icuSources/i18n/unicode/bms.h

   1 /*
   2  * Copyright (C) 1996-2011, International Business Machines Corporation and Others.
   3  * All rights reserved.
   4  */
   5
   6 /**
   7  * \file
   8  * \brief C API: Boyer-Moore StringSearch prototype.
   9  * \internal
  10  */
  11
  12 #ifndef _BMS_H
  13 #define _BMS_H
  14
  15 #include "unicode/utypes.h"
  16
  17 #if !UCONFIG_NO_COLLATION && !UCONFIG_NO_BREAK_ITERATION
  18
  19 #include "unicode/ucol.h"
  20
  21 #ifndef U_HIDE_INTERNAL_API
  22
  23 /**
  24  * A <code>UCD</code> object holds the Collator-specific data needed to
  25  * compute the length of the shortest string that can
  26  * generate a partcular list of CEs.
  27  *
  28  * <code>UCD</code> objects are quite expensive to compute. Because
  29  * of this, they are cached. When you call <code>ucd_open</code> it
  30  * returns a reference counted cached object. When you call <code>ucd_close</code>
  31  * the reference count on the object is decremented but the object is not deleted.
  32  *
  33  * If you do not need to reuse any unreferenced objects in the cache, you can call
  34  * <code>ucd_flushCCache</code>. If you no longer need any <code>UCD</code>
  35  * objects, you can call <code>ucd_freeCache</code>
  36  *
  37  * @internal ICU 4.0.1 technology preview
  38  */
  39 typedef void UCD;
  40
  41 /**
  42  * Open a <code>UCD</code> object.
  43  *
  44  * @param coll - the collator
  45  * @param status - will be set if any errors occur.
  46  *
  47  * @return the <code>UCD</code> object. You must call
  48  *         <code>ucd_close</code> when you are done using the object.
  49  *
  50  * Note: if on return status is set to an error, the only safe
  51  * thing to do with the returned object is to call <code>ucd_close</code>.
  52  *
  53  * @internal ICU 4.0.1 technology preview
  54  */
  55 U_CAPI UCD * U_EXPORT2
  56 ucd_open(UCollator *coll, UErrorCode *status);
  57
  58 /**
  59  * Release a <code>UCD</code> object.
  60  *
  61  * @param ucd - the object
  62  *
  63  * @internal ICU 4.0.1 technology preview
  64  */
  65 U_CAPI void U_EXPORT2
  66 ucd_close(UCD *ucd);
  67
  68 /**
  69  * Get the <code>UCollator</code> object used to create a <code>UCD</code> object.
  70  * The <code>UCollator</code> object returned may not be the exact
  71  * object that was used to create this object, but it will have the
  72  * same behavior.
  73  *
  74  * @param ucd - the <code>UCD</code> object
  75  *
  76  * @return the <code>UCollator</code> used to create the given
  77  *         <code>UCD</code> object.
  78  *
  79  * @internal ICU 4.0.1 technology preview
  80  */
  81 U_CAPI UCollator * U_EXPORT2
  82 ucd_getCollator(UCD *ucd);
  83
  84 /**
  85  * <code>UCD</code> objects are expensive to compute, and so
  86  * may be cached. This routine will free the cached objects and delete
  87  * the cache.
  88  *
  89  * WARNING: Don't call this until you are have called <code>close</code>
  90  * for each <code>UCD</code> object that you have used. also,
  91  * DO NOT call this if another thread may be calling <code>ucd_flushCache</code>
  92  * at the same time.
  93  *
  94  * @internal ICU 4.0.1 technology preview
  95  */
  96 U_CAPI void U_EXPORT2
  97 ucd_freeCache();
  98
  99 /**
 100  * <code>UCD</code> objects are expensive to compute, and so
 101  * may be cached. This routine will remove any unused <code>UCD</code>
 102  * objects from the cache.
 103  *
 104  * @internal 4.0.1 technology preview
 105  */
 106 U_CAPI void U_EXPORT2
 107 ucd_flushCache();
 108
 109 /**
 110  * BMS
 111  *
 112  * This object holds the information needed to do a Collation sensitive Boyer-Moore search. It encapulates
 113  * the pattern, the "bad character" and "good suffix" tables, the Collator-based data needed to compute them,
 114  * and a reference to the text being searched.
 115  *
 116  * To do a search, you first need to get a <code>UCD</code> object by calling <code>ucd_open</code>.
 117  * Then you construct a <code>BMS</code> object from the <code>UCD</code> object, the pattern
 118  * string and the target string. Then you call the <code>search</code> method. Here's a code sample:
 119  *
 120  * <pre>
 121  * void boyerMooreExample(UCollator *collator, UChar *pattern, int32_t patternLen, UChar *target, int32_t targetLength)
 122  * {
 123  *     UErrorCode status = U_ZERO_ERROR;
 124  *     int32_t offset = 0, start = -1, end = -1;
 125  *     UCD *ucd = NULL);
 126  *     BMS *bms = NULL;
 127  *
 128  *     ucd = ucd_open(collator, &status);
 129  *     if (U_FAILURE(status)) {
 130  *         // could not create a UCD object
 131  *         return;
 132  *     }
 133  *
 134  *     BMS *bms = bms_open(ucd, pattern, patternLength, target, targetlength, &status);
 135  *     if (U_FAILURE(status)) {
 136  *         // could not create a BMS object
 137  *         ucd_close(ucd);
 138  *         return;
 139  *     }
 140  *
 141  *
 142  *     // Find all matches
 143  *     while (bms_search(bms, offset, &start, &end)) {
 144  *         // process the match between start and end
 145  *         ...
