icuSources/common/ustrcase.c

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2001-2003, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  ustrcase.c
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2002feb20
  14 *   created by: Markus W. Scherer
  15 *
  16 *   Implementation file for string casing C API functions.
  17 *   Uses functions from uchar.c for basic functionality that requires access
  18 *   to the Unicode Character Database (uprops.dat).
  19 */
  20
  21 #include "unicode/utypes.h"
  22 #include "unicode/ustring.h"
  23 #include "unicode/ubrk.h"
  24 #include "cmemory.h"
  25 #include "unormimp.h"
  26 #include "ustr_imp.h"
  27
  28 /* string casing ------------------------------------------------------------ */
  29
  30 #if !UCONFIG_NO_BREAK_ITERATION
  31
  32 /*
  33  * Internal titlecasing function,
  34  * using u_internalStrToLower() and u_internalToTitle().
  35  *
  36  * Must get titleIter!=NULL.
  37  */
  38 U_CFUNC int32_t
  39 u_internalStrToTitle(UChar *dest, int32_t destCapacity,
  40                      const UChar *src, int32_t srcLength,
  41                      UBreakIterator *titleIter,
  42                      const char *locale,
  43                      UErrorCode *pErrorCode) {
  44     UCharIterator iter;
  45     UChar32 c;
  46     int32_t prev, index, destIndex, length;
  47     UBool isFirstIndex;
  48
  49     /* set up local variables */
  50     uiter_setString(&iter, src, srcLength);
  51     destIndex=0;
  52     prev=0;
  53     isFirstIndex=TRUE;
  54
  55     /* titlecasing loop */
  56     while(prev<srcLength) {
  57         /* find next index where to titlecase */
  58         if(isFirstIndex) {
  59             isFirstIndex=FALSE;
  60             index=ubrk_first(titleIter);
  61         } else {
  62             index=ubrk_next(titleIter);
  63         }
  64         if(index==UBRK_DONE || index>srcLength) {
  65             index=srcLength;
  66         }
  67
  68         /* lowercase [prev..index[ */
  69         if(prev<index) {
  70             if(destIndex<destCapacity) {
  71                 length=u_internalStrToLower(dest+destIndex, destCapacity-destIndex,
  72                                             src, srcLength,
  73                                             prev, index,
  74                                             locale,
  75                                             pErrorCode);
  76             } else {
  77                 length=u_internalStrToLower(NULL, 0,
  78                                             src, srcLength,
  79                                             prev, index,
  80                                             locale,
  81                                             pErrorCode);
  82             }
  83             destIndex+=length;
  84         }
  85
  86         if(index>=srcLength) {
  87             break;
  88         }
  89
  90         /* titlecase the character at the found index */
  91         UTF_NEXT_CHAR(src, index, srcLength, c);
  92         iter.move(&iter, index, UITER_ZERO);
  93         if(destIndex<destCapacity) {
  94             length=u_internalToTitle(c, &iter,
  95                                      dest+destIndex, destCapacity-destIndex,
  96                                      locale);
  97         } else {
  98             length=u_internalToTitle(c, &iter, NULL, 0, locale);
  99         }
 100         if(length<0) {
 101             length=-length;
 102         }
 103         destIndex+=length;
 104
 105         prev=index;
 106     }
 107
 108     return destIndex;
 109 }
 110
 111 #endif
 112
 113 /*
 114  * Implement argument checking and buffer handling
 115  * for string case mapping as a common function.
 116  */
 117 enum {
 118     TO_LOWER,
 119     TO_UPPER,
 120     TO_TITLE,
 121     FOLD_CASE
 122 };
 123
 124 static int32_t
 125 u_strCaseMap(UChar *dest, int32_t destCapacity,
 126              const UChar *src, int32_t srcLength,
 127              UBreakIterator *titleIter,
 128              const char *locale,
 129              uint32_t options,
 130              int32_t toWhichCase,
 131              UErrorCode *pErrorCode) {
 132     UChar buffer[300];
 133     UChar *temp;
 134     int32_t destLength;
 135     UBool ownTitleIter;
 136
 137     /* check argument values */
 138     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
 139         return 0;
 140     }
 141     if( destCapacity<0 ||
 142         (dest==NULL && destCapacity>0) ||
 143         src==NULL ||
 144         srcLength<-1
 145     ) {
 146         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 147         return 0;
 148     }
 149
 150     /* get the string length */
 151     if(srcLength==-1) {
 152         srcLength=u_strlen(src);
 153     }
 154
 155     /* check for overlapping source and destination */
 156     if( dest!=NULL &&
 157         ((src>=dest && src<(dest+destCapacity)) ||
 158          (dest>=src && dest<(src+srcLength)))
 159     ) {
 160         /* overlap: provide a temporary destination buffer and later copy the result */
 161         if(destCapacity<=(sizeof(buffer)/U_SIZEOF_UCHAR)) {
 162             /* the stack buffer is large enough */
 163             temp=buffer;
 164         } else {
 165             /* allocate a buffer */
 166             temp=(UChar *)uprv_malloc(destCapacity*U_SIZEOF_UCHAR);
 167             if(temp==NULL) {
 168                 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
 169                 return 0;
 170             }
 171         }
