OLD | NEW |
(Empty) | |
| 1 /* |
| 2 ** 2015 Aug 04 |
| 3 ** |
| 4 ** The author disclaims copyright to this source code. In place of |
| 5 ** a legal notice, here is a blessing: |
| 6 ** |
| 7 ** May you do good and not evil. |
| 8 ** May you find forgiveness for yourself and forgive others. |
| 9 ** May you share freely, never taking more than you give. |
| 10 ** |
| 11 ****************************************************************************** |
| 12 ** |
| 13 ** This file contains test code only, it is not included in release |
| 14 ** versions of FTS5. It contains the implementation of an FTS5 auxiliary |
| 15 ** function very similar to the FTS4 function matchinfo(): |
| 16 ** |
| 17 ** https://www.sqlite.org/fts3.html#matchinfo |
| 18 ** |
| 19 ** Known differences are that: |
| 20 ** |
| 21 ** 1) this function uses the FTS5 definition of "matchable phrase", which |
| 22 ** excludes any phrases that are part of an expression sub-tree that |
| 23 ** does not match the current row. This comes up for MATCH queries |
| 24 ** such as: |
| 25 ** |
| 26 ** "a OR (b AND c)" |
| 27 ** |
| 28 ** In FTS4, if a single row contains instances of tokens "a" and "c", |
| 29 ** but not "b", all instances of "c" are considered matches. In FTS5, |
| 30 ** they are not (as the "b AND c" sub-tree does not match the current |
| 31 ** row. |
| 32 ** |
| 33 ** 2) For the values returned by 'x' that apply to all rows of the table, |
| 34 ** NEAR constraints are not considered. But for the number of hits in |
| 35 ** the current row, they are. |
| 36 ** |
| 37 ** This file exports a single function that may be called to register the |
| 38 ** matchinfo() implementation with a database handle: |
| 39 ** |
| 40 ** int sqlite3Fts5TestRegisterMatchinfo(sqlite3 *db); |
| 41 */ |
| 42 |
| 43 |
| 44 #ifdef SQLITE_ENABLE_FTS5 |
| 45 |
| 46 #include "fts5.h" |
| 47 #include <assert.h> |
| 48 #include <string.h> |
| 49 |
| 50 typedef struct Fts5MatchinfoCtx Fts5MatchinfoCtx; |
| 51 |
| 52 #ifndef SQLITE_AMALGAMATION |
| 53 typedef unsigned int u32; |
| 54 #endif |
| 55 |
| 56 struct Fts5MatchinfoCtx { |
| 57 int nCol; /* Number of cols in FTS5 table */ |
| 58 int nPhrase; /* Number of phrases in FTS5 query */ |
| 59 char *zArg; /* nul-term'd copy of 2nd arg */ |
| 60 int nRet; /* Number of elements in aRet[] */ |
| 61 u32 *aRet; /* Array of 32-bit unsigned ints to return */ |
| 62 }; |
| 63 |
| 64 |
| 65 |
| 66 /* |
| 67 ** Return a pointer to the fts5_api pointer for database connection db. |
| 68 ** If an error occurs, return NULL and leave an error in the database |
| 69 ** handle (accessible using sqlite3_errcode()/errmsg()). |
