Add the tokendata=1 option to ignore trailing token-data when querying an fts5 table.

FossilOrigin-Name: 122935182ad5869ce3a4c6d796c38a0509f6f3384dd1b3e60a3f2f0f366cc5f5

diff --git a/ext/fts5/fts5Int.h b/ext/fts5/fts5Int.h
index 1687168d5ff657cabb770a978569b4836a08776e..4aa578559b8cd1dec2533d14693a8cf222eb4e05 100644 (file)
--- a/ext/fts5/fts5Int.h
+++ b/ext/fts5/fts5Int.h
@@ -196,6 +196,7 @@ struct Fts5Config {
   char *zContent;                 /* content table */ 
   char *zContentRowid;            /* "content_rowid=" option value */ 
   int bColumnsize;                /* "columnsize=" option value (dflt==1) */
+  int bTokendata;                 /* "tokendata=" option value (dflt==0) */
   int eDetail;                    /* FTS5_DETAIL_XXX value */
   char *zContentExprlist;
   Fts5Tokenizer *pTok;

diff --git a/ext/fts5/fts5_config.c b/ext/fts5/fts5_config.c
index 5d0770502eed19e0277a68e42f93f1a497ebfb1c..d2e8309cd2c26fbe27de30ff29eee2823d942f26 100644 (file)
--- a/ext/fts5/fts5_config.c
+++ b/ext/fts5/fts5_config.c
@@ -398,6 +398,16 @@ static int fts5ConfigParseSpecial(
     return rc;
   }
 
+  if( sqlite3_strnicmp("tokendata", zCmd, nCmd)==0 ){
+    if( (zArg[0]!='0' && zArg[0]!='1') || zArg[1]!='\0' ){
+      *pzErr = sqlite3_mprintf("malformed tokendata=... directive");
+      rc = SQLITE_ERROR;
+    }else{
+      pConfig->bTokendata = (zArg[0]=='1');
+    }
+    return rc;
+  }
+
   *pzErr = sqlite3_mprintf("unrecognized option: \"%.*s\"", nCmd, zCmd);
   return SQLITE_ERROR;
 }

diff --git a/ext/fts5/fts5_expr.c b/ext/fts5/fts5_expr.c
index 745a5d9fa6aa83a662f8d4aee7d752bbd8220e19..a2c6320719af4675d7c6ec07e8d9d9a5f3cd5c1f 100644 (file)
--- a/ext/fts5/fts5_expr.c
+++ b/ext/fts5/fts5_expr.c
@@ -101,7 +101,8 @@ struct Fts5ExprTerm {
   u8 bPrefix;                     /* True for a prefix term */
   u8 bFirst;                      /* True if token must be first in column */
   char *pTerm;                    /* Term data */
-  int nTerm;                      /* Size of term in bytes */
+  int nQueryTerm;                 /* Effective size of term in bytes */
+  int nFullTerm;                  /* Size of term in bytes incl. tokendata */
   Fts5IndexIter *pIter;           /* Iterator for this term */
   Fts5ExprTerm *pSynonym;         /* Pointer to first in list of synonyms */
 };
@@ -968,7 +969,7 @@ static int fts5ExprNearInitAll(
             p->pIter = 0;
           }
           rc = sqlite3Fts5IndexQuery(
-              pExpr->pIndex, p->pTerm, p->nTerm,
+              pExpr->pIndex, p->pTerm, p->nQueryTerm,
               (pTerm->bPrefix ? FTS5INDEX_QUERY_PREFIX : 0) |
               (pExpr->bDesc ? FTS5INDEX_QUERY_DESC : 0),
               pNear->pColset,
@@ -1703,6 +1704,7 @@ Fts5ExprNearset *sqlite3Fts5ParseNearset(
 typedef struct TokenCtx TokenCtx;
 struct TokenCtx {
   Fts5ExprPhrase *pPhrase;
+  Fts5Config *pConfig;
   int rc;
 };
 
