Changeset 16011

Timestamp:

2008-06-16T11:14:30+12:00 (16 years ago)

Author:

kjdon

Message:

moved the -first option to here from ReadTextFile

File:

• gsdl/trunk/perllib/plugins/AutoExtractMetadata.pm (modified) (3 diffs)

gsdl/trunk/perllib/plugins/AutoExtractMetadata.pm

@@ -45 +45 @@
 }
 
-my $arguments = [];
+my $arguments = [
+         {'name' => "first",
+          'desc' => "{AutoExtractMetadata.first}",
+          'type' => "string",
+          'reqd' => "no" }
+         ];
 
 
@@ -96 +101 @@
 }
 
-# here is where we call methods from the supporting plugins - gis and textextract
+# here is where we call methods from the supporting extractor plugins 
 sub auto_extract_metadata {
     my $self = shift(@_);
     my ($doc_obj) = @_;
 
+    if ($self->{'first'}) {
+    my $thissection = $doc_obj->get_top_section();
+    while (defined $thissection) {
+        my $text = $doc_obj->get_text($thissection);
+        $self->extract_first_NNNN_characters (\$text, $doc_obj, $thissection) if $text =~ /./;
+        $thissection = $doc_obj->get_next_section ($thissection);
+    }
+    }
     $self->extract_acronym_metadata($doc_obj);
     $self->extract_keyphrase_metadata($doc_obj);
@@ -107 +120 @@
     $self->extract_gis_metadata($doc_obj);
 
+}
+
+
+# FIRSTNNN: extract the first NNN characters as metadata
+sub extract_first_NNNN_characters {
+    my $self = shift (@_);
+    my ($textref, $doc_obj, $thissection) = @_;
+    
+    foreach my $size (split /,/, $self->{'first'}) {
+    my $tmptext =  $$textref;
+    $tmptext =~ s/^\s+//;
+    $tmptext =~ s/\s+$//;
+    $tmptext =~ s/\s+/ /gs;
+    $tmptext = substr ($tmptext, 0, $size);
+    $tmptext =~ s/\s\S*$/…/;
+    $doc_obj->add_utf8_metadata ($thissection, "First$size", $tmptext);
+    }
 }