| 70 */ |
| 71 static int fts5_api_from_db(sqlite3 *db, fts5_api **ppApi){ |
| 72 sqlite3_stmt *pStmt = 0; |
| 73 int rc; |
| 74 |
| 75 *ppApi = 0; |
| 76 rc = sqlite3_prepare(db, "SELECT fts5()", -1, &pStmt, 0); |
| 77 if( rc==SQLITE_OK ){ |
| 78 if( SQLITE_ROW==sqlite3_step(pStmt) |
| 79 && sizeof(fts5_api*)==sqlite3_column_bytes(pStmt, 0) |
| 80 ){ |
| 81 memcpy(ppApi, sqlite3_column_blob(pStmt, 0), sizeof(fts5_api*)); |
| 82 } |
| 83 rc = sqlite3_finalize(pStmt); |
| 84 } |
| 85 |
| 86 return rc; |
| 87 } |
| 88 |
| 89 |
| 90 /* |
| 91 ** Argument f should be a flag accepted by matchinfo() (a valid character |
| 92 ** in the string passed as the second argument). If it is not, -1 is |
| 93 ** returned. Otherwise, if f is a valid matchinfo flag, the value returned |
| 94 ** is the number of 32-bit integers added to the output array if the |
| 95 ** table has nCol columns and the query nPhrase phrases. |
| 96 */ |
| 97 static int fts5MatchinfoFlagsize(int nCol, int nPhrase, char f){ |
| 98 int ret = -1; |
| 99 switch( f ){ |
| 100 case 'p': ret = 1; break; |
| 101 case 'c': ret = 1; break; |
| 102 case 'x': ret = 3 * nCol * nPhrase; break; |
| 103 case 'y': ret = nCol * nPhrase; break; |
| 104 case 'b': ret = ((nCol + 31) / 32) * nPhrase; break; |
| 105 case 'n': ret = 1; break; |
| 106 case 'a': ret = nCol; break; |
| 107 case 'l': ret = nCol; break; |
| 108 case 's': ret = nCol; break; |
| 109 } |
| 110 return ret; |
| 111 } |
| 112 |
| 113 static int fts5MatchinfoIter( |
| 114 const Fts5ExtensionApi *pApi, /* API offered by current FTS version */ |
| 115 Fts5Context *pFts, /* First arg to pass to pApi functions */ |
| 116 Fts5MatchinfoCtx *p, |
| 117 int(*x)(const Fts5ExtensionApi*,Fts5Context*,Fts5MatchinfoCtx*,char,u32*) |
| 118 ){ |
| 119 int i; |
| 120 int n = 0; |
| 121 int rc = SQLITE_OK; |
| 122 char f; |
| 123 for(i=0; (f = p->zArg[i]); i++){ |
| 124 rc = x(pApi, pFts, p, f, &p->aRet[n]); |
| 125 if( rc!=SQLITE_OK ) break; |
| 126 n += fts5MatchinfoFlagsize(p->nCol, p->nPhrase, f); |
| 127 } |
| 128 return rc; |
| 129 } |
| 130 |
| 131 static int fts5MatchinfoXCb( |
| 132 const Fts5ExtensionApi *pApi, |
| 133 Fts5Context *pFts, |
| 134 void *pUserData |
| 135 ){ |
| 136 Fts5PhraseIter iter; |
| 137 int iCol, iOff; |
| 138 u32 *aOut = (u32*)pUserData; |
| 139 int iPrev = -1; |
| 140 |
| 141 for(pApi->xPhraseFirst(pFts, 0, &iter, &iCol, &iOff); |
| 142 iCol>=0; |
| 143 pApi->xPhraseNext(pFts, &iter, &iCol, &iOff) |
| 144 ){ |
| 145 aOut[iCol*3+1]++; |
| 146 if( iCol!=iPrev ) aOut[iCol*3 + 2]++; |
| 147 iPrev = iCol; |
| 148 } |
| 149 |
| 150 return SQLITE_OK; |