@@ -1737,7 +1739,8 @@ static int fts5ParseTokenize(
     }else{
       memset(pSyn, 0, (size_t)nByte);
       pSyn->pTerm = ((char*)pSyn) + sizeof(Fts5ExprTerm) + sizeof(Fts5Buffer);
-      pSyn->nTerm = nToken;
+      pSyn->nFullTerm = pSyn->nQueryTerm = nToken;
+      if( pCtx->pConfig->bTokendata ) pSyn->nQueryTerm = strlen(pSyn->pTerm);
       memcpy(pSyn->pTerm, pToken, nToken);
       pSyn->pSynonym = pPhrase->aTerm[pPhrase->nTerm-1].pSynonym;
       pPhrase->aTerm[pPhrase->nTerm-1].pSynonym = pSyn;
@@ -1764,7 +1767,8 @@ static int fts5ParseTokenize(
       pTerm = &pPhrase->aTerm[pPhrase->nTerm++];
       memset(pTerm, 0, sizeof(Fts5ExprTerm));
       pTerm->pTerm = sqlite3Fts5Strndup(&rc, pToken, nToken);
-      pTerm->nTerm = nToken;
+      pTerm->nFullTerm = pTerm->nQueryTerm = nToken;
+      if( pCtx->pConfig->bTokendata ) pTerm->nQueryTerm = strlen(pTerm->pTerm);
     }
   }
 
@@ -1831,6 +1835,7 @@ Fts5ExprPhrase *sqlite3Fts5ParseTerm(
 
   memset(&sCtx, 0, sizeof(TokenCtx));
   sCtx.pPhrase = pAppend;
+  sCtx.pConfig = pConfig;
 
   rc = fts5ParseStringFromToken(pToken, &z);
   if( rc==SQLITE_OK ){
@@ -1880,8 +1885,7 @@ int sqlite3Fts5ExprClonePhrase(
   int rc = SQLITE_OK;             /* Return code */
   Fts5ExprPhrase *pOrig;          /* The phrase extracted from pExpr */
   Fts5Expr *pNew = 0;             /* Expression to return via *ppNew */
-  TokenCtx sCtx = {0,0};          /* Context object for fts5ParseTokenize */
-
+  TokenCtx sCtx = {0,0,0};        /* Context object for fts5ParseTokenize */
   pOrig = pExpr->apExprPhrase[iPhrase];
   pNew = (Fts5Expr*)sqlite3Fts5MallocZero(&rc, sizeof(Fts5Expr));
   if( rc==SQLITE_OK ){
@@ -1912,11 +1916,12 @@ int sqlite3Fts5ExprClonePhrase(
 
   if( pOrig->nTerm ){
     int i;                          /* Used to iterate through phrase terms */
+    sCtx.pConfig = pExpr->pConfig;
     for(i=0; rc==SQLITE_OK && i<pOrig->nTerm; i++){
       int tflags = 0;
       Fts5ExprTerm *p;
       for(p=&pOrig->aTerm[i]; p && rc==SQLITE_OK; p=p->pSynonym){
-        rc = fts5ParseTokenize((void*)&sCtx, tflags, p->pTerm, p->nTerm, 0, 0);
+        rc = fts5ParseTokenize((void*)&sCtx, tflags, p->pTerm,p->nFullTerm,0,0);
         tflags = FTS5_TOKEN_COLOCATED;
       }
       if( rc==SQLITE_OK ){
@@ -2298,12 +2303,12 @@ static Fts5ExprNode *fts5ParsePhraseToAnd(
           fts5ExprPhraseFree(pPhrase);
         }else{
           Fts5ExprTerm *p = &pNear->apPhrase[0]->aTerm[ii];
+          Fts5ExprTerm *pTo = &pPhrase->aTerm[0];
           pParse->apPhrase[pParse->nPhrase++] = pPhrase;
           pPhrase->nTerm = 1;
-          pPhrase->aTerm[0].pTerm = sqlite3Fts5Strndup(
-              &pParse->rc, p->pTerm, p->nTerm
-          );
-          pPhrase->aTerm[0].nTerm = p->nTerm;
+          pTo->pTerm = sqlite3Fts5Strndup(&pParse->rc, p->pTerm, p->nFullTerm);
+          pTo->nQueryTerm = p->nQueryTerm;
+          pTo->nFullTerm = p->nFullTerm;
           pRet->apChild[ii] = sqlite3Fts5ParseNode(pParse, FTS5_STRING, 
               0, 0, sqlite3Fts5ParseNearset(pParse, 0, pPhrase)
           );
@@ -2488,7 +2493,7 @@ static char *fts5ExprTermPrint(Fts5ExprTerm *pTerm){
 
   /* Determine the maximum amount of space required. */
   for(p=pTerm; p; p=p->pSynonym){
-    nByte += pTerm->nTerm * 2 + 3 + 2;
+    nByte += pTerm->nQueryTerm * 2 + 3 + 2;
   }
   zQuoted = sqlite3_malloc64(nByte);
 
