• 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) 1996-2015, International Business Machines Corporation and
6  * others. All Rights Reserved.
7  *****************************************************************************
8  */
9 
10 #include "unicode/utypes.h"
11 
12 #if !UCONFIG_NO_NORMALIZATION
13 
14 #include "unicode/caniter.h"
15 #include "unicode/normalizer2.h"
16 #include "unicode/uchar.h"
17 #include "unicode/uniset.h"
18 #include "unicode/usetiter.h"
19 #include "unicode/ustring.h"
20 #include "unicode/utf16.h"
21 #include "cmemory.h"
22 #include "hash.h"
23 #include "normalizer2impl.h"
24 
25 /**
26  * This class allows one to iterate through all the strings that are canonically equivalent to a given
27  * string. For example, here are some sample results:
28 Results for: {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA}
29 1: \u0041\u030A\u0064\u0307\u0327
30  = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA}
31 2: \u0041\u030A\u0064\u0327\u0307
32  = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE}
33 3: \u0041\u030A\u1E0B\u0327
34  = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA}
35 4: \u0041\u030A\u1E11\u0307
36  = {LATIN CAPITAL LETTER A}{COMBINING RING ABOVE}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE}
37 5: \u00C5\u0064\u0307\u0327
38  = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA}
39 6: \u00C5\u0064\u0327\u0307
40  = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE}
41 7: \u00C5\u1E0B\u0327
42  = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA}
43 8: \u00C5\u1E11\u0307
44  = {LATIN CAPITAL LETTER A WITH RING ABOVE}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE}
45 9: \u212B\u0064\u0307\u0327
46  = {ANGSTROM SIGN}{LATIN SMALL LETTER D}{COMBINING DOT ABOVE}{COMBINING CEDILLA}
47 10: \u212B\u0064\u0327\u0307
48  = {ANGSTROM SIGN}{LATIN SMALL LETTER D}{COMBINING CEDILLA}{COMBINING DOT ABOVE}
49 11: \u212B\u1E0B\u0327
50  = {ANGSTROM SIGN}{LATIN SMALL LETTER D WITH DOT ABOVE}{COMBINING CEDILLA}
51 12: \u212B\u1E11\u0307
52  = {ANGSTROM SIGN}{LATIN SMALL LETTER D WITH CEDILLA}{COMBINING DOT ABOVE}
53  *<br>Note: the code is intended for use with small strings, and is not suitable for larger ones,
54  * since it has not been optimized for that situation.
55  *@author M. Davis
56  *@draft
57  */
58 
59 // public
60 
61 U_NAMESPACE_BEGIN
62 
63 // TODO: add boilerplate methods.
64 
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CanonicalIterator)65 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CanonicalIterator)
66 
67 /**
68  *@param source string to get results for
69  */
70 CanonicalIterator::CanonicalIterator(const UnicodeString &sourceStr, UErrorCode &status) :
71     pieces(NULL),
72     pieces_length(0),
73     pieces_lengths(NULL),
74     current(NULL),
75     current_length(0),
76     nfd(*Normalizer2::getNFDInstance(status)),
77     nfcImpl(*Normalizer2Factory::getNFCImpl(status))
78 {
79     if(U_SUCCESS(status) && nfcImpl.ensureCanonIterData(status)) {
80       setSource(sourceStr, status);
81     }
82 }
83 
~CanonicalIterator()84 CanonicalIterator::~CanonicalIterator() {
85   cleanPieces();
86 }
87 
cleanPieces()88 void CanonicalIterator::cleanPieces() {
89     int32_t i = 0;
90     if(pieces != NULL) {
91         for(i = 0; i < pieces_length; i++) {
92             if(pieces[i] != NULL) {
93                 delete[] pieces[i];
94             }
95         }
96         uprv_free(pieces);
97         pieces = NULL;
98         pieces_length = 0;
99     }
100     if(pieces_lengths != NULL) {
101         uprv_free(pieces_lengths);
102         pieces_lengths = NULL;
103     }
104     if(current != NULL) {
105         uprv_free(current);
106         current = NULL;
107         current_length = 0;
108     }
109 }
110 
111 /**
112  *@return gets the source: NOTE: it is the NFD form of source
113  */
getSource()114 UnicodeString CanonicalIterator::getSource() {
115   return source;
116 }
117 
118 /**
119  * Resets the iterator so that one can start again from the beginning.
