Add a configuration option to remap the "rank" column to an auxiliary fts5 function.

FossilOrigin-Name: b5f5971283b9b2f60c16f9675099855af95012cd
This commit is contained in:
dan 2014-12-02 20:18:11 +00:00
parent 89a89560d6
commit 615a9ae5db
7 changed files with 196 additions and 38 deletions

View File

@ -157,9 +157,14 @@ struct Fts5Cursor {
Fts5Sorter *pSorter; /* Sorter for "ORDER BY rank" queries */
int csrflags; /* Mask of cursor flags (see below) */
Fts5Cursor *pNext; /* Next cursor in Fts5Cursor.pCsr list */
Fts5Auxiliary *pRank; /* Rank callback (or NULL) */
char *zSpecial; /* Result of special query */
/* "rank" function. Populated on demand from vtab.xColumn(). */
Fts5Auxiliary *pRank; /* Rank callback (or NULL) */
int nRankArg; /* Number of trailing arguments for rank() */
sqlite3_value **apRankArg; /* Array of trailing arguments */
sqlite3_stmt *pRankArgStmt; /* Origin of objects in apRankArg[] */
/* Variables used by auxiliary functions */
i64 iCsrId; /* Cursor id */
Fts5Auxiliary *pAux; /* Currently executing extension function */
@ -539,6 +544,9 @@ static int fts5CloseMethod(sqlite3_vtab_cursor *pCursor){
for(pp=&pTab->pGlobal->pCsr; (*pp)!=pCsr; pp=&(*pp)->pNext);
*pp = pCsr->pNext;
sqlite3_finalize(pCsr->pRankArgStmt);
sqlite3_free(pCsr->apRankArg);
sqlite3_free(pCsr->zSpecial);
sqlite3_free(pCsr);
return SQLITE_OK;
@ -633,6 +641,7 @@ static int fts5CursorFirstSorted(Fts5Table *pTab, Fts5Cursor *pCsr, int bAsc){
int nByte;
int rc = SQLITE_OK;
char *zSql;
const char *zRank = pConfig->zRank ? pConfig->zRank : FTS5_DEFAULT_RANK;
nPhrase = sqlite3Fts5ExprPhraseCount(pCsr->pExpr);
nByte = sizeof(Fts5Sorter) + sizeof(int) * nPhrase;
@ -648,8 +657,10 @@ static int fts5CursorFirstSorted(Fts5Table *pTab, Fts5Cursor *pCsr, int bAsc){
** table, saving it creates a circular reference.
**
** If SQLite a built-in statement cache, this wouldn't be a problem. */
zSql = sqlite3_mprintf("SELECT rowid, %s FROM %Q.%Q ORDER BY +%s %s",
pConfig->zName, pConfig->zDb, pConfig->zName, FTS5_RANK_NAME,
zSql = sqlite3_mprintf("SELECT rowid, rank FROM %Q.%Q ORDER BY %s(%s%s%s) %s",
pConfig->zDb, pConfig->zName, zRank, pConfig->zName,
(pConfig->zRankArgs ? ", " : ""),
(pConfig->zRankArgs ? pConfig->zRankArgs : ""),
bAsc ? "ASC" : "DESC"
);
if( zSql==0 ){
@ -721,6 +732,74 @@ static int fts5SpecialMatch(
return rc;
}
/*
** Search for an auxiliary function named zName that can be used with table
** pTab. If one is found, return a pointer to the corresponding Fts5Auxiliary
** structure. Otherwise, if no such function exists, return NULL.
*/
static Fts5Auxiliary *fts5FindAuxiliary(Fts5Table *pTab, const char *zName){
Fts5Auxiliary *pAux;
for(pAux=pTab->pGlobal->pAux; pAux; pAux=pAux->pNext){
if( sqlite3_stricmp(zName, pAux->zFunc)==0 ) return pAux;
}
/* No function of the specified name was found. Return 0. */
return 0;
}
static int fts5FindRankFunction(Fts5Cursor *pCsr){
Fts5Table *pTab = (Fts5Table*)(pCsr->base.pVtab);
Fts5Config *pConfig = pTab->pConfig;
const char *zRank = pConfig->zRank;
int rc = SQLITE_OK;
Fts5Auxiliary *pAux;
if( zRank==0 ) zRank = FTS5_DEFAULT_RANK;
if( pTab->pConfig->zRankArgs ){
char *zSql = sqlite3_mprintf("SELECT %s", pTab->pConfig->zRankArgs);
if( zSql==0 ){
rc = SQLITE_NOMEM;
}else{
sqlite3_stmt *pStmt = 0;
rc = sqlite3_prepare_v2(pConfig->db, zSql, -1, &pStmt, 0);
sqlite3_free(zSql);
