Bug fix in the FTS1 snippet generator. Improvements in the way the snippet

generator handles whitespace. (CVS 3448)

FossilOrigin-Name: d3f4ae827582bd0aac54ae3211d272a1429b6523

diff --git a/ext/fts1/fts1.c b/ext/fts1/fts1.c
index 47ce5f8591fa96215a6d144c84fa2da67595c738..4ebe95660fbe75e60ca57812ed6fa9e8f004b7d9 100644 (file)
--- a/ext/fts1/fts1.c
+++ b/ext/fts1/fts1.c
@@ -2298,11 +2298,32 @@ static int wordBoundary(
   return iBreak;
 }
 
+/*
+** If the StringBuffer does not end in white space, add a single
+** space character to the end.
+*/
+static void appendWhiteSpace(StringBuffer *p){
+  if( p->len==0 ) return;
+  if( isspace(p->s[p->len-1]) ) return;
+  append(p, " ");
+}
+
+/*
+** Remove white space from teh end of the StringBuffer
+*/
+static void trimWhiteSpace(StringBuffer *p){
+  while( p->len>0 && isspace(p->s[p->len-1]) ){
+    p->len--;
+  }
+}
+
+
+
 /*
 ** Allowed values for Snippet.aMatch[].snStatus
 */
 #define SNIPPET_IGNORE  0   /* It is ok to omit this match from the snippet */
-#define SNIPPET_DESIRED 1   /* We want to include this match in the snippet */  
+#define SNIPPET_DESIRED 1   /* We want to include this match in the snippet */
 
 /*
 ** Generate the text of a snippet.
@@ -2369,8 +2390,12 @@ static void snippetText(
       wantEllipsis = 0;
       tailEllipsis = 0;
     }
+    if( iCol!=tailCol || iStart!=tailOffset ){
+      appendWhiteSpace(&sb);
+    }
     if( wantEllipsis || tailEllipsis ){
       append(&sb, zEllipsis);
+      appendWhiteSpace(&sb);
     }
     iEnd = aMatch[i].iStart + aMatch[i].nByte + 40;
     iEnd = wordBoundary(iEnd, zDoc, nDoc, aMatch, nMatch, iCol);
@@ -2382,8 +2407,12 @@ static void snippetText(
     }
     while( iMatch<nMatch && aMatch[iMatch].iCol<iCol ){ iMatch++; }
     while( iStart<iEnd ){
-      while( iMatch<nMatch && aMatch[iMatch].iStart<iStart ){ iMatch++; }
-      if( iMatch<nMatch && aMatch[iMatch].iStart<iEnd ){
+      while( iMatch<nMatch && aMatch[iMatch].iStart<iStart
+             && aMatch[iMatch].iCol<=iCol ){
+        iMatch++;
+      }
+      if( iMatch<nMatch && aMatch[iMatch].iStart<iEnd
+             && aMatch[iMatch].iCol==iCol ){
         nappend(&sb, &zDoc[iStart], aMatch[iMatch].iStart - iStart);
         iStart = aMatch[iMatch].iStart;
         append(&sb, zStartMark);
@@ -2405,7 +2434,9 @@ static void snippetText(
     tailCol = iCol;
     tailOffset = iEnd;
   }
+  trimWhiteSpace(&sb);
   if( tailEllipsis ){
+    appendWhiteSpace(&sb);
     append(&sb, zEllipsis);
   }
   pCursor->snippet.zSnippet = sb.s;

