Handle heuristic for fuzzy commit matches better

diff --git a/match-xsa b/match-xsa
index a88a1d031a201fb7f563a810f75f398a292778cd..a07628f8edab58713401406c8547de2f190c8933 100755 (executable)
--- a/match-xsa
+++ b/match-xsa
@@ -36,6 +36,11 @@ my $qemuu_git   = $opt{'QEMUU_GIT'};
 my $qemut_git   = $opt{'QEMUT_GIT'};
 my $xsa         = $XSADIR;
 
+# Make print output redirected to HTML files more easily readable
+my $NL = "\n";
+if ($HTML) {
+   $NL = "<br>\n";
+}
 
 # Always get/update logs
 my $rl .= "./xen-release-logs --logroot $LOGROOT --version $VERSION --major $MAJOR";
@@ -110,9 +115,9 @@ getlogs("$LOGDIR/qemuu_$SHORT", \@LOGQEMUU, \@LOGQEMUU_hash);
 getlogs("$LOGDIR/qemut_$SHORT", \@LOGQEMUT, \@LOGQEMUT_hash);
 
 # Do the actual matching
-print "CHECKING '$XSALISTFILE' against 'xen_$SHORT.log', 'qemuu_$SHORT.log' ".
-      "and 'qemut_$SHORT.log'.\n";
-print "\n";
+print "CHECKING '$XSALISTFILE' against 'xen_$SHORT.log',
+      'qemuu_$SHORT.log' "."and 'qemut_$SHORT.log'.$NL";
+print $NL;
 for ($i=0; $i <= $XSAs; $i++) {
    my $j=0;
    my $i_correct = $i+1;
@@ -561,6 +566,7 @@ sub matchsmart {
    # If there is a match, compare the actual patch
    
    my $index = 0;
+   my @matches = ();
    for my $log_id (@$log_ref) {
       my $ret = matchcommitmsg ($XSA_ID[$i], $log_id);
       if ($ret) {
@@ -568,27 +574,44 @@ sub matchsmart {
       }
       if ($DEBUG != 0) { 
          $debugstring .= $index.": ".$log_id."\n";
-         if ($namefound != 0) {
-            if ($ret == 1) {
-               $debugstring .= ">>> FILENAME MATCHED\n";  
-            } elsif ($ret == 2) {
-               $debugstring .= ">>> FILENAME MATCHED (truncated)\n";             
+      }
+      if ($namefound != 0) {
+         if ($ret == 1) {
+            if ($DEBUG != 0) {
+               $debugstring .= ">>> FILENAME MATCHED\n";
+            }
+            # Ignore any almost matches found before
+            @matches = ($index);
+            $namefound = 1;
+            last;
+         } elsif ($ret == 2) {
+            if ($DEBUG != 0) {
+               $debugstring .= ">>> FILENAME MATCHED (truncated)";
+               $debugstring .= "... keep searching\n";
             }
+            push @matches, $index;
+            # Keep tracking this for a while
+            print "WARNING XSA=".$XSA[$i]."(".$i."); Possible Issue: ".
+               "truncated file match: '".$XSA_ID[$i]."' with '".
+               $log_id."'".$NL;
          }
-      }        
-      
-      if ($namefound != 0) {
+      }
+      $index++;
+   }
+
+   if ($namefound != 0) {
+      for my $index (@matches) {
          chdir($git_dir);
          #my $git_command="git show ".@$loghash_ref[$index].' --format=""';
          my $git_command="git show ".@$loghash_ref[$index];
          my $diff_git = `$git_command`;
-         
+
          if ($XSAFETCH) {
             my $path="";
             # Check whether $XSA_PATCH[$i] contains a path
-            my $index = rindex($XSA_PATCH[$i], "/");
-            if ($index != -1) {
-                $path = "/".substr($XSA_PATCH[$i], 0, $index);
+            my $rindex = rindex($XSA_PATCH[$i], "/");
+            if ($rindex != -1) {
+                $path = "/".substr($XSA_PATCH[$i], 0, $rindex);
             }
             if (! -d "$xsa_dir$path") {
                system("mkdir $xsa_dir$path");
@@ -596,35 +619,35 @@ sub matchsmart {
             system("wget --quiet --no-check-certificate -P $xsa_dir$path ".
                    "$XSAFETCH/$XSA_PATCH[$i]");
          }
-         
+
          my $patch = read_file("$xsa_dir/$XSA_PATCH[$i]");
-         
+
          my $diff_git_n = normalize_diff($diff_git);
          my $patch_n = normalize_diff($patch);
-         
-         if ($DEBUG != 0) { 
+
+         if ($DEBUG != 0) {
             my $gitfile_n   = $debugbase."-git-n.txt";
-            my $gitfile     = $debugbase."-git.txt"; 
+            my $gitfile     = $debugbase."-git.txt";
             my $patchfile_n = $debugbase."-patch-n.txt";
-            my $patchfile   = $debugbase."-patch.txt";  
+            my $patchfile   = $debugbase."-patch.txt";
             my $compfile_n  = $debugbase."-diff-n.txt";
-            my $compfile    = $debugbase."-diff.txt";              
-            
-            write_file($gitfile_n, $diff_git_n);              
+            my $compfile    = $debugbase."-diff.txt";
+
+            write_file($gitfile_n, $diff_git_n);
             write_file($gitfile, $diff_git);
-            write_file($patchfile_n, $patch_n);              
-            write_file($patchfile, $patch);   
+            write_file($patchfile_n, $patch_n);
+            write_file($patchfile, $patch);
             system ("diff $gitfile_n $patchfile_n > $compfile_n");
             system ("diff $gitfile $patchfile > $compfile");
          }
-         
+
          # Check whether the normalized patch and diff are the same
-         if ($diff_git_n eq $patch_n) {            
+         if ($diff_git_n eq $patch_n) {
             $found++;
             if ($DEBUG) {
                my $debuglog = $debugbase."--content-match.txt";
                write_file($debuglog, "matches: ".$found);
-               $debugstring .= ">>> CONTENT MATCHED\n"; 
+               $debugstring .= ">>> CONTENT MATCHED\n";
                $XSA_HASH[$i] = @$loghash_ref[$index];
                if ($type eq "xen") {
                    $XSA_REPO[$i] = "xen.git";
@@ -632,14 +655,11 @@ sub matchsmart {
                    $XSA_REPO[$i] = "qemu-xen.git";
                } elsif ($type eq "qemut") {
                    $XSA_REPO[$i] = "qemu-xen-traditional.git";
-               } 
+               }
             }
          }
-
          last;
-      }   
-      
-      $index++;
+      }
    }
 
    if ($namefound == 0) {