1 /* 2 * Copyright (C) 1996-2010, 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 /** 22 * A <code>UCD</code> object holds the Collator-specific data needed to 23 * compute the length of the shortest string that can 24 * generate a partcular list of CEs. 25 * 26 * <code>UCD</code> objects are quite expensive to compute. Because 27 * of this, they are cached. When you call <code>ucd_open</code> it 28 * returns a reference counted cached object. When you call <code>ucd_close</code> 29 * the reference count on the object is decremented but the object is not deleted. 30 * 31 * If you do not need to reuse any unreferenced objects in the cache, you can call 32 * <code>ucd_flushCCache</code>. If you no longer need any <code>UCD</code> 33 * objects, you can call <code>ucd_freeCache</code> 34 */ 35 typedef void UCD; 36 37 /** 38 * Open a <code>UCD</code> object. 39 * 40 * @param coll - the collator 41 * @param status - will be set if any errors occur. 42 * 43 * @return the <code>UCD</code> object. You must call 44 * <code>ucd_close</code> when you are done using the object. 45 * 46 * Note: if on return status is set to an error, the only safe 47 * thing to do with the returned object is to call <code>ucd_close</code>. 48 * 49 * @internal ICU 4.0.1 technology preview 50 */ 51 U_CAPI UCD * U_EXPORT2 52 ucd_open(UCollator *coll, UErrorCode *status); 53 54 /** 55 * Release a <code>UCD</code> object. 56 * 57 * @param ucd - the object 58 * 59 * @internal ICU 4.0.1 technology preview 60 */ 61 U_CAPI void U_EXPORT2 62 ucd_close(UCD *ucd); 63 64 /** 65 * Get the <code>UCollator</code> object used to create a <code>UCD</code> object. 66 * The <code>UCollator</code> object returned may not be the exact 67 * object that was used to create this object, but it will have the 68 * same behavior. 69 * 70 * @param ucd - the <code>UCD</code> object 71 * 72 * @return the <code>UCollator</code> used to create the given 73 * <code>UCD</code> object. 74 * 75 * @internal ICU 4.0.1 technology preview 76 */ 77 U_CAPI UCollator * U_EXPORT2 78 ucd_getCollator(UCD *ucd); 79 80 /** 81 * <code>UCD</code> objects are expensive to compute, and so 82 * may be cached. This routine will free the cached objects and delete 83 * the cache. 84 * 85 * WARNING: Don't call this until you are have called <code>close</code> 86 * for each <code>UCD</code> object that you have used. also, 87 * DO NOT call this if another thread may be calling <code>ucd_flushCache</code> 88 * at the same time. 89 * 90 * @internal ICU 4.0.1 technology preview 91 */ 92 U_CAPI void U_EXPORT2 93 ucd_freeCache(); 94 95 /** 96 * <code>UCD</code> objects are expensive to compute, and so 97 * may be cached. This routine will remove any unused <code>UCD</code> 98 * objects from the cache. 99 * 100 * @internal 4.0.1 technology preview 101 */ 102 U_CAPI void U_EXPORT2 103 ucd_flushCache(); 104 105 /** 106 * BMS 107 * 108 * This object holds the information needed to do a Collation sensitive Boyer-Moore search. It encapulates 109 * the pattern, the "bad character" and "good suffix" tables, the Collator-based data needed to compute them, 110 * and a reference to the text being searched. 111 * 112 * To do a search, you first need to get a <code>UCD</code> object by calling <code>ucd_open</code>. 113 * Then you construct a <code>BMS</code> object from the <code>UCD</code> object, the pattern 114 * string and the target string. Then you call the <code>search</code> method. Here's a code sample: 115 * 116 * <pre> 117 * void boyerMooreExample(UCollator *collator, UChar *pattern, int32_t patternLen, UChar *target, int32_t targetLength) 118 * { 119 * UErrorCode status = U_ZERO_ERROR; 120 * int32_t offset = 0, start = -1, end = -1; 121 * UCD *ucd = NULL); 122 * BMS *bms = NULL; 123 * 124 * ucd = ucd_open(collator, &status); 125 * if (U_FAILURE(status)) { 126 * // could not create a UCD object 127 * return; 128 * } 129 * 130 * BMS *bms = bms_open(ucd, pattern, patternLength, target, targetlength, &status); 131 * if (U_FAILURE(status)) { 132 * // could not create a BMS object 133 * ucd_close(ucd); 134 * return; 135 * } 136 * 137 * 138 * // Find all matches 139 * while (bms_search(bms, offset, &start, &end)) { 140 * // process the match between start and end 141 * ... 142 * 143 * // advance past the match 144 * offset = end; 145 * } 146 * 147 * // at this point, if offset == 0, there were no matches 148 * if (offset == 0) { 149 * // handle the case of no matches 150 * } 151 * 152 * bms_close(bms); 153 * ucd_close(ucd); 154 * 155 * // UCD objects are cached, so the call to 156 * // ucd_close doesn't delete the object. 