120  */
reset()121 void CanonicalIterator::reset() {
122     done = FALSE;
123     for (int i = 0; i < current_length; ++i) {
124         current[i] = 0;
125     }
126 }
127 
128 /**
129  *@return the next string that is canonically equivalent. The value null is returned when
130  * the iteration is done.
131  */
next()132 UnicodeString CanonicalIterator::next() {
133     int32_t i = 0;
134 
135     if (done) {
136       buffer.setToBogus();
137       return buffer;
138     }
139 
140     // delete old contents
141     buffer.remove();
142 
143     // construct return value
144 
145     for (i = 0; i < pieces_length; ++i) {
146         buffer.append(pieces[i][current[i]]);
147     }
148     //String result = buffer.toString(); // not needed
149 
150     // find next value for next time
151 
152     for (i = current_length - 1; ; --i) {
153         if (i < 0) {
154             done = TRUE;
155             break;
156         }
157         current[i]++;
158         if (current[i] < pieces_lengths[i]) break; // got sequence
159         current[i] = 0;
160     }
161     return buffer;
162 }
163 
164 /**
165  *@param set the source string to iterate against. This allows the same iterator to be used
166  * while changing the source string, saving object creation.
167  */
setSource(const UnicodeString & newSource,UErrorCode & status)168 void CanonicalIterator::setSource(const UnicodeString &newSource, UErrorCode &status) {
169     int32_t list_length = 0;
170     UChar32 cp = 0;
171     int32_t start = 0;
172     int32_t i = 0;
173     UnicodeString *list = NULL;
174 
175     nfd.normalize(newSource, source, status);
176     if(U_FAILURE(status)) {
177       return;
178     }
179     done = FALSE;
180 
181     cleanPieces();
182 
183     // catch degenerate case
184     if (newSource.length() == 0) {
185         pieces = (UnicodeString **)uprv_malloc(sizeof(UnicodeString *));
186         pieces_lengths = (int32_t*)uprv_malloc(1 * sizeof(int32_t));
187         pieces_length = 1;
188         current = (int32_t*)uprv_malloc(1 * sizeof(int32_t));
189         current_length = 1;
190         if (pieces == NULL || pieces_lengths == NULL || current == NULL) {
191             status = U_MEMORY_ALLOCATION_ERROR;
192             goto CleanPartialInitialization;
193         }
194         current[0] = 0;
195         pieces[0] = new UnicodeString[1];
196         pieces_lengths[0] = 1;
197         if (pieces[0] == 0) {
198             status = U_MEMORY_ALLOCATION_ERROR;
199             goto CleanPartialInitialization;
200         }
201         return;
202     }
203 
204 
205     list = new UnicodeString[source.length()];
206     if (list == 0) {
207         status = U_MEMORY_ALLOCATION_ERROR;
208         goto CleanPartialInitialization;
209     }
210 
211     // i should initialy be the number of code units at the
212     // start of the string
213     i = U16_LENGTH(source.char32At(0));
214     //int32_t i = 1;
215     // find the segments
216     // This code iterates through the source string and
217     // extracts segments that end up on a codepoint that
218     // doesn't start any decompositions. (Analysis is done
219     // on the NFD form - see above).
220     for (; i < source.length(); i += U16_LENGTH(cp)) {
221         cp = source.char32At(i);
222         if (nfcImpl.isCanonSegmentStarter(cp)) {
223             source.extract(start, i-start, list[list_length++]); // add up to i
224             start = i;
225         }
226     }
227     source.extract(start, i-start, list[list_length++]); // add last one
228 
229 
230     // allocate the arrays, and find the strings that are CE to each segment
231     pieces = (UnicodeString **)uprv_malloc(list_length * sizeof(UnicodeString *));
232     pieces_length = list_length;
233     pieces_lengths = (int32_t*)uprv_malloc(list_length * sizeof(int32_t));
234     current = (int32_t*)uprv_malloc(list_length * sizeof(int32_t));
235     current_length = list_length;
236     if (pieces == NULL || pieces_lengths == NULL || current == NULL) {
237         status = U_MEMORY_ALLOCATION_ERROR;
238         goto CleanPartialInitialization;
239     }
240 
241     for (i = 0; i < current_length; i++) {
242         current[i] = 0;
243     }
244     // for each segment, get all the combinations that can produce
245     // it after NFD normalization
246     for (i = 0; i < pieces_length; ++i) {
247         //if (PROGRESS) printf("SEGMENT\n");
248         pieces[i] = getEquivalents(list[i], pieces_lengths[i], status);
249     }
250 
251     delete[] list;
252     return;
253 // Common section to cleanup all local variables and reset object variables.