| 151 } |
| 152 |
| 153 static int fts5MatchinfoGlobalCb( |
| 154 const Fts5ExtensionApi *pApi, |
| 155 Fts5Context *pFts, |
| 156 Fts5MatchinfoCtx *p, |
| 157 char f, |
| 158 u32 *aOut |
| 159 ){ |
| 160 int rc = SQLITE_OK; |
| 161 switch( f ){ |
| 162 case 'p': |
| 163 aOut[0] = p->nPhrase; |
| 164 break; |
| 165 |
| 166 case 'c': |
| 167 aOut[0] = p->nCol; |
| 168 break; |
| 169 |
| 170 case 'x': { |
| 171 int i; |
| 172 for(i=0; i<p->nPhrase && rc==SQLITE_OK; i++){ |
| 173 void *pPtr = (void*)&aOut[i * p->nCol * 3]; |
| 174 rc = pApi->xQueryPhrase(pFts, i, pPtr, fts5MatchinfoXCb); |
| 175 } |
| 176 break; |
| 177 } |
| 178 |
| 179 case 'n': { |
| 180 sqlite3_int64 nRow; |
| 181 rc = pApi->xRowCount(pFts, &nRow); |
| 182 aOut[0] = (u32)nRow; |
| 183 break; |
| 184 } |
| 185 |
| 186 case 'a': { |
| 187 sqlite3_int64 nRow = 0; |
| 188 rc = pApi->xRowCount(pFts, &nRow); |
| 189 if( nRow==0 ){ |
| 190 memset(aOut, 0, sizeof(u32) * p->nCol); |
| 191 }else{ |
| 192 int i; |
| 193 for(i=0; rc==SQLITE_OK && i<p->nCol; i++){ |
| 194 sqlite3_int64 nToken; |
| 195 rc = pApi->xColumnTotalSize(pFts, i, &nToken); |
| 196 if( rc==SQLITE_OK){ |
| 197 aOut[i] = (u32)((2*nToken + nRow) / (2*nRow)); |
| 198 } |
| 199 } |
| 200 } |
| 201 break; |
| 202 } |
| 203 |
| 204 } |
| 205 return rc; |
| 206 } |
| 207 |
| 208 static int fts5MatchinfoLocalCb( |
| 209 const Fts5ExtensionApi *pApi, |
| 210 Fts5Context *pFts, |
| 211 Fts5MatchinfoCtx *p, |
| 212 char f, |
| 213 u32 *aOut |
| 214 ){ |
| 215 int i; |
| 216 int rc = SQLITE_OK; |
| 217 |
| 218 switch( f ){ |
| 219 case 'b': { |
| 220 int iPhrase; |
| 221 int nInt = ((p->nCol + 31) / 32) * p->nPhrase; |
| 222 for(i=0; i<nInt; i++) aOut[i] = 0; |
| 223 |
| 224 for(iPhrase=0; iPhrase<p->nPhrase; iPhrase++){ |
| 225 Fts5PhraseIter iter; |
| 226 int iCol; |
| 227 for(pApi->xPhraseFirstColumn(pFts, iPhrase, &iter, &iCol); |
| 228 iCol>=0; |
| 229 pApi->xPhraseNextColumn(pFts, &iter, &iCol) |
| 230 ){ |
| 231 aOut[iPhrase * ((p->nCol+31)/32) + iCol/32] |= ((u32)1 << iCol%32); |
| 232 } |
| 233 } |
| 234 |
| 235 break; |
| 236 } |
| 237 |
| 238 case 'x': |
| 239 case 'y': { |
| 240 int nMul = (f=='x' ? 3 : 1); |
| 241 int iPhrase; |
| 242 |
| 243 for(i=0; i<(p->nCol*p->nPhrase); i++) aOut[i*nMul] = 0; |
| 244 |
| 245 for(iPhrase=0; iPhrase<p->nPhrase; iPhrase++){ |
| 246 Fts5PhraseIter iter; |
| 247 int iOff, iCol; |
| 248 for(pApi->xPhraseFirst(pFts, iPhrase, &iter, &iCol, &iOff); |
| 249 iOff>=0; |
| 250 pApi->xPhraseNext(pFts, &iter, &iCol, &iOff) |
| 251 ){ |
| 252 aOut[nMul * (iCol + iPhrase * p->nCol)]++; |
| 253 } |
| 254 } |
| 255 |
| 256 break; |
