Improve the performance of docid merges in fts5.

FossilOrigin-Name: b2de77a01cc5edcea2f98f7916e64cb33e6bd414

diff --git a/ext/fts5/fts5Int.h b/ext/fts5/fts5Int.h
index 074a2713f4f06952f6688c4f3f1d1674b05dd219..149f6b669460d7873a89d4d4c2bb6fc278a21c25 100644 (file)
--- a/ext/fts5/fts5Int.h
+++ b/ext/fts5/fts5Int.h
@@ -38,6 +38,12 @@ typedef sqlite3_uint64 u64;
 #define MIN(x,y) (((x) < (y)) ? (x) : (y))
 #define MAX(x,y) (((x) > (y)) ? (x) : (y))
 
+/*
+** Constants for the largest and smallest possible 64-bit signed integers.
+*/
+# define LARGEST_INT64  (0xffffffff|(((i64)0x7fffffff)<<32))
+# define SMALLEST_INT64 (((i64)-1) - LARGEST_INT64)
+
 #endif
 
 

diff --git a/ext/fts5/fts5_index.c b/ext/fts5/fts5_index.c
index f2f68baa93ff4dc44f9e63641cac84debdff6869..0b8f137ebfc422123ae271d943ea9a1984a6e858 100644 (file)
--- a/ext/fts5/fts5_index.c
+++ b/ext/fts5/fts5_index.c
@@ -514,6 +514,8 @@ struct Fts5IndexIter {
   int bRev;                       /* True to iterate in reverse order */
   int bSkipEmpty;                 /* True to skip deleted entries */
   int bEof;                       /* True at EOF */
+
+  i64 iSwitchRowid;               /* Firstest rowid of other than aFirst[1] */
   Fts5CResult *aFirst;            /* Current merge state (see above) */
   Fts5SegIter aSeg[1];            /* Array of segment iterators */
 };
@@ -2499,9 +2501,21 @@ static void fts5AssertComparisonResult(
 */
 static void fts5AssertMultiIterSetup(Fts5Index *p, Fts5IndexIter *pIter){
   if( p->rc==SQLITE_OK ){
+    Fts5SegIter *pFirst = &pIter->aSeg[ pIter->aFirst[1].iFirst ];
     int i;
 
-    assert( (pIter->aSeg[ pIter->aFirst[1].iFirst ].pLeaf==0)==pIter->bEof );
+    assert( (pFirst->pLeaf==0)==pIter->bEof );
+
+    /* Check that pIter->iSwitchRowid is set correctly. */
+    for(i=0; i<pIter->nSeg; i++){
+      Fts5SegIter *p1 = &pIter->aSeg[i];
+      assert( p1==pFirst 
+           || p1->pLeaf==0 
+           || fts5BufferCompare(&pFirst->term, &p1->term) 
+           || p1->iRowid==pIter->iSwitchRowid
+           || (p1->iRowid<pIter->iSwitchRowid)==pIter->bRev
+      );
+    }
 
     for(i=0; i<pIter->nSeg; i+=2){
       Fts5SegIter *p1 = &pIter->aSeg[i];
@@ -2719,27 +2733,35 @@ static int fts5MultiIterAdvanceRowid(
   Fts5IndexIter *pIter,           /* Iterator to update aFirst[] array for */
   int iChanged                    /* Index of sub-iterator just advanced */
 ){
-  int i;
   Fts5SegIter *pNew = &pIter->aSeg[iChanged];
-  Fts5SegIter *pOther = &pIter->aSeg[iChanged ^ 0x0001];
-
-  for(i=(pIter->nSeg+iChanged)/2; 1; i=i/2){
-    Fts5CResult *pRes = &pIter->aFirst[i];
-
-    assert( pNew->pLeaf );
-    assert( pRes->bTermEq==0 || pOther->pLeaf );
-    
-    if( pRes->bTermEq ){
-      if( pNew->iRowid==pOther->iRowid ){
-        return 1;
-      }else if( (pOther->iRowid>pNew->iRowid)==pIter->bRev ){
-        pNew = pOther;
+
+  if( pNew->iRowid==pIter->iSwitchRowid
+   || (pNew->iRowid<pIter->iSwitchRowid)==pIter->bRev
+  ){
+    int i;
+    Fts5SegIter *pOther = &pIter->aSeg[iChanged ^ 0x0001];
+    pIter->iSwitchRowid = pIter->bRev ? SMALLEST_INT64 : LARGEST_INT64;
+    for(i=(pIter->nSeg+iChanged)/2; 1; i=i/2){
+      Fts5CResult *pRes = &pIter->aFirst[i];
+
+      assert( pNew->pLeaf );
+      assert( pRes->bTermEq==0 || pOther->pLeaf );
+
+      if( pRes->bTermEq ){
+        if( pNew->iRowid==pOther->iRowid ){
+          return 1;
+        }else if( (pOther->iRowid>pNew->iRowid)==pIter->bRev ){
+          pIter->iSwitchRowid = pOther->iRowid;
+          pNew = pOther;
+        }else if( (pOther->iRowid>pIter->iSwitchRowid)==pIter->bRev ){
+          pIter->iSwitchRowid = pOther->iRowid;
+        }
       }
-    }
-    pRes->iFirst = (pNew - pIter->aSeg);
-    if( i==1 ) break;
+      pRes->iFirst = (pNew - pIter->aSeg);
+      if( i==1 ) break;
 