assert( rc==SQLITE_OK || pCsr->pRankArgStmt==0 );
if( rc==SQLITE_OK ){
if( SQLITE_ROW==sqlite3_step(pStmt) ){
int nByte;
pCsr->nRankArg = sqlite3_column_count(pStmt);
nByte = sizeof(sqlite3_value*)*pCsr->nRankArg;
pCsr->apRankArg = (sqlite3_value**)sqlite3Fts5MallocZero(&rc, nByte);
if( rc==SQLITE_OK ){
int i;
for(i=0; i<pCsr->nRankArg; i++){
pCsr->apRankArg[i] = sqlite3_column_value(pStmt, i);
}
}
pCsr->pRankArgStmt = pStmt;
}else{
rc = sqlite3_finalize(pStmt);
assert( rc!=SQLITE_OK );
}
}
}
}
if( rc==SQLITE_OK ){
pAux = fts5FindAuxiliary(pTab, zRank);
if( pAux==0 ){
assert( pTab->base.zErrMsg==0 );
pTab->base.zErrMsg = sqlite3_mprintf("no such function: %s", zRank);
rc = SQLITE_ERROR;
}
}
pCsr->pRank = pAux;
return rc;
}
/*
** This is the xFilter interface for the virtual table. See
@ -753,7 +832,6 @@ static int fts5FilterMethod(
** fts5CursorFirstSorted() above. */
assert( FTS5_PLAN(idxNum)==FTS5_PLAN_SCAN );
pCsr->idxNum = FTS5_PLAN_SOURCE;
pCsr->pRank = pTab->pSortCsr->pRank;
pCsr->pExpr = pTab->pSortCsr->pExpr;
rc = fts5CursorFirst(pTab, pCsr, bAsc);
}else{
@ -769,7 +847,6 @@ static int fts5FilterMethod(
rc = fts5SpecialMatch(pTab, pCsr, &zExpr[1]);
}else{
char **pzErr = &pTab->base.zErrMsg;
pCsr->pRank = pTab->pGlobal->pAux;
rc = sqlite3Fts5ExprNew(pTab->pConfig, zExpr, &pCsr->pExpr, pzErr);
if( rc==SQLITE_OK ){
if( ePlan==FTS5_PLAN_MATCH ){
@ -1092,7 +1169,7 @@ static int fts5CacheInstArray(Fts5Cursor *pCsr){
aIter = (Fts5PoslistReader*)sqlite3Fts5MallocZero(&rc, nByte);
if( aIter ){
Fts5Buffer buf = {0, 0, 0}; /* Build up aInst[] here */
int nInst; /* Number instances seen so far */
int nInst = 0; /* Number instances seen so far */
int i;
/* Initialize all iterators */
@ -1426,19 +1503,23 @@ static int fts5ColumnMethod(
}else
if( iCol==pConfig->nCol ){
/* User is requesting the value of the special column with the same name
** as the table. Return the cursor integer id number. This value is only
** useful in that it may be passed as the first argument to an FTS5
** auxiliary function. */
sqlite3_result_int64(pCtx, pCsr->iCsrId);
}else if( iCol==pConfig->nCol+1 ){
/* The value of the "rank" column. */
if( FTS5_PLAN(pCsr->idxNum)==FTS5_PLAN_SOURCE ){
fts5PoslistBlob(pCtx, pCsr);
}else{
/* User is requesting the value of the special column with the same name
** as the table. Return the cursor integer id number. This value is only
** useful in that it may be passed as the first argument to an FTS5
** auxiliary function. */
sqlite3_result_int64(pCtx, pCsr->iCsrId);
}
}else if( iCol==pConfig->nCol+1 ){
/* The value of the "rank" column. */
if( pCsr->pRank ){
fts5ApiInvoke(pCsr->pRank, pCsr, pCtx, 0, 0);
}else if(
FTS5_PLAN(pCsr->idxNum)==FTS5_PLAN_MATCH
|| FTS5_PLAN(pCsr->idxNum)==FTS5_PLAN_SORTED_MATCH
){
if( pCsr->pRank || SQLITE_OK==(rc = fts5FindRankFunction(pCsr)) ){
fts5ApiInvoke(pCsr->pRank, pCsr, pCtx, pCsr->nRankArg, pCsr->apRankArg);
}
}
}else{
rc = fts5SeekCursor(pCsr);
@ -1464,12 +1545,11 @@ static int fts5FindFunctionMethod(
Fts5Table *pTab = (Fts5Table*)pVtab;
Fts5Auxiliary *pAux;
for(pAux=pTab->pGlobal->pAux; pAux; pAux=pAux->pNext){
if( sqlite3_stricmp(zName, pAux->zFunc)==0 ){
*pxFunc = fts5ApiCallback;
*ppArg = (void*)pAux;
return 1;
}
pAux = fts5FindAuxiliary(pTab, zName);
if( pAux ){
*pxFunc = fts5ApiCallback;
*ppArg = (void*)pAux;
return 1;
}
/* No function of the specified name was found. Return 0. */

