Annotation of embedaddon/sqlite3/ext/fts3/fts3_tokenizer.c, revision 1.1
1.1 ! misho 1: /*
! 2: ** 2007 June 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 is part of an SQLite module implementing full-text search.
! 14: ** This particular file implements the generic tokenizer interface.
! 15: */
! 16:
! 17: /*
! 18: ** The code in this file is only compiled if:
! 19: **
! 20: ** * The FTS3 module is being built as an extension
! 21: ** (in which case SQLITE_CORE is not defined), or
! 22: **
! 23: ** * The FTS3 module is being built into the core of
! 24: ** SQLite (in which case SQLITE_ENABLE_FTS3 is defined).
! 25: */
! 26: #include "fts3Int.h"
! 27: #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3)
! 28:
! 29: #include <assert.h>
! 30: #include <string.h>
! 31:
! 32: /*
! 33: ** Implementation of the SQL scalar function for accessing the underlying
! 34: ** hash table. This function may be called as follows:
! 35: **
! 36: ** SELECT <function-name>(<key-name>);
! 37: ** SELECT <function-name>(<key-name>, <pointer>);
! 38: **
! 39: ** where <function-name> is the name passed as the second argument
! 40: ** to the sqlite3Fts3InitHashTable() function (e.g. 'fts3_tokenizer').
! 41: **
! 42: ** If the <pointer> argument is specified, it must be a blob value
! 43: ** containing a pointer to be stored as the hash data corresponding
! 44: ** to the string <key-name>. If <pointer> is not specified, then
! 45: ** the string <key-name> must already exist in the has table. Otherwise,
! 46: ** an error is returned.
! 47: **
! 48: ** Whether or not the <pointer> argument is specified, the value returned
! 49: ** is a blob containing the pointer stored as the hash data corresponding
! 50: ** to string <key-name> (after the hash-table is updated, if applicable).
! 51: */
! 52: static void scalarFunc(
! 53: sqlite3_context *context,
! 54: int argc,
! 55: sqlite3_value **argv
! 56: ){
! 57: Fts3Hash *pHash;
! 58: void *pPtr = 0;
! 59: const unsigned char *zName;
! 60: int nName;
! 61:
! 62: assert( argc==1 || argc==2 );
! 63:
! 64: pHash = (Fts3Hash *)sqlite3_user_data(context);
! 65:
! 66: zName = sqlite3_value_text(argv[0]);
! 67: nName = sqlite3_value_bytes(argv[0])+1;
! 68:
! 69: if( argc==2 ){
! 70: void *pOld;
! 71: int n = sqlite3_value_bytes(argv[1]);
! 72: if( n!=sizeof(pPtr) ){
! 73: sqlite3_result_error(context, "argument type mismatch", -1);
! 74: return;
! 75: }
! 76: pPtr = *(void **)sqlite3_value_blob(argv[1]);
! 77: pOld = sqlite3Fts3HashInsert(pHash, (void *)zName, nName, pPtr);
! 78: if( pOld==pPtr ){
! 79: sqlite3_result_error(context, "out of memory", -1);
! 80: return;
! 81: }
! 82: }else{
! 83: pPtr = sqlite3Fts3HashFind(pHash, zName, nName);
! 84: if( !pPtr ){
! 85: char *zErr = sqlite3_mprintf("unknown tokenizer: %s", zName);
! 86: sqlite3_result_error(context, zErr, -1);
! 87: sqlite3_free(zErr);
! 88: return;
! 89: }
! 90: }
! 91:
! 92: sqlite3_result_blob(context, (void *)&pPtr, sizeof(pPtr), SQLITE_TRANSIENT);
! 93: }
! 94:
! 95: int sqlite3Fts3IsIdChar(char c){
! 96: static const char isFtsIdChar[] = {
! 97: 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 0x */
! 98: 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 1x */
! 99: 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 2x */
! 100: 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, /* 3x */
! 101: 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 4x */
! 102: 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, /* 5x */
