Changeset 315 for trunk/gsdl/perllib/classify.pm

Timestamp:

1999-06-30T15:35:07+12:00 (25 years ago)

Author:

sjboddie

Message:

• removed old infodb stuff
• changed the way classifiers work
• added maxdocs and allclassifications options
• no longer get doctype from collect.cfg but instead set it directly in plugins that don't use the default

File:

• trunk/gsdl/perllib/classify.pm (modified) (4 diffs)

trunk/gsdl/perllib/classify.pm

@@ -6 +6 @@
 
 
-$next_classify_num = "0";
+$next_classify_num = 1;
 
 # load_classifier will load one classifier. $classinfo is an
@@ -39 +39 @@
 
 # init_classifiers resets all the classifiers and readys them to process
-# the documents. They are each given a unique classification OID
+# the documents.
 sub init_classifiers {
     my ($classifiers) = @_;
@@ -45 +45 @@
     foreach $classobj (@$classifiers) {
     $classobj->init();
-    $classobj->set_OID ("CL$next_classify_num");
-    $next_classify_num++;
     }
 }
@@ -62 +60 @@
 # to the gdbm
 sub output_classify_info {
-    my ($classifiers, $handle) = @_;
+    my ($classifiers, $handle, $allclassifications) = @_;
 #    $handle = "main::STDOUT";
 
-    my $contains_str = "";
-    my $first = 1;
-    
-    # output each of the classifications
+    # create a classification containing all the info
+    my $classifyinfo = {'classifyOID'=>'browse',
+            'contains'=>[]};
+
+    # get each of the classifications
     foreach $classobj (@$classifiers) {
-    $classobj->output_classify_info($handle);
-    $contains_str .= ";" unless $first;
-    $first = 0;
-    $contains_str .= $classobj->get_OID();
+    my $tempinfo = $classobj->get_classify_info();
+    $tempinfo->{'classifyOID'} = "CL$next_classify_num";
+    $next_classify_num++;
+    push (@{$classifyinfo->{'contains'}}, $tempinfo);
     }
 
-    # output information about each of the classifications
-    print $handle "[browse]\n";
-    print $handle "<doctype>classify\n";
-    print $handle "<hastxt>0\n";
-    print $handle "<contains>$contains_str\n" 
-    unless $contains_str eq "";
-    print $handle '-' x 70, "\n";
+    &print_classify_info ($handle, $classifyinfo, "", $allclassifications);
 }
 
+sub print_classify_info {
+    my ($handle, $classifyinfo, $OID, $allclassifications) = @_;
+    $OID =~ s/^\.+//; # just for good luck
+
+    # book information is printed elsewhere
+    return if (defined ($classifyinfo->{'OID'}));
+ 
+    # don't want empty classifications
+    if ($allclassifications || &clean_contents ($classifyinfo)) {
+   
+    $OID = $classifyinfo->{'classifyOID'} if defined ($classifyinfo->{'classifyOID'});
+    
+    my $outputtext = "[$OID]\n";
+    $outputtext .= "<doctype>classify\n";
+    $outputtext .= "<hastxt>0\n";
+    $outputtext .= "<classifytype>$classifyinfo->{'classifytype'}\n" if defined $classifyinfo->{'classifytype'};
+    $outputtext .= "<Title>$classifyinfo->{'Title'}\n" if defined $classifyinfo->{'Title'};
+    
+    $outputtext .= "<contains>";
+    
+    my $next_subOID = 1;
+    my $first = 1;
+    foreach $tempinfo (@{$classifyinfo->{'contains'}}) {
+        # empty contents were made undefined by clean_contents()
+        next unless defined $tempinfo;
+        
+        $outputtext .= ";" unless $first;
+        $first = 0;
+        
+        if (defined ($tempinfo->{'classifyOID'})) {
+        $outputtext .= $tempinfo->{'classifyOID'};
+        &print_classify_info ($handle, $tempinfo, $tempinfo->{'classifyOID'},
+                      $allclassifications);
+        } elsif (defined ($tempinfo->{'OID'})) {
+        $outputtext .= $tempinfo->{'OID'};
+        # note: we don't want to print the contents of the books
+        } else {
+        $outputtext .= "\".$next_subOID";
+        &print_classify_info ($handle, $tempinfo, "$OID.$next_subOID",
+                      $allclassifications);
+        $next_subOID++;
+        }
+    }
+    $outputtext .= "\n";
+    $outputtext .= '-' x 70 . "\n";
+    
+    print $handle $outputtext;
+    }
+}
+
+sub clean_contents {
+    my ($classifyinfo) = @_;
+    my $has_content = 0;
+
+    foreach $content (@{$classifyinfo->{'contains'}}) {
+    if (defined $content->{'OID'}) {
+        # found a book
+        $has_content = 1;
+    } elsif (&clean_contents ($content)) {
+        # there's a book somewhere below
+        $has_content = 1;
+    } else {
+        # section contains no books so we want to remove 
+        # it from its parents contents
+        $content = undef;
+    }
+    }
+    return $has_content;
+}
 
 1;