 172     } else {
 173         temp=dest;
 174     }
 175
 176     ownTitleIter=FALSE;
 177     destLength=0;
 178
 179     if(toWhichCase==TO_LOWER) {
 180         destLength=u_internalStrToLower(temp, destCapacity,
 181                                         src, srcLength,
 182                                         0, srcLength,
 183                                         locale, pErrorCode);
 184     } else if(toWhichCase==TO_UPPER) {
 185         destLength=u_internalStrToUpper(temp, destCapacity, src, srcLength,
 186                                         locale, pErrorCode);
 187 #if !UCONFIG_NO_BREAK_ITERATION
 188     } else if(toWhichCase==TO_TITLE) {
 189         if(titleIter==NULL) {
 190             titleIter=ubrk_open(UBRK_WORD, locale,
 191                                 src, srcLength,
 192                                 pErrorCode);
 193             ownTitleIter=(UBool)U_SUCCESS(*pErrorCode);
 194         }
 195         if(U_SUCCESS(*pErrorCode)) {
 196             destLength=u_internalStrToTitle(temp, destCapacity, src, srcLength,
 197                                             titleIter, locale, pErrorCode);
 198         }
 199 #endif
 200     } else {
 201         destLength=u_internalStrFoldCase(temp, destCapacity, src, srcLength,
 202                                          options, pErrorCode);
 203     }
 204     if(temp!=dest) {
 205         /* copy the result string to the destination buffer */
 206         if(destLength>0) {
 207             int32_t copyLength= destLength<=destCapacity ? destLength : destCapacity;
 208             if(copyLength>0) {
 209                 uprv_memmove(dest, temp, copyLength*U_SIZEOF_UCHAR);
 210             }
 211         }
 212         if(temp!=buffer) {
 213             uprv_free(temp);
 214         }
 215     }
 216
 217 #if !UCONFIG_NO_BREAK_ITERATION
 218     if(ownTitleIter) {
 219         ubrk_close(titleIter);
 220     }
 221 #endif
 222
 223     return u_terminateUChars(dest, destCapacity, destLength, pErrorCode);
 224 }
 225
 226 U_CAPI int32_t U_EXPORT2
 227 u_strToLower(UChar *dest, int32_t destCapacity,
 228              const UChar *src, int32_t srcLength,
 229              const char *locale,
 230              UErrorCode *pErrorCode) {
 231     return u_strCaseMap(dest, destCapacity,
 232                         src, srcLength,
 233                         NULL, locale, 0,
 234                         TO_LOWER, pErrorCode);
 235 }
 236
 237 U_CAPI int32_t U_EXPORT2
 238 u_strToUpper(UChar *dest, int32_t destCapacity,
 239              const UChar *src, int32_t srcLength,
 240              const char *locale,
 241              UErrorCode *pErrorCode) {
 242     return u_strCaseMap(dest, destCapacity,
 243                         src, srcLength,
 244                         NULL, locale, 0,
 245                         TO_UPPER, pErrorCode);
 246 }
 247
 248 U_CAPI int32_t U_EXPORT2
 249 u_strToTitle(UChar *dest, int32_t destCapacity,
 250              const UChar *src, int32_t srcLength,
 251              UBreakIterator *titleIter,
 252              const char *locale,
 253              UErrorCode *pErrorCode) {
 254     return u_strCaseMap(dest, destCapacity,
 255                         src, srcLength,
 256                         titleIter, locale, 0,
 257                         TO_TITLE, pErrorCode);
 258 }
 259
 260 U_CAPI int32_t U_EXPORT2
 261 u_strFoldCase(UChar *dest, int32_t destCapacity,
 262               const UChar *src, int32_t srcLength,
 263               uint32_t options,
 264               UErrorCode *pErrorCode) {
 265     return u_strCaseMap(dest, destCapacity,
 266                         src, srcLength,
 267                         NULL, NULL, options,
 268                         FOLD_CASE, pErrorCode);
 269 }
 270
 271 /* case-insensitive string comparisons */
 272
 273 U_CAPI int32_t U_EXPORT2
 274 u_strCaseCompare(const UChar *s1, int32_t length1,
 275                  const UChar *s2, int32_t length2,
 276                  uint32_t options,
 277                  UErrorCode *pErrorCode) {
 278     /* argument checking */
 279     if(pErrorCode==0 || U_FAILURE(*pErrorCode)) {
 280         return 0;
 281     }
 282     if(s1==NULL || length1<-1 || s2==NULL || length2<-1) {
 283         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
 284         return 0;
 285     }
 286     return unorm_cmpEquivFold(s1, length1, s2, length2,
 287                               options|U_COMPARE_IGNORE_CASE,
 288                               pErrorCode);
 289 }
 290
 291 U_CAPI int32_t U_EXPORT2
 292 u_strcasecmp(const UChar *s1, const UChar *s2, uint32_t options) {
 293     UErrorCode errorCode=U_ZERO_ERROR;
 294     return unorm_cmpEquivFold(s1, -1, s2, -1,
 295                               options|U_COMPARE_IGNORE_CASE,
 296                               &errorCode);
 297 }
 298
 299 U_CAPI int32_t U_EXPORT2
 300 u_memcasecmp(const UChar *s1, const UChar *s2, int32_t length, uint32_t options) {
 301     UErrorCode errorCode=U_ZERO_ERROR;
 302     return unorm_cmpEquivFold(s1, length, s2, length,
 303                               options|U_COMPARE_IGNORE_CASE,
 304                               &errorCode);
 305 }
 306
 307 U_CAPI int32_t U_EXPORT2
 308 u_strncasecmp(const UChar *s1, const UChar *s2, int32_t n, uint32_t options) {
 309     UErrorCode errorCode=U_ZERO_ERROR;
 310     return unorm_cmpEquivFold(s1, n, s2, n,
 311                               options|(U_COMPARE_IGNORE_CASE|_STRNCMP_STYLE),
 312                               &errorCode);
 313 }