 146  *
 147  *         // advance past the match
 148  *         offset = end;
 149  *     }
 150  *
 151  *     // at this point, if offset == 0, there were no matches
 152  *     if (offset == 0) {
 153  *         // handle the case of no matches
 154  *     }
 155  *
 156  *     bms_close(bms);
 157  *     ucd_close(ucd);
 158  *
 159  *     // UCD objects are cached, so the call to
 160  *     // ucd_close doesn't delete the object.
 161  *     // Call this if you don't need the object any more.
 162  *     ucd_flushCache();
 163  * }
 164  * </pre>
 165  *
 166  * NOTE: This is a technology preview. The final version of this API may not bear any resenblence to this API.
 167  *
 168  * Knows linitations:
 169  *   1) Backwards searching has not been implemented.
 170  *
 171  *   2) For Han and Hangul characters, this code ignores any Collation tailorings. In general,
 172  *      this isn't a problem, but in Korean locals, at strength 1, Hangul characters are tailored
 173  *      to be equal to Han characters with the same pronounciation. Because this code ignroes
 174  *      tailorings, searching for a Hangul character will not find a Han character and visa-versa.
 175  *
 176  *   3) In some cases, searching for a pattern that needs to be normalized and ends
 177  *      in a discontiguous contraction may fail. The only known cases of this are with
 178  *      the Tibetan script. For example searching for the pattern
 179  *      "\u0F7F\u0F80\u0F81\u0F82\u0F83\u0F84\u0F85" will fail. (This case is artificial. We've
 180  *      been unable to find a pratical, real-world example of this failure.)
 181  *
 182  * NOTE: This is a technology preview. The final version of this API may not bear any resenblence to this API.
 183  *
 184  * @internal ICU 4.0.1 technology preview
 185  */
 186 struct BMS;
 187 typedef struct BMS BMS; /**< @see BMS */
 188
 189 /**
 190  * Construct a <code>MBS</code> object.
 191  *
 192  * @param ucd - A <code>UCD</code> object holding the Collator-sensitive data
 193  * @param pattern - the string for which to search
 194  * @param patternLength - the length of the string for which to search
 195  * @param target - the string in which to search
 196  * @param targetLength - the length of the string in which to search
 197  * @param status - will be set if any errors occur.
 198  *
 199  * @return the <code>BMS</code> object.
 200  *
 201  * Note: if on return status is set to an error, the only safe
 202  * thing to do with the returned object is to call
 203  * <code>bms_close</code>.
 204  *
 205  * @internal ICU 4.0.1 technology preview
 206  */
 207 U_CAPI BMS * U_EXPORT2
 208 bms_open(UCD *ucd,
 209          const UChar *pattern, int32_t patternLength,
 210          const UChar *target,  int32_t targetLength,
 211          UErrorCode  *status);
 212
 213 /**
 214  * Close a <code>BMS</code> object and release all the
 215  * storage associated with it.
 216  *
 217  * @param bms - the <code>BMS</code> object to close.
 218  * @internal ICU 4.0.1 technology preview
 219  */
 220 U_CAPI void U_EXPORT2
 221 bms_close(BMS *bms);
 222
 223 /**
 224  * Test the pattern to see if it generates any CEs.
 225  *
 226  * @param bms - the <code>BMS</code> object
 227  * @return <code>TRUE</code> if the pattern string did not generate any CEs
 228  *
 229  * @internal ICU 4.0.1 technology preview
 230  */
 231 U_CAPI UBool U_EXPORT2
 232 bms_empty(BMS *bms);
 233
 234 /**
 235  * Get the <code>UCD</code> object used to create
 236  * a given <code>BMS</code> object.
 237  *
 238  * @param bms - the <code>BMS</code> object
 239  *
 240  * @return - the <code>UCD</code> object used to create
 241  *           the given <code>BMS</code> object.
 242  *
 243  * @internal ICU 4.0.1 technology preview
 244  */
 245 U_CAPI UCD * U_EXPORT2
 246 bms_getData(BMS *bms);
 247
 248 /**
 249  * Search for the pattern string in the target string.
 250  *
 251  * @param bms - the <code>BMS</code> object
 252  * @param offset - the offset in the target string at which to begin the search
 253  * @param start - will be set to the starting offset of the match, or -1 if there's no match
 254  * @param end - will be set to the ending offset of the match, or -1 if there's no match
 255  *
 256  * @return <code>TRUE</code> if the match succeeds, <code>FALSE</code> otherwise.
 257  *
 258  * @internal ICU 4.0.1 technology preview
 259  */
 260 U_CAPI UBool U_EXPORT2
 261 bms_search(BMS *bms, int32_t offset, int32_t *start, int32_t *end);
 262
 263 /**
 264  * Set the target string for the match.
 265  *
 266  * @param bms - the <code>BMS</code> object
 267  * @param target - the new target string
 268  * @param targetLength - the length of the new target string
 269  * @param status - will be set if any errors occur.
 270  *
 271  * @internal ICU 4.0.1 technology preview
 272  */
 273 U_CAPI void U_EXPORT2
 274 bms_setTargetString(BMS *bms, const UChar *target, int32_t targetLength, UErrorCode *status);
 275
 276 #endif  /* U_HIDE_INTERNAL_API */
 277
 278 #endif
 279
 280 #endif /* _BMS_H */