OLD | NEW |
(Empty) | |
| 1 /* |
| 2 ** 2013 Apr 22 |
| 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 code for the "fts3tokenize" virtual table module. |
| 14 ** An fts3tokenize virtual table is created as follows: |
| 15 ** |
| 16 ** CREATE VIRTUAL TABLE <tbl> USING fts3tokenize( |
| 17 ** <tokenizer-name>, <arg-1>, ... |
| 18 ** ); |
| 19 ** |
| 20 ** The table created has the following schema: |
| 21 ** |
| 22 ** CREATE TABLE <tbl>(input, token, start, end, position) |
| 23 ** |
| 24 ** When queried, the query must include a WHERE clause of type: |
| 25 ** |
| 26 ** input = <string> |
| 27 ** |
| 28 ** The virtual table module tokenizes this <string>, using the FTS3 |
| 29 ** tokenizer specified by the arguments to the CREATE VIRTUAL TABLE |
| 30 ** statement and returns one row for each token in the result. With |
| 31 ** fields set as follows: |
| 32 ** |
| 33 ** input: Always set to a copy of <string> |
| 34 ** token: A token from the input. |
| 35 ** start: Byte offset of the token within the input <string>. |
| 36 ** end: Byte offset of the byte immediately following the end of the |
| 37 ** token within the input string. |
| 38 ** pos: Token offset of token within input. |
| 39 ** |
| 40 */ |
| 41 #include "fts3Int.h" |
| 42 #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) |
| 43 |
| 44 #include <string.h> |
| 45 #include <assert.h> |
| 46 |
| 47 typedef struct Fts3tokTable Fts3tokTable; |
| 48 typedef struct Fts3tokCursor Fts3tokCursor; |
| 49 |
| 50 /* |
| 51 ** Virtual table structure. |
| 52 */ |
| 53 struct Fts3tokTable { |
| 54 sqlite3_vtab base; /* Base class used by SQLite core */ |
| 55 const sqlite3_tokenizer_module *pMod; |
| 56 sqlite3_tokenizer *pTok; |
| 57 }; |
| 58 |
| 59 /* |
| 60 ** Virtual table cursor structure. |
| 61 */ |
| 62 struct Fts3tokCursor { |
| 63 sqlite3_vtab_cursor base; /* Base class used by SQLite core */ |
| 64 char *zInput; /* Input string */ |
| 65 sqlite3_tokenizer_cursor *pCsr; /* Cursor to iterate through zInput */ |
| 66 int iRowid; /* Current 'rowid' value */ |
| 67 const char *zToken; /* Current 'token' value */ |
| 68 int nToken; /* Size of zToken in bytes */ |
| 69 int iStart; /* Current 'start' value */ |
| 70 int iEnd; /* Current 'end' value */ |
| 71 int iPos; /* Current 'pos' value */ |
| 72 }; |
| 73 |
| 74 /* |
| 75 ** Query FTS for the tokenizer implementation named zName. |
| 76 */ |
| 77 static int fts3tokQueryTokenizer( |
| 78 Fts3Hash *pHash, |
| 79 const char *zName, |
| 80 const sqlite3_tokenizer_module **pp, |
| 81 char **pzErr |
| 82 ){ |
| 83 sqlite3_tokenizer_module *p; |
| 84 int nName = (int)strlen(zName); |
| 85 |
| 86 p = (sqlite3_tokenizer_module *)sqlite3Fts3HashFind(pHash, zName, nName+1); |
| 87 if( !p ){ |
| 88 *pzErr = sqlite3_mprintf("unknown tokenizer: %s", zName); |
| 89 return SQLITE_ERROR; |
| 90 } |
| 91 |
| 92 *pp = p; |
| 93 return SQLITE_OK; |
| 94 } |
| 95 |
| 96 /* |
