Be a bit more susicious of invalid results from the tokenizer.

Backports check-in (4514) from fts3. (CVS 5459)

FossilOrigin-Name: 311aeb9c2b75c420a37198a93e353c72e9166747

diff --git a/ext/fts2/fts2.c b/ext/fts2/fts2.c
index a8425b680b7e98406e558d48a79b66b3495b7884..d3a3596b8e9bf60ca5f96b11611fb5fbd82dacd2 100644 (file)
--- a/ext/fts2/fts2.c
+++ b/ext/fts2/fts2.c
@@ -3883,17 +3883,18 @@ static int buildTerms(fulltext_vtab *v, sqlite_int64 iDocid,
   if( rc!=SQLITE_OK ) return rc;
 
   pCursor->pTokenizer = pTokenizer;
-  while( SQLITE_OK==pTokenizer->pModule->xNext(pCursor,
-                                               &pToken, &nTokenBytes,
-                                               &iStartOffset, &iEndOffset,
-                                               &iPosition) ){
+  while( SQLITE_OK==(rc=pTokenizer->pModule->xNext(pCursor,
+                                                   &pToken, &nTokenBytes,
+                                                   &iStartOffset, &iEndOffset,
+                                                   &iPosition)) ){
     DLCollector *p;
     int nData;                   /* Size of doclist before our update. */
 
-    /* Positions can't be negative; we use -1 as a terminator internally. */
-    if( iPosition<0 ){
-      pTokenizer->pModule->xClose(pCursor);
-      return SQLITE_ERROR;
+    /* Positions can't be negative; we use -1 as a terminator
+     * internally.  Token can't be NULL or empty. */
+    if( iPosition<0 || pToken == NULL || nTokenBytes == 0 ){
+      rc = SQLITE_ERROR;
+      break;
     }
 
     p = fts2HashFind(&v->pendingTerms, pToken, nTokenBytes);
@@ -3922,6 +3923,7 @@ static int buildTerms(fulltext_vtab *v, sqlite_int64 iDocid,
   ** not durable.  *ponder*
   */
   pTokenizer->pModule->xClose(pCursor);
+  if( SQLITE_DONE == rc ) return SQLITE_OK;
   return rc;
 }
 

diff --git a/manifest b/manifest
index 0759a3df803d16596e90f635ad0e80a57433cbae..16f4b2b68cc0eda298194501510197a2be4c08e5 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Implement\soptimize()\sfunction.\r\nBackports\scheck-in\s(5417)\sfrom\sfts3.\s(CVS\s5458)
-D 2008-07-22T23:49:44
+C Be\sa\sbit\smore\ssusicious\sof\sinvalid\sresults\sfrom\sthe\stokenizer.\r\nBackports\scheck-in\s(4514)\sfrom\sfts3.\s(CVS\s5459)
+D 2008-07-22T23:54:51
 F Makefile.arm-wince-mingw32ce-gcc fcd5e9cd67fe88836360bb4f9ef4cb7f8e2fb5a0
 F Makefile.in 77ff156061bb870aa0a8b3d545c670d08070f7e6
 F Makefile.linux-gcc d53183f4aa6a9192d249731c90dbdffbd2c68654
@@ -39,7 +39,7 @@ F ext/fts1/simple_tokenizer.c 1844d72f7194c3fd3d7e4173053911bf0661b70d
 F ext/fts1/tokenizer.h 0c53421b832366d20d720d21ea3e1f6e66a36ef9
 F ext/fts2/README.tokenizers 21e3684ea5a095b55d70f6878b4ce6af5932dfb7
 F ext/fts2/README.txt 8c18f41574404623b76917b9da66fcb0ab38328d
-F ext/fts2/fts2.c ff1d7646d48810d08b5a081be3495b16c3a603f1
+F ext/fts2/fts2.c 5f6f8fa8f756d37a5eb271ea42e56a103f965ef5
 F ext/fts2/fts2.h da5f76c65163301d1068a971fd32f4119e3c95fa
 F ext/fts2/fts2_hash.c 2689e42e1107ea67207f725cf69cf8972d00cf93
 F ext/fts2/fts2_hash.h 9a5b1be94664139f93217a0770d7144425cffb3a
@@ -611,7 +611,7 @@ F tool/speedtest16.c c8a9c793df96db7e4933f0852abb7a03d48f2e81
 F tool/speedtest2.tcl ee2149167303ba8e95af97873c575c3e0fab58ff
 F tool/speedtest8.c 1dbced29de5f59ba2ebf877edcadf171540374d1
 F tool/speedtest8inst1.c 293327bc76823f473684d589a8160bde1f52c14e
-P 4c98179be258319f441ae4e123cf59af77e96409
-R 7d4ed449896aa36fee287e4efa2f50e0
+P c16900dc7603cab30f8729b25361bc88bb37ae43
+R 2823196dc71e81c3c7ae664b26b8eaee
 U shess
-Z 5157ad4962dd96eaaf179dc694e76db5
+Z 401522a3652c6757103a76d2da0f5923

diff --git a/manifest.uuid b/manifest.uuid
index 19de4acade17bc513afb3f68bf19bd40c11540c5..e38f0fcbbf5dfa8694e0f4bbdc3e136fc7fd7582 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-c16900dc7603cab30f8729b25361bc88bb37ae43
\ No newline at end of file
+311aeb9c2b75c420a37198a93e353c72e9166747
\ No newline at end of file