Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
Comment: | Fix a problem handling OOM conditions within fts5 queries that feature synonyms. |
---|---|
Downloads: | Tarball | ZIP archive |
Timelines: | family | ancestors | descendants | both | fts5-incompatible |
Files: | files | file ages | folders |
SHA1: |
11fa980897c6c7be218bbd9c4cd82532 |
User & Date: | dan 2015-09-02 14:17:38.670 |
Context
2015-09-02
| ||
17:34 | Fix a problem with fts5 synonyms and phrase queries. Also fix an OOM handling bug in fts5. (check-in: a4c35fa2c9 user: dan tags: fts5-incompatible) | |
14:17 | Fix a problem handling OOM conditions within fts5 queries that feature synonyms. (check-in: 11fa980897 user: dan tags: fts5-incompatible) | |
08:22 | Fix a problem with fts5 synonyms and the xQueryPhrase() auxiliary function API. (check-in: cf3e45e76d user: dan tags: fts5-incompatible) | |
Changes
Changes to ext/fts5/fts5_expr.c.
︙ | ︙ | |||
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 | } return pRet; } typedef struct TokenCtx TokenCtx; struct TokenCtx { Fts5ExprPhrase *pPhrase; }; /* ** Callback for tokenizing terms used by ParseTerm(). */ static int fts5ParseTokenize( void *pContext, /* Pointer to Fts5InsertCtx object */ int tflags, /* Mask of FTS5_TOKEN_* flags */ const char *pToken, /* Buffer containing token */ int nToken, /* Size of token in bytes */ int iUnused1, /* Start offset of token */ int iUnused2 /* End offset of token */ ){ int rc = SQLITE_OK; const int SZALLOC = 8; TokenCtx *pCtx = (TokenCtx*)pContext; Fts5ExprPhrase *pPhrase = pCtx->pPhrase; assert( pPhrase==0 || pPhrase->nTerm>0 ); if( pPhrase && (tflags & FTS5_TOKEN_COLOCATED) ){ Fts5ExprTerm *pSyn; int nByte = sizeof(Fts5ExprTerm) + nToken+1; pSyn = (Fts5ExprTerm*)sqlite3_malloc(nByte); | > > > > | > > | | | | | > | > > | | | | | > > | | | | > > > | 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 | } return pRet; } typedef struct TokenCtx TokenCtx; struct TokenCtx { Fts5ExprPhrase *pPhrase; int rc; }; /* ** Callback for tokenizing terms used by ParseTerm(). */ static int fts5ParseTokenize( void *pContext, /* Pointer to Fts5InsertCtx object */ int tflags, /* Mask of FTS5_TOKEN_* flags */ const char *pToken, /* Buffer containing token */ int nToken, /* Size of token in bytes */ int iUnused1, /* Start offset of token */ int iUnused2 /* End offset of token */ ){ int rc = SQLITE_OK; const int SZALLOC = 8; TokenCtx *pCtx = (TokenCtx*)pContext; Fts5ExprPhrase *pPhrase = pCtx->pPhrase; /* If an error has already occurred, this is a no-op */ if( pCtx->rc!