| 97 ** The second argument, argv[], is an array of pointers to nul-terminated |
| 98 ** strings. This function makes a copy of the array and strings into a |
| 99 ** single block of memory. It then dequotes any of the strings that appear |
| 100 ** to be quoted. |
| 101 ** |
| 102 ** If successful, output parameter *pazDequote is set to point at the |
| 103 ** array of dequoted strings and SQLITE_OK is returned. The caller is |
| 104 ** responsible for eventually calling sqlite3_free() to free the array |
| 105 ** in this case. Or, if an error occurs, an SQLite error code is returned. |
| 106 ** The final value of *pazDequote is undefined in this case. |
| 107 */ |
| 108 static int fts3tokDequoteArray( |
| 109 int argc, /* Number of elements in argv[] */ |
| 110 const char * const *argv, /* Input array */ |
| 111 char ***pazDequote /* Output array */ |
| 112 ){ |
| 113 int rc = SQLITE_OK; /* Return code */ |
| 114 if( argc==0 ){ |
| 115 *pazDequote = 0; |
| 116 }else{ |
| 117 int i; |
| 118 int nByte = 0; |
| 119 char **azDequote; |
| 120 |
| 121 for(i=0; i<argc; i++){ |
| 122 nByte += (int)(strlen(argv[i]) + 1); |
| 123 } |
| 124 |
| 125 *pazDequote = azDequote = sqlite3_malloc(sizeof(char *)*argc + nByte); |
| 126 if( azDequote==0 ){ |
| 127 rc = SQLITE_NOMEM; |
| 128 }else{ |
| 129 char *pSpace = (char *)&azDequote[argc]; |
| 130 for(i=0; i<argc; i++){ |
| 131 int n = (int)strlen(argv[i]); |
| 132 azDequote[i] = pSpace; |
| 133 memcpy(pSpace, argv[i], n+1); |
| 134 sqlite3Fts3Dequote(pSpace); |
| 135 pSpace += (n+1); |
| 136 } |
| 137 } |
| 138 } |
| 139 |
| 140 return rc; |
| 141 } |
| 142 |
| 143 /* |
| 144 ** Schema of the tokenizer table. |
| 145 */ |
| 146 #define FTS3_TOK_SCHEMA "CREATE TABLE x(input, token, start, end, position)" |
| 147 |
| 148 /* |
| 149 ** This function does all the work for both the xConnect and xCreate methods. |
| 150 ** These tables have no persistent representation of their own, so xConnect |
| 151 ** and xCreate are identical operations. |
| 152 ** |
| 153 ** argv[0]: module name |
| 154 ** argv[1]: database name |
| 155 ** argv[2]: table name |
| 156 ** argv[3]: first argument (tokenizer name) |
| 157 */ |
| 158 static int fts3tokConnectMethod( |
| 159 sqlite3 *db, /* Database connection */ |
| 160 void *pHash, /* Hash table of tokenizers */ |
| 161 int argc, /* Number of elements in argv array */ |
| 162 const char * const *argv, /* xCreate/xConnect argument array */ |
| 163 sqlite3_vtab **ppVtab, /* OUT: New sqlite3_vtab object */ |
| 164 char **pzErr /* OUT: sqlite3_malloc'd error message */ |
| 165 ){ |
| 166 Fts3tokTable *pTab; |
| 167 const sqlite3_tokenizer_module *pMod = 0; |
| 168 sqlite3_tokenizer *pTok = 0; |
| 169 int rc; |
| 170 char **azDequote = 0; |
| 171 int nDequote; |
| 172 |
| 173 rc = sqlite3_declare_vtab(db, FTS3_TOK_SCHEMA); |
| 174 if( rc!=SQLITE_OK ) return rc; |
| 175 |
| 176 nDequote = argc-3; |