! 103: 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 6x */
! 104: 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 7x */
! 105: };
! 106: return (c&0x80 || isFtsIdChar[(int)(c)]);
! 107: }
! 108:
! 109: const char *sqlite3Fts3NextToken(const char *zStr, int *pn){
! 110: const char *z1;
! 111: const char *z2 = 0;
! 112:
! 113: /* Find the start of the next token. */
! 114: z1 = zStr;
! 115: while( z2==0 ){
! 116: char c = *z1;
! 117: switch( c ){
! 118: case '\0': return 0; /* No more tokens here */
! 119: case '\'':
! 120: case '"':
! 121: case '`': {
! 122: z2 = z1;
! 123: while( *++z2 && (*z2!=c || *++z2==c) );
! 124: break;
! 125: }
! 126: case '[':
! 127: z2 = &z1[1];
! 128: while( *z2 && z2[0]!=']' ) z2++;
! 129: if( *z2 ) z2++;
! 130: break;
! 131:
! 132: default:
! 133: if( sqlite3Fts3IsIdChar(*z1) ){
! 134: z2 = &z1[1];
! 135: while( sqlite3Fts3IsIdChar(*z2) ) z2++;
! 136: }else{
! 137: z1++;
! 138: }
! 139: }
! 140: }
! 141:
! 142: *pn = (int)(z2-z1);
! 143: return z1;
! 144: }
! 145:
! 146: int sqlite3Fts3InitTokenizer(
! 147: Fts3Hash *pHash, /* Tokenizer hash table */
! 148: const char *zArg, /* Tokenizer name */
! 149: sqlite3_tokenizer **ppTok, /* OUT: Tokenizer (if applicable) */
! 150: char **pzErr /* OUT: Set to malloced error message */
! 151: ){
! 152: int rc;
! 153: char *z = (char *)zArg;
! 154: int n = 0;
! 155: char *zCopy;
! 156: char *zEnd; /* Pointer to nul-term of zCopy */
! 157: sqlite3_tokenizer_module *m;
! 158:
! 159: zCopy = sqlite3_mprintf("%s", zArg);
! 160: if( !zCopy ) return SQLITE_NOMEM;
! 161: zEnd = &zCopy[strlen(zCopy)];
! 162:
! 163: z = (char *)sqlite3Fts3NextToken(zCopy, &n);
! 164: z[n] = '\0';
! 165: sqlite3Fts3Dequote(z);
! 166:
! 167: m = (sqlite3_tokenizer_module *)sqlite3Fts3HashFind(pHash,z,(int)strlen(z)+1);
! 168: if( !m ){
! 169: *pzErr = sqlite3_mprintf("unknown tokenizer: %s", z);
! 170: rc = SQLITE_ERROR;
! 171: }else{
! 172: char const **aArg = 0;
! 173: int iArg = 0;
! 174: z = &z[n+1];
! 175: while( z<zEnd && (NULL!=(z = (char *)sqlite3Fts3NextToken(z, &n))) ){
! 176: int nNew = sizeof(char *)*(iArg+1);
! 177: char const **aNew = (const char **)sqlite3_realloc((void *)aArg, nNew);
! 178: if( !aNew ){
! 179: sqlite3_free(zCopy);
! 180: sqlite3_free((void *)aArg);
! 181: return SQLITE_NOMEM;
! 182: }
! 183: aArg = aNew;
! 184: aArg[iArg++] = z;
! 185: z[n] = '\0';
! 186: sqlite3Fts3Dequote(z);
! 187: z = &z[n+1];
! 188: }
! 189: rc = m->xCreate(iArg, aArg, ppTok);
! 190: assert( rc!=SQLITE_OK || *ppTok );
! 191: if( rc!=SQLITE_OK ){
! 192: *pzErr = sqlite3_mprintf("unknown tokenizer");
! 193: }else{
! 194: (*ppTok)->pModule = m;
! 195: }
! 196: sqlite3_free((void *)aArg);
! 197: }
! 198:
! 199: sqlite3_free(zCopy);
! 200: return rc;
! 201: }
! 202:
! 203:
! 204: #ifdef SQLITE_TEST
! 205:
! 206: #include <tcl.h>
! 207: #include <string.h>
! 208:
! 209: /*
! 210: ** Implementation of a special SQL scalar function for testing tokenizers
! 211: ** designed to be used in concert with the Tcl testing framework. This
! 212: ** function must be called with two arguments:
! 213: **
! 214: ** SELECT <function-name>(<key-name>, <input-string>);
! 215: ** SELECT <function-name>(<key-name>, <pointer>);
! 216: **
! 217: ** where <function-name> is the name passed as the second argument
! 218: ** to the sqlite3Fts3InitHashTable() function (e.g. 'fts3_tokenizer')
! 219: ** concatenated with the string '_test' (e.g. 'fts3_tokenizer_test').