| 257 } |
| 258 |
| 259 case 'l': { |
| 260 for(i=0; rc==SQLITE_OK && i<p->nCol; i++){ |
| 261 int nToken; |
| 262 rc = pApi->xColumnSize(pFts, i, &nToken); |
| 263 aOut[i] = (u32)nToken; |
| 264 } |
| 265 break; |
| 266 } |
| 267 |
| 268 case 's': { |
| 269 int nInst; |
| 270 |
| 271 memset(aOut, 0, sizeof(u32) * p->nCol); |
| 272 |
| 273 rc = pApi->xInstCount(pFts, &nInst); |
| 274 for(i=0; rc==SQLITE_OK && i<nInst; i++){ |
| 275 int iPhrase, iOff, iCol = 0; |
| 276 int iNextPhrase; |
| 277 int iNextOff; |
| 278 u32 nSeq = 1; |
| 279 int j; |
| 280 |
| 281 rc = pApi->xInst(pFts, i, &iPhrase, &iCol, &iOff); |
| 282 iNextPhrase = iPhrase+1; |
| 283 iNextOff = iOff+pApi->xPhraseSize(pFts, 0); |
| 284 for(j=i+1; rc==SQLITE_OK && j<nInst; j++){ |
| 285 int ip, ic, io; |
| 286 rc = pApi->xInst(pFts, j, &ip, &ic, &io); |
| 287 if( ic!=iCol || io>iNextOff ) break; |
| 288 if( ip==iNextPhrase && io==iNextOff ){ |
| 289 nSeq++; |
| 290 iNextPhrase = ip+1; |
| 291 iNextOff = io + pApi->xPhraseSize(pFts, ip); |
| 292 } |
| 293 } |
| 294 |
| 295 if( nSeq>aOut[iCol] ) aOut[iCol] = nSeq; |
| 296 } |
| 297 |
| 298 break; |
| 299 } |
| 300 } |
| 301 return rc; |
| 302 } |
| 303 |
| 304 static Fts5MatchinfoCtx *fts5MatchinfoNew( |
| 305 const Fts5ExtensionApi *pApi, /* API offered by current FTS version */ |
| 306 Fts5Context *pFts, /* First arg to pass to pApi functions */ |
| 307 sqlite3_context *pCtx, /* Context for returning error message */ |
| 308 const char *zArg /* Matchinfo flag string */ |
| 309 ){ |
| 310 Fts5MatchinfoCtx *p; |
| 311 int nCol; |
| 312 int nPhrase; |
| 313 int i; |
| 314 int nInt; |
| 315 int nByte; |
| 316 int rc; |
| 317 |
| 318 nCol = pApi->xColumnCount(pFts); |
| 319 nPhrase = pApi->xPhraseCount(pFts); |
| 320 |
| 321 nInt = 0; |
| 322 for(i=0; zArg[i]; i++){ |
| 323 int n = fts5MatchinfoFlagsize(nCol, nPhrase, zArg[i]); |
| 324 if( n<0 ){ |
| 325 char *zErr = sqlite3_mprintf("unrecognized matchinfo flag: %c", zArg[i]); |
| 326 sqlite3_result_error(pCtx, zErr, -1); |
| 327 sqlite3_free(zErr); |
| 328 return 0; |
| 329 } |
| 330 nInt += n; |
| 331 } |
| 332 |
| 333 nByte = sizeof(Fts5MatchinfoCtx) /* The struct itself */ |
| 334 + sizeof(u32) * nInt /* The p->aRet[] array */ |
| 335 + (i+1); /* The p->zArg string */ |
| 336 p = (Fts5MatchinfoCtx*)sqlite3_malloc(nByte); |
| 337 if( p==0 ){ |
| 338 sqlite3_result_error_nomem(pCtx); |
| 339 return 0; |
| 340 } |
| 341 memset(p, 0, nByte); |
| 342 |
| 343 p->nCol = nCol; |
| 344 p->nPhrase = nPhrase; |
| 345 p->aRet = (u32*)&p[1]; |
| 346 p->nRet = nInt; |
| 347 p->zArg = (char*)&p->aRet[nInt]; |
| 348 memcpy(p->zArg, zArg, i); |
