Turn around from backward checking.
365         start = pos;
366         if(pos == limit) {
367             state = CHECK_FWD;  // Check forward.
368         } else {  // pos < limit
369             state = IN_FCD_SEGMENT;  // Stay in FCD segment.
370         }
371     } else {
372         // Reached the end of the FCD segment.
373         if(state == IN_FCD_SEGMENT) {
374             // The input text segment is FCD, extend it forward.
375         } else {
376             // The input text segment needed to be normalized.
377             // Switch to checking forward from it.
378             start = pos = limit;
379         }
380         state = CHECK_FWD;
381     }
382 }
383 
384 UBool
nextSegment(UErrorCode & errorCode)385 FCDUTF8CollationIterator::nextSegment(UErrorCode &errorCode) {
386     if(U_FAILURE(errorCode)) { return FALSE; }
387     U_ASSERT(state == CHECK_FWD && pos != length);
388     // The input text [start..pos[ passes the FCD check.
389     int32_t segmentStart = pos;
390     // Collect the characters being checked, in case they need to be normalized.
391     UnicodeString s;
392     uint8_t prevCC = 0;
393     for(;;) {
394         // Fetch the next character and its fcd16 value.
395         int32_t cpStart = pos;
396         UChar32 c;
397         U8_NEXT_OR_FFFD(u8, pos, length, c);
398         uint16_t fcd16 = nfcImpl.getFCD16(c);
399         uint8_t leadCC = (uint8_t)(fcd16 >> 8);
400         if(leadCC == 0 && cpStart != segmentStart) {
401             // FCD boundary before this character.
402             pos = cpStart;
403             break;
404         }
405         s.append(c);
406         if(leadCC != 0 && (prevCC > leadCC || CollationFCD::isFCD16OfTibetanCompositeVowel(fcd16))) {
407             // Fails FCD check. Find the next FCD boundary and normalize.
408             while(pos != length) {
409                 cpStart = pos;
410                 U8_NEXT_OR_FFFD(u8, pos, length, c);
411                 if(nfcImpl.getFCD16(c) <= 0xff) {
412                     pos = cpStart;
413                     break;
414                 }
415                 s.append(c);
416             }
417             if(!normalize(s, errorCode)) { return FALSE; }
418             start = segmentStart;
419             limit = pos;
420             state = IN_NORMALIZED;
421             pos = 0;
422             return TRUE;
423         }
424         prevCC = (uint8_t)fcd16;
425         if(pos == length || prevCC == 0) {
426             // FCD boundary after the last character.
427             break;
428         }
429     }
430     limit = pos;
431     pos = segmentStart;
432     U_ASSERT(pos != limit);
433     state = IN_FCD_SEGMENT;
434     return TRUE;
435 }
436 
437 void
switchToBackward()438 FCDUTF8CollationIterator::switchToBackward() {
439     U_ASSERT(state == CHECK_FWD ||
440              (state == IN_FCD_SEGMENT && pos == start) ||
441              (state >= IN_NORMALIZED && pos == 0));
442     if(state == CHECK_FWD) {
443         // Turn around from forward checking.
444         limit = pos;
445         if(pos == start) {
446             state = CHECK_BWD;  // Check backward.
447         } else {  // pos > start
448             state = IN_FCD_SEGMENT;  // Stay in FCD segment.
449         }
450     } else {
451         // Reached the start of the FCD segment.
452         if(state == IN_FCD_SEGMENT) {
453             // The input text segment is FCD, extend it backward.
454         } else {
455             // The input text segment needed to be normalized.
456             // Switch to checking backward from it.
457             limit = pos = start;
458         }
459         state = CHECK_BWD;
460     }
461 }
462 
463 UBool
previousSegment(UErrorCode & errorCode)464 FCDUTF8CollationIterator::previousSegment(UErrorCode &errorCode) {
465     if(U_FAILURE(errorCode)) { return FALSE; }
466     U_ASSERT(state == CHECK_BWD && pos != 0);
467     // The input text [pos..limit[ passes the FCD check.
468     int32_t segmentLimit = pos;
469     // Collect the characters being checked, in case they need to be normalized.
470     UnicodeString s;
471     uint8_t nextCC = 0;
472     for(;;) {
473         // Fetch the previous character and its fcd16 value.
474         int32_t cpLimit = pos;
475         UChar32 c;
476         U8_PREV_OR_FFFD(u8, 0, pos, c);
477         uint16_t fcd16 = nfcImpl.getFCD16(c);
478         uint8_t trailCC = (uint8_t)fcd16;
479         if(trailCC == 0 && cpLimit != segmentLimit) {
480             // FCD boundary after this character.
481             pos = cpLimit;
482             break;
483         }
484         s.append(c);
485         if(trailCC != 0 && ((nextCC != 0 && trailCC > nextCC) ||
486                             CollationFCD::isFCD16OfTibetanCompositeVowel(fcd16))) {
487             // Fails FCD check. Find the previous FCD boundary and normalize.
488             while(fcd16 > 0xff && pos != 0) {
489                 cpLimit = pos;
490                 U8_PREV_OR_FFFD(u8, 0, pos, c);
491                 fcd16 = nfcImpl.getFCD16(c);
492                 if(fcd16 == 0) {
493                     pos = cpLimit;
494                     break;
495                 }
496                 s.append(c);
497             }
498             s.reverse();
499             if(!normalize(s, errorCode)) { return FALSE; }
500             limit = segmentLimit;
501             start = pos;
502             state = IN_NORMALIZED;
503             pos = normalized.length();
504             return TRUE;
505         }
506         nextCC = (uint8_t)(fcd16 >> 8);
507         if(pos == 0 || nextCC == 0) {
508             // FCD boundary before the following character.
509             break;
510         }
511     }
512     start = pos;
513     pos = segmentLimit;
514     U_ASSERT(pos != start);
515     state = IN_FCD_SEGMENT;
516     return TRUE;
517 }
518 
519 UBool
normalize(const UnicodeString & s,UErrorCode & errorCode)520 FCDUTF8CollationIterator::normalize(const UnicodeString &s, UErrorCode &errorCode) {
521     // NFD without argument checking.
522     U_ASSERT(U_SUCCESS(errorCode));
523     nfcImpl.decompose(s, normalized, errorCode);
524     return U_SUCCESS(errorCode);
525 }
526 
527 U_NAMESPACE_END
528 
529 #endif  // !UCONFIG_NO_COLLATION
530