254 CleanPartialInitialization:
255     if (list != NULL) {
256         delete[] list;
257     }
258     cleanPieces();
259 }
260 
261 /**
262  * Dumb recursive implementation of permutation.
263  * TODO: optimize
264  * @param source the string to find permutations for
265  * @return the results in a set.
266  */
permute(UnicodeString & source,UBool skipZeros,Hashtable * result,UErrorCode & status)267 void U_EXPORT2 CanonicalIterator::permute(UnicodeString &source, UBool skipZeros, Hashtable *result, UErrorCode &status) {
268     if(U_FAILURE(status)) {
269         return;
270     }
271     //if (PROGRESS) printf("Permute: %s\n", UToS(Tr(source)));
272     int32_t i = 0;
273 
274     // optimization:
275     // if zero or one character, just return a set with it
276     // we check for length < 2 to keep from counting code points all the time
277     if (source.length() <= 2 && source.countChar32() <= 1) {
278         UnicodeString *toPut = new UnicodeString(source);
279         /* test for NULL */
280         if (toPut == 0) {
281             status = U_MEMORY_ALLOCATION_ERROR;
282             return;
283         }
284         result->put(source, toPut, status);
285         return;
286     }
287 
288     // otherwise iterate through the string, and recursively permute all the other characters
289     UChar32 cp;
290     Hashtable subpermute(status);
291     if(U_FAILURE(status)) {
292         return;
293     }
294     subpermute.setValueDeleter(uprv_deleteUObject);
295 
296     for (i = 0; i < source.length(); i += U16_LENGTH(cp)) {
297         cp = source.char32At(i);
298         const UHashElement *ne = NULL;
299         int32_t el = UHASH_FIRST;
300         UnicodeString subPermuteString = source;
301 
302         // optimization:
303         // if the character is canonical combining class zero,
304         // don't permute it
305         if (skipZeros && i != 0 && u_getCombiningClass(cp) == 0) {
306             //System.out.println("Skipping " + Utility.hex(UTF16.valueOf(source, i)));
307             continue;
308         }
309 
310         subpermute.removeAll();
311 
312         // see what the permutations of the characters before and after this one are
313         //Hashtable *subpermute = permute(source.substring(0,i) + source.substring(i + UTF16.getCharCount(cp)));
314         permute(subPermuteString.remove(i, U16_LENGTH(cp)), skipZeros, &subpermute, status);
315         /* Test for buffer overflows */
316         if(U_FAILURE(status)) {
317             return;
318         }
319         // The upper remove is destructive. The question is do we have to make a copy, or we don't care about the contents
320         // of source at this point.
321 
322         // prefix this character to all of them
323         ne = subpermute.nextElement(el);
324         while (ne != NULL) {
325             UnicodeString *permRes = (UnicodeString *)(ne->value.pointer);
326             UnicodeString *chStr = new UnicodeString(cp);
327             //test for  NULL
328             if (chStr == NULL) {
329                 status = U_MEMORY_ALLOCATION_ERROR;
330                 return;
331             }
332             chStr->append(*permRes); //*((UnicodeString *)(ne->value.pointer));
333             //if (PROGRESS) printf("  Piece: %s\n", UToS(*chStr));
334             result->put(*chStr, chStr, status);
335             ne = subpermute.nextElement(el);
336         }
337     }
338     //return result;
339 }
340 
341 // privates
342 
343 // we have a segment, in NFD. Find all the strings that are canonically equivalent to it.
getEquivalents(const UnicodeString & segment,int32_t & result_len,UErrorCode & status)344 UnicodeString* CanonicalIterator::getEquivalents(const UnicodeString &segment, int32_t &result_len, UErrorCode &status) {
345     Hashtable result(status);
346     Hashtable permutations(status);
347     Hashtable basic(status);
348     if (U_FAILURE(status)) {
349         return 0;
350     }
351     result.setValueDeleter(uprv_deleteUObject);
352     permutations.setValueDeleter(uprv_deleteUObject);
353     basic.setValueDeleter(uprv_deleteUObject);
354 
355     UChar USeg[256];
356     int32_t segLen = segment.extract(USeg, 256, status);
357     getEquivalents2(&basic, USeg, segLen, status);
358 
359     // now get all the permutations
360     // add only the ones that are canonically equivalent
361     // TODO: optimize by not permuting any class zero.