! 220: **
! 221: ** The return value is a string that may be interpreted as a Tcl
! 222: ** list. For each token in the <input-string>, three elements are
! 223: ** added to the returned list. The first is the token position, the
! 224: ** second is the token text (folded, stemmed, etc.) and the third is the
! 225: ** substring of <input-string> associated with the token. For example,
! 226: ** using the built-in "simple" tokenizer:
! 227: **
! 228: ** SELECT fts_tokenizer_test('simple', 'I don't see how');
! 229: **
! 230: ** will return the string:
! 231: **
! 232: ** "{0 i I 1 dont don't 2 see see 3 how how}"
! 233: **
! 234: */
! 235: static void testFunc(
! 236: sqlite3_context *context,
! 237: int argc,
! 238: sqlite3_value **argv
! 239: ){
! 240: Fts3Hash *pHash;
! 241: sqlite3_tokenizer_module *p;
! 242: sqlite3_tokenizer *pTokenizer = 0;
! 243: sqlite3_tokenizer_cursor *pCsr = 0;
! 244:
! 245: const char *zErr = 0;
! 246:
! 247: const char *zName;
! 248: int nName;
! 249: const char *zInput;
! 250: int nInput;
! 251:
! 252: const char *zArg = 0;
! 253:
! 254: const char *zToken;
! 255: int nToken;
! 256: int iStart;
! 257: int iEnd;
! 258: int iPos;
! 259:
! 260: Tcl_Obj *pRet;
! 261:
! 262: assert( argc==2 || argc==3 );
! 263:
! 264: nName = sqlite3_value_bytes(argv[0]);
! 265: zName = (const char *)sqlite3_value_text(argv[0]);
! 266: nInput = sqlite3_value_bytes(argv[argc-1]);
! 267: zInput = (const char *)sqlite3_value_text(argv[argc-1]);
! 268:
! 269: if( argc==3 ){
! 270: zArg = (const char *)sqlite3_value_text(argv[1]);
! 271: }
! 272:
! 273: pHash = (Fts3Hash *)sqlite3_user_data(context);
! 274: p = (sqlite3_tokenizer_module *)sqlite3Fts3HashFind(pHash, zName, nName+1);
! 275:
! 276: if( !p ){
! 277: char *zErr = sqlite3_mprintf("unknown tokenizer: %s", zName);
! 278: sqlite3_result_error(context, zErr, -1);
! 279: sqlite3_free(zErr);
! 280: return;
! 281: }
! 282:
! 283: pRet = Tcl_NewObj();
! 284: Tcl_IncrRefCount(pRet);
! 285:
! 286: if( SQLITE_OK!=p->xCreate(zArg ? 1 : 0, &zArg, &pTokenizer) ){
! 287: zErr = "error in xCreate()";
! 288: goto finish;
! 289: }
! 290: pTokenizer->pModule = p;
! 291: if( SQLITE_OK!=p->xOpen(pTokenizer, zInput, nInput, &pCsr) ){
! 292: zErr = "error in xOpen()";
! 293: goto finish;
! 294: }
! 295: pCsr->pTokenizer = pTokenizer;
! 296:
! 297: while( SQLITE_OK==p->xNext(pCsr, &zToken, &nToken, &iStart, &iEnd, &iPos) ){
! 298: Tcl_ListObjAppendElement(0, pRet, Tcl_NewIntObj(iPos));
! 299: Tcl_ListObjAppendElement(0, pRet, Tcl_NewStringObj(zToken, nToken));
