1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/anytrans.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,386 @@
1.4 +/*
1.5 +*****************************************************************
1.6 +* Copyright (c) 2002-2011, International Business Machines Corporation
1.7 +* and others. All Rights Reserved.
1.8 +*****************************************************************
1.9 +* Date Name Description
1.10 +* 06/06/2002 aliu Creation.
1.11 +*****************************************************************
1.12 +*/
1.13 +
1.14 +#include "unicode/utypes.h"
1.15 +
1.16 +#if !UCONFIG_NO_TRANSLITERATION
1.17 +
1.18 +#include "unicode/uobject.h"
1.19 +#include "unicode/uscript.h"
1.20 +#include "nultrans.h"
1.21 +#include "anytrans.h"
1.22 +#include "uvector.h"
1.23 +#include "tridpars.h"
1.24 +#include "hash.h"
1.25 +#include "putilimp.h"
1.26 +#include "uinvchar.h"
1.27 +
1.28 +//------------------------------------------------------------
1.29 +// Constants
1.30 +
1.31 +static const UChar TARGET_SEP = 45; // '-'
1.32 +static const UChar VARIANT_SEP = 47; // '/'
1.33 +static const UChar ANY[] = {65,110,121,0}; // "Any"
1.34 +static const UChar NULL_ID[] = {78,117,108,108,0}; // "Null"
1.35 +static const UChar LATIN_PIVOT[] = {45,76,97,116,105,110,59,76,97,116,105,110,45,0}; // "-Latin;Latin-"
1.36 +
1.37 +//------------------------------------------------------------
1.38 +
1.39 +U_CDECL_BEGIN
1.40 +/**
1.41 + * Deleter function for Transliterator*.
1.42 + */
1.43 +static void U_CALLCONV
1.44 +_deleteTransliterator(void *obj) {
1.45 + delete (icu::Transliterator*) obj;
1.46 +}
1.47 +U_CDECL_END
1.48 +
1.49 +//------------------------------------------------------------
1.50 +
1.51 +U_NAMESPACE_BEGIN
1.52 +
1.53 +//------------------------------------------------------------
1.54 +// ScriptRunIterator
1.55 +
1.56 +/**
1.57 + * Returns a series of ranges corresponding to scripts. They will be
1.58 + * of the form:
1.59 + *
1.60 + * ccccSScSSccccTTcTcccc - c = common, S = first script, T = second
1.61 + * | | - first run (start, limit)
1.62 + * | | - second run (start, limit)
1.63 + *
1.64 + * That is, the runs will overlap. The reason for this is so that a
1.65 + * transliterator can consider common characters both before and after
1.66 + * the scripts.
1.67 + */
1.68 +class ScriptRunIterator : public UMemory {
1.69 +private:
1.70 + const Replaceable& text;
1.71 + int32_t textStart;
1.72 + int32_t textLimit;
1.73 +
1.74 +public:
1.75 + /**
1.76 + * The code of the current run, valid after next() returns. May
1.77 + * be USCRIPT_INVALID_CODE if and only if the entire text is
1.78 + * COMMON/INHERITED.
1.79 + */
1.80 + UScriptCode scriptCode;
1.81 +
1.82 + /**
1.83 + * The start of the run, inclusive, valid after next() returns.
1.84 + */
1.85 + int32_t start;
1.86 +
1.87 + /**
1.88 + * The end of the run, exclusive, valid after next() returns.
1.89 + */
1.90 + int32_t limit;
1.91 +
1.92 + /**
1.93 + * Constructs a run iterator over the given text from start
1.94 + * (inclusive) to limit (exclusive).
1.95 + */
1.96 + ScriptRunIterator(const Replaceable& text, int32_t start, int32_t limit);
1.97 +
1.98 + /**
1.99 + * Returns TRUE if there are any more runs. TRUE is always
1.100 + * returned at least once. Upon return, the caller should
1.101 + * examine scriptCode, start, and limit.
1.102 + */
1.103 + UBool next();
1.104 +
1.105 + /**
1.106 + * Adjusts internal indices for a change in the limit index of the
1.107 + * given delta. A positive delta means the limit has increased.
