Changeset 1411

Timestamp:

2000-08-16T18:07:29+12:00 (24 years ago)

Author:

dmm9

Message:

added the options for the date extractor

File:

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

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -34 +34 @@
 use doc;
 use diagnostics;
+use DateExtract;
 
 sub print_general_usage {
@@ -84 +85 @@
     my $self = {};
     my $encodings = "^(iso_8859_1|Latin1|ascii|gb|iso_8859_6|windows_1256|Arabic|utf8|unicode)\$";
-
+    my $year = (localtime)[5]+1900;
     # general options available to all plugins
     if (!parsargv::parse(\@_,
@@ -93 +94 @@
              q^markup_acronyms^, \$self->{'markup_acronyms'},
              q^extract_language^, \$self->{'extract_language'},
+             q^date_extract^, \$self->{'date_extract'},
+             "maximum_date/\\d{4}/$year", \$self->{'max_year'},
+             q^no_bibliography^, \$self->{'no_biblio'},
+             "maximum_century/-?\\d{1,2}( ?B\\.C\\.E\\.)?/-1",
+             \$self->{'max_century'},
              "allow_extra_options")) {
-
+    
     print STDERR "\nThe $plugin_name plugin uses an incorrect general option (general options are those\n";
     print STDERR "available to all plugins). Check your collect.cfg configuration file.\n";
@@ -329 +335 @@
     }
     }
+
+     if($self->{'date_extract'}) {
+     my $thissection = $doc_obj->get_top_section();
+     while (defined $thissection) {
+         
+         my $text = $doc_obj->get_text($thissection);
+         &DateExtract::get_date_metadata($text, $doc_obj, 
+                         $thissection, 
+                         $self->{'no_biblio'}, 
+                         $self->{'max_year'}, 
+                         $self->{'max_century'});
+         $thissection = $doc_obj->get_next_section ($thissection);
+         }
+     }
 
     if ($self->{'extract_language'}) {