1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ***************************************************************** |
5 | * Copyright (c) 2002-2014, International Business Machines Corporation |
6 | * and others. All Rights Reserved. |
7 | ***************************************************************** |
8 | * Date Name Description |
9 | * 06/06/2002 aliu Creation. |
10 | ***************************************************************** |
11 | */ |
12 | |
13 | #include "unicode/utypes.h" |
14 | |
15 | #if !UCONFIG_NO_TRANSLITERATION |
16 | |
17 | #include "unicode/uobject.h" |
18 | #include "unicode/uscript.h" |
19 | |
20 | #include "anytrans.h" |
21 | #include "hash.h" |
22 | #include "mutex.h" |
23 | #include "nultrans.h" |
24 | #include "putilimp.h" |
25 | #include "tridpars.h" |
26 | #include "uinvchar.h" |
27 | #include "uvector.h" |
28 | |
29 | //------------------------------------------------------------ |
30 | // Constants |
31 | |
32 | static const UChar TARGET_SEP = 45; // '-' |
33 | static const UChar VARIANT_SEP = 47; // '/' |
34 | static const UChar ANY[] = {0x41,0x6E,0x79,0}; // "Any" |
35 | static const UChar NULL_ID[] = {78,117,108,108,0}; // "Null" |
36 | static const UChar LATIN_PIVOT[] = {0x2D,0x4C,0x61,0x74,0x6E,0x3B,0x4C,0x61,0x74,0x6E,0x2D,0}; // "-Latn;Latn-" |
37 | |
38 | // initial size for an Any-XXXX transform's cache of script-XXXX transforms |
39 | // (will grow as necessary, but we don't expect to have source text with more than 7 scripts) |
40 | #define ANY_TRANS_CACHE_INIT_SIZE 7 |
41 | |
42 | //------------------------------------------------------------ |
43 | |
44 | U_CDECL_BEGIN |
45 | /** |
46 | * Deleter function for Transliterator*. |
47 | */ |
48 | static void U_CALLCONV |
49 | _deleteTransliterator(void *obj) { |
50 | delete (icu::Transliterator*) obj; |
51 | } |
52 | U_CDECL_END |
53 | |
54 | //------------------------------------------------------------ |
55 | |
56 | U_NAMESPACE_BEGIN |
57 | |
58 | //------------------------------------------------------------ |
59 | // ScriptRunIterator |
60 | |
61 | /** |
62 | * Returns a series of ranges corresponding to scripts. They will be |
63 | * of the form: |
64 | * |
65 | * ccccSScSSccccTTcTcccc - c = common, S = first script, T = second |
66 | * | | - first run (start, limit) |
67 | * | | - second run (start, limit) |
68 | * |
69 | * That is, the runs will overlap. The reason for this is so that a |
70 | * transliterator can consider common characters both before and after |
71 | * the scripts. |
72 | */ |
73 | class ScriptRunIterator : public UMemory { |
74 | private: |
75 | const Replaceable& text; |
76 | int32_t textStart; |
77 | int32_t textLimit; |
78 | |
79 | public: |
80 | /** |
81 | * The code of the current run, valid after next() returns. May |
82 | * be USCRIPT_INVALID_CODE if and only if the entire text is |
83 | * COMMON/INHERITED. |
84 | */ |
85 | UScriptCode scriptCode; |
86 | |
87 | /** |
88 | * The start of the run, inclusive, valid after next() returns. |
89 | */ |
90 | int32_t start; |
91 | |
92 | /** |
93 | * The end of the run, exclusive, valid after next() returns. |
94 | */ |
95 | int32_t limit; |
96 | |
97 | /** |
98 | * Constructs a run iterator over the given text from start |
99 | * (inclusive) to limit (exclusive). |
100 | */ |
101 | ScriptRunIterator(const Replaceable& text, int32_t start, int32_t limit); |
102 | |
103 | /** |
104 | * Returns TRUE if there are any more runs. TRUE is always |