-    pOther = &pIter->aSeg[ pIter->aFirst[i ^ 0x0001].iFirst ];
+      pOther = &pIter->aSeg[ pIter->aFirst[i ^ 0x0001].iFirst ];
+    }
   }
 
   return 0;
@@ -2749,7 +2771,9 @@ static int fts5MultiIterAdvanceRowid(
 ** Set the pIter->bEof variable based on the state of the sub-iterators.
 */
 static void fts5MultiIterSetEof(Fts5IndexIter *pIter){
-  pIter->bEof = pIter->aSeg[ pIter->aFirst[1].iFirst ].pLeaf==0;
+  Fts5SegIter *pSeg = &pIter->aSeg[ pIter->aFirst[1].iFirst ];
+  pIter->bEof = pSeg->pLeaf==0;
+  pIter->iSwitchRowid = pSeg->iRowid;
 }
 
 /*

diff --git a/ext/fts5/fts5_main.c b/ext/fts5/fts5_main.c
index 05ae3444795722a92416cc97562dfbe69bb7e14d..ccb94bd8d6dfcb6ac2e9bfd1cde4a1a85b4c7168 100644 (file)
--- a/ext/fts5/fts5_main.c
+++ b/ext/fts5/fts5_main.c
@@ -227,14 +227,6 @@ struct Fts5Cursor {
 #define BitFlagAllTest(x,y) (((x) & (y))==(y))
 #define BitFlagTest(x,y)    (((x) & (y))!=0)
 
-/*
-** Constants for the largest and smallest possible 64-bit signed integers.
-** These are copied from sqliteInt.h.
-*/
-#ifndef SQLITE_AMALGAMATION
-# define LARGEST_INT64  (0xffffffff|(((i64)0x7fffffff)<<32))
-# define SMALLEST_INT64 (((i64)-1) - LARGEST_INT64)
-#endif
 
 /*
 ** Macros to Set(), Clear() and Test() cursor flags.

diff --git a/manifest b/manifest
index 78ab54d28af42cedd45c9f5bd63a4bba1dd42092..56929d7dc1ae0445a3594bf7966c5253ac6ea051 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Reduce\sthe\snumber\sof\scalls\sto\smalloc()\smade\sby\sfts5.
-D 2015-07-09T19:02:19.523
+C Improve\sthe\sperformance\sof\sdocid\smerges\sin\sfts5.
+D 2015-07-09T20:46:35.829
 F Makefile.arm-wince-mingw32ce-gcc d6df77f1f48d690bd73162294bbba7f59507c72f
 F Makefile.in 017bf0511d1b2dd1db5e16488fbf75a17b526cbc
 F Makefile.linux-gcc 91d710bdc4998cb015f39edf3cb314ec4f4d7e23
@@ -106,14 +106,14 @@ F ext/fts3/unicode/mkunicode.tcl 95cf7ec186e48d4985e433ff8a1c89090a774252
 F ext/fts3/unicode/parseunicode.tcl da577d1384810fb4e2b209bf3313074353193e95
 F ext/fts5/extract_api_docs.tcl 55a6d648d516f35d9a1e580ac00de27154e1904a
 F ext/fts5/fts5.h 81d1a92fc2b4bd477af7e4e0b38b456f3e199fba
-F ext/fts5/fts5Int.h 582eb73064b5a2837d44e72907f84f94e86a2f79
+F ext/fts5/fts5Int.h 8d9bce1847a10df2e4ed9492ea4f3868276748fb
 F ext/fts5/fts5_aux.c 7cd0e2858171dacf505fea4e2e84ee6126854c3d
 F ext/fts5/fts5_buffer.c 80f9ba4431848cb857e3d2158f5280093dcd8015
 F ext/fts5/fts5_config.c b2456e9625bca41c51d54c363e369c6356895c90
 F ext/fts5/fts5_expr.c d2e148345639c5a5583e0daa39a639bf298ae6a7
 F ext/fts5/fts5_hash.c 219f4edd72e5cf95b19c33f1058809a18fad5229
-F ext/fts5/fts5_index.c ba397fe86f0b5ffdafe8eff71d5dbc85cc1180da
-F ext/fts5/fts5_main.c 37b0055cb4036c4b4bb4eb36e30ebd1c21c63939
+F ext/fts5/fts5_index.c 1a1fd996dfe2d632df1dd00689553bc0d205497d
+F ext/fts5/fts5_main.c 2e43726b3ef40b3d5efc0adc7c279d857b1c74fe
 F ext/fts5/fts5_storage.c 4cae85b5287b159d9d98174a4e70adf872b0930a
 F ext/fts5/fts5_tcl.c 85eb4e0d0fefa9420b78151496ad4599a1783e20
 F ext/fts5/fts5_tokenize.c 30f97a8c74683797b4cd233790444fbefb3b0708
@@ -1365,7 +1365,7 @@ F tool/vdbe_profile.tcl 67746953071a9f8f2f668b73fe899074e2c6d8c1
 F tool/warnings-clang.sh f6aa929dc20ef1f856af04a730772f59283631d4
 F tool/warnings.sh 48bd54594752d5be3337f12c72f28d2080cb630b
 F tool/win/sqlite.vsix deb315d026cc8400325c5863eef847784a219a2f
-P 372c1db2475f367d54270d5801aff0503745bff4
-R cd5eb7bfe39c7cb521d188bde2e79a4e
+P 898618ccf61d3ad166d9fc742e132d518338b5c3
+R 67cbb00207cd0d10d2422bd26a9385a8
 U dan
-Z 5190374d0e2aa885aa84b71a9146c526
+Z ab9be791b136092de6b55e47372c8b3c

diff --git a/manifest.uuid b/manifest.uuid
index fb8de8c58012bc44903b8fddf95ea76863e4f8d4..b82d7bb75b363ee3ca0e64d409c1a55c2a95a997 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-898618ccf61d3ad166d9fc742e132d518338b5c3
\ No newline at end of file
+b2de77a01cc5edcea2f98f7916e64cb33e6bd414
\ No newline at end of file