@@ -2496,7 +2501,7 @@ static char *fts5ExprTermPrint(Fts5ExprTerm *pTerm){
     int i = 0;
     for(p=pTerm; p; p=p->pSynonym){
       char *zIn = p->pTerm;
-      char *zEnd = &zIn[p->nTerm];
+      char *zEnd = &zIn[p->nQueryTerm];
       zQuoted[i++] = '"';
       while( zIn<zEnd ){
         if( *zIn=='"' ) zQuoted[i++] = '"';
@@ -2578,7 +2583,7 @@ static char *fts5ExprPrintTcl(
       for(iTerm=0; zRet && iTerm<pPhrase->nTerm; iTerm++){
         Fts5ExprTerm *p = &pPhrase->aTerm[iTerm];
         zRet = fts5PrintfAppend(zRet, "%s%.*s", iTerm==0?"":" ", 
-            p->nTerm, p->pTerm
+            p->nQueryTerm, p->pTerm
         );
         if( pPhrase->aTerm[iTerm].bPrefix ){
           zRet = fts5PrintfAppend(zRet, "*");
@@ -2997,11 +3002,11 @@ static int fts5ExprPopulatePoslistsCb(
   if( nToken>FTS5_MAX_TOKEN_SIZE ) nToken = FTS5_MAX_TOKEN_SIZE;
   if( (tflags & FTS5_TOKEN_COLOCATED)==0 ) p->iOff++;
   for(i=0; i<pExpr->nPhrase; i++){
-    Fts5ExprTerm *pTerm;
+    Fts5ExprTerm *pT;
     if( p->aPopulator[i].bOk==0 ) continue;
-    for(pTerm=&pExpr->apExprPhrase[i]->aTerm[0]; pTerm; pTerm=pTerm->pSynonym){
-      if( (pTerm->nTerm==nToken || (pTerm->nTerm<nToken && pTerm->bPrefix))
-       && memcmp(pTerm->pTerm, pToken, pTerm->nTerm)==0
+    for(pT=&pExpr->apExprPhrase[i]->aTerm[0]; pT; pT=pT->pSynonym){
+      if( (pT->nFullTerm==nToken || (pT->nFullTerm<nToken && pT->bPrefix))
+       && memcmp(pT->pTerm, pToken, pT->nFullTerm)==0
       ){
         int rc = sqlite3Fts5PoslistWriterAppend(
             &pExpr->apExprPhrase[i]->poslist, &p->aPopulator[i].writer, p->iOff

diff --git a/ext/fts5/fts5_index.c b/ext/fts5/fts5_index.c
index 6cf30b5a003ee9dec528f3f458e5eab0058d0dc5..b90a66b88f4a0b438d334418c3c751e42516006e 100644 (file)
--- a/ext/fts5/fts5_index.c
+++ b/ext/fts5/fts5_index.c
@@ -6037,11 +6037,12 @@ static void fts5MergePrefixLists(
 static void fts5SetupPrefixIter(
   Fts5Index *p,                   /* Index to read from */
   int bDesc,                      /* True for "ORDER BY rowid DESC" */
+  int bTokenscan,
   int iIdx,                       /* Index to scan for data */
   u8 *pToken,                     /* Buffer containing prefix to match */
   int nToken,                     /* Size of buffer pToken in bytes */
   Fts5Colset *pColset,            /* Restrict matches to these columns */
-  Fts5Iter **ppIter          /* OUT: New iterator */
+  Fts5Iter **ppIter               /* OUT: New iterator */
 ){
   Fts5Structure *pStruct;
   Fts5Buffer *aBuf;
@@ -6060,6 +6061,8 @@ static void fts5SetupPrefixIter(
     xAppend = fts5AppendPoslist;
   }
 
+  assert( bTokenscan==0 || iIdx==0 );
+
   aBuf = (Fts5Buffer*)fts5IdxMalloc(p, sizeof(Fts5Buffer)*nBuf);
   pStruct = fts5StructureRead(p);
 