! 300: zToken = &zInput[iStart];
! 301: nToken = iEnd-iStart;
! 302: Tcl_ListObjAppendElement(0, pRet, Tcl_NewStringObj(zToken, nToken));
! 303: }
! 304:
! 305: if( SQLITE_OK!=p->xClose(pCsr) ){
! 306: zErr = "error in xClose()";
! 307: goto finish;
! 308: }
! 309: if( SQLITE_OK!=p->xDestroy(pTokenizer) ){
! 310: zErr = "error in xDestroy()";
! 311: goto finish;
! 312: }
! 313:
! 314: finish:
! 315: if( zErr ){
! 316: sqlite3_result_error(context, zErr, -1);
! 317: }else{
! 318: sqlite3_result_text(context, Tcl_GetString(pRet), -1, SQLITE_TRANSIENT);
! 319: }
! 320: Tcl_DecrRefCount(pRet);
! 321: }
! 322:
! 323: static
! 324: int registerTokenizer(
! 325: sqlite3 *db,
! 326: char *zName,
! 327: const sqlite3_tokenizer_module *p
! 328: ){
! 329: int rc;
! 330: sqlite3_stmt *pStmt;
! 331: const char zSql[] = "SELECT fts3_tokenizer(?, ?)";
! 332:
! 333: rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0);
! 334: if( rc!=SQLITE_OK ){
! 335: return rc;
! 336: }
! 337:
! 338: sqlite3_bind_text(pStmt, 1, zName, -1, SQLITE_STATIC);
! 339: sqlite3_bind_blob(pStmt, 2, &p, sizeof(p), SQLITE_STATIC);
! 340: sqlite3_step(pStmt);
! 341:
! 342: return sqlite3_finalize(pStmt);
! 343: }
! 344:
! 345: static
! 346: int queryTokenizer(
! 347: sqlite3 *db,
! 348: char *zName,
! 349: const sqlite3_tokenizer_module **pp
! 350: ){
! 351: int rc;
! 352: sqlite3_stmt *pStmt;
! 353: const char zSql[] = "SELECT fts3_tokenizer(?)";
! 354:
! 355: *pp = 0;
! 356: rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0);
! 357: if( rc!=SQLITE_OK ){
! 358: return rc;
! 359: }
! 360:
! 361: sqlite3_bind_text(pStmt, 1, zName, -1, SQLITE_STATIC);
! 362: if( SQLITE_ROW==sqlite3_step(pStmt) ){
! 363: if( sqlite3_column_type(pStmt, 0)==SQLITE_BLOB ){
! 364: memcpy((void *)pp, sqlite3_column_blob(pStmt, 0), sizeof(*pp));
! 365: }
! 366: }
! 367:
! 368: return sqlite3_finalize(pStmt);
! 369: }
! 370:
! 371: void sqlite3Fts3SimpleTokenizerModule(sqlite3_tokenizer_module const**ppModule);
! 372:
! 373: /*
! 374: ** Implementation of the scalar function fts3_tokenizer_internal_test().
! 375: ** This function is used for testing only, it is not included in the
! 376: ** build unless SQLITE_TEST is defined.
! 377: **
! 378: ** The purpose of this is to test that the fts3_tokenizer() function
! 379: ** can be used as designed by the C-code in the queryTokenizer and
! 380: ** registerTokenizer() functions above. These two functions are repeated
! 381: ** in the README.tokenizer file as an example, so it is important to
! 382: ** test them.