1.108 + */
1.109 + void adjustLimit(int32_t delta);
1.110 +
1.111 +private:
1.112 + ScriptRunIterator(const ScriptRunIterator &other); // forbid copying of this class
1.113 + ScriptRunIterator &operator=(const ScriptRunIterator &other); // forbid copying of this class
1.114 +};
1.115 +
1.116 +ScriptRunIterator::ScriptRunIterator(const Replaceable& theText,
1.117 + int32_t myStart, int32_t myLimit) :
1.118 + text(theText)
1.119 +{
1.120 + textStart = myStart;
1.121 + textLimit = myLimit;
1.122 + limit = myStart;
1.123 +}
1.124 +
1.125 +UBool ScriptRunIterator::next() {
1.126 + UChar32 ch;
1.127 + UScriptCode s;
1.128 + UErrorCode ec = U_ZERO_ERROR;
1.129 +
1.130 + scriptCode = USCRIPT_INVALID_CODE; // don't know script yet
1.131 + start = limit;
1.132 +
1.133 + // Are we done?
1.134 + if (start == textLimit) {
1.135 + return FALSE;
1.136 + }
1.137 +
1.138 + // Move start back to include adjacent COMMON or INHERITED
1.139 + // characters
1.140 + while (start > textStart) {
1.141 + ch = text.char32At(start - 1); // look back
1.142 + s = uscript_getScript(ch, &ec);
1.143 + if (s == USCRIPT_COMMON || s == USCRIPT_INHERITED) {
1.144 + --start;
1.145 + } else {
1.146 + break;
1.147 + }
1.148 + }
1.149 +
1.150 + // Move limit ahead to include COMMON, INHERITED, and characters
1.151 + // of the current script.
1.152 + while (limit < textLimit) {
1.153 + ch = text.char32At(limit); // look ahead
1.154 + s = uscript_getScript(ch, &ec);
1.155 + if (s != USCRIPT_COMMON && s != USCRIPT_INHERITED) {
1.156 + if (scriptCode == USCRIPT_INVALID_CODE) {
1.157 + scriptCode = s;
1.158 + } else if (s != scriptCode) {
1.159 + break;
1.160 + }
1.161 + }
1.162 + ++limit;
1.163 + }
1.164 +
1.165 + // Return TRUE even if the entire text is COMMON / INHERITED, in
1.166 + // which case scriptCode will be USCRIPT_INVALID_CODE.
1.167 + return TRUE;
1.168 +}
1.169 +
1.170 +void ScriptRunIterator::adjustLimit(int32_t delta) {
1.171 + limit += delta;
1.172 + textLimit += delta;
1.173 +}
1.174 +
1.175 +//------------------------------------------------------------
1.176 +// AnyTransliterator
1.177 +
1.178 +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(AnyTransliterator)
1.179 +
1.180 +AnyTransliterator::AnyTransliterator(const UnicodeString& id,
1.181 + const UnicodeString& theTarget,
1.182 + const UnicodeString& theVariant,
1.183 + UScriptCode theTargetScript,
1.184 + UErrorCode& ec) :
1.185 + Transliterator(id, NULL),
1.186 + targetScript(theTargetScript)
1.187 +{
1.188 + cache = uhash_open(uhash_hashLong, uhash_compareLong, NULL, &ec);
1.189 + if (U_FAILURE(ec)) {
1.190 + return;
1.191 + }
1.192 + uhash_setValueDeleter(cache, _deleteTransliterator);
1.193 +
1.194 + target = theTarget;
1.195 + if (theVariant.length() > 0) {
1.196 + target.append(VARIANT_SEP).append(theVariant);
1.197 + }
1.198 +}
1.199 +
1.200 +AnyTransliterator::~AnyTransliterator() {
1.201 + uhash_close(cache);
1.202 +}
1.203 +
1.204 +/**
1.205 + * Copy constructor.