@@ -6075,6 +6078,12 @@ static void fts5SetupPrefixIter(
     int bNewTerm = 1;
 
     memset(&doclist, 0, sizeof(doclist));
+
+    /* If iIdx is non-zero, then it is the number of a prefix-index for
+    ** prefixes 1 character longer than the prefix being queried for. That
+    ** index contains all the doclists required, except for the one
+    ** corresponding to the prefix itself. That one is extracted from the
+    ** main term index here.  */
     if( iIdx!=0 ){
       int dummy = 0;
       const int f2 = FTS5INDEX_QUERY_SKIPEMPTY|FTS5INDEX_QUERY_NOOUTPUT;
@@ -6110,6 +6119,7 @@ static void fts5SetupPrefixIter(
       assert_nc( memcmp(pToken, pTerm, MIN(nToken, nTerm))<=0 );
       if( bNewTerm ){
         if( nTerm<nToken || memcmp(pToken, pTerm, nToken) ) break;
+        if( bTokenscan && nTerm>nToken && pTerm[nToken]!=0x00 ) break;
       }
 
       if( p1->base.nData==0 ) continue;
@@ -6438,7 +6448,7 @@ int sqlite3Fts5IndexQuery(
       }
     }
 
-    if( iIdx<=pConfig->nPrefix ){
+    if( iIdx<=pConfig->nPrefix && (pConfig->bTokendata==0 || iIdx!=0) ){
       /* Straight index lookup */
       Fts5Structure *pStruct = fts5StructureRead(p);
       buf.p[0] = (u8)(FTS5_MAIN_PREFIX + iIdx);
@@ -6451,7 +6461,10 @@ int sqlite3Fts5IndexQuery(
     }else{
       /* Scan multiple terms in the main index */
       int bDesc = (flags & FTS5INDEX_QUERY_DESC)!=0;
-      fts5SetupPrefixIter(p, bDesc, iPrefixIdx, buf.p, nToken+1, pColset,&pRet);
+      int bTokenscan = (iIdx==0);
+      fts5SetupPrefixIter(
+          p, bDesc, bTokenscan, iPrefixIdx, buf.p, nToken+1, pColset, &pRet
+      );
       if( pRet==0 ){
         assert( p->rc!=SQLITE_OK );
       }else{

diff --git a/ext/fts5/test/fts5_common.tcl b/ext/fts5/test/fts5_common.tcl
index 9c012932da0c190a8a9a7fb5b783ca2d8e400bff..001cad1de23ea3cbc3c2997998ddd308bf4e50c0 100644 (file)
--- a/ext/fts5/test/fts5_common.tcl
+++ b/ext/fts5/test/fts5_common.tcl
@@ -438,6 +438,20 @@ proc detail_is_none {} { detail_check ; expr {$::detail == "none"} }
 proc detail_is_col {}  { detail_check ; expr {$::detail == "col" } }
 proc detail_is_full {} { detail_check ; expr {$::detail == "full"} }
 
+proc foreach_tokenizer_mode {prefix script} {
+  set saved $::testprefix
+  foreach {d mapping} {
+    ""              {}
+    "-origintext"   {, tokenize="origintext unicode61", tokendata=1}
+  } {
+    set s [string map [list %TOKENIZER% $mapping] $script]
+    set ::testprefix "$prefix$d"
+    reset_db
+    sqlite3_fts5_register_origintext db
+    uplevel $s
+  }
+  set ::testprefix $saved
+}
 
 #-------------------------------------------------------------------------
 # Convert a poslist of the type returned by fts5_test_poslist() to a 

diff --git a/ext/fts5/test/fts5aa.test b/ext/fts5/test/fts5aa.test
index e1551fc51602d5617e552ad0881292de97792ad9..a80a307a49cc24b1f337a4b75ccc21dbca8ba7b8 100644 (file)
--- a/ext/fts5/test/fts5aa.test
+++ b/ext/fts5/test/fts5aa.test
@@ -22,6 +22,7 @@ ifcapable !fts5 {
 }
 
 foreach_detail_mode $::testprefix {
+foreach_tokenizer_mode $::testprefix {
 
 do_execsql_test 1.0 {
   CREATE VIRTUAL TABLE t1 USING fts5(a, b, c);
@@ -44,7 +45,7 @@ do_execsql_test 1.1 {
 #
 
 do_execsql_test 2.0 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x, y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x, y, detail=%DETAIL% %TOKENIZER%);
 }
 do_execsql_test 2.1 {
   INSERT INTO t1 VALUES('a b c', 'd e f');
@@ -73,8 +74,9 @@ do_execsql_test 2.4 {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 3.0 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL% %TOKENIZER%);
 }
 foreach {i x y} {
    1  {g f d b f} {h h e i a}
@@ -97,8 +99,9 @@ foreach {i x y} {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 4.0 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
 }
 foreach {i x y} {
@@ -121,8 +124,9 @@ foreach {i x y} {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 5.0 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
 }
 foreach {i x y} {
@@ -145,8 +149,9 @@ foreach {i x y} {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 6.0 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
 }
 
