2
**********************************************************************
3
* Copyright (c) 2001-2011, International Business Machines
4
* Corporation and others. All Rights Reserved.
5
**********************************************************************
6
* Date Name Description
7
* 11/19/2001 aliu Creation.
8
**********************************************************************
11
#include "unicode/utypes.h"
13
#if !UCONFIG_NO_TRANSLITERATION
15
#include "unicode/utf16.h"
21
static const UChar UNIPRE[] = {85,43,0}; // "U+"
22
static const UChar BS_u[] = {92,117,0}; // "\\u"
23
static const UChar BS_U[] = {92,85,0}; // "\\U"
24
static const UChar XMLPRE[] = {38,35,120,0}; // "&#x"
25
static const UChar XML10PRE[] = {38,35,0}; // "&#"
26
static const UChar PERLPRE[] = {92,120,123,0}; // "\\x{"
27
static const UChar SEMI[] = {59,0}; // ";"
28
static const UChar RBRACE[] = {125,0}; // "}"
30
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(EscapeTransliterator)
35
static Transliterator* _createEscUnicode(const UnicodeString& ID, Transliterator::Token /*context*/) {
36
// Unicode: "U+10FFFF" hex, min=4, max=6
37
return new EscapeTransliterator(ID, UnicodeString(TRUE, UNIPRE, 2), UnicodeString(), 16, 4, TRUE, NULL);
39
static Transliterator* _createEscJava(const UnicodeString& ID, Transliterator::Token /*context*/) {
40
// Java: "\\uFFFF" hex, min=4, max=4
41
return new EscapeTransliterator(ID, UnicodeString(TRUE, BS_u, 2), UnicodeString(), 16, 4, FALSE, NULL);
43
static Transliterator* _createEscC(const UnicodeString& ID, Transliterator::Token /*context*/) {
44
// C: "\\uFFFF" hex, min=4, max=4; \\U0010FFFF hex, min=8, max=8
45
return new EscapeTransliterator(ID, UnicodeString(TRUE, BS_u, 2), UnicodeString(), 16, 4, TRUE,
46
new EscapeTransliterator(UnicodeString(), UnicodeString(TRUE, BS_U, 2), UnicodeString(), 16, 8, TRUE, NULL));
48
static Transliterator* _createEscXML(const UnicodeString& ID, Transliterator::Token /*context*/) {
49
// XML: "" hex, min=1, max=6
50
return new EscapeTransliterator(ID, UnicodeString(TRUE, XMLPRE, 3), UnicodeString(SEMI[0]), 16, 1, TRUE, NULL);
52
static Transliterator* _createEscXML10(const UnicodeString& ID, Transliterator::Token /*context*/) {
53
// XML10: "&1114111;" dec, min=1, max=7 (not really "Any-Hex")
54
return new EscapeTransliterator(ID, UnicodeString(TRUE, XML10PRE, 2), UnicodeString(SEMI[0]), 10, 1, TRUE, NULL);
56
static Transliterator* _createEscPerl(const UnicodeString& ID, Transliterator::Token /*context*/) {
57
// Perl: "\\x{263A}" hex, min=1, max=6
58
return new EscapeTransliterator(ID, UnicodeString(TRUE, PERLPRE, 3), UnicodeString(RBRACE[0]), 16, 1, TRUE, NULL);
62
* Registers standard variants with the system. Called by
63
* Transliterator during initialization.
65
void EscapeTransliterator::registerIDs() {
66
Token t = integerToken(0);
68
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/Unicode"), _createEscUnicode, t);
70
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/Java"), _createEscJava, t);
72
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/C"), _createEscC, t);
74
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/XML"), _createEscXML, t);
76
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/XML10"), _createEscXML10, t);
78
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex/Perl"), _createEscPerl, t);
80
Transliterator::_registerFactory(UNICODE_STRING_SIMPLE("Any-Hex"), _createEscJava, t);
84
* Constructs an escape transliterator with the given ID and
85
* parameters. See the class member documentation for details.
87
EscapeTransliterator::EscapeTransliterator(const UnicodeString& newID,
88
const UnicodeString& _prefix, const UnicodeString& _suffix,
89
int32_t _radix, int32_t _minDigits,
90
UBool _grokSupplementals,
91
EscapeTransliterator* adoptedSupplementalHandler) :
92
Transliterator(newID, NULL)
94
this->prefix = _prefix;
95
this->suffix = _suffix;
97
this->minDigits = _minDigits;
98
this->grokSupplementals = _grokSupplementals;
99
this->supplementalHandler = adoptedSupplementalHandler;
105
EscapeTransliterator::EscapeTransliterator(const EscapeTransliterator& o) :
110
minDigits(o.minDigits),
111
grokSupplementals(o.grokSupplementals) {
112
supplementalHandler = (o.supplementalHandler != 0) ?
113
new EscapeTransliterator(*o.supplementalHandler) : NULL;
116
EscapeTransliterator::~EscapeTransliterator() {
117
delete supplementalHandler;
121
* Transliterator API.
123
Transliterator* EscapeTransliterator::clone() const {
124
return new EscapeTransliterator(*this);
128
* Implements {@link Transliterator#handleTransliterate}.
130
void EscapeTransliterator::handleTransliterate(Replaceable& text,
132
UBool /*isIncremental*/) const
134
/* TODO: Verify that isIncremental can be ignored */
135
int32_t start = pos.start;
136
int32_t limit = pos.limit;
138
UnicodeString buf(prefix);
139
int32_t prefixLen = prefix.length();
140
UBool redoPrefix = FALSE;
142
while (start < limit) {
143
int32_t c = grokSupplementals ? text.char32At(start) : text.charAt(start);
144
int32_t charLen = grokSupplementals ? U16_LENGTH(c) : 1;
146
if ((c & 0xFFFF0000) != 0 && supplementalHandler != NULL) {
148
buf.append(supplementalHandler->prefix);
149
ICU_Utility::appendNumber(buf, c, supplementalHandler->radix,
150
supplementalHandler->minDigits);
151
buf.append(supplementalHandler->suffix);
159
buf.truncate(prefixLen);
161
ICU_Utility::appendNumber(buf, c, radix, minDigits);
165
text.handleReplaceBetween(start, start + charLen, buf);
166
start += buf.length();
167
limit += buf.length() - charLen;
170
pos.contextLimit += limit - pos.limit;
177
#endif /* #if !UCONFIG_NO_TRANSLITERATION */