| 349 |
| 350 rc = fts5MatchinfoIter(pApi, pFts, p, fts5MatchinfoGlobalCb); |
| 351 if( rc!=SQLITE_OK ){ |
| 352 sqlite3_result_error_code(pCtx, rc); |
| 353 sqlite3_free(p); |
| 354 p = 0; |
| 355 } |
| 356 |
| 357 return p; |
| 358 } |
| 359 |
| 360 static void fts5MatchinfoFunc( |
| 361 const Fts5ExtensionApi *pApi, /* API offered by current FTS version */ |
| 362 Fts5Context *pFts, /* First arg to pass to pApi functions */ |
| 363 sqlite3_context *pCtx, /* Context for returning result/error */ |
| 364 int nVal, /* Number of values in apVal[] array */ |
| 365 sqlite3_value **apVal /* Array of trailing arguments */ |
| 366 ){ |
| 367 const char *zArg; |
| 368 Fts5MatchinfoCtx *p; |
| 369 int rc = SQLITE_OK; |
| 370 |
| 371 if( nVal>0 ){ |
| 372 zArg = (const char*)sqlite3_value_text(apVal[0]); |
| 373 }else{ |
| 374 zArg = "pcx"; |
| 375 } |
| 376 |
| 377 p = (Fts5MatchinfoCtx*)pApi->xGetAuxdata(pFts, 0); |
| 378 if( p==0 || sqlite3_stricmp(zArg, p->zArg) ){ |
| 379 p = fts5MatchinfoNew(pApi, pFts, pCtx, zArg); |
| 380 if( p==0 ){ |
| 381 rc = SQLITE_NOMEM; |
| 382 }else{ |
| 383 rc = pApi->xSetAuxdata(pFts, p, sqlite3_free); |
| 384 } |
| 385 } |
| 386 |
| 387 if( rc==SQLITE_OK ){ |
| 388 rc = fts5MatchinfoIter(pApi, pFts, p, fts5MatchinfoLocalCb); |
| 389 } |
| 390 if( rc!=SQLITE_OK ){ |
| 391 sqlite3_result_error_code(pCtx, rc); |
| 392 }else{ |
| 393 /* No errors has occured, so return a copy of the array of integers. */ |
| 394 int nByte = p->nRet * sizeof(u32); |
| 395 sqlite3_result_blob(pCtx, (void*)p->aRet, nByte, SQLITE_TRANSIENT); |
| 396 } |
| 397 } |
| 398 |
| 399 int sqlite3Fts5TestRegisterMatchinfo(sqlite3 *db){ |
| 400 int rc; /* Return code */ |
| 401 fts5_api *pApi; /* FTS5 API functions */ |
| 402 |
| 403 /* Extract the FTS5 API pointer from the database handle. The |
| 404 ** fts5_api_from_db() function above is copied verbatim from the |
| 405 ** FTS5 documentation. Refer there for details. */ |
| 406 rc = fts5_api_from_db(db, &pApi); |
| 407 if( rc!=SQLITE_OK ) return rc; |
| 408 |
| 409 /* If fts5_api_from_db() returns NULL, then either FTS5 is not registered |
| 410 ** with this database handle, or an error (OOM perhaps?) has occurred. |
| 411 ** |
| 412 ** Also check that the fts5_api object is version 2 or newer. |
| 413 */ |
| 414 if( pApi==0 || pApi->iVersion<2 ){ |
| 415 return SQLITE_ERROR; |
| 416 } |
| 417 |
| 418 /* Register the implementation of matchinfo() */ |
| 419 rc = pApi->xCreateFunction(pApi, "matchinfo", 0, fts5MatchinfoFunc, 0); |
| 420 |
| 421 return rc; |
| 422 } |
| 423 |
| 424 #endif /* SQLITE_ENABLE_FTS5 */ |
| 425 |
OLD | NEW |