Changeset 1393 for trunk/gsdl/perllib/plugins/BasPlug.pm

Timestamp:

2000-08-11T11:12:15+12:00 (24 years ago)

Author:

say1

Message:

acronym markup functionality

File:

• trunk/gsdl/perllib/plugins/BasPlug.pm (modified) (6 diffs)

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -67 +67 @@
     print STDERR "                     file extensions.\n";
     print STDERR "   -extract_acronyms Extract acronyms from within text and set as metadata\n\n";
+    print STDERR "   -markup_acronyms  Added acronym metadata into document text\n\n";
     print STDERR "   -extract_langauge Identify the language of the text and set as metadata\n\n";
 }
@@ -90 +91 @@
              q^block_exp/.*/^, \$self->{'block_exp'},
              q^extract_acronyms^, \$self->{'extract_acronyms'},
+             q^markup_acronyms^, \$self->{'markup_acronyms'},
              q^extract_language^, \$self->{'extract_language'},
              "allow_extra_options")) {
@@ -297 +299 @@
     }
 
+    if ($self->{'markup_acronyms'}) {
+    my $thissection = $doc_obj->get_top_section();
+    while (defined $thissection) {
+        my $text = $doc_obj->get_text($thissection);
+        $text = $self->markup_acronyms ($text, $doc_obj, $thissection);
+        $doc_obj->delete_text($thissection);
+        $doc_obj->add_text($thissection, $text);
+        $thissection = $doc_obj->get_next_section ($thissection);
+    }
+    }
+
     if ($self->{'extract_language'}) {
     my $thissection = $doc_obj->get_top_section();
@@ -341 +354 @@
     my ($textref, $doc_obj, $thissection) = @_;
 
-    print STDERR " checking for acronyms ...\n" 
+    print STDERR " extracting acronyms ...\n" 
     if ($self->{'verbosity'} >= 2);
 
@@ -362 +375 @@
     if ($seen_before eq "false")
     {
+        #write it to the file ...
+        $acro->write_to_file();
+
         #do the normal acronym
         $doc_obj->add_utf8_metadata($thissection, "Acronym",  $acro->to_string());
@@ -367 +383 @@
             if ($self->{'verbosity'} >= 1);
         
-        # do the KWIC (Key Word In Context) acronym
-        my @kwic = $acro->to_string_kwic();
-        foreach my $kwic (@kwic) {
-        $doc_obj->add_utf8_metadata($thissection, "AcronymKWIC",  $kwic);
-        print STDERR "   adding ".  $kwic . "\n" 
-            if ($self->{'verbosity'} >= 2);
-        }
-    }
-    }
-    print STDERR " done with acronyms. \n" 
+#       # do the KWIC (Key Word In Context) acronym
+#       my @kwic = $acro->to_string_kwic();
+#       foreach my $kwic (@kwic) {
+#       $doc_obj->add_utf8_metadata($thissection, "AcronymKWIC",  $kwic);
+#       print STDERR "   adding ".  $kwic . "\n" 
+#           if ($self->{'verbosity'} >= 2);
+#       }
+    }
+    }
+    print STDERR " done extracting acronyms. \n" 
     if ($self->{'verbosity'} >= 2);
 }
 
+sub markup_acronyms {
+    my $self = shift (@_);
+    my ($text, $doc_obj, $thissection) = @_;
+
+    print STDERR " marking up acronyms ...\n" 
+    if ($self->{'verbosity'} >= 2);
+
+    #self is passed in to check for verbosity ...
+    $text = &acronym::markup_acronyms($text, $self);
+
+    print STDERR " done marking up acronyms. \n" 
+    if ($self->{'verbosity'} >= 2);
+
+    return $text;
+}
+
 1;