Changeset 11069

Timestamp:

2006-01-19T16:00:06+13:00 (18 years ago)

Author:

mdewsnip

Message:

Added an option to use Kea 4.0 -- this isn't included with Greenstone, but it will tell you where to get it.

Location:

trunk/gsdl/perllib

Files:

• Kea.pm (modified) (3 diffs)
• plugins/BasPlug.pm (modified) (3 diffs)
• strings.rb (modified) (2 diffs)

trunk/gsdl/perllib/Kea.pm

@@ -14 +14 @@
 # appropriate format.
 
+sub get_Kea_directory
+{
+    my $kea_version = shift(@_);
+    return &util::filename_cat($ENV{'GSDLHOME'}, "packages", "kea", "kea-$kea_version");
+}
+
 # returns a string containing comma-separated keyphrases
 sub extract_KeyPhrases {
 
     # Parsing arguments of the function
+    my $kea_version = shift(@_);
     my $doc = shift(@_); # documents text  
     my $args = shift(@_); # any options
@@ -23 +30 @@
 
     # Specifying directory names
-    my $gsdlhome = $ENV{'GSDLHOME'}; 
-    my $keahome = "$gsdlhome/packages/kea/kea-3.0";
-    my $defaultmodel = "$keahome/CSTR";
-    
+    my $keahome = &get_Kea_directory($kea_version);
+    my $defaultmodel = &util::filename_cat($keahome, "CSTR-20");
+    if ($kea_version eq "4.0") {
+    # Use a different model for Kea 4.0
+    $defaultmodel = &util::filename_cat($keahome, "FAO-20docs");
+    }
+
     # Initializing variables:
     my $command = ""; 
@@ -91 +101 @@
 
     # Write text to a temporary file doc.txt 
+    my $gsdlhome = $ENV{'GSDLHOME'}; 
     open(OUT, ">$gsdlhome/tmp/doc.txt") or die "In Kea.pm doc.txt could not be created\n";  
     print OUT $doc;

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -135 +135 @@
       { 'name' => "extract_keyphrases",
     'desc' => "{BasPlug.extract_keyphrases}",
+    'type' => "flag",
+    'reqd' => "no" },
+      { 'name' => "extract_keyphrases_kea4",
+    'desc' => "{BasPlug.extract_keyphrases_kea4}",
     'type' => "flag",
     'reqd' => "no" },
@@ -1289 +1293 @@
     } 
 
-    #adding kea keyphrases
-    if ($self->{'extract_keyphrases'}) {  
-    
-    my $thissection = $doc_obj->get_top_section();
-    my $text = "";
-    my $list;
-
-    #loop through sections to gather whole doc
-    while (defined $thissection) { 
-        my $sectiontext = $doc_obj->get_text($thissection);   
-        $text = $text.$sectiontext;
-        $thissection = $doc_obj->get_next_section ($thissection);
-    } 
-       
-    if($self->{'extract_keyphrase_options'}) { #if kea options flag is set, call Kea with specified options 
-        $list = &Kea::extract_KeyPhrases ($text, $self->{'extract_keyphrase_options'}); 
-    } else { #otherwise call Kea with no options
-        $list = &Kea::extract_KeyPhrases ($text); 
-    }
-     
-    if ($list){
-        # if a list of kea keyphrases was returned (ie not empty)
-        if ($self->{'verbosity'}) {
-        gsprintf(STDERR, "{BasPlug.keyphrases}: $list\n");
-        }
-
-        #add metadata to top section
-        $thissection = $doc_obj->get_top_section(); 
-
-        # add all key phrases as one metadata
-        $doc_obj->add_metadata($thissection, "Keyphrases", $list);
-
-        # add individual key phrases as multiple metadata
-        foreach my $keyphrase (split(',', $list)) {
-            $keyphrase =~ s/^\s+|\s+$//g;
-        $doc_obj->add_metadata($thissection, "Keyphrase", $keyphrase);
-        }
-    }
-    } #end of kea 
+    if ($self->{'extract_keyphrases'} || $self->{'extract_keyphrases_kea4'}) {
+    $self->extract_keyphrases($doc_obj);
+    }
 
     if ($self->{'first'}) {
@@ -1372 +1340 @@
     }
 }
+
+
+#adding kea keyphrases
+sub extract_keyphrases
+{
+    my $self = shift(@_);
+    my $doc_obj = shift(@_);
+
+    # Use Kea 3.0 unless 4.0 has been specified
+    my $kea_version = "3.0";
+    if ($self->{'extract_keyphrases_kea4'}) {
+    $kea_version = "4.0";
+    }
+
+    # Check that Kea exists, and tell the user where to get it if not
+    my $keahome = &Kea::get_Kea_directory($kea_version);
+    if (!-e $keahome) {
+    gsprintf(STDERR, "{BasPlug.missing_kea}\n", $keahome, $kea_version);
+    return;
+    }
+
+    my $thissection = $doc_obj->get_top_section();
+    my $text = "";
+    my $list;
+
+    #loop through sections to gather whole doc
+    while (defined $thissection) { 
+    my $sectiontext = $doc_obj->get_text($thissection);   
+    $text = $text.$sectiontext;
+    $thissection = $doc_obj->get_next_section ($thissection);
+    } 
+   
+    if($self->{'extract_keyphrase_options'}) { #if kea options flag is set, call Kea with specified options 
+    $list = &Kea::extract_KeyPhrases ($kea_version, $text, $self->{'extract_keyphrase_options'});
+    } else { #otherwise call Kea with no options
+    $list = &Kea::extract_KeyPhrases ($kea_version, $text);
+    }
+ 
+    if ($list){
+    # if a list of kea keyphrases was returned (ie not empty)
+    if ($self->{'verbosity'}) {
+        gsprintf(STDERR, "{BasPlug.keyphrases}: $list\n");
+    }
+
+    #add metadata to top section
+    $thissection = $doc_obj->get_top_section(); 
+
+    # add all key phrases as one metadata
+    $doc_obj->add_metadata($thissection, "Keyphrases", $list);
+
+    # add individual key phrases as multiple metadata
+    foreach my $keyphrase (split(',', $list)) {
+        $keyphrase =~ s/^\s+|\s+$//g;
+        $doc_obj->add_metadata($thissection, "Keyphrase", $keyphrase);
+    }
+    }
+}
+
 
 # extract acronyms from a section in a document. progress is 

trunk/gsdl/perllib/strings.rb

@@ -571 +571 @@
 BasPlug.extract_keyphrases:Extract keyphrases automatically with Kea (default settings).
 
+BasPlug.extract_keyphrases_kea4:Extract keyphrases automatically with Kea 4.0 (default settings). Kea 4.0 is a new version of Kea that has been developed for controlled indexing of documents in the domain of agriculture.
+
 BasPlug.extract_keyphrase_options:Options for keyphrase extraction with Kea. For example: mALIWEB - use ALIWEB extraction model; n5 - extract 5 keyphrase;, eGBK - use GBK encoding.
 
@@ -596 +598 @@
 
 BasPlug.maximum_year:The maximum historical date to be used as metadata (in a Common Era date, such as 1950).
+
+BasPlug.missing_kea:Error: The Kea software could not be found at %s. Please download Kea %s from http://www.nzdl.org/Kea and install it in this directory.
 
 BasPlug.must_be_implemented:BasPlug::read function must be implemented in sub-class for recursive plugins