diff --git a/manifest b/manifest
index 3a46379121d6a2ca30a368bfebf81bfde1efb3fe..91923b1cc12e68e5210abb5615b4f4d7c8b1a304 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Avoid\ssegfaults\swhen\sinserted\sNULL\svalues\sinto\sFTS1.\s(CVS\s3447)
-D 2006-09-28T11:41:41
+C Bug\sfix\sin\sthe\sFTS1\ssnippet\sgenerator.\s\sImprovements\sin\sthe\sway\sthe\ssnippet\ngenerator\shandles\swhitespace.\s(CVS\s3448)
+D 2006-09-28T18:37:16
 F Makefile.in cabd42d34340f49260bc2a7668c38eba8d4cfd99
 F Makefile.linux-gcc 2d8574d1ba75f129aba2019f0b959db380a90935
 F README 9c4e2d6706bdcc3efdd773ce752a8cdab4f90028
@@ -21,7 +21,7 @@ F ext/README.txt 913a7bd3f4837ab14d7e063304181787658b14e1
 F ext/fts1/README.txt 20ac73b006a70bcfd80069bdaf59214b6cf1db5e
 F ext/fts1/ft_hash.c 3927bd880e65329bdc6f506555b228b28924921b
 F ext/fts1/ft_hash.h 1a35e654a235c2c662d3ca0dfc3138ad60b8b7d5
-F ext/fts1/fts1.c 5cb7829d331502a3151d6648cadd358ac97b5161
+F ext/fts1/fts1.c 8531a2a86cc9a29d0b47762243e04f8e45655d33
 F ext/fts1/fts1.h 6060b8f62c1d925ea8356cb1a6598073eb9159a6
 F ext/fts1/fts1_hash.c 3196cee866edbebb1c0521e21672e6d599965114
 F ext/fts1/fts1_hash.h 957d378355ed29f672cd5add012ce8b088a5e089
@@ -192,7 +192,7 @@ F test/fkey1.test 153004438d51e6769fb1ce165f6313972d6263ce
 F test/format4.test bf3bed3b13c63abfb3cfec232597a319a31d0bcc
 F test/fts1a.test 8fa66282a2a297faebdab76279c1693eb7920a3b
 F test/fts1b.test 5d8a01aefbecc8b7442b36c94c05eb7a845462d5
-F test/fts1c.test a57cb192d59ddacba64d17c326ff99393c181dc6
+F test/fts1c.test 3e1ad5a8b9118ba2970e6e4572cbc0f0d952d184
 F test/func.test 0ed54b5aeaad319f68016c033acfebef56f5874a
 F test/hook.test 7e7645fd9a033f79cce8fdff151e32715e7ec50a
 F test/in.test 369cb2aa1eab02296b4ec470732fe8c131260b1d
@@ -399,7 +399,7 @@ F www/tclsqlite.tcl bb0d1357328a42b1993d78573e587c6dcbc964b9
 F www/vdbe.tcl 87a31ace769f20d3627a64fa1fade7fed47b90d0
 F www/version3.tcl 890248cf7b70e60c383b0e84d77d5132b3ead42b
 F www/whentouse.tcl 97e2b5cd296f7d8057e11f44427dea8a4c2db513
-P 944d5f812004e2dc0065fce267accf8907235a50
-R a3b9f1ec905300375312507d7a2e50ec
+P 165645d30115f3171fc45489823f85639fe2bfcd
+R 02e1bfc3de0db0fdbc150a5e5d77d065
 U drh
-Z e44db4ed3c8a0588eb2a24973c8a730d
+Z 838ab44cedca04109db513f0d883f765

diff --git a/manifest.uuid b/manifest.uuid
index 3144d7b27c5ca7c315df55b56aa1b098ac7f12b3..2288e204ed6a50442b71a33e20817e73770c9c30 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-165645d30115f3171fc45489823f85639fe2bfcd
\ No newline at end of file
+d3f4ae827582bd0aac54ae3211d272a1429b6523
\ No newline at end of file

diff --git a/test/fts1c.test b/test/fts1c.test
index 3003f5fc64353ce54a5f2036032ef4d7caa2c34f..dc88ddb14005ebc09bd039151255ca5e45c0d88a 100644 (file)
--- a/test/fts1c.test
+++ b/test/fts1c.test
@@ -11,7 +11,7 @@
 # This file implements regression tests for SQLite library.  The
 # focus of this script is testing the FTS1 module.
 #
-# $Id: fts1c.test,v 1.7 2006/09/21 02:03:11 drh Exp $
+# $Id: fts1c.test,v 1.8 2006/09/28 18:37:16 drh Exp $
 #
 
 set testdir [file dirname $argv0]
@@ -1131,31 +1131,31 @@ do_test fts1c-4.2 {
     SELECT snippet(email) FROM email
      WHERE email MATCH 'christmas candlelight'
   }
-} {{<b>...</b>place.? What do you think about going here <b>Christmas</b> 
+} {{<b>...</b> place.? What do you think about going here <b>Christmas</b> 
 eve?? They have an 11:00 a.m. service and a <b>candlelight</b> service at 5:00 p.m., 
-among others.
-
-<b>...</b>}}
+among others. <b>...</b>}}
 
 do_test fts1c-4.3 {
   execsql {
     SELECT snippet(email) FROM email
      WHERE email MATCH 'deal sheet potential reuse'
   }
-} {{EOL-Accenture <b>Deal</b> <b>Sheet</b><b>...</b>intent
+} {{EOL-Accenture <b>Deal</b> <b>Sheet</b> <b>...</b> intent
      Review Enron asset base for <b>potential</b> <b>reuse</b>/ licensing
-     Contract negotiations
-
-<b>...</b>}}
+     Contract negotiations <b>...</b>}}
 do_test fts1c-4.4 {
   execsql {
     SELECT snippet(email,'<<<','>>>',' ') FROM email
      WHERE email MATCH 'deal sheet potential reuse'
   }
-} {{EOL-Accenture <<<Deal>>> <<<Sheet>>> intent
+} {{EOL-Accenture <<<Deal>>> <<<Sheet>>>  intent
      Review Enron asset base for <<<potential>>> <<<reuse>>>/ licensing
-     Contract negotiations
-
- }}
+     Contract negotiations  }}
+do_test fts1c-4.5 {
+  execsql {
+    SELECT snippet(email,'<<<','>>>',' ') FROM email
+     WHERE email MATCH 'first things'
+  }
+} {{Re: <<<First>>> Polish Deal! Congrats!  <<<Things>>> seem to be building rapidly now on the  }}
 
 finish_test