1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/esctrn.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,179 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (c) 2001-2011, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +**********************************************************************
1.9 +* Date Name Description
1.10 +* 11/19/2001 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/utf16.h"
1.19 +#include "esctrn.h"
1.20 +#include "util.h"
1.21 +
1.22 +U_NAMESPACE_BEGIN
1.23 +
1.24 +static const UChar UNIPRE[] = {85,43,0}; // "U+"
1.25 +static const UChar BS_u[] = {92,117,0}; // "\\u"
1.26 +static const UChar BS_U[] = {92,85,0}; // "\\U"
1.27 +static const UChar XMLPRE[] = {38,35,120,0}; // "&#x"
1.28 +static const UChar XML10PRE[] = {38,35,0}; // "&#"
1.29 +static const UChar PERLPRE[] = {92,120,123,0}; // "\\x{"
1.30 +static const UChar SEMI[] = {59,0}; // ";"
1.31 +static const UChar RBRACE[] = {125,0}; // "}"
1.32 +
1.33 +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(EscapeTransliterator)
1.34 +
1.35 +/**
1.36 + * Factory methods
1.37 + */
1.38 +static Transliterator* _createEscUnicode(const UnicodeString& ID, Transliterator::Token /*context*/) {
1.39 + // Unicode: "U+10FFFF" hex, min=4, max=6
1.40 + return new EscapeTransliterator(ID, UnicodeString(TRUE, UNIPRE, 2), UnicodeString(), 16, 4, TRUE, NULL);
1.41 +}
1.42 +static Transliterator* _createEscJava(const UnicodeString& ID, Transliterator::Token /*context*/) {
1.43 + // Java: "\\uFFFF" hex, min=4, max=4
1.44 + return new EscapeTransliterator(ID, UnicodeString(TRUE, BS_u, 2), UnicodeString(), 16, 4, FALSE, NULL);
1.45 +}
1.46 +static Transliterator* _createEscC(const UnicodeString& ID, Transliterator::Token /*context*/) {
1.47 + // C: "\\uFFFF" hex, min=4, max=4; \\U0010FFFF hex, min=8, max=8
1.48 + return new EscapeTransliterator(ID, UnicodeString(TRUE, BS_u, 2), UnicodeString(), 16, 4, TRUE,
1.49 + new EscapeTransliterator(UnicodeString(), UnicodeString(TRUE, BS_U, 2), UnicodeString(), 16, 8, TRUE, NULL));
1.50 +}
1.51 +static Transliterator* _createEscXML(const UnicodeString& ID, Transliterator::Token /*context*/) {
1.52 + // XML: "" hex, min=1, max=6
1.53 + return new EscapeTransliterator(ID, UnicodeString(TRUE, XMLPRE, 3), UnicodeString(SEMI[0]), 16, 1, TRUE, NULL);
1.54 +}
1.55 +static Transliterator* _createEscXML10(const UnicodeString& ID, Transliterator::Token /*context*/) {
1.56 + // XML10: "&1114111;" dec, min=1, max=7 (not really "Any-Hex")
1.57 + return new EscapeTransliterator(ID, UnicodeString(TRUE, XML10PRE, 2), UnicodeString(SEMI[0]), 10, 1, TRUE, NULL);
1.58 +}
1.59 +static Transliterator* _createEscPerl(const UnicodeString& ID, Transliterator::Token /*context*/) {
1.60 + // Perl: "\\x{263A}" hex, min=1, max=6
1.61 + return new EscapeTransliterator(ID, UnicodeString(TRUE, PERLPRE, 3), UnicodeString(RBRACE[0]), 16, 1, TRUE, NULL);
1.62 +}
1.63 +
1.64 +/**
1.65 + * Registers standard variants with the system. Called by
1.66 + * Transliterator during initialization.
1.67 + */
1.68 +void EscapeTransliterator::registerIDs() {
1.69 + Token t = integerToken(0);
1.70 +
1.71 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/Unicode"), _createEscUnicode, t);
1.72 +
1.73 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/Java"), _createEscJava, t);
1.74 +
1.75 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/C"), _createEscC, t);
1.76 +
1.77 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/XML"), _createEscXML, t);
1.78 +
1.79 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/XML10"), _createEscXML10, t);
1.80 +
1.81 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/Perl"), _createEscPerl, t);
1.82 +
1.83 + Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex"), _createEscJava, t);
1.84 +}
1.85 +
1.86 +/**
1.87 + * Constructs an escape transliterator with the given ID and
1.88 + * parameters. See the class member documentation for details.