| 177 rc = fts3tokDequoteArray(nDequote, &argv[3], &azDequote); |
| 178 |
| 179 if( rc==SQLITE_OK ){ |
| 180 const char *zModule; |
| 181 if( nDequote<1 ){ |
| 182 zModule = "simple"; |
| 183 }else{ |
| 184 zModule = azDequote[0]; |
| 185 } |
| 186 rc = fts3tokQueryTokenizer((Fts3Hash*)pHash, zModule, &pMod, pzErr); |
| 187 } |
| 188 |
| 189 assert( (rc==SQLITE_OK)==(pMod!=0) ); |
| 190 if( rc==SQLITE_OK ){ |
| 191 const char * const *azArg = (const char * const *)&azDequote[1]; |
| 192 rc = pMod->xCreate((nDequote>1 ? nDequote-1 : 0), azArg, &pTok); |
| 193 } |
| 194 |
| 195 if( rc==SQLITE_OK ){ |
| 196 pTab = (Fts3tokTable *)sqlite3_malloc(sizeof(Fts3tokTable)); |
| 197 if( pTab==0 ){ |
| 198 rc = SQLITE_NOMEM; |
| 199 } |
| 200 } |
| 201 |
| 202 if( rc==SQLITE_OK ){ |
| 203 memset(pTab, 0, sizeof(Fts3tokTable)); |
| 204 pTab->pMod = pMod; |
| 205 pTab->pTok = pTok; |
| 206 *ppVtab = &pTab->base; |
| 207 }else{ |
| 208 if( pTok ){ |
| 209 pMod->xDestroy(pTok); |
| 210 } |
| 211 } |
| 212 |
| 213 sqlite3_free(azDequote); |
| 214 return rc; |
| 215 } |
| 216 |
| 217 /* |
| 218 ** This function does the work for both the xDisconnect and xDestroy methods. |
| 219 ** These tables have no persistent representation of their own, so xDisconnect |
| 220 ** and xDestroy are identical operations. |
| 221 */ |
| 222 static int fts3tokDisconnectMethod(sqlite3_vtab *pVtab){ |
| 223 Fts3tokTable *pTab = (Fts3tokTable *)pVtab; |
| 224 |
| 225 pTab->pMod->xDestroy(pTab->pTok); |
| 226 sqlite3_free(pTab); |
| 227 return SQLITE_OK; |
| 228 } |
| 229 |
| 230 /* |
| 231 ** xBestIndex - Analyze a WHERE and ORDER BY clause. |
| 232 */ |
| 233 static int fts3tokBestIndexMethod( |
| 234 sqlite3_vtab *pVTab, |
| 235 sqlite3_index_info *pInfo |
| 236 ){ |
| 237 int i; |
| 238 UNUSED_PARAMETER(pVTab); |
| 239 |
| 240 for(i=0; i<pInfo->nConstraint; i++){ |
| 241 if( pInfo->aConstraint[i].usable |
| 242 && pInfo->aConstraint[i].iColumn==0 |
| 243 && pInfo->aConstraint[i].op==SQLITE_INDEX_CONSTRAINT_EQ |
| 244 ){ |
| 245 pInfo->idxNum = 1; |
| 246 pInfo->aConstraintUsage[i].argvIndex = 1; |
| 247 pInfo->aConstraintUsage[i].omit = 1; |
| 248 pInfo->estimatedCost = 1; |
| 249 return SQLITE_OK; |
| 250 } |
| 251 } |
| 252 |
| 253 pInfo->idxNum = 0; |
| 254 assert( pInfo->estimatedCost>1000000.0 ); |
| 255 |
| 256 return SQLITE_OK; |
| 257 } |
| 258 |
| 259 /* |
| 260 ** xOpen - Open a cursor. |
| 261 */ |
| 262 static int fts3tokOpenMethod(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCsr){ |
| 263 Fts3tokCursor *pCsr; |
| 264 UNUSED_PARAMETER(pVTab); |
| 265 |
| 266 pCsr = (Fts3tokCursor *)sqlite3_malloc(sizeof(Fts3tokCursor)); |
| 267 if( pCsr==0 ){ |
| 268 return SQLITE_NOMEM; |
| 269 } |
| 270 memset(pCsr, 0, sizeof(Fts3tokCursor)); |
| 271 |
| 272 *ppCsr = (sqlite3_vtab_cursor *)pCsr; |
| 273 return SQLITE_OK; |
| 274 } |
| 275 |
| 276 /* |
| 277 ** Reset the tokenizer cursor passed as the only argument. As if it had |
