Wed, 31 Dec 2014 07:22:50 +0100
Correct previous dual key logic pending first delivery installment.
michael@0 | 1 | /* |
michael@0 | 2 | ******************************************************************************* |
michael@0 | 3 | * |
michael@0 | 4 | * Copyright (C) 1999-2011, International Business Machines |
michael@0 | 5 | * Corporation and others. All Rights Reserved. |
michael@0 | 6 | * |
michael@0 | 7 | ******************************************************************************* |
michael@0 | 8 | * file name: unistr_case.cpp |
michael@0 | 9 | * encoding: US-ASCII |
michael@0 | 10 | * tab size: 8 (not used) |
michael@0 | 11 | * indentation:2 |
michael@0 | 12 | * |
michael@0 | 13 | * created on: 2004aug19 |
michael@0 | 14 | * created by: Markus W. Scherer |
michael@0 | 15 | * |
michael@0 | 16 | * Case-mapping functions moved here from unistr.cpp |
michael@0 | 17 | */ |
michael@0 | 18 | |
michael@0 | 19 | #include "unicode/utypes.h" |
michael@0 | 20 | #include "unicode/putil.h" |
michael@0 | 21 | #include "cstring.h" |
michael@0 | 22 | #include "cmemory.h" |
michael@0 | 23 | #include "unicode/ustring.h" |
michael@0 | 24 | #include "unicode/unistr.h" |
michael@0 | 25 | #include "unicode/uchar.h" |
michael@0 | 26 | #include "uelement.h" |
michael@0 | 27 | #include "ustr_imp.h" |
michael@0 | 28 | |
michael@0 | 29 | U_NAMESPACE_BEGIN |
michael@0 | 30 | |
michael@0 | 31 | //======================================== |
michael@0 | 32 | // Read-only implementation |
michael@0 | 33 | //======================================== |
michael@0 | 34 | |
michael@0 | 35 | int8_t |
michael@0 | 36 | UnicodeString::doCaseCompare(int32_t start, |
michael@0 | 37 | int32_t length, |
michael@0 | 38 | const UChar *srcChars, |
michael@0 | 39 | int32_t srcStart, |
michael@0 | 40 | int32_t srcLength, |
michael@0 | 41 | uint32_t options) const |
michael@0 | 42 | { |
michael@0 | 43 | // compare illegal string values |
michael@0 | 44 | // treat const UChar *srcChars==NULL as an empty string |
michael@0 | 45 | if(isBogus()) { |
michael@0 | 46 | return -1; |
michael@0 | 47 | } |
michael@0 | 48 | |
michael@0 | 49 | // pin indices to legal values |
michael@0 | 50 | pinIndices(start, length); |
michael@0 | 51 | |
michael@0 | 52 | if(srcChars == NULL) { |
michael@0 | 53 | srcStart = srcLength = 0; |
michael@0 | 54 | } |
michael@0 | 55 | |
michael@0 | 56 | // get the correct pointer |
michael@0 | 57 | const UChar *chars = getArrayStart(); |
michael@0 | 58 | |
michael@0 | 59 | chars += start; |
michael@0 | 60 | if(srcStart!=0) { |
michael@0 | 61 | srcChars += srcStart; |
michael@0 | 62 | } |
michael@0 | 63 | |
michael@0 | 64 | if(chars != srcChars) { |
michael@0 | 65 | UErrorCode errorCode=U_ZERO_ERROR; |
michael@0 | 66 | int32_t result=u_strcmpFold(chars, length, srcChars, srcLength, |
michael@0 | 67 | options|U_COMPARE_IGNORE_CASE, &errorCode); |
michael@0 | 68 | if(result!=0) { |
michael@0 | 69 | return (int8_t)(result >> 24 | 1); |
michael@0 | 70 | } |
michael@0 | 71 | } else { |
michael@0 | 72 | // get the srcLength if necessary |
michael@0 | 73 | if(srcLength < 0) { |
michael@0 | 74 | srcLength = u_strlen(srcChars + srcStart); |
michael@0 | 75 | } |
michael@0 | 76 | if(length != srcLength) { |
michael@0 | 77 | return (int8_t)((length - srcLength) >> 24 | 1); |
michael@0 | 78 | } |
michael@0 | 79 | } |
michael@0 | 80 | return 0; |
michael@0 | 81 | } |
michael@0 | 82 | |
michael@0 | 83 | //======================================== |
michael@0 | 84 | // Write implementation |
michael@0 | 85 | //======================================== |
michael@0 | 86 | |
michael@0 | 87 | UnicodeString & |
michael@0 | 88 | UnicodeString::caseMap(const UCaseMap *csm, |
michael@0 | 89 | UStringCaseMapper *stringCaseMapper) { |
michael@0 | 90 | if(isEmpty() || !isWritable()) { |
michael@0 | 91 | // nothing to do |
michael@0 | 92 | return *this; |
michael@0 | 93 | } |
michael@0 | 94 | |
michael@0 | 95 | // We need to allocate a new buffer for the internal string case mapping function. |