1.206 + */
1.207 +AnyTransliterator::AnyTransliterator(const AnyTransliterator& o) :
1.208 + Transliterator(o),
1.209 + target(o.target),
1.210 + targetScript(o.targetScript)
1.211 +{
1.212 + // Don't copy the cache contents
1.213 + UErrorCode ec = U_ZERO_ERROR;
1.214 + cache = uhash_open(uhash_hashLong, uhash_compareLong, NULL, &ec);
1.215 + if (U_FAILURE(ec)) {
1.216 + return;
1.217 + }
1.218 + uhash_setValueDeleter(cache, _deleteTransliterator);
1.219 +}
1.220 +
1.221 +/**
1.222 + * Transliterator API.
1.223 + */
1.224 +Transliterator* AnyTransliterator::clone() const {
1.225 + return new AnyTransliterator(*this);
1.226 +}
1.227 +
1.228 +/**
1.229 + * Implements {@link Transliterator#handleTransliterate}.
1.230 + */
1.231 +void AnyTransliterator::handleTransliterate(Replaceable& text, UTransPosition& pos,
1.232 + UBool isIncremental) const {
1.233 + int32_t allStart = pos.start;
1.234 + int32_t allLimit = pos.limit;
1.235 +
1.236 + ScriptRunIterator it(text, pos.contextStart, pos.contextLimit);
1.237 +
1.238 + while (it.next()) {
1.239 + // Ignore runs in the ante context
1.240 + if (it.limit <= allStart) continue;
1.241 +
1.242 + // Try to instantiate transliterator from it.scriptCode to
1.243 + // our target or target/variant
1.244 + Transliterator* t = getTransliterator(it.scriptCode);
1.245 +
1.246 + if (t == NULL) {
1.247 + // We have no transliterator. Do nothing, but keep
1.248 + // pos.start up to date.
1.249 + pos.start = it.limit;
1.250 + continue;
1.251 + }
1.252 +
1.253 + // If the run end is before the transliteration limit, do
1.254 + // a non-incremental transliteration. Otherwise do an
1.255 + // incremental one.
1.256 + UBool incremental = isIncremental && (it.limit >= allLimit);
1.257 +
1.258 + pos.start = uprv_max(allStart, it.start);
1.259 + pos.limit = uprv_min(allLimit, it.limit);
1.260 + int32_t limit = pos.limit;
1.261 + t->filteredTransliterate(text, pos, incremental);
1.262 + int32_t delta = pos.limit - limit;
1.263 + allLimit += delta;
1.264 + it.adjustLimit(delta);
1.265 +
1.266 + // We're done if we enter the post context
1.267 + if (it.limit >= allLimit) break;
1.268 + }
1.269 +
1.270 + // Restore limit. pos.start is fine where the last transliterator
1.271 + // left it, or at the end of the last run.
1.272 + pos.limit = allLimit;
1.273 +}
1.274 +
1.275 +Transliterator* AnyTransliterator::getTransliterator(UScriptCode source) const {
1.276 +
1.277 + if (source == targetScript || source == USCRIPT_INVALID_CODE) {
1.278 + return NULL;
1.279 + }
1.280 +
1.281 + Transliterator* t = (Transliterator*) uhash_iget(cache, (int32_t) source);
1.282 + if (t == NULL) {
1.283 + UErrorCode ec = U_ZERO_ERROR;
1.284 + UnicodeString sourceName(uscript_getName(source), -1, US_INV);
1.285 + UnicodeString id(sourceName);
1.286 + id.append(TARGET_SEP).append(target);
1.287 +
1.288 + t = Transliterator::createInstance(id, UTRANS_FORWARD, ec);
1.289 + if (U_FAILURE(ec) || t == NULL) {
1.290 + delete t;
1.291 +
1.292 + // Try to pivot around Latin, our most common script
1.293 + id = sourceName;
1.294 + id.append(LATIN_PIVOT, -1).append(target);
1.295 + t = Transliterator::createInstance(id, UTRANS_FORWARD, ec);
1.296 + if (U_FAILURE(ec) || t == NULL) {
1.297 + delete t;
1.298 + t = NULL;
1.299 + }
1.300 + }
1.301 +
1.302 + if (t != NULL) {
1.303 + uhash_iput(cache, (int32_t) source, t, &ec);
1.304 + }
1.305 + }
1.306 +
1.307 + return t;
1.308 +}
1.309 +
1.310 +/**
1.311 + * Return the script code for a given name, or -1 if not found.