| 278 ** just been returned by fts3tokOpenMethod(). |
| 279 */ |
| 280 static void fts3tokResetCursor(Fts3tokCursor *pCsr){ |
| 281 if( pCsr->pCsr ){ |
| 282 Fts3tokTable *pTab = (Fts3tokTable *)(pCsr->base.pVtab); |
| 283 pTab->pMod->xClose(pCsr->pCsr); |
| 284 pCsr->pCsr = 0; |
| 285 } |
| 286 sqlite3_free(pCsr->zInput); |
| 287 pCsr->zInput = 0; |
| 288 pCsr->zToken = 0; |
| 289 pCsr->nToken = 0; |
| 290 pCsr->iStart = 0; |
| 291 pCsr->iEnd = 0; |
| 292 pCsr->iPos = 0; |
| 293 pCsr->iRowid = 0; |
| 294 } |
| 295 |
| 296 /* |
| 297 ** xClose - Close a cursor. |
| 298 */ |
| 299 static int fts3tokCloseMethod(sqlite3_vtab_cursor *pCursor){ |
| 300 Fts3tokCursor *pCsr = (Fts3tokCursor *)pCursor; |
| 301 |
| 302 fts3tokResetCursor(pCsr); |
| 303 sqlite3_free(pCsr); |
| 304 return SQLITE_OK; |
| 305 } |
| 306 |
| 307 /* |
| 308 ** xNext - Advance the cursor to the next row, if any. |
| 309 */ |
| 310 static int fts3tokNextMethod(sqlite3_vtab_cursor *pCursor){ |
| 311 Fts3tokCursor *pCsr = (Fts3tokCursor *)pCursor; |
| 312 Fts3tokTable *pTab = (Fts3tokTable *)(pCursor->pVtab); |
| 313 int rc; /* Return code */ |
| 314 |
| 315 pCsr->iRowid++; |
| 316 rc = pTab->pMod->xNext(pCsr->pCsr, |
| 317 &pCsr->zToken, &pCsr->nToken, |
| 318 &pCsr->iStart, &pCsr->iEnd, &pCsr->iPos |
| 319 ); |
| 320 |
| 321 if( rc!=SQLITE_OK ){ |
| 322 fts3tokResetCursor(pCsr); |
| 323 if( rc==SQLITE_DONE ) rc = SQLITE_OK; |
| 324 } |
| 325 |
| 326 return rc; |
| 327 } |
| 328 |
| 329 /* |
| 330 ** xFilter - Initialize a cursor to point at the start of its data. |
| 331 */ |
| 332 static int fts3tokFilterMethod( |
| 333 sqlite3_vtab_cursor *pCursor, /* The cursor used for this query */ |
| 334 int idxNum, /* Strategy index */ |
| 335 const char *idxStr, /* Unused */ |
| 336 int nVal, /* Number of elements in apVal */ |
| 337 sqlite3_value **apVal /* Arguments for the indexing scheme */ |
| 338 ){ |
| 339 int rc = SQLITE_ERROR; |
| 340 Fts3tokCursor *pCsr = (Fts3tokCursor *)pCursor; |
| 341 Fts3tokTable *pTab = (Fts3tokTable *)(pCursor->pVtab); |
| 342 UNUSED_PARAMETER(idxStr); |
| 343 UNUSED_PARAMETER(nVal); |
| 344 |
| 345 fts3tokResetCursor(pCsr); |
| 346 if( idxNum==1 ){ |
| 347 const char *zByte = (const char *)sqlite3_value_text(apVal[0]); |
| 348 int nByte = sqlite3_value_bytes(apVal[0]); |
| 349 pCsr->zInput = sqlite3_malloc(nByte+1); |
| 350 if( pCsr->zInput==0 ){ |
| 351 rc = SQLITE_NOMEM; |
| 352 }else{ |
| 353 memcpy(pCsr->zInput, zByte, nByte); |
| 354 pCsr->zInput[nByte] = 0; |
| 355 rc = pTab->pMod->xOpen(pTab->pTok, pCsr->zInput, nByte, &pCsr->pCsr); |
| 356 if( rc==SQLITE_OK ){ |
| 357 pCsr->pCsr->pTokenizer = pTab->pTok; |
| 358 } |
| 359 } |
| 360 } |
| 361 |
| 362 if( rc!=SQLITE_OK ) return rc; |
| 363 return fts3tokNextMethod(pCursor); |
| 364 } |
| 365 |
| 366 /* |
| 367 ** xEof - Return true if the cursor is at EOF, or false otherwise. |
| 368 */ |