=SQLITE_OK ) return pCtx->rc; assert( pPhrase==0 || pPhrase->nTerm>0 ); if( pPhrase && (tflags & FTS5_TOKEN_COLOCATED) ){ Fts5ExprTerm *pSyn; int nByte = sizeof(Fts5ExprTerm) + nToken+1; pSyn = (Fts5ExprTerm*)sqlite3_malloc(nByte); if( pSyn==0 ){ rc = SQLITE_NOMEM; }else{ memset(pSyn, 0, nByte); pSyn->zTerm = (char*)&pSyn[1]; memcpy(pSyn->zTerm, pToken, nToken); pSyn->pSynonym = pPhrase->aTerm[pPhrase->nTerm-1].pSynonym; pPhrase->aTerm[pPhrase->nTerm-1].pSynonym = pSyn; } }else{ Fts5ExprTerm *pTerm; if( pPhrase==0 || (pPhrase->nTerm % SZALLOC)==0 ){ Fts5ExprPhrase *pNew; int nNew = SZALLOC + (pPhrase ? pPhrase->nTerm : 0); pNew = (Fts5ExprPhrase*)sqlite3_realloc(pPhrase, sizeof(Fts5ExprPhrase) + sizeof(Fts5ExprTerm) * nNew ); if( pNew==0 ){ rc = SQLITE_NOMEM; }else{ if( pPhrase==0 ) memset(pNew, 0, sizeof(Fts5ExprPhrase)); pCtx->pPhrase = pPhrase = pNew; pNew->nTerm = nNew - SZALLOC; } } if( rc==SQLITE_OK ){ pTerm = &pPhrase->aTerm[pPhrase->nTerm++]; memset(pTerm, 0, sizeof(Fts5ExprTerm)); pTerm->zTerm = sqlite3Fts5Strndup(&rc, pToken, nToken); } } pCtx->rc = rc; return rc; } /* ** Free the phrase object passed as the only argument. */ |
︙ | ︙ | |||
1569 1570 1571 1572 1573 1574 1575 | int flags = FTS5_TOKENIZE_QUERY | (bPrefix ? FTS5_TOKENIZE_QUERY : 0); int n; sqlite3Fts5Dequote(z); n = strlen(z); rc = sqlite3Fts5Tokenize(pConfig, flags, z, n, &sCtx, fts5ParseTokenize); } sqlite3_free(z); | | | 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 | int flags = FTS5_TOKENIZE_QUERY | (bPrefix ? FTS5_TOKENIZE_QUERY : 0); int n; sqlite3Fts5Dequote(z); n = strlen(z); rc = sqlite3Fts5Tokenize(pConfig, flags, z, n, &sCtx, fts5ParseTokenize); } sqlite3_free(z); if( rc || (rc = sCtx.rc) ){ pParse->rc = rc; fts5ExprPhraseFree(sCtx.pPhrase); sCtx.pPhrase = 0; }else if( sCtx.pPhrase ){ if( pAppend==0 ){ if( (pParse->nPhrase % 8)==0 ){ |
︙ | ︙ | |||
1618 1619 1620 1621 1622 1623 1624 | int i; /* Used to iterate through phrase terms */ Fts5Expr *pNew = 0; /* Expression to return via *ppNew */ Fts5ExprPhrase **apPhrase; /* pNew->apPhrase */ Fts5ExprNode *pNode; /* pNew->pRoot */ Fts5ExprNearset *pNear; /* pNew->pRoot->pNear */ | | | 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 | int i; /* Used to iterate through phrase terms */ Fts5Expr *pNew = 0; /* Expression to return via *ppNew */ Fts5ExprPhrase **apPhrase; /* pNew->apPhrase */ Fts5ExprNode *pNode; /* pNew->pRoot */ Fts5ExprNearset *pNear; /* pNew->pRoot->pNear */ TokenCtx sCtx = {0,0}; /* Context object for fts5ParseTokenize */ pOrig = pExpr->apExprPhrase[iPhrase]; pNew = (Fts5Expr*)sqlite3Fts5MallocZero(&rc, sizeof(Fts5Expr)); if( rc==SQLITE_OK ){ pNew->apExprPhrase = (Fts5ExprPhrase**)sqlite3Fts5MallocZero(&rc, |
︙ | ︙ |
Changes to ext/fts5/fts5_test_mi.c.