View File

@ -27,6 +27,7 @@
#define FTS5_MAX_PREFIX_INDEXES 31
#define FTS5_DEFAULT_NEARDIST 10
#define FTS5_DEFAULT_RANK "bm25"
/* Name of rank column */
#define FTS5_RANK_NAME "rank"

View File

@ -468,11 +468,13 @@ static int fts5ConfigParseRank(
p++;
}
if( rc==SQLITE_OK ){
const char *pArgs = p;
const char *pArgs;
p = fts5ConfigSkipWhitespace(p);
pArgs = p;
p = fts5ConfigSkipArgs(p);
if( p==0 ){
rc = SQLITE_ERROR;
}else{
}else if( p!=pArgs ){
zRankArgs = sqlite3Fts5MallocZero(&rc, 1 + p - pArgs);
if( zRankArgs ) memcpy(zRankArgs, pArgs, p-pArgs);
}

View File

@ -1,5 +1,5 @@
C Add\scode\sto\sparse\sa\srank()\sfunction\sspecification.\sAnd\sa\stcl\sinterface\sto\sadd\sauxiliary\sfunctions\sto\sfts5.
D 2014-12-01T20:05:00.761
C Add\sa\sconfiguration\soption\sto\sremap\sthe\s"rank"\scolumn\sto\san\sauxiliary\sfts5\sfunction.
D 2014-12-02T20:18:11.604
F Makefile.arm-wince-mingw32ce-gcc d6df77f1f48d690bd73162294bbba7f59507c72f
F Makefile.in b03432313a3aad96c706f8164fb9f5307eaf19f5
F Makefile.linux-gcc 91d710bdc4998cb015f39edf3cb314ec4f4d7e23
@ -104,12 +104,12 @@ F ext/fts3/unicode/CaseFolding.txt 8c678ca52ecc95e16bc7afc2dbf6fc9ffa05db8c
F ext/fts3/unicode/UnicodeData.txt cd07314edb62d49fde34debdaf92fa2aa69011e7
F ext/fts3/unicode/mkunicode.tcl dc6f268eb526710e2c6e496c372471d773d0c368
F ext/fts5/extract_api_docs.tcl 6320db4a1d0722a4e2069e661381ad75e9889786
F ext/fts5/fts5.c 07f81ce7ebbffdd0acdad9eb090ff506fa503a10
F ext/fts5/fts5.c 572bd5d4d272ca562240dc1905538f060783ab78
F ext/fts5/fts5.h 72fc1e9995b1ddc254a487b9528614a83bd3dfb6
F ext/fts5/fts5Int.h e16cf2213ae748ccc2c890f404fc341eb941d10b
F ext/fts5/fts5Int.h 9dbf415de032b1cc770dcedaa5a8e434d88ca90c
F ext/fts5/fts5_aux.c 0e3e5fea6bf5772805afe14c95cb5f16e03e4b3f
F ext/fts5/fts5_buffer.c c79d67a5a611521f1f3b9d495981f22c02ef4bdb
F ext/fts5/fts5_config.c bb87c2b915ae94002d94d02a6b1f81a0dac9c6db
