Improve performance of editdist3() by keeping the costs in sorted order.

Also add a new regression test to editdist3().

FossilOrigin-Name: dc734c5b61464dfd6bfa7963f2ecce32e405a0c2ba1ef6f453ec9389da080256

diff --git a/ext/misc/spellfix.c b/ext/misc/spellfix.c
index 8b528562a844d0d7b7c43ccb94e680b7c13b2aef..8651bb7a9d3ac5c2baf5e008e0b9b5dd94b15805 100644 (file)
--- a/ext/misc/spellfix.c
+++ b/ext/misc/spellfix.c
@@ -658,6 +658,79 @@ static void editDist3ConfigDelete(void *pIn){
   sqlite3_free(p);
 }
 
+/* Compare the FROM values of two EditDist3Cost objects, for sorting.
+** Return negative, zero, or positive if the A is less than, equal to,
+** or greater than B.
+*/
+static int editDist3CostCompare(EditDist3Cost *pA, EditDist3Cost *pB){
+  int n = pA->nFrom;
+  int rc;
+  if( n>pB->nFrom ) n = pB->nFrom;
+  rc = strncmp(pA->a, pB->a, n);
+  if( rc==0 ) rc = pA->nFrom - pB->nFrom;
+  return rc;
+}
+
+/*
+** Merge together two sorted lists of EditDist3Cost objects, in order
+** of increasing FROM.
+*/
+static EditDist3Cost *editDist3CostMerge(
+  EditDist3Cost *pA,
+  EditDist3Cost *pB
+){
+  EditDist3Cost *pHead = 0;
+  EditDist3Cost **ppTail = &pHead;
+  EditDist3Cost *p;
+  while( pA && pB ){
+    if( editDist3CostCompare(pA,pB)<=0 ){
+      p = pA;
+      pA = pA->pNext;
+    }else{
+      p = pB;
+      pB = pB->pNext;
+    }
+    *ppTail = p;
+    ppTail =  &p->pNext;
+  }
+  if( pA ){
+    *ppTail = pA;
+  }else{
+    *ppTail = pB;
+  }
+  return pHead;
+}
+
+/*
+** Sort a list of EditDist3Cost objects into order of increasing FROM
+*/
+static EditDist3Cost *editDist3CostSort(EditDist3Cost *pList){
+  EditDist3Cost *ap[60], *p;
+  int i;
+  int mx = 0;
+  ap[0] = 0;
+  ap[1] = 0;
+  while( pList ){
+    p = pList;
+    pList = p->pNext;
+    p->pNext = 0;
+    for(i=0; ap[i]; i++){
+      p = editDist3CostMerge(ap[i],p);
+      ap[i] = 0;
+    }
+    ap[i] = p;
+    if( i>mx ){
+      mx = i;
+      ap[i+1] = 0;
+    }
+  }
+  p = 0;
+  for(i=0; i<=mx; i++){
+    if( ap[i] ) p = editDist3CostMerge(p,ap[i]);
+  }
+  return p;
+}
+
 /*
 ** Load all edit-distance weights from a table.
 */
@@ -729,6 +802,12 @@ static int editDist3ConfigLoad(
   }
   rc2 = sqlite3_finalize(pStmt);
   if( rc==SQLITE_OK ) rc = rc2;
+  if( rc==SQLITE_OK ){
+    int iLang;
+    for(iLang=0; iLang<p->nLang; iLang++){
+      p->a[iLang].pCost = editDist3CostSort(p->a[iLang].pCost);
+    }
+  }
   return rc;
 }
 
@@ -943,8 +1022,9 @@ static int editDist3Core(
     a2[i2].nByte = utf8Len((unsigned char)z2[i2], n2-i2);
     for(p=pLang->pCost; p; p=p->pNext){
       EditDist3Cost **apNew;
-      if( p->nFrom>0 ) continue;
+      if( p->nFrom>0 ) break;
       if( i2+p->nTo>n2 ) continue;
+      if( p->a[0]>z2[i2] ) break;
       if( matchTo(p, z2+i2, n2-i2)==0 ) continue;
       a2[i2].nIns++;
       apNew = sqlite3_realloc64(a2[i2].apIns, sizeof(*apNew)*a2[i2].nIns);

