Changeset 4962

Timestamp:

09/17/09 15:27:21 (3 years ago)

Author:

AndrewJones

Message:

Item1363: add ability to skip topics

Location:

trunk/SearchEngineKinoSearchAddOn

Files:

• lib/Foswiki/Contrib/SearchEngineKinoSearchAddOn/Config.spec (modified) (1 diff)
• lib/Foswiki/Contrib/SearchEngineKinoSearchAddOn/Index.pm (modified) (5 diffs)
• lib/Foswiki/Contrib/SearchEngineKinoSearchAddOn/KinoSearch.pm (modified) (3 diffs)
• test/unit/SearchEngineKinoSearchAddOn/KinoSearchTests.pm (modified) (2 diffs)

trunk/SearchEngineKinoSearchAddOn/lib/Foswiki/Contrib/SearchEngineKinoSearchAddOn/Config.spec

@@ -14 +14 @@
 # For example: Web.SomeTopic.AnAttachment.txt, Web.OtherTopic.OtherAttachment.pdf
 $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipAttachments} = '';
+
+# **STRING**
+# List of topics to skip.
+# Topics can be in the form of Web.MyTopic, or if you want a topic to be excluded from all webs just enter MyTopic.
+# For example: Main.WikiUsers, WebStatistics
+$Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipTopics} = '';
 
 # **STRING**

trunk/SearchEngineKinoSearchAddOn/lib/Foswiki/Contrib/SearchEngineKinoSearchAddOn/Index.pm

@@ -47 +47 @@
 
     # FIXME: Make it more clear.
-    $self->{Debug} = $debug;
+    $self->{Debug} = $debug || $Foswiki::cfg{SearchEngineKinoSearchAddOn}{Debug} || 0;
 
     $self->log( "Indexing started", 1 );
@@ -59 +59 @@
 
     my @webs = $self->websToIndex();
+    
+    # get the list of topics not to be indexed
+    my %skipTopics = $self->skipTopics;
 
     foreach my $web (@webs) {
@@ -66 +69 @@
 
         foreach my $topic ( Foswiki::Func::getTopicList($web) ) {
+            next if ( ( $skipTopics{"$web.$topic"} ) || ( $skipTopics{$topic} ) );
+            
             $self->log("Indexing topic | $web.$topic");
             $self->indexTopic( $invindexer, $web, $topic, %fldNames );
@@ -206 +211 @@
     my @topicsToUpdate;
 
+    # get the list of topics not to be indexed
+    my %skipTopics = $self->skipTopics;
+    
     # do not process the same topic twice
     my %exclude;
@@ -216 +224 @@
         my ( $topicName, $userName, $changeTime, $revision ) =
           split( /\t/, $change );
+          
+        next if ( ( $skipTopics{"$web.$topicName"} ) || ( $skipTopics{$topicName} ) );
 
         if ( ( !%exclude ) || ( !$exclude{$topicName} ) ) {

trunk/SearchEngineKinoSearchAddOn/lib/Foswiki/Contrib/SearchEngineKinoSearchAddOn/KinoSearch.pm

@@ -62 +62 @@
         $self->{Log}->print( "| $logtime | $logString\n");
 
-        print STDERR "$logString\n";
+        #print STDERR "$logString\n";
     }
+    $self->{Debug} && Foswiki::Func::writeDebug( $logString );
 }
 
@@ -125 +126 @@
 # QS
 sub skipAttachments {
-    # SMELL: whats the best way to define skipped attachments?
     my $to_skip = $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipAttachments} || '';
     my %skipattachments;
@@ -136 +136 @@
 }
 
+# List of topics to be skipped.
+sub skipTopics {
+    my $to_skip = $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipTopics} || '';
+    my %skiptopics;
+
+    foreach my $t ( split( /\,\s+/, $to_skip ) ) {
+        $skiptopics{$t} = 1;
+    }
+    
+    return %skiptopics;
+}
+
 # List of file extensions to be indexed
 # QS
 sub indexExtensions {
-    my $extensions = $Foswiki::cfg{SearchEngineKinoSearchAddOn}{IndexExtensions} || ".pdf, .doc, .xml, .html, .txt, .xls, .ppt";
+    my $extensions = $Foswiki::cfg{SearchEngineKinoSearchAddOn}{IndexExtensions} || ".txt, .html, .xml, .doc, .docx, .xls, .xlsx, .ppt, .pptx, .pdf";
     my %indexextensions;
 

trunk/SearchEngineKinoSearchAddOn/test/unit/SearchEngineKinoSearchAddOn/KinoSearchTests.pm

@@ -118 +118 @@
 
     my @config_atts  = ("att1", "att2");
-    my $a_att;
 
     $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipAttachments} = "";
@@ -127 +126 @@
     $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipAttachments} = "att1, att2";
     %atts = $ks->skipAttachments();
-    foreach $a_att (@config_atts) {
+    foreach my $a_att (@config_atts) {
         $this->assert($atts{$a_att}, "Attachment $a_att not skipped in config.")
+        }
+}
+
+sub test_skipTopics {
+    my $this = shift;
+    my $ks = new Foswiki::Contrib::SearchEngineKinoSearchAddOn::KinoSearch("index");
+
+    my @config_atts  = ("Web.MyTopic", "MyOtherTopic");
+
+    $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipTopics} = "";
+    my %atts = $ks->skipTopics();
+    my $num = %atts;
+    $this->assert($num == 0, "List of skipped attachments not empty. : $num");
+
+    $Foswiki::cfg{SearchEngineKinoSearchAddOn}{SkipTopics} = "Web.MyTopic, MyOtherTopic";
+    %atts = $ks->skipTopics();
+    foreach my $t (@config_atts) {
+        $this->assert($atts{$t}, "Topic $t not skipped in config.")
         }
 }