@@ -181,6 +186,7 @@ do_execsql_test 6.6 {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 expr srand(0)
 do_execsql_test 7.0 {
   CREATE VIRTUAL TABLE t1 USING fts5(x,y,z);
@@ -222,6 +228,7 @@ for {set i 1} {$i <= 10} {incr i} {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 8.0 {
   CREATE VIRTUAL TABLE t1 USING fts5(x, prefix="1,2,3");
   INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
@@ -236,6 +243,7 @@ do_execsql_test 8.1 {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 
 expr srand(0)
 
@@ -280,8 +288,9 @@ for {set i 1} {$i <= 10} {incr i} {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 10.0 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x,y, detail=%DETAIL% %TOKENIZER%);
 }
 set d10 {
    1  {g f d b f} {h h e i a}
@@ -314,19 +323,19 @@ do_execsql_test 10.4.2 { INSERT INTO t1(t1) VALUES('integrity-check') }
 #-------------------------------------------------------------------------
 #
 do_catchsql_test 11.1 {
-  CREATE VIRTUAL TABLE t2 USING fts5(a, b, c, rank, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t2 USING fts5(a, b, c, rank, detail=%DETAIL% %TOKENIZER%);
 } {1 {reserved fts5 column name: rank}}
 do_catchsql_test 11.2 {
-  CREATE VIRTUAL TABLE rank USING fts5(a, b, c, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE rank USING fts5(a, b, c, detail=%DETAIL% %TOKENIZER%);
 } {1 {reserved fts5 table name: rank}}
 do_catchsql_test 11.3 {
-  CREATE VIRTUAL TABLE t2 USING fts5(a, b, c, rowid, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t2 USING fts5(a, b, c, rowid, detail=%DETAIL% %TOKENIZER%);
 } {1 {reserved fts5 column name: rowid}}
 
 #-------------------------------------------------------------------------
 #
 do_execsql_test 12.1 {
-  CREATE VIRTUAL TABLE t2 USING fts5(x,y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t2 USING fts5(x,y, detail=%DETAIL% %TOKENIZER%);
 } {}
 
 do_catchsql_test 12.2 {
@@ -341,8 +350,9 @@ do_test 12.3 {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 13.1 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1(rowid, x) VALUES(1, 'o n e'), (2, 't w o');
 } {}
 
@@ -365,8 +375,9 @@ do_execsql_test 13.6 {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 14.1 {
-  CREATE VIRTUAL TABLE t1 USING fts5(x, y, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(x, y, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
   WITH d(x,y) AS (
     SELECT NULL, 'xyz xyz xyz xyz xyz xyz'
@@ -449,8 +460,9 @@ do_catchsql_test 16.2 {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 17.1 {
-  CREATE VIRTUAL TABLE b2 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE b2 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO b2 VALUES('a');
   INSERT INTO b2 VALUES('b');
   INSERT INTO b2 VALUES('c');
@@ -466,8 +478,9 @@ do_test 17.2 {
 
 if {[string match n* %DETAIL%]==0} {
   reset_db
+  sqlite3_fts5_register_origintext db
   do_execsql_test 17.3 {
-    CREATE VIRTUAL TABLE c2 USING fts5(x, y, detail=%DETAIL%);
+    CREATE VIRTUAL TABLE c2 USING fts5(x, y, detail=%DETAIL% %TOKENIZER%);
     INSERT INTO c2 VALUES('x x x', 'x x x');
     SELECT rowid FROM c2 WHERE c2 MATCH 'y:x';
   } {1}
@@ -476,8 +489,9 @@ if {[string match n* %DETAIL%]==0} {
 #-------------------------------------------------------------------------
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 17.1 {
-  CREATE VIRTUAL TABLE uio USING fts5(ttt, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE uio USING fts5(ttt, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO uio VALUES(NULL);
   INSERT INTO uio SELECT NULL FROM uio;
   INSERT INTO uio SELECT NULL FROM uio;
@@ -524,8 +538,8 @@ do_execsql_test 17.9 {
 #--------------------------------------------------------------------
 #
 do_execsql_test 18.1 {
-  CREATE VIRTUAL TABLE t1 USING fts5(a, b, detail=%DETAIL%);
-  CREATE VIRTUAL TABLE t2 USING fts5(c, d, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t1 USING fts5(a, b, detail=%DETAIL% %TOKENIZER%);
+  CREATE VIRTUAL TABLE t2 USING fts5(c, d, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1 VALUES('abc*', NULL);
   INSERT INTO t2 VALUES(1, 'abcdefg');
 }
@@ -540,8 +554,9 @@ do_execsql_test 18.3 {
 # fts5 table in the temp schema.
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 19.0 {
-  CREATE VIRTUAL TABLE temp.t1 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE temp.t1 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t1 VALUES('x y z');
   INSERT INTO t1 VALUES('w x 1');
   SELECT rowid FROM t1 WHERE t1 MATCH 'x';
@@ -551,8 +566,9 @@ do_execsql_test 19.0 {
 # Test that 6 and 7 byte varints can be read.
 #
 reset_db
+sqlite3_fts5_register_origintext db
 do_execsql_test 20.0 {
-  CREATE VIRTUAL TABLE temp.tmp USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE temp.tmp USING fts5(x, detail=%DETAIL% %TOKENIZER%);
 }
 set ::ids [list \
   0 [expr 1<<36] [expr 2<<36] [expr 1<<43] [expr 2<<43]
@@ -570,7 +586,7 @@ do_test 20.1 {
 #
 do_execsql_test 21.0 {
   CREATE TEMP TABLE t8(a, b);
-  CREATE VIRTUAL TABLE ft USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE ft USING fts5(x, detail=%DETAIL% %TOKENIZER%);
 }
 
