Speed up parsing of very long fts3 query expressions.

FossilOrigin-Name: 2dd5b6895a3b23c2b9cbf0c1c1e802faf8f2b41ef60819eea25d609755266e64

diff --git a/ext/fts3/fts3_expr.c b/ext/fts3/fts3_expr.c
index ca857835e2edc89ced7e340b55bc26f0ac394ef7..1372cd933bc8654300b110c730755b1b3b0888a6 100644 (file)
--- a/ext/fts3/fts3_expr.c
+++ b/ext/fts3/fts3_expr.c
@@ -161,6 +161,23 @@ int sqlite3Fts3OpenTokenizer(
 */
 static int fts3ExprParse(ParseContext *, const char *, int, Fts3Expr **, int *);
 
+/*
+** Search buffer z[], size n, for a '"' character. Or, if enable_parenthesis
+** is defined, search for '(' and ')' as well. Return the index of the first
+** such character in the buffer. If there is no such character, return -1.
+*/
+static int findBarredChar(const char *z, int n){
+  int ii;
+  for(ii=0; ii<n; ii++){
+    if( (z[ii]=='"')
+     || (sqlite3_fts3_enable_parentheses && (z[ii]=='(' || z[ii]==')'))
+    ){
+      return ii;
+    }
+  }
+  return -1;
+}
+
 /*
 ** Extract the next token from buffer z (length n) using the tokenizer
 ** and other information (column names etc.) in pParse. Create an Fts3Expr
@@ -185,16 +202,9 @@ static int getNextToken(
   int rc;
   sqlite3_tokenizer_cursor *pCursor;
   Fts3Expr *pRet = 0;
-  int i = 0;
-
-  /* Set variable i to the maximum number of bytes of input to tokenize. */
-  for(i=0; i<n; i++){
-    if( sqlite3_fts3_enable_parentheses && (z[i]=='(' || z[i]==')') ) break;
-    if( z[i]=='"' ) break;
-  }
 
-  *pnConsumed = i;
-  rc = sqlite3Fts3OpenTokenizer(pTokenizer, pParse->iLangid, z, i, &pCursor);
+  *pnConsumed = n;
+  rc = sqlite3Fts3OpenTokenizer(pTokenizer, pParse->iLangid, z, n, &pCursor);
   if( rc==SQLITE_OK ){
     const char *zToken;
     int nToken = 0, iStart = 0, iEnd = 0, iPosition = 0;
@@ -202,6 +212,18 @@ static int getNextToken(
 
     rc = pModule->xNext(pCursor, &zToken, &nToken, &iStart, &iEnd, &iPosition);
     if( rc==SQLITE_OK ){
+
+      /* Check that this tokenization did not gobble up any " characters. Or,
+      ** if enable_parenthesis is true, that it did not gobble up any 
+      ** open or close parenthesis characters either. If it did, call
+      ** getNextToken() again, but pass only that part of the input buffer
+      ** up to the first such character.  */
+      int iBarred = findBarredChar(z, iEnd);
+      if( iBarred>=0 ){
+        pModule->xClose(pCursor);
+        return getNextToken(pParse, iCol, z, iBarred, ppExpr, pnConsumed);
+      }
+
       nByte = sizeof(Fts3Expr) + sizeof(Fts3Phrase) + nToken;
       pRet = (Fts3Expr *)sqlite3Fts3MallocZero(nByte);
       if( !pRet ){
@@ -236,7 +258,11 @@ static int getNextToken(
 
       }
       *pnConsumed = iEnd;
-    }else if( i && rc==SQLITE_DONE ){
+    }else if( n && rc==SQLITE_DONE ){
+      int iBarred = findBarredChar(z, n);
+      if( iBarred>=0 ){
+        *pnConsumed = iBarred;
+      }
       rc = SQLITE_OK;
     }
 