︙ | ︙ | |||
348 349 350 351 352 353 354 | Fts5Context *pFts, /* First arg to pass to pApi functions */ sqlite3_context *pCtx, /* Context for returning result/error */ int nVal, /* Number of values in apVal[] array */ sqlite3_value **apVal /* Array of trailing arguments */ ){ const char *zArg; Fts5MatchinfoCtx *p; | | > > > | < | | > > | > | 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 | Fts5Context *pFts, /* First arg to pass to pApi functions */ sqlite3_context *pCtx, /* Context for returning result/error */ int nVal, /* Number of values in apVal[] array */ sqlite3_value **apVal /* Array of trailing arguments */ ){ const char *zArg; Fts5MatchinfoCtx *p; int rc = SQLITE_OK; if( nVal>0 ){ zArg = (const char*)sqlite3_value_text(apVal[0]); }else{ zArg = "pcx"; } p = (Fts5MatchinfoCtx*)pApi->xGetAuxdata(pFts, 0); if( p==0 || sqlite3_stricmp(zArg, p->zArg) ){ p = fts5MatchinfoNew(pApi, pFts, pCtx, zArg); if( p==0 ){ rc = SQLITE_NOMEM; }else{ rc = pApi->xSetAuxdata(pFts, p, sqlite3_free); } } if( rc==SQLITE_OK ){ rc = fts5MatchinfoIter(pApi, pFts, p, fts5MatchinfoLocalCb); } if( rc!=SQLITE_OK ){ sqlite3_result_error_code(pCtx, rc); }else{ /* No errors has occured, so return a copy of the array of integers. */ int nByte = p->nRet * sizeof(u32); sqlite3_result_blob(pCtx, (void*)p->aRet, nByte, SQLITE_TRANSIENT); } |
︙ | ︙ |
Changes to ext/fts5/test/fts5_common.tcl.
︙ | ︙ | |||
290 291 292 293 294 295 296 297 | proc OR {args} { sort_poslist [concat {*}$args] } proc NOT {a b} { if {[llength $b]>0} { return [list] } return $a } | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 | proc OR {args} { sort_poslist [concat {*}$args] } proc NOT {a b} { if {[llength $b]>0} { return [list] } return $a } #------------------------------------------------------------------------- # This command is similar to [split], except that it also provides the # start and end offsets of each token. For example: # # [fts5_tokenize_split "abc d ef"] -> {abc 0 3 d 4 5 ef 6 8} # proc gobble_whitespace {textvar} { upvar $textvar t regexp {([ ]*)(.*)} $t -> space t return [string length $space] } proc gobble_text {textvar wordvar} { upvar $textvar t upvar $wordvar w regexp {([^ ]*)(.*)} $t -> w t return [string length $w] } proc fts5_tokenize_split {text} { set token "" set ret [list] set iOff [gobble_whitespace text] while {[set nToken [gobble_text text word]]} { lappend ret $word $iOff [expr $iOff+$nToken] incr iOff $nToken incr iOff [gobble_whitespace text] } set ret } |
Changes to ext/fts5/test/fts5fault6.test.
︙ | ︙ | |||
143 144 145 146 147 148 149 150 151 152 | } -body { db eval { CREATE VIRTUAL TABLE yu USING fts5(x, tokenize="unicode61 separators abc"); } } -test { faultsim_test_result {0 {}} } finish_test | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 | } -body { db eval { CREATE VIRTUAL TABLE yu USING fts5(x, tokenize="unicode61 separators abc"); } } -test { faultsim_test_result {0 {}} } #------------------------------------------------------------------------- # OOM while running a query that includes synonyms and matchinfo(). # proc mit {blob} { set scan(littleEndian) i* set scan(bigEndian) I* binary scan $blob $scan($::tcl_platform(byteOrder)) r return $r } proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd if {$tflags=="query" && [string length $w]==1} { for {set i 2} {$i < 7} {incr i} { sqlite3_fts5_token -colo [string repeat $w $i] $iStart $iEnd } } } } proc tcl_create {args} { return "tcl_tokenize" } reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create db func mit mit sqlite3_fts5_register_matchinfo db do_test 5.0 { execsql { CREATE VIRTUAL TABLE t1 USING fts5(a, tokenize=tcl) } foreach {rowid text} { 1 {aaaa cc b aaaaa cc aa} 2 {aa aa bb a bbb} 3 {bb aaaaa aaaaa b aaaa aaaaa} 4 {aa a b aaaa aa} 5 {aa b ccc aaaaa cc} 6 {aa aaaaa bbbb cc aaa} 7 {aaaaa aa aa ccccc bb} 8 {ccc bbbbb ccccc bbb c} 9 {cccccc bbbb a aaa cccc c} } { execsql { INSERT INTO t1(rowid, a) VALUES($rowid, $text) } } } {} set res [list {*}{ 1 {3 24 8 2 12 6} 5 {2 24 8 2 12 6} 6 {3 24 8 1 12 6} 7 {3 24 8 1 12 6} 9 {2 24 8 3 12 6} }] do_execsql_test 5.1 { SELECT rowid, mit(matchinfo(t1, 'x')) FROM t1 WHERE t1 MATCH 'a AND c' } $res faultsim_save_and_close do_faultsim_test 5.2 -faults oom* -prep { faultsim_restore_and_reopen sqlite3_fts5_create_tokenizer db tcl tcl_create sqlite3_fts5_register_matchinfo db db func mit mit } -body { db eval { SELECT rowid, mit(matchinfo(t1, 'x')) FROM t1 WHERE t1 MATCH 'a AND c' } } -test { faultsim_test_result [list 0 $::res] } finish_test |
Changes to ext/fts5/test/fts5synonym.test.