362 
363     const UHashElement *ne = NULL;
364     int32_t el = UHASH_FIRST;
365     //Iterator it = basic.iterator();
366     ne = basic.nextElement(el);
367     //while (it.hasNext())
368     while (ne != NULL) {
369         //String item = (String) it.next();
370         UnicodeString item = *((UnicodeString *)(ne->value.pointer));
371 
372         permutations.removeAll();
373         permute(item, CANITER_SKIP_ZEROES, &permutations, status);
374         const UHashElement *ne2 = NULL;
375         int32_t el2 = UHASH_FIRST;
376         //Iterator it2 = permutations.iterator();
377         ne2 = permutations.nextElement(el2);
378         //while (it2.hasNext())
379         while (ne2 != NULL) {
380             //String possible = (String) it2.next();
381             //UnicodeString *possible = new UnicodeString(*((UnicodeString *)(ne2->value.pointer)));
382             UnicodeString possible(*((UnicodeString *)(ne2->value.pointer)));
383             UnicodeString attempt;
384             nfd.normalize(possible, attempt, status);
385 
386             // TODO: check if operator == is semanticaly the same as attempt.equals(segment)
387             if (attempt==segment) {
388                 //if (PROGRESS) printf("Adding Permutation: %s\n", UToS(Tr(*possible)));
389                 // TODO: use the hashtable just to catch duplicates - store strings directly (somehow).
390                 result.put(possible, new UnicodeString(possible), status); //add(possible);
391             } else {
392                 //if (PROGRESS) printf("-Skipping Permutation: %s\n", UToS(Tr(*possible)));
393             }
394 
395             ne2 = permutations.nextElement(el2);
396         }
397         ne = basic.nextElement(el);
398     }
399 
400     /* Test for buffer overflows */
401     if(U_FAILURE(status)) {
402         return 0;
403     }
404     // convert into a String[] to clean up storage
405     //String[] finalResult = new String[result.size()];
406     UnicodeString *finalResult = NULL;
407     int32_t resultCount;
408     if((resultCount = result.count()) != 0) {
409         finalResult = new UnicodeString[resultCount];
410         if (finalResult == 0) {
411             status = U_MEMORY_ALLOCATION_ERROR;
412             return NULL;
413         }
414     }
415     else {
416         status = U_ILLEGAL_ARGUMENT_ERROR;
417         return NULL;
418     }
419     //result.toArray(finalResult);
420     result_len = 0;
421     el = UHASH_FIRST;
422     ne = result.nextElement(el);
423     while(ne != NULL) {
424         finalResult[result_len++] = *((UnicodeString *)(ne->value.pointer));
425         ne = result.nextElement(el);
426     }
427 
428 
429     return finalResult;
430 }
431 
getEquivalents2(Hashtable * fillinResult,const UChar * segment,int32_t segLen,UErrorCode & status)432 Hashtable *CanonicalIterator::getEquivalents2(Hashtable *fillinResult, const UChar *segment, int32_t segLen, UErrorCode &status) {
433 
434     if (U_FAILURE(status)) {
435         return NULL;
436     }
437 
438     //if (PROGRESS) printf("Adding: %s\n", UToS(Tr(segment)));
439 
440     UnicodeString toPut(segment, segLen);
441 
442     fillinResult->put(toPut, new UnicodeString(toPut), status);
443 
444     UnicodeSet starts;
445 
446     // cycle through all the characters
447     UChar32 cp;
448     for (int32_t i = 0; i < segLen; i += U16_LENGTH(cp)) {
449         // see if any character is at the start of some decomposition
450         U16_GET(segment, 0, i, segLen, cp);
451         if (!nfcImpl.getCanonStartSet(cp, starts)) {
452             continue;
453         }
454         // if so, see which decompositions match
455         UnicodeSetIterator iter(starts);
456         while (iter.next()) {
457             UChar32 cp2 = iter.getCodepoint();
458             Hashtable remainder(status);
459             remainder.setValueDeleter(uprv_deleteUObject);
460             if (extract(&remainder, cp2, segment, segLen, i, status) == NULL) {
461                 continue;
462             }
463 
464             // there were some matches, so add all the possibilities to the set.