 do_execsql_test 21.1 {
@@ -581,7 +597,7 @@ do_execsql_test 21.1 {
 }
 
 do_execsql_test 22.0 {
-  CREATE VIRTUAL TABLE t9 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t9 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t9(rowid, x) VALUES(2, 'bbb');
   BEGIN;
     INSERT INTO t9(rowid, x) VALUES(1, 'aaa');
@@ -596,7 +612,7 @@ do_execsql_test 22.1 {
 
 #-------------------------------------------------------------------------
 do_execsql_test 23.0 {
-  CREATE VIRTUAL TABLE t10 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t10 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
   CREATE TABLE t11(x);
 }
 do_execsql_test 23.1 {
@@ -608,7 +624,7 @@ do_execsql_test 23.2 {
 
 #-------------------------------------------------------------------------
 do_execsql_test 24.0 {
-  CREATE VIRTUAL TABLE t12 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t12 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
   INSERT INTO t12 VALUES('aaaa');
 }
 do_execsql_test 24.1 {
@@ -618,6 +634,9 @@ do_execsql_test 24.1 {
     INSERT INTO t12 VALUES('aaaa');
   END;
 }
+execsql_pp {
+  SELECT rowid, hex(block) FROM t12_data
+}
 do_execsql_test 24.2 {
   INSERT INTO t12(t12) VALUES('integrity-check');
 }
@@ -627,7 +646,7 @@ do_execsql_test 24.3 {
 
 #-------------------------------------------------------------------------
 do_execsql_test 25.0 {
-  CREATE VIRTUAL TABLE t13 USING fts5(x, detail=%DETAIL%);
+  CREATE VIRTUAL TABLE t13 USING fts5(x, detail=%DETAIL% %TOKENIZER%);
 }
 do_execsql_test 25.1 {
   BEGIN;
@@ -638,6 +657,7 @@ SELECT * FROM t13('BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
 }
 
 
+}
 }
 
 expand_all_sql db

diff --git a/ext/fts5/test/fts5origintext.test b/ext/fts5/test/fts5origintext.test
index 155d74c0257b49818573a85b86eb9fbfa5944c29..3fd5f17e7bd0536d1ca967a76c478507950e33a3 100644 (file)
--- a/ext/fts5/test/fts5origintext.test
+++ b/ext/fts5/test/fts5origintext.test
@@ -112,9 +112,50 @@ do_execsql_test 2.5 {
   INSERT INTO ft(ft) VALUES('optimize');
 }
 
