Changeset 10404

Timestamp:

2005-08-03T13:26:08+12:00 (19 years ago)

Author:

chi

Message:

remove the plugin arguments to WordPlug.

File:

• trunk/gsdl/perllib/plugins/StructuredHTMLPlug.pm (modified) (5 diffs)

trunk/gsdl/perllib/plugins/StructuredHTMLPlug.pm

@@ -2 +2 @@
 #
 # StructuredHTMLPlug.pm -- html plugin with extra facilities for teasing out 
-# hierarchical structure (such as h1, h2, h3 tags) in an HTML document
+# hierarchical structure (such as h1, h2, h3, or user-defined tags) in an 
+# HTML document
 #
 # A component of the Greenstone digital library software
@@ -25 +26 @@
 #
 ###########################################################################
+# This plugin is to process an HTML file where sections are divided by 
+# user-defined headings tags. As it is difficult to predict what user's definition
+# this plugin allows to detect the user-defined titles up to three levels (level1, level2, level3...)
+# as well as allows to get rid of user-defined Table of Content (TOC)...
+# format:e.g. level1 (Abstract_title|ChapterTitle|Referencing Heading) level2(SectionHeading)...
+
 package StructuredHTMLPlug;
 
@@ -30 +37 @@
 use ImagePlug;
 
+#use strict; # every perl program should have this!
+#no strict 'refs'; # make an exception so we can use variables as filehandles
+
 sub BEGIN {
-    @ISA = ('HTMLPlug');
-}
-
-my $arguments =
-    [ { 'name' => "check_toc",
-    'desc' => "StructuredHTMLPlug.check_toc'}",
-    'type' => "flag",
-    'reqd' => "no"},
-      { 'name' => "title_header",
-    'desc' => "{StructuredHTMLPlug.title_header}",
-    'type' => "regexp",
-    'reqd' => "no"},
-      { 'name' => "level1_header",
-    'desc' => "{StructuredHTMLPlug.level1_header}",
-    'type' => "regexp",
-    'reqd' => "no"},
-      { 'name' => "level2_header",
-    'desc' => "{StructuredHTMLPlug.level2_header}",
-    'type' => "regexp",
-    'reqd' => "no"},
-      { 'name' => "level3_header",
-    'desc' => "{StructuredHTMLPlug.level3_header}",
-    'type' => "regexp",
-    'reqd' => "no"},
-      { 'name' => "toc_header",
-    'desc' => "{StructuredHTMLPlug.toc_header}",
-    'type' => "regexp",
-    'reqd' => "no"},
-      { 'name' => "tof_header",
-    'desc' => "{StructuredHTMLPlug.tof_header}",
-    'type' => "regexp",
-    'reqd' => "no"}];
+    @StructuredHTMLPlug::ISA = ('HTMLPlug');
+}
+
+my $arguments = [];
 
 my $options = { 'name'     => "StructuredHTMLPlug",
@@ -105 +87 @@
 sub process {
     my $self = shift (@_);
-    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj) = @_;
+    #my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj) = @_;
+    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
     my $outhandle = $self->{'outhandle'};
 
@@ -114 +97 @@
     my $head = shift(@head_and_body);
     my $body_text = join("<body", @head_and_body);
-
+    
+    # If checkout_toc is enables, it means to get rid of toc and tof contents.
     # get rid of TOC and TOF sections and their title
-    # If check_toc is enables, it means to get rid of toc and tof contents.
-    if ($self->{'check_toc'}){
+    if ($self->{'checkout_toc'}){
     #line-height:150%;mso-ansi-language:FR'>Contents<o:p></o:p></span></b></p>
     # get rid of Table of Contents title and Table of Figures 
-    $body_text =~ s/<p[^>]*><b><span[^>]*>(Table of Content.|Content.)<o:p><\/o:p><\/span><\/b><\/p>//isg;
-    $body_text =~ s/<p[^>]*><b><span[^>]*>(Table of Figure.|Figure.)<o:p><\/o:p><\/span><\/b><\/p>//isg;
+    #$body_text =~ s/<p[^>]*><b><span[^>]*>(Table of Content.|Content.)<o:p><\/o:p><\/span><\/b><\/p>//isg;
+    #$body_text =~ s/<p[^>]*><b><span[^>]*>(Table of Figure.|Figure.)<o:p><\/o:p><\/span><\/b><\/p>//isg;
     $body_text =~ s/<p class=(($self->{'toc_header'})[^>]*)>(.+?)<\/p>//isg;
     $body_text =~ s/<p class=(($self->{'tof_header'})[^>]*)>(.+?)<\/p>//isg;