! 383: **
! 384: ** To run the tests, evaluate the fts3_tokenizer_internal_test() scalar
! 385: ** function with no arguments. An assert() will fail if a problem is
! 386: ** detected. i.e.:
! 387: **
! 388: ** SELECT fts3_tokenizer_internal_test();
! 389: **
! 390: */
! 391: static void intTestFunc(
! 392: sqlite3_context *context,
! 393: int argc,
! 394: sqlite3_value **argv
! 395: ){
! 396: int rc;
! 397: const sqlite3_tokenizer_module *p1;
! 398: const sqlite3_tokenizer_module *p2;
! 399: sqlite3 *db = (sqlite3 *)sqlite3_user_data(context);
! 400:
! 401: UNUSED_PARAMETER(argc);
! 402: UNUSED_PARAMETER(argv);
! 403:
! 404: /* Test the query function */
! 405: sqlite3Fts3SimpleTokenizerModule(&p1);
! 406: rc = queryTokenizer(db, "simple", &p2);
! 407: assert( rc==SQLITE_OK );
! 408: assert( p1==p2 );
! 409: rc = queryTokenizer(db, "nosuchtokenizer", &p2);
! 410: assert( rc==SQLITE_ERROR );
! 411: assert( p2==0 );
! 412: assert( 0==strcmp(sqlite3_errmsg(db), "unknown tokenizer: nosuchtokenizer") );
! 413:
! 414: /* Test the storage function */
! 415: rc = registerTokenizer(db, "nosuchtokenizer", p1);
! 416: assert( rc==SQLITE_OK );
! 417: rc = queryTokenizer(db, "nosuchtokenizer", &p2);
! 418: assert( rc==SQLITE_OK );
! 419: assert( p2==p1 );
! 420:
! 421: sqlite3_result_text(context, "ok", -1, SQLITE_STATIC);
! 422: }
! 423:
! 424: #endif
! 425:
! 426: /*
! 427: ** Set up SQL objects in database db used to access the contents of
! 428: ** the hash table pointed to by argument pHash. The hash table must
! 429: ** been initialised to use string keys, and to take a private copy
! 430: ** of the key when a value is inserted. i.e. by a call similar to:
! 431: **
! 432: ** sqlite3Fts3HashInit(pHash, FTS3_HASH_STRING, 1);
! 433: **
! 434: ** This function adds a scalar function (see header comment above
! 435: ** scalarFunc() in this file for details) and, if ENABLE_TABLE is
! 436: ** defined at compilation time, a temporary virtual table (see header
! 437: ** comment above struct HashTableVtab) to the database schema. Both
! 438: ** provide read/write access to the contents of *pHash.
! 439: **
! 440: ** The third argument to this function, zName, is used as the name
! 441: ** of both the scalar and, if created, the virtual table.
! 442: */
! 443: int sqlite3Fts3InitHashTable(
! 444: sqlite3 *db,
! 445: Fts3Hash *pHash,
! 446: const char *zName
! 447: ){
! 448: int rc = SQLITE_OK;
! 449: void *p = (void *)pHash;
! 450: const int any = SQLITE_ANY;
! 451:
! 452: #ifdef SQLITE_TEST
! 453: char *zTest = 0;
! 454: char *zTest2 = 0;
! 455: void *pdb = (void *)db;
! 456: zTest = sqlite3_mprintf("%s_test", zName);
! 457: zTest2 = sqlite3_mprintf("%s_internal_test", zName);
! 458: if( !zTest || !zTest2 ){
! 459: rc = SQLITE_NOMEM;
! 460: }
! 461: #endif
! 462:
! 463: if( SQLITE_OK==rc ){
! 464: rc = sqlite3_create_function(db, zName, 1, any, p, scalarFunc, 0, 0);
! 465: }
! 466: if( SQLITE_OK==rc ){
! 467: rc = sqlite3_create_function(db, zName, 2, any, p, scalarFunc, 0, 0);
! 468: }
! 469: #ifdef SQLITE_TEST
! 470: if( SQLITE_OK==rc ){
! 471: rc = sqlite3_create_function(db, zTest, 2, any, p, testFunc, 0, 0);
! 472: }
! 473: if( SQLITE_OK==rc ){
! 474: rc = sqlite3_create_function(db, zTest, 3, any, p, testFunc, 0, 0);
! 475: }
! 476: if( SQLITE_OK==rc ){
! 477: rc = sqlite3_create_function(db, zTest2, 0, any, pdb, intTestFunc, 0, 0);
! 478: }
! 479: #endif
! 480:
! 481: #ifdef SQLITE_TEST
! 482: sqlite3_free(zTest);
! 483: sqlite3_free(zTest2);
! 484: #endif
! 485:
! 486: return rc;
! 487: }
! 488:
! 489: #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>