F ext/fts5/fts5_config.c 664fdc8519b55753f5c24d7b45176f05586b7965
F ext/fts5/fts5_expr.c d317be07d70223a6865444f17982570260b690a5
F ext/fts5/fts5_hash.c 63fa8379c5f2ac107d47c2b7d9ac04c95ef8a279
F ext/fts5/fts5_index.c 7e7023f3a29f104b44df2ca2474b296b8dfe447c
@ -603,14 +603,14 @@ F test/fts5aa.test 27c7d3c865e144a0501dcbfbd6d2ae87f77602ea
F test/fts5ab.test 52f6b9223372ff70b0edb5a3054fbd7bc7fcfefc
F test/fts5ac.test 60302196b7711176ce872fe2e4c73c75ac2c4038
F test/fts5ad.test ed60fdafc73d879b42573abcfa6ede7e02e07c19
F test/fts5ae.test 6decf7634acd161af9583ce32ab7197b0113c5cd
F test/fts5ae.test 5de775469d45a2f8218fc89b8d6d5176c226d05e
F test/fts5af.test d24e3b0f879998ef5f60087272f8ab7b3a8fd4dc
F test/fts5ag.test 1c6c188d1bdc41b2277db3f4ddfea7d90bf44ceb
F test/fts5ah.test 788e923e60b5e7a559f672cfbf262b8b260ea176
F test/fts5ai.test aa2b5fd0f8d2cf59ac0211111e63cbca3b40ed7d
F test/fts5aj.test bc3d91bd012c7ca175cdf266c2074920bb5fa5ba
F test/fts5ak.test e55bb0f3fac1291d32bc9485a3ee55a7d76f4d5f
F test/fts5al.test d716a933bb88eb6986b02b985924fa42960b6eec
F test/fts5al.test 61b067f3b0b61679ab164a8a855882dfd313988d
F test/fts5ea.test afaf3497b43add578384dc1fd26b0342738abe87
F test/full.test 6b3c8fb43c6beab6b95438c1675374b95fab245d
F test/func.test ae97561957aba6ca9e3a7b8a13aac41830d701ef
@ -1207,7 +1207,7 @@ F tool/vdbe_profile.tcl 67746953071a9f8f2f668b73fe899074e2c6d8c1
F tool/warnings-clang.sh f6aa929dc20ef1f856af04a730772f59283631d4
F tool/warnings.sh 0abfd78ceb09b7f7c27c688c8e3fe93268a13b32
F tool/win/sqlite.vsix deb315d026cc8400325c5863eef847784a219a2f
P bb4a37b53de60da9ec8b9317eec14afa99690828
R efa8336057fcd1502b8cbf6d797345c7
P 9c1697a2aa1f601e6eb11704abe63a73c8105447
R 30808e5592c3e61509564bec30e4914f
U dan
Z dc9192af5fedea55ad78c651e89e8c7b
Z 9589e0356694de369bd9f49ee042fc35

View File

@ -1 +1 @@
9c1697a2aa1f601e6eb11704abe63a73c8105447
b5f5971283b9b2f60c16f9675099855af95012cd