105 | * returned at least once. Upon return, the caller should |
106 | * examine scriptCode, start, and limit. |
107 | */ |
108 | UBool next(); |
109 | |
110 | /** |
111 | * Adjusts internal indices for a change in the limit index of the |
112 | * given delta. A positive delta means the limit has increased. |
113 | */ |
114 | void adjustLimit(int32_t delta); |
115 | |
116 | private: |
117 | ScriptRunIterator(const ScriptRunIterator &other); // forbid copying of this class |
118 | ScriptRunIterator &operator=(const ScriptRunIterator &other); // forbid copying of this class |
119 | }; |
120 | |
121 | ScriptRunIterator::ScriptRunIterator(const Replaceable& theText, |
122 | int32_t myStart, int32_t myLimit) : |
123 | text(theText) |
124 | { |
125 | textStart = myStart; |
126 | textLimit = myLimit; |
127 | limit = myStart; |
128 | } |
129 | |
130 | UBool ScriptRunIterator::next() { |
131 | UChar32 ch; |
132 | UScriptCode s; |
133 | UErrorCode ec = U_ZERO_ERROR; |
134 | |
135 | scriptCode = USCRIPT_INVALID_CODE; // don't know script yet |
136 | start = limit; |
137 | |
138 | // Are we done? |
139 | if (start == textLimit) { |
140 | return FALSE; |
141 | } |
142 | |
143 | // Move start back to include adjacent COMMON or INHERITED |
144 | // characters |
145 | while (start > textStart) { |
146 | ch = text.char32At(start - 1); // look back |
147 | s = uscript_getScript(ch, &ec); |
148 | if (s == USCRIPT_COMMON || s == USCRIPT_INHERITED) { |
149 | --start; |
150 | } else { |
151 | break; |
152 | } |
153 | } |
154 | |
155 | // Move limit ahead to include COMMON, INHERITED, and characters |
156 | // of the current script. |
157 | while (limit < textLimit) { |
158 | ch = text.char32At(limit); // look ahead |
159 | s = uscript_getScript(ch, &ec); |
160 | if (s != USCRIPT_COMMON && s != USCRIPT_INHERITED) { |
161 | if (scriptCode == USCRIPT_INVALID_CODE) { |
162 | scriptCode = s; |
163 | } else if (s != scriptCode) { |
164 | break; |
165 | } |
166 | } |
167 | ++limit; |
168 | } |
169 | |
170 | // Return TRUE even if the entire text is COMMON / INHERITED, in |
171 | // which case scriptCode will be USCRIPT_INVALID_CODE. |
172 | return TRUE; |
173 | } |
174 | |
175 | void ScriptRunIterator::adjustLimit(int32_t delta) { |
176 | limit += delta; |
177 | textLimit += delta; |
178 | } |
179 | |
180 | //------------------------------------------------------------ |
181 | // AnyTransliterator |
182 | |
183 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(AnyTransliterator) |
184 | |
185 | AnyTransliterator::AnyTransliterator(const UnicodeString& id, |
186 | const UnicodeString& theTarget, |
187 | const UnicodeString& theVariant, |
188 | UScriptCode theTargetScript, |
189 | UErrorCode& ec) : |
190 | Transliterator(id, NULL), |
191 | targetScript(theTargetScript) |
192 | { |
193 | cache = uhash_openSize(uhash_hashLong, uhash_compareLong, NULL, ANY_TRANS_CACHE_INIT_SIZE, &ec); |
194 | if (U_FAILURE(ec)) { |
195 | return; |
196 | } |
197 | uhash_setValueDeleter(cache, _deleteTransliterator); |
198 | |
199 | target = theTarget; |
200 | if (theVariant.length() > 0) { |
201 | target.append(VARIANT_SEP).append(theVariant); |
202 | } |
203 | } |
204 | |
205 | AnyTransliterator::~AnyTransliterator() { |
206 | uhash_close(cache); |
207 | } |
208 | |
209 | /** |
210 | * Copy constructor. |
211 | */ |
212 | AnyTransliterator::AnyTransliterator(const AnyTransliterator& o) : |
213 | Transliterator(o), |
214 | target(o.target), |
