From: drh <drh@noemail.net>
Date: Thu, 13 Apr 2017 21:29:02 +0000 (+0000)
Subject: Fix the skip-ahead-distinct optimization so that it works with indexes that
X-Git-Tag: version-3.19.0~59^2~1
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=172806e4898834428dc4af8877b65c29504cccd9;p=thirdparty%2Fsqlite.git

Fix the skip-ahead-distinct optimization so that it works with indexes that
have repeated columns with different collating sequences.

FossilOrigin-Name: ce1e2b88777e00a82c04abe5ba35eec81b5f324e462f099cd00b21054f369688
---

diff --git a/manifest b/manifest
index b74a3d4f87..19f451a683 100644
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Simplification\sof\sthe\sskip-ahead-distinct\slogic.\s\sThere\sis\sstill\san\sissue\nwith\shandling\sCOLLATE.
-D 2017-04-13T19:48:29.675
+C Fix\sthe\sskip-ahead-distinct\soptimization\sso\sthat\sit\sworks\swith\sindexes\sthat\nhave\srepeated\scolumns\swith\sdifferent\scollating\ssequences.
+D 2017-04-13T21:29:02.792
 F Makefile.in 1cc758ce3374a32425e4d130c2fe7b026b20de5b8843243de75f087c0a2661fb
 F Makefile.linux-gcc 7bc79876b875010e8c8f9502eb935ca92aa3c434
 F Makefile.msc a4c0613a18663bda56d8cf76079ab6590a7c3602e54befb4bbdef76bcaa38b6a
@@ -482,8 +482,8 @@ F src/vxworks.h d2988f4e5a61a4dfe82c6524dd3d6e4f2ce3cdb9
 F src/wal.c 40c543f0a2195d1b0dc88ef12142bea690009344
 F src/wal.h 06b2a0b599cc0f53ea97f497cf8c6b758c999f71
 F src/walker.c b71a992b413b3a022572eccf29ef4b4890223791
-F src/where.c 2b88a9abe28584178a395f286479dda0b103a758c3bf3c70e52a0560f40b467d
-F src/whereInt.h 7a21ef633e26acbf46df04add2eba6e0a2100c78dc5879049e93f981fc3344df
+F src/where.c 911bf344a45e63528933d92ba2574a8382ed2dcd980960888ad809e1b76a1068
+F src/whereInt.h 2a4b634d63ce488b46d4b0da8f2eaa8f9aeab202bc25ef76f007de5e3fba1f20
 F src/wherecode.c 943e32e9dccd0af802e0683ae11071c8bd808364e5908a5fb66758bd404c8681
 F src/whereexpr.c e913aaa7b73ffcce66abcea5f197e2c538d48b5df78d0b7bba8ff4d73cc2e745
 F test/8_3_names.test ebbb5cd36741350040fd28b432ceadf495be25b2
@@ -658,7 +658,7 @@ F test/descidx2.test 9f1a0c83fd57f8667c82310ca21b30a350888b5d
 F test/descidx3.test 09ddbe3f5295f482d2f8b687cf6db8bad7acd9a2
 F test/diskfull.test 106391384780753ea6896b7b4f005d10e9866b6e
 F test/distinct.test a1783b960ad8c15a77cd9f207be072898db1026c
-F test/distinct2.test 5bf7cb4233222cf275b72bcde9149107d300297dd88711aa7f7902d4884b6730
+F test/distinct2.test faef8a3f27424e2cfbe19b2a40752294de3b5d957e049e3336be53ec0476cb58
 F test/distinctagg.test 1a6ef9c87a58669438fc771450d7a72577417376
 F test/e_blobbytes.test 439a945953b35cb6948a552edaec4dc31fd70a05
 F test/e_blobclose.test 4b3c8c60c2171164d472059c73e9f3c1844bb66d
@@ -1572,7 +1572,7 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 70ac9ea1a6cb2f4906f00f1a04f668e5ba5eeed8d4d0fa4be57a9c9eb0683697
-R a148756c1ff7d66b58143c3257e91f62
+P 57c5173b6376a76013b0611ce9770100bd8c04e80402e35d821d8614709f4795
+R 74434244bfed2daca1fdf1468b8ad643
 U drh
-Z 90de4c6922e024a076aa3621c058524e
+Z 9e01ec1a39b84d7c7feff809a6e0c2eb
diff --git a/manifest.uuid b/manifest.uuid
index b718a572c9..46b4331194 100644
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-57c5173b6376a76013b0611ce9770100bd8c04e80402e35d821d8614709f4795
\ No newline at end of file
+ce1e2b88777e00a82c04abe5ba35eec81b5f324e462f099cd00b21054f369688
\ No newline at end of file
diff --git a/src/where.c b/src/where.c
index dbb8111d14..f5ef089b89 100644
--- a/src/where.c
+++ b/src/where.c
@@ -3532,6 +3532,8 @@ static i8 wherePathSatisfiesOrderBy(
     if( pLoop->wsFlags & WHERE_VIRTUALTABLE ){
       if( pLoop->u.vtab.isOrdered ) obSat = obDone;
       break;
+    }else{
+      pLoop->u.btree.nIdxCol = 0;
     }
     iCur = pWInfo->pTabList->a[pLoop->iTab].iCursor;
 