michael@0 | 96 | // This is very similar to how doReplace() keeps the old array pointer |
michael@0 | 97 | // and deletes the old array itself after it is done. |
michael@0 | 98 | // In addition, we are forcing cloneArrayIfNeeded() to always allocate a new array. |
michael@0 | 99 | UChar oldStackBuffer[US_STACKBUF_SIZE]; |
michael@0 | 100 | UChar *oldArray; |
michael@0 | 101 | int32_t oldLength; |
michael@0 | 102 | |
michael@0 | 103 | if(fFlags&kUsingStackBuffer) { |
michael@0 | 104 | // copy the stack buffer contents because it will be overwritten |
michael@0 | 105 | u_memcpy(oldStackBuffer, fUnion.fStackBuffer, fShortLength); |
michael@0 | 106 | oldArray = oldStackBuffer; |
michael@0 | 107 | oldLength = fShortLength; |
michael@0 | 108 | } else { |
michael@0 | 109 | oldArray = getArrayStart(); |
michael@0 | 110 | oldLength = length(); |
michael@0 | 111 | } |
michael@0 | 112 | |
michael@0 | 113 | int32_t capacity; |
michael@0 | 114 | if(oldLength <= US_STACKBUF_SIZE) { |
michael@0 | 115 | capacity = US_STACKBUF_SIZE; |
michael@0 | 116 | } else { |
michael@0 | 117 | capacity = oldLength + 20; |
michael@0 | 118 | } |
michael@0 | 119 | int32_t *bufferToDelete = 0; |
michael@0 | 120 | if(!cloneArrayIfNeeded(capacity, capacity, FALSE, &bufferToDelete, TRUE)) { |
michael@0 | 121 | return *this; |
michael@0 | 122 | } |
michael@0 | 123 | |
michael@0 | 124 | // Case-map, and if the result is too long, then reallocate and repeat. |
michael@0 | 125 | UErrorCode errorCode; |
michael@0 | 126 | int32_t newLength; |
michael@0 | 127 | do { |
michael@0 | 128 | errorCode = U_ZERO_ERROR; |
michael@0 | 129 | newLength = stringCaseMapper(csm, getArrayStart(), getCapacity(), |
michael@0 | 130 | oldArray, oldLength, &errorCode); |
michael@0 | 131 | setLength(newLength); |
michael@0 | 132 | } while(errorCode==U_BUFFER_OVERFLOW_ERROR && cloneArrayIfNeeded(newLength, newLength, FALSE)); |
michael@0 | 133 | |
michael@0 | 134 | if (bufferToDelete) { |
michael@0 | 135 | uprv_free(bufferToDelete); |
michael@0 | 136 | } |
michael@0 | 137 | if(U_FAILURE(errorCode)) { |
michael@0 | 138 | setToBogus(); |
michael@0 | 139 | } |
michael@0 | 140 | return *this; |
michael@0 | 141 | } |
michael@0 | 142 | |
michael@0 | 143 | UnicodeString & |
michael@0 | 144 | UnicodeString::foldCase(uint32_t options) { |
michael@0 | 145 | UCaseMap csm=UCASEMAP_INITIALIZER; |
michael@0 | 146 | csm.csp=ucase_getSingleton(); |
michael@0 | 147 | csm.options=options; |
michael@0 | 148 | return caseMap(&csm, ustrcase_internalFold); |
michael@0 | 149 | } |
michael@0 | 150 | |
michael@0 | 151 | U_NAMESPACE_END |
michael@0 | 152 | |
michael@0 | 153 | // Defined here to reduce dependencies on break iterator |
michael@0 | 154 | U_CAPI int32_t U_EXPORT2 |
michael@0 | 155 | uhash_hashCaselessUnicodeString(const UElement key) { |
michael@0 | 156 | U_NAMESPACE_USE |
michael@0 | 157 | const UnicodeString *str = (const UnicodeString*) key.pointer; |
michael@0 | 158 | if (str == NULL) { |
michael@0 | 159 | return 0; |
michael@0 | 160 | } |
michael@0 | 161 | // Inefficient; a better way would be to have a hash function in |
michael@0 | 162 | // UnicodeString that does case folding on the fly. |
michael@0 | 163 | UnicodeString copy(*str); |
michael@0 | 164 | return copy.foldCase().hashCode(); |
michael@0 | 165 | } |
michael@0 | 166 | |
michael@0 | 167 | // Defined here to reduce dependencies on break iterator |
michael@0 | 168 | U_CAPI UBool U_EXPORT2 |
michael@0 | 169 | uhash_compareCaselessUnicodeString(const UElement key1, const UElement key2) { |
michael@0 | 170 | U_NAMESPACE_USE |
michael@0 | 171 | const UnicodeString *str1 = (const UnicodeString*) key1.pointer; |
michael@0 | 172 | const UnicodeString *str2 = (const UnicodeString*) key2.pointer; |
michael@0 | 173 | if (str1 == str2) { |
michael@0 | 174 | return TRUE; |
michael@0 | 175 | } |
michael@0 | 176 | if (str1 == NULL || str2 == NULL) { |
michael@0 | 177 | return FALSE; |
michael@0 | 178 | } |
michael@0 | 179 | return str1->caseCompare(*str2, U_FOLD_CASE_DEFAULT) == 0; |
michael@0 | 180 | } |