215 | targetScript(o.targetScript) |
216 | { |
217 | // Don't copy the cache contents |
218 | UErrorCode ec = U_ZERO_ERROR; |
219 | cache = uhash_openSize(uhash_hashLong, uhash_compareLong, NULL, ANY_TRANS_CACHE_INIT_SIZE, &ec); |
220 | if (U_FAILURE(ec)) { |
221 | return; |
222 | } |
223 | uhash_setValueDeleter(cache, _deleteTransliterator); |
224 | } |
225 | |
226 | /** |
227 | * Transliterator API. |
228 | */ |
229 | AnyTransliterator* AnyTransliterator::clone() const { |
230 | return new AnyTransliterator(*this); |
231 | } |
232 | |
233 | /** |
234 | * Implements {@link Transliterator#handleTransliterate}. |
235 | */ |
236 | void AnyTransliterator::handleTransliterate(Replaceable& text, UTransPosition& pos, |
237 | UBool isIncremental) const { |
238 | int32_t allStart = pos.start; |
239 | int32_t allLimit = pos.limit; |
240 | |
241 | ScriptRunIterator it(text, pos.contextStart, pos.contextLimit); |
242 | |
243 | while (it.next()) { |
244 | // Ignore runs in the ante context |
245 | if (it.limit <= allStart) continue; |
246 | |
247 | // Try to instantiate transliterator from it.scriptCode to |
248 | // our target or target/variant |
249 | Transliterator* t = getTransliterator(it.scriptCode); |
250 | |
251 | if (t == NULL) { |
252 | // We have no transliterator. Do nothing, but keep |
253 | // pos.start up to date. |
254 | pos.start = it.limit; |
255 | continue; |
256 | } |
257 | |
258 | // If the run end is before the transliteration limit, do |
259 | // a non-incremental transliteration. Otherwise do an |
260 | // incremental one. |
261 | UBool incremental = isIncremental && (it.limit >= allLimit); |
262 | |
263 | pos.start = uprv_max(allStart, it.start); |
264 | pos.limit = uprv_min(allLimit, it.limit); |
265 | int32_t limit = pos.limit; |
266 | t->filteredTransliterate(text, pos, incremental); |
267 | int32_t delta = pos.limit - limit; |
268 | allLimit += delta; |
269 | it.adjustLimit(delta); |
270 | |
271 | // We're done if we enter the post context |
272 | if (it.limit >= allLimit) break; |
273 | } |
274 | |
275 | // Restore limit. pos.start is fine where the last transliterator |
276 | // left it, or at the end of the last run. |
277 | pos.limit = allLimit; |
278 | } |
279 | |
280 | Transliterator* AnyTransliterator::getTransliterator(UScriptCode source) const { |
281 | |
282 | if (source == targetScript || source == USCRIPT_INVALID_CODE) { |
283 | return NULL; |
284 | } |
285 | |
286 | Transliterator* t = NULL; |
287 | { |
288 | Mutex m(NULL); |
289 | t = (Transliterator*) uhash_iget(cache, (int32_t) source); |
290 | } |
291 | if (t == NULL) { |
292 | UErrorCode ec = U_ZERO_ERROR; |
293 | UnicodeString sourceName(uscript_getShortName(source), -1, US_INV); |
294 | UnicodeString id(sourceName); |
295 | id.append(TARGET_SEP).append(target); |
296 | |
297 | t = Transliterator::createInstance(id, UTRANS_FORWARD, ec); |
298 | if (U_FAILURE(ec) || t == NULL) { |
299 | delete t; |
300 | |
301 | // Try to pivot around Latin, our most common script |
302 | id = sourceName; |
303 | id.append(LATIN_PIVOT, -1).append(target); |
304 | t = Transliterator::createInstance(id, UTRANS_FORWARD, ec); |
305 | if (U_FAILURE(ec) || t == NULL) { |
306 | delete t; |
307 | t = NULL; |
308 | } |
309 | } |
310 | |
311 | if (t != NULL) { |
312 | Transliterator *rt = NULL; |
313 | { |
314 | Mutex m(NULL); |
315 | rt = static_cast<Transliterator *> (uhash_iget(cache, (int32_t) source)); |
316 | if (rt == NULL) { |
317 | // Common case, no race to cache this new transliterator. |