@@ -1239,7 +1265,6 @@ static void fts3ExprTestCommon(
   }
 
   if( rc!=SQLITE_OK && rc!=SQLITE_NOMEM ){
-    sqlite3Fts3ExprFree(pExpr);
     sqlite3_result_error(context, "Error parsing expression", -1);
   }else if( rc==SQLITE_NOMEM || !(zBuf = exprToString(pExpr, 0)) ){
     sqlite3_result_error_nomem(context);

diff --git a/manifest b/manifest
index 3b99e06a1dd0ba2e7510ecad33168bce7d3f24b5..07cc260a22160fe277dbebbfaae3d20d013a27f0 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Configure-internal\sbuild\scleanups\s(no\sfunctional\schanges).\sAdd\sEXTRA_SRC\sto\sthe\sdeps\sof\ssqlite3.c.
-D 2025-03-15T15:19:42.945
+C Speed\sup\sparsing\sof\svery\slong\sfts3\squery\sexpressions.
+D 2025-03-15T16:58:39.639
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md e108e1e69ae8e8a59e93c455654b8ac9356a11720d3345df2a4743e9590fb20d
@@ -82,7 +82,7 @@ F ext/fts3/fts3.c fd2a8642fa4701ef5dd6bce7947ecb3c7ae472e1d44022772454a8b74a1315
 F ext/fts3/fts3.h 3a10a0af180d502cecc50df77b1b22df142817fe
 F ext/fts3/fts3Int.h 75b9cf37c93d3c56d8e569d64527c927cb54a5279afb3823740ca1e29e481c15
 F ext/fts3/fts3_aux.c 7eab82a9cf0830f6551ba3abfdbe73ed39e322a4d3940ee82fbf723674ecd9f3
-F ext/fts3/fts3_expr.c 673bf600655f5080239ff0e1e80eaae0176389f7e7d3af54c6d51491280ca360
+F ext/fts3/fts3_expr.c b8ff0d3775f33eddae559b444df2000d48768d7d9bdb642e3f6434c9f2543ffc
 F ext/fts3/fts3_hash.c d9dba473741445789330c7513d4f65737c92df23c3212784312931641814672a
 F ext/fts3/fts3_hash.h 39cf6874dc239d6b4e30479b1975fe5b22a3caaf
 F ext/fts3/fts3_icu.c 305ce7fb6036484085b5556a9c8e62acdc7763f0f4cdf5fd538212a9f3720116
@@ -2213,8 +2213,8 @@ F tool/version-info.c 3b36468a90faf1bbd59c65fd0eb66522d9f941eedd364fabccd7227350
 F tool/warnings-clang.sh bbf6a1e685e534c92ec2bfba5b1745f34fb6f0bc2a362850723a9ee87c1b31a7
 F tool/warnings.sh 49a486c5069de041aedcbde4de178293e0463ae9918ecad7539eedf0ec77a139
 F tool/win/sqlite.vsix deb315d026cc8400325c5863eef847784a219a2f
-P fa6f6ccdffc50024624306900efd2538c7415d8bdd0f02835b2e9c05adab3cf1
-R fa3f42bd795b87d6c693892aa92ef559
-U stephan
-Z 60eb85208863d91adaf5e3aa8394c9ef
+P 8afb8bbce8654d6f76207fb136e79dc52b6724a71eae82a4c098690a68eb75a1
+R 1bd8f3cfa90570231383a500e0910fe3
+U dan
+Z 1c83f58e6b1181c1bb790bbf7879f1cb
 # Remove this line to create a well-formed Fossil manifest.

diff --git a/manifest.uuid b/manifest.uuid
index a3c26300555d554c0c2b34c8d3ada8c99b071ac8..a2320ead9e384ac9fd7f43359e25238cd882df92 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-8afb8bbce8654d6f76207fb136e79dc52b6724a71eae82a4c098690a68eb75a1
+2dd5b6895a3b23c2b9cbf0c1c1e802faf8f2b41ef60819eea25d609755266e64