@@ -3677,6 +3679,7 @@ static i8 wherePathSatisfiesOrderBy(
             if( !pColl ) pColl = db->pDfltColl;
             if( sqlite3StrICmp(pColl->zName, pIndex->azColl[j])!=0 ) continue;
           }
+          if( pLoop->u.btree.nIdxCol<=j ) pLoop->u.btree.nIdxCol = j+1;
           isMatch = 1;
           break;
         }
@@ -4850,53 +4853,20 @@ void sqlite3WhereEnd(WhereInfo *pWInfo){
     if( pLevel->op!=OP_Noop ){
 #ifndef SQLITE_DISABLE_SKIPAHEAD_DISTINCT
       int addrSeek = 0;
-      int n = -1;
-      int j, k, op;
-      int r1 = pParse->nMem+1;
       Index *pIdx;
+      int n;
       if( pWInfo->eDistinct==WHERE_DISTINCT_ORDERED
        && (pLoop->wsFlags & WHERE_INDEXED)!=0
        && OptimizationEnabled(db, SQLITE_SkipAhead)
        && (pIdx = pLoop->u.btree.pIndex)->hasStat1
+       && pIdx->aiRowLogEst[(n = pLoop->u.btree.nIdxCol)-1]>=36
       ){
-        /* This is the Skip-ahead optimization.  When doing a DISTINCT query
-        ** that has WHERE_DISTINCT_ORDERED, use OP_SkipGT/OP_SkipLT to skip
-        ** over all duplicate entries, rather than visiting all duplicates
-        ** using OP_Next/OP_Prev. */
-        ExprList *pX = pWInfo->pResultSet;
-        for(j=0; j<pX->nExpr; j++){
-          Expr *pE = sqlite3ExprSkipCollate(pX->a[j].pExpr);
-          if( pE->op==TK_COLUMN ){
-            if( pE->iTable==pLevel->iIdxCur ){
-              k = pE->iColumn+1;
-            }else{
-              if( pE->iTable!=pLevel->iTabCur ) continue;
-              k = 1+sqlite3ColumnOfIndex(pIdx, pE->iColumn);
-            }
-            if( k>n ) n = k;
-          }
-#ifdef SQLITE_DEBUG
-          /* Any expressions in the result-set that match columns of the
-          ** index should have already been transformed to TK_COLUMN 
-          ** operations by whereIndexExprTrans().  */
-          else if( pIdx->aColExpr ){
-            for(k=0; k<pIdx->nKeyCol; k++){
-              Expr *pI = pIdx->aColExpr->a[k].pExpr;
-              assert( pI==0 || sqlite3ExprCompare(pE, pI, pE->iTable)!=0 );
-            }
-          }
-#endif
-        }
-      }
-      /* TUNING: Only try to skip ahead using OP_Seek if we expect to
-      ** skip over 11 or more rows.  Otherwise, OP_Next is just as fast.
-      */
-      assert( 36==sqlite3LogEst(12) );
-      if( n>0 && pIdx->aiRowLogEst[n]>=36 ){
+        int r1 = pParse->nMem+1;
+        int j, op;
         for(j=0; j<n; j++){
           sqlite3VdbeAddOp3(v, OP_Column, pLevel->iIdxCur, j, r1+j);
         }
-        pParse->nMem += n;
+        pParse->nMem += n+1;
         op = pLevel->op==OP_Prev ? OP_SeekLT : OP_SeekGT;
         addrSeek = sqlite3VdbeAddOp4Int(v, op, pLevel->iIdxCur, 0, r1, n);
         VdbeCoverageIf(v, op==OP_SeekLT);
diff --git a/src/whereInt.h b/src/whereInt.h
index 6bd4aab2b0..435f4b3efa 100644
--- a/src/whereInt.h
+++ b/src/whereInt.h
@@ -124,6 +124,7 @@ struct WhereLoop {
       u16 nEq;               /* Number of equality constraints */
       u16 nBtm;              /* Size of BTM vector */
       u16 nTop;              /* Size of TOP vector */
+      u16 nIdxCol;           /* Index column used for ORDER BY */
       Index *pIndex;         /* Index used, or NULL */
     } btree;
     struct {               /* Information for virtual tables */
diff --git a/test/distinct2.test b/test/distinct2.test
index 53f4729b90..a7d59db705 100644
--- a/test/distinct2.test
+++ b/test/distinct2.test
@@ -139,6 +139,45 @@ do_execsql_test 830 {
   SELECT DISTINCT a, b, c FROM t8 WHERE b=3;
 } {1 3 1}
 