157 * // Call this if you don't need the object any more. 158 * ucd_flushCache(); 159 * } 160 * </pre> 161 * 162 * NOTE: This is a technology preview. The final version of this API may not bear any resenblence to this API. 163 * 164 * Knows linitations: 165 * 1) Backwards searching has not been implemented. 166 * 167 * 2) For Han and Hangul characters, this code ignores any Collation tailorings. In general, 168 * this isn't a problem, but in Korean locals, at strength 1, Hangul characters are tailored 169 * to be equal to Han characters with the same pronounciation. Because this code ignroes 170 * tailorings, searching for a Hangul character will not find a Han character and visa-versa. 171 * 172 * 3) In some cases, searching for a pattern that needs to be normalized and ends 173 * in a discontiguous contraction may fail. The only known cases of this are with 174 * the Tibetan script. For example searching for the pattern 175 * "\u0F7F\u0F80\u0F81\u0F82\u0F83\u0F84\u0F85" will fail. (This case is artificial. We've 176 * been unable to find a pratical, real-world example of this failure.) 177 * 178 * NOTE: This is a technology preview. The final version of this API may not bear any resenblence to this API. 179 * 180 * @internal ICU 4.0.1 technology preview 181 */ 182 struct BMS; 183 typedef struct BMS BMS; /**< @see BMS */ 184 185 /** 186 * Construct a <code>MBS</code> object. 187 * 188 * @param ucd - A <code>UCD</code> object holding the Collator-sensitive data 189 * @param pattern - the string for which to search 190 * @param patternLength - the length of the string for which to search 191 * @param target - the string in which to search 192 * @param targetLength - the length of the string in which to search 193 * @param status - will be set if any errors occur. 194 * 195 * @return the <code>BMS</code> object. 196 * 197 * Note: if on return status is set to an error, the only safe 198 * thing to do with the returned object is to call 199 * <code>bms_close</code>. 200 * 201 * @internal ICU 4.0.1 technology preview 202 */ 203 U_CAPI BMS * U_EXPORT2 204 bms_open(UCD *ucd, 205 const UChar *pattern, int32_t patternLength, 206 const UChar *target, int32_t targetLength, 207 UErrorCode *status); 208 209 /** 210 * Close a <code>BMS</code> object and release all the 211 * storage associated with it. 212 * 213 * @param bms - the <code>BMS</code> object to close. 214 * @internal ICU 4.0.1 technology preview 215 */ 216 U_CAPI void U_EXPORT2 217 bms_close(BMS *bms); 218 219 /** 220 * Test the pattern to see if it generates any CEs. 221 * 222 * @param bms - the <code>BMS</code> object 223 * @return <code>TRUE</code> if the pattern string did not generate any CEs 224 * 225 * @internal ICU 4.0.1 technology preview 226 */ 227 U_CAPI UBool U_EXPORT2 228 bms_empty(BMS *bms); 229 230 /** 231 * Get the <code>UCD</code> object used to create 232 * a given <code>BMS</code> object. 233 * 234 * @param bms - the <code>BMS</code> object 235 * 236 * @return - the <code>UCD</code> object used to create 237 * the given <code>BMS</code> object. 238 * 239 * @internal ICU 4.0.1 technology preview 240 */ 241 U_CAPI UCD * U_EXPORT2 242 bms_getData(BMS *bms); 243 244 /** 245 * Search for the pattern string in the target string. 246 * 247 * @param bms - the <code>BMS</code> object 248 * @param offset - the offset in the target string at which to begin the search 249 * @param start - will be set to the starting offset of the match, or -1 if there's no match 250 * @param end - will be set to the ending offset of the match, or -1 if there's no match 251 * 252 * @return <code>TRUE</code> if the match succeeds, <code>FALSE</code> otherwise. 253 * 254 * @internal ICU 4.0.1 technology preview 255 */ 256 U_CAPI UBool U_EXPORT2 257 bms_search(BMS *bms, int32_t offset, int32_t *start, int32_t *end); 258 259 /** 260 * Set the target string for the match. 261 * 262 * @param bms - the <code>BMS</code> object 263 * @param target - the new target string 264 * @param targetLength - the length of the new target string 265 * @param status - will be set if any errors occur. 266 * 267 * @internal ICU 4.0.1 technology preview 268 */ 269 U_CAPI void U_EXPORT2 270 bms_setTargetString(BMS *bms, const UChar *target, int32_t targetLength, UErrorCode *status); 271 272 #endif 273 274 #endif /* _BMS_H */ 275