diff --git a/manifest b/manifest
index 411cad769d25cb40bcaf4500b8bfc091b4b488d6..dc00c62eff48a10925607bb2271321553e472617 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Reduce\sthe\snumber\sof\scalls\sto\sstrncmp()\srequired\sto\srun\seditDist3Core().
-D 2018-02-15T03:05:56.872
+C Improve\sperformance\sof\seditdist3()\sby\skeeping\sthe\scosts\sin\ssorted\sorder.\nAlso\sadd\sa\snew\sregression\stest\sto\seditdist3().
+D 2018-02-15T03:56:33.574
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F Makefile.in 7a3f714b4fcf793108042b7b0a5c720b0b310ec84314d61ba7f3f49f27e550ea
@@ -294,7 +294,7 @@ F ext/misc/series.c f3c0dba5c5c749ce1782b53076108f87cf0b71041eb6023f727a9c50681d
 F ext/misc/sha1.c 0b9e9b855354910d3ca467bf39099d570e73db56
 F ext/misc/shathree.c 9e960ba50483214c6a7a4b1517f8d8cef799e9db381195178c3fd3ad207e10c0
 F ext/misc/showauth.c 732578f0fe4ce42d577e1c86dc89dd14a006ab52
-F ext/misc/spellfix.c ad435d8d84a64a7287f67ba618d5fc4a775e73d8489b1c7ae85a339ed5d20c34
+F ext/misc/spellfix.c b3a644285cb008f3c10ed4cf04e17adcbc7d62c3911c79d786dfc91c177534f0
 F ext/misc/sqlar.c 57d5bc45cd5492208e451f697404be88f8612527d64c9d42f96b325b64983d74
 F ext/misc/stmt.c 6f16443abb3551e3f5813bb13ba19a30e7032830015b0f92fe0c0453045c0a11
 F ext/misc/totype.c 4a167594e791abeed95e0a8db028822b5e8fe512
@@ -1265,7 +1265,7 @@ F test/speedtest1.c a5faf4cbe5769eee4b721b3875cb3f12520a9b99d9026b1063b47c396033
 F test/spellfix.test 951a6405d49d1a23d6b78027d3877b4a33eeb8221dcab5704b499755bb4f552e
 F test/spellfix2.test dfc8f519a3fc204cb2dfa8b4f29821ae90f6f8c3
 F test/spellfix3.test 0f9efaaa502a0e0a09848028518a6fb096c8ad33
-F test/spellfix4.test eaae7a334516e09896a684fb552d338fa2844c65db9817d900701fe7598327e6
+F test/spellfix4.test 51c7c26514ade169855c66bcf130bd5acfb4d7fd090cc624645ab275ae6a41fb
 F test/sqldiff1.test 28cd737cf1b0078b1ec1bbf425e674c47785835e
 F test/sqllimits1.test a74ee2a3740b9f9c2437c246d8fb77354862a142
 F test/sqllog.test 6af6cb0b09f4e44e1917e06ce85be7670302517a
@@ -1706,7 +1706,7 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 00707f2f2f746a6421f3e2de995e68cc8adba7225a04db6b28db52944e7e988e
-R bf16a0c80a0f655a42b9a2dc09fe803f
+P afd6fbc01052ccfc9bd29fb8f934b291b8f56af44fcae870da7e1355fe95c29a
+R 8c5288c6ebac589573d48d7520074806
 U drh
-Z 2ab397c1e9ab0ece50ec14a24b630ce7
+Z 65b4bdcfd9180aa4d82e9c5cf2ce0fa5

diff --git a/manifest.uuid b/manifest.uuid
index df373242554bf8dbe5ef1876dc7c4b1be6886985..a1afa0904978b27d6794e734f3d51750c63b0233 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-afd6fbc01052ccfc9bd29fb8f934b291b8f56af44fcae870da7e1355fe95c29a
\ No newline at end of file
+dc734c5b61464dfd6bfa7963f2ecce32e405a0c2ba1ef6f453ec9389da080256
\ No newline at end of file

diff --git a/test/spellfix4.test b/test/spellfix4.test
index 7054bb98effd1d41a0ce25ca0e21368eb01c74e0..caf6d5139a67ac93c80474037362da750e168cae 100644 (file)
--- a/test/spellfix4.test
+++ b/test/spellfix4.test
@@ -97,22 +97,22 @@ do_execsql_test 300 {
     (0, 'a', 'e', 50),
     (0, 'a', 'i', 70),
     (0, 'a', 'o', 75),
-    (0, 'a', 'u', 85),
+    (0, 'a', 'u', 81),
     (0, 'e', 'a', 50),
-    (0, 'e', 'i', 50),
-    (0, 'e', 'o', 75),
-    (0, 'e', 'u', 85),
+    (0, 'e', 'i', 52),
+    (0, 'e', 'o', 72),
+    (0, 'e', 'u', 82),
     (0, 'i', 'a', 70),
-    (0, 'i', 'e', 50),
+    (0, 'i', 'e', 52),
     (0, 'i', 'o', 75),
-    (0, 'i', 'u', 85),
+    (0, 'i', 'u', 83),
     (0, 'o', 'a', 75),
-    (0, 'o', 'e', 75),
+    (0, 'o', 'e', 72),
     (0, 'o', 'i', 75),
     (0, 'o', 'u', 40),
-    (0, 'u', 'a', 85),
-    (0, 'u', 'e', 85),
-    (0, 'u', 'i', 85),
+    (0, 'u', 'a', 81),
+    (0, 'u', 'e', 82),
+    (0, 'u', 'i', 83),
     (0, 'u', 'o', 40),
     (0, 'm', 'n', 45),
     (0, 'n', 'm', 45)
@@ -340,7 +340,14 @@ do_execsql_test 310 {
    WHERE a.x<b.x
    ORDER BY 1, 2
    LIMIT 20
-} {139 bucket pocket 149 manual mental 150 meter motor 169 crack trick 173 sinatra sonata 174 edition emotion 174 major motor 174 risk rose 174 state stone 194 deal detail 196 alert talent 196 analog catalog 196 deal legal 196 ford forum 196 risk trick 196 stone strong 197 china tina 197 congo logo 197 diana tina 197 florida gloria}
-
+} {139 bucket pocket 144 meter motor 149 manual mental 169 crack trick 173 sinatra sonata 174 edition emotion 174 major motor 174 risk rose 174 state stone 194 deal detail 196 alert talent 196 analog catalog 196 deal legal 196 ford forum 196 risk trick 196 stone strong 197 china tina 197 congo logo 197 diana tina 197 florida gloria}
+do_execsql_test 320 {
+  SELECT md5sum(ed||'/'||sx||'/'||sy||',') FROM (
+      SELECT editdist3(a.x,b.x) AS ed, a.x AS sx, b.x AS sy
+        FROM words a, words b
+       WHERE a.x<b.x
+       ORDER BY 1, 2
+  )
+} {69d0a31872203a775e19325ea98cd053}
 
 finish_test