1 /* 2 ********************************************************************** 3 * Copyright (C) 2001-2011, International Business Machines 4 * Corporation and others. All Rights Reserved. 5 ********************************************************************** 6 * Date Name Description 7 * 06/07/01 aliu Creation. 8 ********************************************************************** 9 */ 10 11 #include "unicode/utypes.h" 12 13 #if !UCONFIG_NO_TRANSLITERATION 14 15 #include "unicode/unifilt.h" 16 #include "unicode/uchar.h" 17 #include "unicode/uniset.h" 18 #include "cmemory.h" 19 #include "name2uni.h" 20 #include "patternprops.h" 21 #include "uprops.h" 22 #include "uinvchar.h" 23 #include "util.h" 24 25 U_NAMESPACE_BEGIN 26 27 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(NameUnicodeTransliterator) 28 29 static const UChar OPEN[] = {92,78,126,123,126,0}; // "\N~{~" 30 static const UChar OPEN_DELIM = 92; // '\\' first char of OPEN 31 static const UChar CLOSE_DELIM = 125; // '}' 32 static const UChar SPACE = 32; // ' ' 33 34 U_CDECL_BEGIN 35 36 // USetAdder implementation 37 // Does not use uset.h to reduce code dependencies 38 static void U_CALLCONV 39 _set_add(USet *set, UChar32 c) { 40 uset_add(set, c); 41 } 42 43 // These functions aren't used. 44 /*static void U_CALLCONV 45 _set_addRange(USet *set, UChar32 start, UChar32 end) { 46 ((UnicodeSet *)set)->add(start, end); 47 } 48 49 static void U_CALLCONV 50 _set_addString(USet *set, const UChar *str, int32_t length) { 51 ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length)); 52 }*/ 53 54 U_CDECL_END 55 56 /** 57 * Constructs a transliterator with the default delimiters '{' and 58 * '}'. 59 */ 60 NameUnicodeTransliterator::NameUnicodeTransliterator(UnicodeFilter* adoptedFilter) : 61 Transliterator(UNICODE_STRING("Name-Any", 8), adoptedFilter) { 62 63 UnicodeSet *legalPtr = &legal; 64 // Get the legal character set 65 USetAdder sa = { 66 (USet *)legalPtr, // USet* == UnicodeSet* 67 _set_add, 68 NULL, // Don't need _set_addRange 69 NULL, // Don't need _set_addString 70 NULL, // Don't need remove() 71 NULL 72 }; 73 uprv_getCharNameCharacters(&sa); 74 } 75 76 /** 77 * Destructor. 78 */ 79 NameUnicodeTransliterator::~NameUnicodeTransliterator() {} 80 81 /** 82 * Copy constructor. 83 */ 84 NameUnicodeTransliterator::NameUnicodeTransliterator(const NameUnicodeTransliterator& o) : 85 Transliterator(o), legal(o.legal) {} 86 87 /** 88 * Assignment operator. 89 */ 90 /*NameUnicodeTransliterator& NameUnicodeTransliterator::operator=( 91 const NameUnicodeTransliterator& o) { 92 Transliterator::operator=(o); 93 // not necessary: the legal sets should all be the same -- legal=o.legal; 94 return *this; 95 }*/ 96 97 /** 98 * Transliterator API. 99 */ 100 Transliterator* NameUnicodeTransliterator::clone(void) const { 101 return new NameUnicodeTransliterator(*this); 102 } 103 104 /** 105 * Implements {@link Transliterator#handleTransliterate}. 106 */ 107 void NameUnicodeTransliterator::handleTransliterate(Replaceable& text, UTransPosition& offsets, 108 UBool isIncremental) const { 109 // The failure mode, here and below, is to behave like Any-Null, 110 // if either there is no name data (max len == 0) or there is no 111 // memory (malloc() => NULL). 