-proc b {x} { string map [list "\0" "."] $x }
-db func b b
+#-------------------------------------------------------------------------
+reset_db
+
+sqlite3_fts5_register_origintext db
+do_execsql_test 3.0 {
+  CREATE VIRTUAL TABLE ft USING fts5(x, tokenize="origintext unicode61");
+  CREATE VIRTUAL TABLE vocab USING fts5vocab(ft, instance);
+
+  INSERT INTO ft(rowid, x) VALUES(1, 'hello');
+  INSERT INTO ft(rowid, x) VALUES(2, 'Hello');
+  INSERT INTO ft(rowid, x) VALUES(3, 'HELLO');
+}
+
+#proc b {x} { string map [list "\0" "."] $x }
+#db func b b
+#execsql_pp { SELECT b(term) FROM vocab }
+
+do_execsql_test 3.1.1 { SELECT rowid FROM ft('hello') } 1
+do_execsql_test 3.1.2 { SELECT rowid FROM ft('Hello') } 2
+do_execsql_test 3.1.3 { SELECT rowid FROM ft('HELLO') } 3
+
+do_execsql_test 3.0 {
+  CREATE VIRTUAL TABLE ft2 USING fts5(x, 
+      tokenize="origintext unicode61", 
+      tokendata=1
+  );
+  CREATE VIRTUAL TABLE vocab2 USING fts5vocab(ft2, instance);
+
+  INSERT INTO ft2(rowid, x) VALUES(1, 'hello');
+  INSERT INTO ft2(rowid, x) VALUES(2, 'Hello');
+  INSERT INTO ft2(rowid, x) VALUES(3, 'HELLO');
+
+  INSERT INTO ft2(rowid, x) VALUES(10, 'helloooo');
+}
+
+#proc b {x} { string map [list "\0" "."] $x }
+#db func b b
 #execsql_pp { SELECT b(term) FROM vocab }
 
+do_execsql_test 3.1.1 { SELECT rowid FROM ft2('hello') } {1 2 3}
+do_execsql_test 3.1.2 { SELECT rowid FROM ft2('Hello') } {1 2 3}
+do_execsql_test 3.1.3 { SELECT rowid FROM ft2('HELLO') } {1 2 3}
+
+do_execsql_test 3.1.4 { SELECT rowid FROM ft2('hello*') } {1 2 3 10}
+
 finish_test
 