| 369 static int fts3tokEofMethod(sqlite3_vtab_cursor *pCursor){ |
| 370 Fts3tokCursor *pCsr = (Fts3tokCursor *)pCursor; |
| 371 return (pCsr->zToken==0); |
| 372 } |
| 373 |
| 374 /* |
| 375 ** xColumn - Return a column value. |
| 376 */ |
| 377 static int fts3tokColumnMethod( |
| 378 sqlite3_vtab_cursor *pCursor, /* Cursor to retrieve value from */ |
| 379 sqlite3_context *pCtx, /* Context for sqlite3_result_xxx() calls */ |
| 380 int iCol /* Index of column to read value from */ |
| 381 ){ |
| 382 Fts3tokCursor *pCsr = (Fts3tokCursor *)pCursor; |
| 383 |
| 384 /* CREATE TABLE x(input, token, start, end, position) */ |
| 385 switch( iCol ){ |
| 386 case 0: |
| 387 sqlite3_result_text(pCtx, pCsr->zInput, -1, SQLITE_TRANSIENT); |
| 388 break; |
| 389 case 1: |
| 390 sqlite3_result_text(pCtx, pCsr->zToken, pCsr->nToken, SQLITE_TRANSIENT); |
| 391 break; |
| 392 case 2: |
| 393 sqlite3_result_int(pCtx, pCsr->iStart); |
| 394 break; |
| 395 case 3: |
| 396 sqlite3_result_int(pCtx, pCsr->iEnd); |
| 397 break; |
| 398 default: |
| 399 assert( iCol==4 ); |
| 400 sqlite3_result_int(pCtx, pCsr->iPos); |
| 401 break; |
| 402 } |
| 403 return SQLITE_OK; |
| 404 } |
| 405 |
| 406 /* |
| 407 ** xRowid - Return the current rowid for the cursor. |
| 408 */ |
| 409 static int fts3tokRowidMethod( |
| 410 sqlite3_vtab_cursor *pCursor, /* Cursor to retrieve value from */ |
| 411 sqlite_int64 *pRowid /* OUT: Rowid value */ |
| 412 ){ |
| 413 Fts3tokCursor *pCsr = (Fts3tokCursor *)pCursor; |
| 414 *pRowid = (sqlite3_int64)pCsr->iRowid; |
| 415 return SQLITE_OK; |
| 416 } |
| 417 |
| 418 /* |
| 419 ** Register the fts3tok module with database connection db. Return SQLITE_OK |
| 420 ** if successful or an error code if sqlite3_create_module() fails. |
| 421 */ |
| 422 int sqlite3Fts3InitTok(sqlite3 *db, Fts3Hash *pHash){ |
| 423 static const sqlite3_module fts3tok_module = { |
| 424 0, /* iVersion */ |
| 425 fts3tokConnectMethod, /* xCreate */ |
| 426 fts3tokConnectMethod, /* xConnect */ |
| 427 fts3tokBestIndexMethod, /* xBestIndex */ |
| 428 fts3tokDisconnectMethod, /* xDisconnect */ |
| 429 fts3tokDisconnectMethod, /* xDestroy */ |
| 430 fts3tokOpenMethod, /* xOpen */ |
| 431 fts3tokCloseMethod, /* xClose */ |
| 432 fts3tokFilterMethod, /* xFilter */ |
| 433 fts3tokNextMethod, /* xNext */ |
| 434 fts3tokEofMethod, /* xEof */ |
| 435 fts3tokColumnMethod, /* xColumn */ |
| 436 fts3tokRowidMethod, /* xRowid */ |
| 437 0, /* xUpdate */ |
| 438 0, /* xBegin */ |
| 439 0, /* xSync */ |
| 440 0, /* xCommit */ |
| 441 0, /* xRollback */ |
| 442 0, /* xFindFunction */ |
| 443 0, /* xRename */ |
| 444 0, /* xSavepoint */ |
| 445 0, /* xRelease */ |
| 446 0 /* xRollbackTo */ |
| 447 }; |
| 448 int rc; /* Return code */ |
| 449 |
| 450 rc = sqlite3_create_module(db, "fts3tokenize", &fts3tok_module, (void*)pHash); |
| 451 return rc; |
| 452 } |
| 453 |
| 454 #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
OLD | NEW |