Changeset 2025 for trunk/gsdl

Timestamp:

2001-02-20T15:09:22+13:00 (23 years ago)

Author:

paynter

Message:

You can now have several phind classifiers on one collection. This
required numbering the classifiers (1,2,3...), creating a phind directory
(phind1, phind2...) and filehandles for each, and adding parameter
information to classifier info. The parameter information lets us
differentiate between different classifiers when we construct the Phind web
pages.

File:

• trunk/gsdl/perllib/classify/phind.pm (modified) (12 diffs)

trunk/gsdl/perllib/classify/phind.pm

@@ -155 +155 @@
     # Parse classifier arguments
     my $builddir = "";
-    my $phinddir = "";
     if (!parsargv::parse(\@_, 
              q^text/.*/section:Title,section:text^, \$self->{'indexes'},
@@ -185 +184 @@
     }
     $self->{'builddir'} = $builddir;
-    $self->{'phinddir'} = &util::filename_cat($builddir, "phind");
    
     return bless $self, $class;
@@ -201 +199 @@
 
     # create phind directory
-    my $phinddir = $self->{'phinddir'};
-    if (-e "$phinddir") {
-    &util::rm_r("$phinddir");
+    my $phnumber = 1;
+    my $phinddir = &util::filename_cat($builddir, "phind1");
+    while (-e "$phinddir") {
+    $phnumber++;
+    $phinddir = &util::filename_cat($builddir, "phind$phnumber");
     }
     &util::mk_dir("$phinddir");
+    $self->{'phinddir'} = $phinddir;
+    $self->{'phindnumber'} = $phnumber;
 
     # open filehandles for documents and text
     my $clausefile =  &util::filename_cat("$phinddir", "clauses");
     &util::rm($clausefile) if (-e $clausefile);
-    open(TEXT, ">$clausefile") || die "Cannot open $clausefile: $!";
-    $self->{'txthandle'} = TEXT;
+
+    my $txthandle = 'TEXT' . $phnumber;
+    open($txthandle, ">$clausefile") || die "Cannot open $clausefile: $!";
+    $self->{'txthandle'} = $txthandle;
 
     my $docfile = &util::filename_cat("$phinddir", "docs.txt");
     &util::rm($docfile) if (-e $docfile);
-    open(DOCS, ">$docfile") || die "Cannot open $docfile: $!";
-    $self->{'dochandle'} = DOCS;
+
+    my $dochandle = 'DOC' . $phnumber;
+    open($dochandle, ">$docfile") || die "Cannot open $docfile: $!";
+    $self->{'dochandle'} = $dochandle;
     
 }
@@ -229 +235 @@
 sub classify {
     my $self = shift (@_);
-    my ($doc_obj) = @_;
+    my $doc_obj = shift @_;
 
     my $verbosity = $self->{'verbosity'};
@@ -242 +248 @@
     my $doclanguage = $doc_obj->get_metadata_element ($top_section, "Language");
     my $phrlanguage = $self->{'language_exp'};
-
-    print STDERR "+ CLASSIFY - doclanguage: $doclanguage, phrlanguage $phrlanguage \n";
-
     return if ($doclanguage && ($doclanguage !~ /$phrlanguage/i));
     
@@ -274 +277 @@
     ($level, $field) = split(/:/, $part);
     die unless ($level && $field);
-
+    
     # Extract the text from every section
     # (In phind, document:text and section:text are equivalent)
@@ -301 +304 @@
         $section = $doc_obj->get_top_section();
         while (defined($section)) {
-        $dataref .= $doc_obj->get_metadata($section, $field);
-        $data .= join("\n", $$dataref) . "\n";
+        $dataref = $doc_obj->get_metadata($section, $field);
+        $data .= join("\n", @$dataref) . "\n";
         $section = $doc_obj->get_next_section($section);
         }
@@ -314 +317 @@
     
     }
-    
+
     # output the text
     $text =~ tr/\n//s;
@@ -338 +341 @@
     if ($verbosity) {
     print $out "\n*** phind.pm generating indexes for ", $self->{'indexes'}, "\n";
+    print $out "***          in", $self->{'phinddir'}, "\n";
     }
 
@@ -354 +358 @@
 
     # Create the phrase file and put phrase numbers in phind/phrases
-    print $out "\nSorting and Renumbering phrases for input to mgpp\n" if $verbosity;
+    print $out "\nSorting and renumbering phrases for input to mgpp\n" if $verbosity;
     &renumber_phrases($self);
     
@@ -416 +420 @@
 
 
-    # Insert the classifier into.... what?
+    # Return the information about the classifier that we'll later want to
+    # use to create macros when the Phind classifier document is displayed.
+    my %classifyinfo = ('thistype'=>'Invisible',
+                        'childtype'=>'Phind', 
+                        'Title'=>$self->{'buttonname'},
+                        'parameters'=>"phindnumber=$self->{'phindnumber'}",
+            'contains'=>[]);
+    
     my $collection = $self->{'collection'};
     my $url = "library?a=p&p=phind&c=$collection";
-
-    my %classifyinfo = ('thistype'=>'Invisible',
-            'childtype'=>'Phind',
-            'Title'=>$self->{'buttonname'},
-            'contains'=>[]);
-    
     push (@{$classifyinfo{'contains'}}, {'OID'=>$url});
+   
     return \%classifyinfo;
 }
@@ -435 +441 @@
     my ($language_exp, $text) = @_;
 
-    print STDERR "+ tokenising in $language_exp\n";
-    
     if ($language_exp =~ /en/) {
     return &convert_gml_to_tokens_EN($text);