Changeset 1954 for trunk/gsdl/perllib/plugins/BasPlug.pm

Timestamp:

2001-02-13T10:58:26+13:00 (23 years ago)

Author:

jmt14

Message:

* empty log message *

File:

• trunk/gsdl/perllib/plugins/BasPlug.pm (modified) (7 diffs)

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -25 +25 @@
 
 package BasPlug;
-
+use Kea;
 use parsargv;
 use multiread;
@@ -130 +130 @@
     $self->{'outhandle'} = STDERR;
     my $year = (localtime)[5]+1900;
-
+    
+  
     # general options available to all plugins
     if (!parsargv::parse(\@_,
              q^process_exp/.*/^, \$self->{'process_exp'},
              q^block_exp/.*/^, \$self->{'block_exp'},
+             q^extract_acronyms^, \$self->{'extract_acronyms'}, 
+             q^extract_keyphrases^, \$self->{'kea'}, #with extra options 
+             q^extract_keyphrase_options/.*/^, \$self->{'kea_options'}, #no extra options
              qq^input_encoding/$enc/auto^, \$self->{'input_encoding'},
              qq^default_encoding/$denc/iso_8859_1^, \$self->{'default_encoding'},
-             q^extract_acronyms^, \$self->{'extract_acronyms'},
              q^extract_email^, \$self->{'extract_email'},
              q^markup_acronyms^, \$self->{'markup_acronyms'},
-             q^extract_language^, \$self->{'extract_language'},
              q^default_language/.{2}/en^, \$self->{'default_language'},
              q^first/.*/^, \$self->{'first'},
@@ -233 +235 @@
 
 sub read {
-    my $self = shift (@_);
+    my $self = shift (@_);  
+  
     my ($pluginfo, $base_dir, $file, $metadata, $processor, $maxdocs) = @_;
 
@@ -283 +286 @@
     return 0;
     }
-
+   
     # include any metadata passed in from previous plugins 
     # note that this metadata is associated with the top level section
@@ -290 +293 @@
     # do plugin specific processing of doc_obj
     return undef unless defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj));
-
+   
     # do any automatic metadata extraction
     $self->auto_extract_metadata ($doc_obj);
-
+   
     # add an OID
     $doc_obj->set_OID();
@@ -501 +504 @@
 
 # extract metadata
-sub auto_extract_metadata {
+sub auto_extract_metadata { 
+ 
+
     my $self = shift (@_);
     my ($doc_obj) = @_;
@@ -512 +517 @@
         $thissection = $doc_obj->get_next_section ($thissection);
     }
-    }   
+    } 
+
+
+#adding kea keyphrases
+    if ($self->{'kea'}) {  
+    
+    my $thissection = $doc_obj->get_top_section();
+    my $text = "";
+    my @list;
+
+    while (defined $thissection) { #loop through sections to gather whole doc
+        my $sectiontext = $doc_obj->get_text($thissection);   
+        $text = $text.$sectiontext;
+        $thissection = $doc_obj->get_next_section ($thissection);
+    } 
+       
+    if($self->{'kea_options'}) { #if kea options flag is set, call Kea with specified options 
+        @list = &Kea::extract_KeyPhrases ($text, $self->{'kea_options'}); 
+    } else { #otherwise call Kea with no options
+        @list = &Kea::extract_KeyPhrases ($text); 
+    }
+     
+    if(@list){ #if a list of kea keyphrases was returned (ie not empty)
+        my $keyphrases = $list[0]; #first arg is keyphrase list
+        my $stems = $list[1]; #second  arg is stemmed keyphrase list
+        print STDERR "keyphrases: $keyphrases\n"; 
+        print STDERR "stems: $stems\n";
+        $thissection = $doc_obj->get_top_section(); #add metadata to top section
+        $doc_obj->add_metadata($thissection, "kea", $keyphrases);
+        $doc_obj->add_metadata($thissection, "stems", $stems); 
+    }
+    } #end of kea 
+
     if ($self->{'first'}) {
     my $thissection = $doc_obj->get_top_section();