︙ | ︙ | |||
36 37 38 39 40 41 42 | foreach s $S { set o [list] foreach x $S {if {$x!=$s} {lappend o $x}} set ::syn($s) $o } } | < < < < < < < < < < < < < < < < < < < < < < < < < < | | 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 | foreach s $S { set o [list] foreach x $S {if {$x!=$s} {lappend o $x}} set ::syn($s) $o } } proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd } } proc tcl_create {args} { return "tcl_tokenize" } |
︙ | ︙ | |||
91 92 93 94 95 96 97 | #------------------------------------------------------------------------- # Test a tokenizer that supports synonyms by adding extra entries to the # FTS index. # proc tcl_tokenize {tflags text} { | | | 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 | #------------------------------------------------------------------------- # Test a tokenizer that supports synonyms by adding extra entries to the # FTS index. # proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd if {$tflags=="document" && [info exists ::syn($w)]} { foreach s $::syn($w) { sqlite3_fts5_token -colo $s $iStart $iEnd } } } |
︙ | ︙ | |||
133 134 135 136 137 138 139 | # 3.2.*: A tokenizer that reports two identical tokens at the same position. # This is allowed. # reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { set bColo 1 | | | 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 | # 3.2.*: A tokenizer that reports two identical tokens at the same position. # This is allowed. # reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { set bColo 1 foreach {w iStart iEnd} [fts5_tokenize_split $text] { if {$bColo} { sqlite3_fts5_token -colo $w $iStart $iEnd set bColo 0 } { sqlite3_fts5_token $w $iStart $iEnd } } |
︙ | ︙ | |||
156 157 158 159 160 161 162 | } do_execsql_test 3.1.1 { INSERT INTO ft(ft) VALUES('integrity-check'); } {} proc tcl_tokenize {tflags text} { | | | | 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 | } do_execsql_test 3.1.1 { INSERT INTO ft(ft) VALUES('integrity-check'); } {} proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd } } do_execsql_test 3.1.2 { SELECT rowid FROM ft WHERE ft MATCH 'one two three' } {1} reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd sqlite3_fts5_token -colo $w $iStart $iEnd } } do_execsql_test 3.2.0 { CREATE VIRTUAL TABLE ft USING fts5(x, tokenize = tcl); INSERT INTO ft VALUES('one one two three'); |
︙ | ︙ | |||
203 204 205 206 207 208 209 | #------------------------------------------------------------------------- # Check that expressions with synonyms can be parsed and executed. # reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { | | | 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 | #------------------------------------------------------------------------- # Check that expressions with synonyms can be parsed and executed. # reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd if {$tflags=="query" && [info exists ::syn($w)]} { foreach s $::syn($w) { sqlite3_fts5_token -colo $s $iStart $iEnd } } } |
︙ | ︙ | |||
326 327 328 329 330 331 332 | #------------------------------------------------------------------------- # Test terms with more than 4 synonyms. # reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { | | | 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 | #------------------------------------------------------------------------- # Test terms with more than 4 synonyms. # reset_db sqlite3_fts5_create_tokenizer db tcl tcl_create proc tcl_tokenize {tflags text} { foreach {w iStart iEnd} [fts5_tokenize_split $text] { sqlite3_fts5_token $w $iStart $iEnd if {$tflags=="query" && [string length $w]==1} { for {set i 2} {$i<=10} {incr i} { sqlite3_fts5_token -colo [string repeat $w $i] $iStart $iEnd } } } |
︙ | ︙ |