112 113 int32_t maxLen = uprv_getMaxCharNameLength(); 114 if (maxLen == 0) { 115 offsets.start = offsets.limit; 116 return; 117 } 118 119 // Accomodate the longest possible name 120 ++maxLen; // allow for temporary trailing space 121 char* cbuf = (char*) uprv_malloc(maxLen); 122 if (cbuf == NULL) { 123 offsets.start = offsets.limit; 124 return; 125 } 126 127 UnicodeString openPat(TRUE, OPEN, -1); 128 UnicodeString str, name; 129 130 int32_t cursor = offsets.start; 131 int32_t limit = offsets.limit; 132 133 // Modes: 134 // 0 - looking for open delimiter 135 // 1 - after open delimiter 136 int32_t mode = 0; 137 int32_t openPos = -1; // open delim candidate pos 138 139 UChar32 c; 140 while (cursor < limit) { 141 c = text.char32At(cursor); 142 143 switch (mode) { 144 case 0: // looking for open delimiter 145 if (c == OPEN_DELIM) { // quick check first 146 openPos = cursor; 147 int32_t i = 148 ICU_Utility::parsePattern(openPat, text, cursor, limit); 149 if (i >= 0 && i < limit) { 150 mode = 1; 151 name.truncate(0); 152 cursor = i; 153 continue; // *** reprocess char32At(cursor) 154 } 155 } 156 break; 157 158 case 1: // after open delimiter 159 // Look for legal chars. If \s+ is found, convert it 160 // to a single space. If closeDelimiter is found, exit 161 // the loop. If any other character is found, exit the 162 // loop. If the limit is reached, exit the loop. 163 164 // Convert \s+ => SPACE. This assumes there are no 165 // runs of >1 space characters in names. 166 if (PatternProps::isWhiteSpace(c)) { 167 // Ignore leading whitespace 168 if (name.length() > 0 && 169 name.charAt(name.length()-1) != SPACE) { 170 name.append(SPACE); 171 // If we are too long then abort. maxLen includes 172 // temporary trailing space, so use '>'. 173 if (name.length() > maxLen) { 174 mode = 0; 175 } 176 } 177 break; 178 } 179 180 if (c == CLOSE_DELIM) { 181 int32_t len = name.length(); 182 183 // Delete trailing space, if any 184 if (len > 0 && 185 name.charAt(len-1) == SPACE) { 186 --len; 187 } 188 189 if (uprv_isInvariantUString(name.getBuffer(), len)) { 190 name.extract(0, len, cbuf, maxLen, US_INV); 191 192 UErrorCode status = U_ZERO_ERROR; 193 c = u_charFromName(U_EXTENDED_CHAR_NAME, cbuf, &status); 194 if (U_SUCCESS(status)) { 195 // Lookup succeeded 196 197 // assert(UTF_CHAR_LENGTH(CLOSE_DELIM) == 1); 198 cursor++; // advance over CLOSE_DELIM 199 200 str.truncate(0); 201 str.append(c); 202 text.handleReplaceBetween(openPos, cursor, str); 203 204 // Adjust indices for the change in the length of 205 // the string. Do not assume that str.length() == 206 // 1, in case of surrogates. 207 int32_t delta = cursor - openPos - str.length(); 208 cursor -= delta; 209 limit -= delta; 210 // assert(cursor == openPos + str.length()); 211 } 212 } 213 // If the lookup failed, we leave things as-is and 214 // still switch to mode 0 and continue. 215 mode = 0; 216 openPos = -1; // close off candidate 217 continue; // *** reprocess char32At(cursor) 218 } 219 220 // Check if c is a legal char. We assume here that 221 // legal.contains(OPEN_DELIM) is FALSE, so when we abort a 222 // name, we don't have to go back to openPos+1. 223 if (legal.contains(c)) { 224 name.append(c); 225 // If we go past the longest possible name then abort. 226 // maxLen includes temporary trailing space, so use '>='. 227 if (name.length() >= maxLen) { 228 mode = 0; 229 } 230 } 231 232 // Invalid character 233 else { 234 --cursor; // Backup and reprocess this character 235 mode = 0; 236 } 237 238 break; 239 } 240 241 cursor += UTF_CHAR_LENGTH(c); 242 } 243 244 offsets.contextLimit += limit - offsets.limit; 245 offsets.limit = limit; 246 // In incremental mode, only advance the cursor up to the last 247 // open delimiter candidate. 248 offsets.start = (isIncremental && openPos >= 0) ? openPos : cursor; 249 250 uprv_free(cbuf); 251 } 252 253 U_NAMESPACE_END 254 255 #endif /* #if !UCONFIG_NO_TRANSLITERATION */ 256