diff --git a/manifest b/manifest
index e0c37bf1e275738ab7c64654795143b8337106fc..92f3bfcf68eb6ea53efe89086ca4d9ab96e7b65b 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Fixes\sfor\sfts5\sexpression\sparser\smodule\sto\sallow\sembedded\s0x00\sbytes\sin\stokens.
-D 2023-10-03T19:06:52.966
+C Add\sthe\stokendata=1\soption\sto\signore\strailing\stoken-data\swhen\squerying\san\sfts5\stable.
+D 2023-10-11T21:08:12.656
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724
@@ -88,13 +88,13 @@ F ext/fts3/unicode/mkunicode.tcl d5aebf022fa4577ee8cdf27468f0d847879993959101f6d
 F ext/fts3/unicode/parseunicode.tcl a981bd6466d12dd17967515801c3ff23f74a281be1a03cf1e6f52a6959fc77eb
 F ext/fts5/extract_api_docs.tcl a36e54ec777172ddd3f9a88daf593b00848368e0
 F ext/fts5/fts5.h 05501612cc655504c5dce8ba765ab621d50fc478490089beaa0d75e00b23e520
-F ext/fts5/fts5Int.h 66a38b285e2b860baa29745d8eff27f5b0809268e7820498494d9acfaccf8a5c
+F ext/fts5/fts5Int.h a21eb1cf036ac9eb943e45ed307762901ea86f0159bf0848baa2079a112ddc2f
 F ext/fts5/fts5_aux.c 572d5ec92ba7301df2fea3258576332f2f4d2dfd66d8263afd157d9deceac480
 F ext/fts5/fts5_buffer.c 3001fbabb585d6de52947b44b455235072b741038391f830d6b729225eeaf6a5
-F ext/fts5/fts5_config.c 054359543566cbff1ba65a188330660a5457299513ac71c53b3a07d934c7b081
-F ext/fts5/fts5_expr.c cc215d39714b428523d2f2ef42b713c83095a28a67bc7f6f2dc4ac036a29f460
+F ext/fts5/fts5_config.c 8072a207034b51ae9b7694121d1b5715c794e94b275e088f70ae532378ca5cdf
+F ext/fts5/fts5_expr.c fd091d0558fda2517602ed5886ec615ce3e1bd76fb0bb0e5d1aa85ba8db287a8
 F ext/fts5/fts5_hash.c 76765856397eff56f526b0640b23a1677d737d35e07bc00e4b4b2e0fc5fda60d
-F ext/fts5/fts5_index.c e472083d371f420d52ec80445b9d2a99b16b23548205cb4064ddcd41bd79f63e
+F ext/fts5/fts5_index.c 79a8e45771d0be24f0399b12268299f132ce0970ade941ba8a2d40b1d1aee4d7
 F ext/fts5/fts5_main.c 799ec88d2309055f6406bddb0bd6ed80148c5da5eb14594c3c5309a6e944d489
 F ext/fts5/fts5_storage.c 3c9b41fce41b6410f2e8f82eb035c6a29b2560483f773e6dc98cf3cb2e4ddbb5
 F ext/fts5/fts5_tcl.c 0d2bb0ff7bf6ee136015be118167f0bd956ddd05a8f02c68bd34299b50648f9f
@@ -106,8 +106,8 @@ F ext/fts5/fts5_varint.c e64d2113f6e1bfee0032972cffc1207b77af63319746951bf1d0988
 F ext/fts5/fts5_vocab.c 12138e84616b56218532e3e8feb1d3e0e7ae845e33408dbe911df520424dc9d6
 F ext/fts5/fts5parse.y eb526940f892ade5693f22ffd6c4f2702543a9059942772526eac1fde256bb05
 F ext/fts5/mkportersteps.tcl 5acf962d2e0074f701620bb5308155fa1e4a63ba
-F ext/fts5/test/fts5_common.tcl a9de9c2209cc4e7ae3c753e783504e67206c6c1467d08f209cd0c5923d3e8d8b
-F ext/fts5/test/fts5aa.test ba5158eba7d61359becdfca895ef471072c7bf7b20e5e60dcb4d024c8419c926
+F ext/fts5/test/fts5_common.tcl 8b1848ac2baad10e444e4183034a52050b52d20b3796d9d30e78f01ab0d05583
+F ext/fts5/test/fts5aa.test 4db81519863244a3cab35795fe65ab6b592e7970c7409eba098b23ebbfc08d95
 F ext/fts5/test/fts5ab.test bd932720c748383277456b81f91bc00453de2174f9762cd05f95d0495dc50390
 F ext/fts5/test/fts5ac.test a7aa7e1fefc6e1918aa4d3111d5c44a09177168e962c5fd2cca9620de8a7ed6d
 F ext/fts5/test/fts5ad.test e8cf959dfcd57c8e46d6f5f25665686f3b6627130a9a981371dafdf6482790de
@@ -187,7 +187,7 @@ F ext/fts5/test/fts5onepass.test f9b7d9b2c334900c6542a869760290e2ab5382af8fbd618
 F ext/fts5/test/fts5optimize.test 36a752d24c818792032e4ff502936fc9cc5ef938721696396fdc79214b2717f1
 F ext/fts5/test/fts5optimize2.test 93e742c36b487d8874621360af5b1ce4d39b04fb9e71ce9bc34015c5fc811785
 F ext/fts5/test/fts5optimize3.test bf9c91bb927d0fb2b9a06318a217a0419183ac5913842e062c7e0b98ea5d0fca
-F ext/fts5/test/fts5origintext.test 3e1ac3230f65a0d644e9bf0738bebb09b4db9d9f123e1307d8630e42269b4afb
+F ext/fts5/test/fts5origintext.test 646df137f1aa5b3d7032374ebe82bfdbe88d9f825d73ce8d44bead480317a9c5
 F ext/fts5/test/fts5phrase.test 13e5d8e9083077b3d9c74315b3c92ec723cc6eb37c8155e0bfe1bba00559f07b
 F ext/fts5/test/fts5plan.test b65cfcca9ddd6fdaa118c61e17aeec8e8433bc5b6bb307abd116514f79c49c5a
 F ext/fts5/test/fts5porter.test 8d08010c28527db66bc3feebd2b8767504aaeb9b101a986342fa7833d49d0d15
@@ -2123,8 +2123,8 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P e051120067fd87f57b498e505e3960cf4d14e8e33bad940618cc0823253254f7
-R 5cce41f02eae121cb66e72942ef56113
+P 342c8d0783f449817d3f565ff6b9f010a6c690beeea32f1861640810490a8b5f
+R 28168382a793dd4cf732de7d6442ef72
 U dan
-Z 80b8e2664e768ed2ac03913fcf0180ea
+Z cded15b3b6aeefc28aa13f7856e47e8d
 # Remove this line to create a well-formed Fossil manifest.

diff --git a/manifest.uuid b/manifest.uuid
index 4798938837a0ca957af4f6595e2763256ef47672..8bc375eea26aebb138ba577b466affc62e04682f 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-342c8d0783f449817d3f565ff6b9f010a6c690beeea32f1861640810490a8b5f
\ No newline at end of file
+122935182ad5869ce3a4c6d796c38a0509f6f3384dd1b3e60a3f2f0f366cc5f5
\ No newline at end of file