1.89 + */
1.90 +EscapeTransliterator::EscapeTransliterator(const UnicodeString& newID,
1.91 + const UnicodeString& _prefix, const UnicodeString& _suffix,
1.92 + int32_t _radix, int32_t _minDigits,
1.93 + UBool _grokSupplementals,
1.94 + EscapeTransliterator* adoptedSupplementalHandler) :
1.95 + Transliterator(newID, NULL)
1.96 +{
1.97 + this->prefix = _prefix;
1.98 + this->suffix = _suffix;
1.99 + this->radix = _radix;
1.100 + this->minDigits = _minDigits;
1.101 + this->grokSupplementals = _grokSupplementals;
1.102 + this->supplementalHandler = adoptedSupplementalHandler;
1.103 +}
1.104 +
1.105 +/**
1.106 + * Copy constructor.
1.107 + */
1.108 +EscapeTransliterator::EscapeTransliterator(const EscapeTransliterator& o) :
1.109 + Transliterator(o),
1.110 + prefix(o.prefix),
1.111 + suffix(o.suffix),
1.112 + radix(o.radix),
1.113 + minDigits(o.minDigits),
1.114 + grokSupplementals(o.grokSupplementals) {
1.115 + supplementalHandler = (o.supplementalHandler != 0) ?
1.116 + new EscapeTransliterator(*o.supplementalHandler) : NULL;
1.117 +}
1.118 +
1.119 +EscapeTransliterator::~EscapeTransliterator() {
1.120 + delete supplementalHandler;
1.121 +}
1.122 +
1.123 +/**
1.124 + * Transliterator API.
1.125 + */
1.126 +Transliterator* EscapeTransliterator::clone() const {
1.127 + return new EscapeTransliterator(*this);
1.128 +}
1.129 +
1.130 +/**
1.131 + * Implements {@link Transliterator#handleTransliterate}.
1.132 + */
1.133 +void EscapeTransliterator::handleTransliterate(Replaceable& text,
1.134 + UTransPosition& pos,
1.135 + UBool /*isIncremental*/) const
1.136 +{
1.137 + /* TODO: Verify that isIncremental can be ignored */
1.138 + int32_t start = pos.start;
1.139 + int32_t limit = pos.limit;
1.140 +
1.141 + UnicodeString buf(prefix);
1.142 + int32_t prefixLen = prefix.length();
1.143 + UBool redoPrefix = FALSE;
1.144 +
1.145 + while (start < limit) {
1.146 + int32_t c = grokSupplementals ? text.char32At(start) : text.charAt(start);
1.147 + int32_t charLen = grokSupplementals ? U16_LENGTH(c) : 1;
1.148 +
1.149 + if ((c & 0xFFFF0000) != 0 && supplementalHandler != NULL) {
1.150 + buf.truncate(0);
1.151 + buf.append(supplementalHandler->prefix);
1.152 + ICU_Utility::appendNumber(buf, c, supplementalHandler->radix,
1.153 + supplementalHandler->minDigits);
1.154 + buf.append(supplementalHandler->suffix);
1.155 + redoPrefix = TRUE;
1.156 + } else {
1.157 + if (redoPrefix) {
1.158 + buf.truncate(0);
1.159 + buf.append(prefix);
1.160 + redoPrefix = FALSE;
1.161 + } else {
1.162 + buf.truncate(prefixLen);
1.163 + }
1.164 + ICU_Utility::appendNumber(buf, c, radix, minDigits);
1.165 + buf.append(suffix);
1.166 + }
1.167 +
1.168 + text.handleReplaceBetween(start, start + charLen, buf);
1.169 + start += buf.length();
1.170 + limit += buf.length() - charLen;
1.171 + }
1.172 +
1.173 + pos.contextLimit += limit - pos.limit;
1.174 + pos.limit = limit;
1.175 + pos.start = start;
1.176 +}
1.177 +
1.178 +U_NAMESPACE_END
1.179 +
1.180 +#endif /* #if !UCONFIG_NO_TRANSLITERATION */
1.181 +
1.182 +//eof
