Changeset 9398 for trunk/gsdl/perllib/plugins/BasPlug.pm

Timestamp:

2005-03-14T09:44:10+13:00 (19 years ago)

Author:

davidb

Message:

Introduction of GISBasPlug for Geographic Informatoin System support.
It was decided to put GISBasPlug in the main code, even though the rest
of what is required is bundled as an extension that must be installed
separately. GISBasPlug (included through BasPlug) only becomes active
if it can see the mapdata directory that is installed as part of the
extension. In becoming active, it adds extra -minus options to the
plugin.

File:

• trunk/gsdl/perllib/plugins/BasPlug.pm (modified) (18 diffs)

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -26 +26 @@
 package BasPlug;
 
+BEGIN {
+    die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
+}
+
 eval {require bytes};
 
@@ -48 +52 @@
 use printusage;
 
+use GISBasPlug;
+
+@ISA = ( GISBasPlug );
 
 my $unicode_list =
@@ -106 +113 @@
     'desc' => "{BasPlug.markup_acronyms}",
     'type' => "flag",
-    'reqd' => "no" },  
+    'reqd' => "no" },
       { 'name' => "extract_keyphrases",
     'desc' => "{BasPlug.extract_keyphrases}",
@@ -147 +154 @@
     'reqd' => "no" } ];
 
+my $gis_arguments = 
+    [ { 'name' => "extract_placenames",
+    'desc' => "{GISBasPlug.extract_placenames}",
+    'type' => "flag",
+    'reqd' => "no" },
+      { 'name' => "gazetteer",
+    'desc' => "{GISBasPlug.gazetteer}",
+    'type' => "string",
+    'reqd' => "no" },
+      { 'name' => "place_list",
+    'desc' => "{GISBasPlug.place_list}",
+    'type' => "flag",
+    'reqd' => "no" } ];
+
+
 my $options = { 'name'     => "BasPlug",
         'desc'     => "{BasPlug.desc}",
@@ -301 +323 @@
     my $self = {};
     $self->{'plugin_type'} = "BasPlug";
+
+    if (GISBasPlug::has_mapdata()) {
+    push(@$arguments,@$gis_arguments);
+    }
+
     my $enc = "^(";
     map {$enc .= "$_|";} keys %$encodings::encodings;
@@ -328 +355 @@
              q^extract_language^, \$self->{'extract_language'},
              q^extract_acronyms^, \$self->{'extract_acronyms'}, 
-             q^extract_keyphrases^, \$self->{'kea'}, 
-             q^extract_keyphrase_options/.*/^, \$self->{'kea_options'},
+             q^extract_keyphrases^, \$self->{'kea'}, #with extra options (UNDOCUMENTED)
+             q^extract_keyphrase_options/.*/^, \$self->{'kea_options'}, #no extra options (UNDOCUMENTED)
              qq^input_encoding/$enc/auto^, \$self->{'input_encoding'},
              qq^default_encoding/$denc/utf8^, \$self->{'default_encoding'},
              q^extract_email^, \$self->{'extract_email'},
+             q^extract_placenames^, \$self->{'extract_placenames'}, 
+             q^gazetteer/.*/^, \$self->{'gazetteer'},
+             q^place_list^, \$self->{'place_list'}, 
              q^markup_acronyms^, \$self->{'markup_acronyms'},
              q^default_language/.{2}/en^, \$self->{'default_language'},
@@ -367 +397 @@
 
     $self->{'cover_image'} = 0 if ($no_cover_image);
-    
+
+    if ($self->{'extract_placenames'}) {
+
+    my $outhandle = $self->{'outhandle'};
+
+    my $places_ref 
+        = GISBasPlug::loadGISDatabase($outhandle,$self->{'gazetteer'});
+
+    if (!defined $places_ref) {
+        print $outhandle "Warning: Error loading mapdata gazetteer \"$self->{'gazetteer'}\"\n";
+        print $outhandle "         No placename extraction will take place.\n";
+        $self->{'extract_placenames'} = undef;
+    }
+    else {
+        $self->{'places'} = $places_ref;
+    }
+    }   
     return bless $self, $class;
 }
@@ -626 +672 @@
     my $smart_block = $self->{'smart_block'};
     my $smart_block_BN = $self->{'smart_block_BN'};
-    
+
     my $filename = $file;
     $filename = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
@@ -663 +709 @@
 
     my ($filemeta) = $file =~ /([^\\\/]+)$/;
-
     # how do we know what encoding the filename is in?
     $doc_obj->add_metadata($doc_obj->get_top_section(), "Source", &ghtml::dmsafe($filemeta));
@@ -690 +735 @@
 
     $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
-    
+
     # do plugin specific processing of doc_obj
     unless (defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli))) {
@@ -699 +744 @@
     $text='';
     undef $text;
-
+   
     # do any automatic metadata extraction
     $self->auto_extract_metadata ($doc_obj);
@@ -716 +761 @@
     $processor->process($doc_obj);
 
+    if(defined($self->{'places_filename'})){
+    &util::rm($self->{'places_filename'});
+    $self->{'places_filename'} = undef;
+    }
+
     $self->{'num_processed'} ++;
     undef $doc_obj;
@@ -784 +834 @@
     return $file_derived_title;
 }
+
 
 sub title_fallback
@@ -952 +1003 @@
     foreach my $field (keys(%$metadata)) {
     # $metadata->{$field} may be an array reference
-
     if ($field eq "gsdlassocfile_tobe") {
         # 'gsdlassocfile_tobe' is artificially introduced metadata
@@ -1063 +1113 @@
     }
     } 
-
+    if ($self->{'extract_placenames'}) {
+    my $thissection = $doc_obj->get_top_section();
+    while (defined $thissection) {
+        my $text = $doc_obj->get_text($thissection);
+        $self->extract_placenames (\$text, $doc_obj, $thissection) if $text =~ /./;
+        $thissection = $doc_obj->get_next_section ($thissection);
+    }
+    } 
 
     # adding kea keyphrases
-
-    if ($self->{'kea'}) {  
+    if ($self->{'kea'}) {
     
     my $thissection = $doc_obj->get_top_section();
@@ -1079 +1135 @@
         $thissection = $doc_obj->get_next_section ($thissection);
     } 
-        
-
-    if($self->{'kea_options'}) {
+       
+    if ($self->{'kea_options'}) {
         #if kea options flag is set, call Kea with specified options 
         $list = &Kea::extract_KeyPhrases ($text, $self->{'kea_options'}); 
@@ -1088 +1143 @@
         $list = &Kea::extract_KeyPhrases ($text); 
     }
-    if($list){ 
+    if ($list){
         # if a list of kea keyphrases was returned (ie not empty)
         &gsprintf(STDERR, "{BasPlug.keyphrases}: $list\n");
@@ -1104 +1159 @@
         }
     }
-    }
- 
-    #end of kea 
+    } #end of kea 
 
     if ($self->{'first'}) {