318 | uhash_iput(cache, (int32_t) source, t, &ec); |
319 | } else { |
320 | // Race case, some other thread beat us to caching this transliterator. |
321 | Transliterator *temp = rt; |
322 | rt = t; // Our newly created transliterator that lost the race & now needs deleting. |
323 | t = temp; // The transliterator from the cache that we will return. |
324 | } |
325 | } |
326 | delete rt; // will be non-null only in case of races. |
327 | } |
328 | } |
329 | return t; |
330 | } |
331 | |
332 | /** |
333 | * Return the script code for a given name, or -1 if not found. |
334 | */ |
335 | static UScriptCode scriptNameToCode(const UnicodeString& name) { |
336 | char buf[128]; |
337 | UScriptCode code; |
338 | UErrorCode ec = U_ZERO_ERROR; |
339 | int32_t nameLen = name.length(); |
340 | UBool isInvariant = uprv_isInvariantUString(name.getBuffer(), nameLen); |
341 | |
342 | if (isInvariant) { |
343 | name.extract(0, nameLen, buf, (int32_t)sizeof(buf), US_INV); |
344 | buf[127] = 0; // Make sure that we NULL terminate the string. |
345 | } |
346 | if (!isInvariant || uscript_getCode(buf, &code, 1, &ec) != 1 || U_FAILURE(ec)) |
347 | { |
348 | code = USCRIPT_INVALID_CODE; |
349 | } |
350 | return code; |
351 | } |
352 | |
353 | /** |
354 | * Registers standard transliterators with the system. Called by |
355 | * Transliterator during initialization. Scan all current targets and |
356 | * register those that are scripts T as Any-T/V. |
357 | */ |
358 | void AnyTransliterator::registerIDs() { |
359 | |
360 | UErrorCode ec = U_ZERO_ERROR; |
361 | Hashtable seen(TRUE, ec); |
362 | |
363 | int32_t sourceCount = Transliterator::_countAvailableSources(); |
364 | for (int32_t s=0; s<sourceCount; ++s) { |
365 | UnicodeString source; |
366 | Transliterator::_getAvailableSource(s, source); |
367 | |
368 | // Ignore the "Any" source |
369 | if (source.caseCompare(ANY, 3, 0 /*U_FOLD_CASE_DEFAULT*/) == 0) continue; |
370 | |
371 | int32_t targetCount = Transliterator::_countAvailableTargets(source); |
372 | for (int32_t t=0; t<targetCount; ++t) { |
373 | UnicodeString target; |
374 | Transliterator::_getAvailableTarget(t, source, target); |
375 | |
376 | // Only process each target once |
377 | if (seen.geti(target) != 0) continue; |
378 | ec = U_ZERO_ERROR; |
379 | seen.puti(target, 1, ec); |
380 | |
381 | // Get the script code for the target. If not a script, ignore. |
382 | UScriptCode targetScript = scriptNameToCode(target); |
383 | if (targetScript == USCRIPT_INVALID_CODE) continue; |
384 | |
385 | int32_t variantCount = Transliterator::_countAvailableVariants(source, target); |
386 | // assert(variantCount >= 1); |
387 | for (int32_t v=0; v<variantCount; ++v) { |
388 | UnicodeString variant; |
389 | Transliterator::_getAvailableVariant(v, source, target, variant); |
390 | |
391 | UnicodeString id; |
392 | TransliteratorIDParser::STVtoID(UnicodeString(TRUE, ANY, 3), target, variant, id); |
393 | ec = U_ZERO_ERROR; |
394 | AnyTransliterator* tl = new AnyTransliterator(id, target, variant, |
395 | targetScript, ec); |
396 | if (U_FAILURE(ec)) { |
397 | delete tl; |
398 | } else { |
399 | Transliterator::_registerInstance(tl); |
400 | Transliterator::_registerSpecialInverse(target, UnicodeString(TRUE, NULL_ID, 4), FALSE); |
401 | } |
402 | } |
403 | } |
404 | } |
405 | } |
406 | |
407 | U_NAMESPACE_END |
408 | |
409 | #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
410 | |
411 | //eof |
412 | |