+do_execsql_test 900 {
+  CREATE TABLE t9(v);
+  INSERT INTO t9 VALUES 
+    ('abcd'), ('Abcd'), ('aBcd'), ('ABcd'), ('abCd'), ('AbCd'), ('aBCd'), 
+    ('ABCd'), ('abcD'), ('AbcD'), ('aBcD'), ('ABcD'), ('abCD'), ('AbCD'), 
+    ('aBCD'), ('ABCD'),
+    ('wxyz'), ('Wxyz'), ('wXyz'), ('WXyz'), ('wxYz'), ('WxYz'), ('wXYz'), 
+    ('WXYz'), ('wxyZ'), ('WxyZ'), ('wXyZ'), ('WXyZ'), ('wxYZ'), ('WxYZ'), 
+    ('wXYZ'), ('WXYZ');
+}
 
-finish_test
+do_execsql_test 910 {
+  SELECT DISTINCT v COLLATE NOCASE, v FROM t9 ORDER BY +v;
+} {
+  ABCD ABCD ABCd ABCd ABcD ABcD ABcd ABcd AbCD
+  AbCD AbCd AbCd AbcD AbcD Abcd Abcd
+  WXYZ WXYZ WXYz WXYz WXyZ WXyZ WXyz WXyz WxYZ
+  WxYZ WxYz WxYz WxyZ WxyZ Wxyz Wxyz
+  aBCD aBCD aBCd aBCd aBcD aBcD aBcd aBcd abCD
+  abCD abCd abCd abcD abcD abcd abcd
+  wXYZ wXYZ wXYz wXYz wXyZ wXyZ wXyz wXyz wxYZ
+  wxYZ wxYz wxYz wxyZ wxyZ wxyz wxyz
+}
+
+do_execsql_test 920 {
+  CREATE INDEX i9 ON t9(v COLLATE NOCASE, v);
+  ANALYZE;
 
+  SELECT DISTINCT v COLLATE NOCASE, v FROM t9 ORDER BY +v;
+} {
+  ABCD ABCD ABCd ABCd ABcD ABcD ABcd ABcd AbCD
+  AbCD AbCd AbCd AbcD AbcD Abcd Abcd
+  WXYZ WXYZ WXYz WXYz WXyZ WXyZ WXyz WXyz WxYZ
+  WxYZ WxYz WxYz WxyZ WxyZ Wxyz Wxyz
+  aBCD aBCD aBCd aBCd aBcD aBcD aBcd aBcd abCD
+  abCD abCd abCd abcD abcD abcd abcd
+  wXYZ wXYZ wXYz wXYz wXyZ wXyZ wXyz wXyz wxYZ
+  wxYZ wxYz wxYz wxyZ wxyZ wxyz wxyz
+}
+
+
+finish_test