1.312 + */
1.313 +static UScriptCode scriptNameToCode(const UnicodeString& name) {
1.314 + char buf[128];
1.315 + UScriptCode code;
1.316 + UErrorCode ec = U_ZERO_ERROR;
1.317 + int32_t nameLen = name.length();
1.318 + UBool isInvariant = uprv_isInvariantUString(name.getBuffer(), nameLen);
1.319 +
1.320 + if (isInvariant) {
1.321 + name.extract(0, nameLen, buf, (int32_t)sizeof(buf), US_INV);
1.322 + buf[127] = 0; // Make sure that we NULL terminate the string.
1.323 + }
1.324 + if (!isInvariant || uscript_getCode(buf, &code, 1, &ec) != 1 || U_FAILURE(ec))
1.325 + {
1.326 + code = USCRIPT_INVALID_CODE;
1.327 + }
1.328 + return code;
1.329 +}
1.330 +
1.331 +/**
1.332 + * Registers standard transliterators with the system. Called by
1.333 + * Transliterator during initialization. Scan all current targets and
1.334 + * register those that are scripts T as Any-T/V.
1.335 + */
1.336 +void AnyTransliterator::registerIDs() {
1.337 +
1.338 + UErrorCode ec = U_ZERO_ERROR;
1.339 + Hashtable seen(TRUE, ec);
1.340 +
1.341 + int32_t sourceCount = Transliterator::_countAvailableSources();
1.342 + for (int32_t s=0; s<sourceCount; ++s) {
1.343 + UnicodeString source;
1.344 + Transliterator::_getAvailableSource(s, source);
1.345 +
1.346 + // Ignore the "Any" source
1.347 + if (source.caseCompare(ANY, 3, 0 /*U_FOLD_CASE_DEFAULT*/) == 0) continue;
1.348 +
1.349 + int32_t targetCount = Transliterator::_countAvailableTargets(source);
1.350 + for (int32_t t=0; t<targetCount; ++t) {
1.351 + UnicodeString target;
1.352 + Transliterator::_getAvailableTarget(t, source, target);
1.353 +
1.354 + // Only process each target once
1.355 + if (seen.geti(target) != 0) continue;
1.356 + ec = U_ZERO_ERROR;
1.357 + seen.puti(target, 1, ec);
1.358 +
1.359 + // Get the script code for the target. If not a script, ignore.
1.360 + UScriptCode targetScript = scriptNameToCode(target);
1.361 + if (targetScript == USCRIPT_INVALID_CODE) continue;
1.362 +
1.363 + int32_t variantCount = Transliterator::_countAvailableVariants(source, target);
1.364 + // assert(variantCount >= 1);
1.365 + for (int32_t v=0; v<variantCount; ++v) {
1.366 + UnicodeString variant;
1.367 + Transliterator::_getAvailableVariant(v, source, target, variant);
1.368 +
1.369 + UnicodeString id;
1.370 + TransliteratorIDParser::STVtoID(UnicodeString(TRUE, ANY, 3), target, variant, id);
1.371 + ec = U_ZERO_ERROR;
1.372 + AnyTransliterator* t = new AnyTransliterator(id, target, variant,
1.373 + targetScript, ec);
1.374 + if (U_FAILURE(ec)) {
1.375 + delete t;
1.376 + } else {
1.377 + Transliterator::_registerInstance(t);
1.378 + Transliterator::_registerSpecialInverse(target, UnicodeString(TRUE, NULL_ID, 4), FALSE);
1.379 + }
1.380 + }
1.381 + }
1.382 + }
1.383 +}
1.384 +
1.385 +U_NAMESPACE_END
1.386 +
1.387 +#endif /* #if !UCONFIG_NO_TRANSLITERATION */
1.388 +
1.389 +//eof