465             UnicodeString prefix(segment, i);
466             prefix += cp2;
467 
468             int32_t el = UHASH_FIRST;
469             const UHashElement *ne = remainder.nextElement(el);
470             while (ne != NULL) {
471                 UnicodeString item = *((UnicodeString *)(ne->value.pointer));
472                 UnicodeString *toAdd = new UnicodeString(prefix);
473                 /* test for NULL */
474                 if (toAdd == 0) {
475                     status = U_MEMORY_ALLOCATION_ERROR;
476                     return NULL;
477                 }
478                 *toAdd += item;
479                 fillinResult->put(*toAdd, toAdd, status);
480 
481                 //if (PROGRESS) printf("Adding: %s\n", UToS(Tr(*toAdd)));
482 
483                 ne = remainder.nextElement(el);
484             }
485         }
486     }
487 
488     /* Test for buffer overflows */
489     if(U_FAILURE(status)) {
490         return NULL;
491     }
492     return fillinResult;
493 }
494 
495 /**
496  * See if the decomposition of cp2 is at segment starting at segmentPos
497  * (with canonical rearrangment!)
498  * If so, take the remainder, and return the equivalents
499  */
extract(Hashtable * fillinResult,UChar32 comp,const UChar * segment,int32_t segLen,int32_t segmentPos,UErrorCode & status)500 Hashtable *CanonicalIterator::extract(Hashtable *fillinResult, UChar32 comp, const UChar *segment, int32_t segLen, int32_t segmentPos, UErrorCode &status) {
501 //Hashtable *CanonicalIterator::extract(UChar32 comp, const UnicodeString &segment, int32_t segLen, int32_t segmentPos, UErrorCode &status) {
502     //if (PROGRESS) printf(" extract: %s, ", UToS(Tr(UnicodeString(comp))));
503     //if (PROGRESS) printf("%s, %i\n", UToS(Tr(segment)), segmentPos);
504 
505     if (U_FAILURE(status)) {
506         return NULL;
507     }
508 
509     UnicodeString temp(comp);
510     int32_t inputLen=temp.length();
511     UnicodeString decompString;
512     nfd.normalize(temp, decompString, status);
513     if (U_FAILURE(status)) {
514         return NULL;
515     }
516     if (decompString.isBogus()) {
517         status = U_MEMORY_ALLOCATION_ERROR;
518         return NULL;
519     }
520     const UChar *decomp=decompString.getBuffer();
521     int32_t decompLen=decompString.length();
522 
523     // See if it matches the start of segment (at segmentPos)
524     UBool ok = FALSE;
525     UChar32 cp;
526     int32_t decompPos = 0;
527     UChar32 decompCp;
528     U16_NEXT(decomp, decompPos, decompLen, decompCp);
529 
530     int32_t i = segmentPos;
531     while(i < segLen) {
532         U16_NEXT(segment, i, segLen, cp);
533 
534         if (cp == decompCp) { // if equal, eat another cp from decomp
535 
536             //if (PROGRESS) printf("  matches: %s\n", UToS(Tr(UnicodeString(cp))));
537 
538             if (decompPos == decompLen) { // done, have all decomp characters!
539                 temp.append(segment+i, segLen-i);
540                 ok = TRUE;
541                 break;
542             }
543             U16_NEXT(decomp, decompPos, decompLen, decompCp);
544         } else {
545             //if (PROGRESS) printf("  buffer: %s\n", UToS(Tr(UnicodeString(cp))));
546 
547             // brute force approach
548             temp.append(cp);
549 
550             /* TODO: optimize
551             // since we know that the classes are monotonically increasing, after zero
552             // e.g. 0 5 7 9 0 3
553             // we can do an optimization
554             // there are only a few cases that work: zero, less, same, greater
555             // if both classes are the same, we fail
556             // if the decomp class < the segment class, we fail
557 
558             segClass = getClass(cp);
559             if (decompClass <= segClass) return null;
560             */
561         }
562     }
563     if (!ok)
564         return NULL; // we failed, characters left over
565 
566     //if (PROGRESS) printf("Matches\n");
567 
568     if (inputLen == temp.length()) {
569         fillinResult->put(UnicodeString(), new UnicodeString(), status);
570         return fillinResult; // succeed, but no remainder
571     }
572 
573     // brute force approach
574     // check to make sure result is canonically equivalent
575     UnicodeString trial;
576     nfd.normalize(temp, trial, status);
577     if(U_FAILURE(status) || trial.compare(segment+segmentPos, segLen - segmentPos) != 0) {
578         return NULL;
579     }
580 
581     return getEquivalents2(fillinResult, temp.getBuffer()+inputLen, temp.length()-inputLen, status);
582 }
583 
584 U_NAMESPACE_END
585 
586 #endif /* #if !UCONFIG_NO_NORMALIZATION */
587