View File

@ -274,6 +274,5 @@ foreach {tn q res} {
} $res
}
finish_test

View File

@ -80,6 +80,7 @@ foreach {tn defn} {
}
#-------------------------------------------------------------------------
# Assorted tests of the tcl interface for creating extension functions.
#
do_execsql_test 3.1 {
@ -134,11 +135,86 @@ proc coltest {cmd} {
}
sqlite3_fts5_create_function db coltest coltest
do_execsql_test 3.4.1 {
do_execsql_test 3.5.1 {
SELECT coltest(t1) FROM t1 WHERE t1 MATCH 'q'
} {
{6 {y t r e w q}} {6 {q w e r t y}}
}
#-------------------------------------------------------------------------
# Tests for remapping the "rank" column.
#
# 4.1.*: Mapped to a function with no arguments.
# 4.2.*: Mapped to a function with one or more arguments.
#
do_execsql_test 4.0 {
CREATE VIRTUAL TABLE t2 USING fts5(a, b);
INSERT INTO t2 VALUES('a s h g s b j m r h', 's b p a d b b a o e');
INSERT INTO t2 VALUES('r h n t a g r d d i', 'l d n j r c f t o q');
INSERT INTO t2 VALUES('q k n i k c a a e m', 'c h n j p g s c i t');
INSERT INTO t2 VALUES('h j g t r e l s g s', 'k q k c i i c k n s');
INSERT INTO t2 VALUES('b l k h d n n n m i', 'p t i a r b t q o l');
INSERT INTO t2 VALUES('k r i l j b g i p a', 't q c h a i m g n l');
INSERT INTO t2 VALUES('a e c q n m o m d g', 'l c t g i s q g q e');
INSERT INTO t2 VALUES('b o j h f o g b p e', 'r t l h s b g i c p');
INSERT INTO t2 VALUES('s q k f q b j g h f', 'n m a o p e i e k t');
INSERT INTO t2 VALUES('o q g g q c o k a b', 'r t k p t f t h p c');
}
proc firstinst {cmd} {
foreach {p c o} [$cmd xInst 0] {}
expr $c*100 + $o
}
sqlite3_fts5_create_function db firstinst firstinst
do_execsql_test 4.1.1 {
SELECT rowid, firstinst(t2) FROM t2 WHERE t2 MATCH 'a' ORDER BY rowid ASC
} {
1 0 2 4 3 6 5 103
6 9 7 0 9 102 10 8
}
do_execsql_test 4.1.2 {
INSERT INTO t2(t2, rank) VALUES('rank', 'firstinst()');
SELECT rowid, rank FROM t2 WHERE t2 MATCH 'a' ORDER BY rowid ASC
} {
1 0 2 4 3 6 5 103
6 9 7 0 9 102 10 8
}
do_execsql_test 4.1.3 {
SELECT rowid, rank FROM t2 WHERE t2 MATCH 'a' ORDER BY rank DESC
} {
5 103 9 102 6 9 10 8 3 6 2 4 7 0 1 0
}
do_execsql_test 4.1.4 {
INSERT INTO t2(t2, rank) VALUES('rank', 'firstinst ( ) ');
SELECT rowid, rank FROM t2 WHERE t2 MATCH 'a' ORDER BY rank DESC
} {
5 103 9 102 6 9 10 8 3 6 2 4 7 0 1 0
}
proc rowidplus {cmd ival} {
expr [$cmd xRowid] + $ival
}
sqlite3_fts5_create_function db rowidplus rowidplus
do_execsql_test 4.2.1 {
INSERT INTO t2(t2, rank) VALUES('rank', 'rowidplus(100) ');
SELECT rowid, rank FROM t2 WHERE t2 MATCH 'o + q + g'
} {
10 110
}
do_execsql_test 4.2.2 {
INSERT INTO t2(t2, rank) VALUES('rank', 'rowidplus(111) ');
SELECT rowid, rank FROM t2 WHERE t2 MATCH 'o + q + g'
} {
10 121
}
finish_test