Changeset 10279 for trunk/gsdl/perllib/plugins/WordPlug.pm

Timestamp:

2005-07-25T14:16:44+12:00 (19 years ago)

Author:

chi

Message:

A modification to allow a secondary plugin setting through ConvertToPlug

File:

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

trunk/gsdl/perllib/plugins/WordPlug.pm

@@ -29 +29 @@
 
 use ConvertToPlug;
+require StructuredHTMLPlug;
 use strict;
 no strict 'refs'; # allow filehandles to be variables and viceversa
 
+
 sub BEGIN {
-    @WordPlug::ISA = ('ConvertToPlug');
+    @WordPlug::ISA = ('ConvertToPlug','StructuredHTMLPlug');
 }
 
@@ -41 +43 @@
     'type' => "regexp",
     'deft' => &get_default_process_exp(),
-    'reqd' => "no" } ];
+    'reqd' => "no" },
+      { 'name' => "title_header",
+    'desc' => "{WordPlug.title_header}",
+    'type' => "string" },
+      { 'name' => "check_toc",
+    'desc' => "{WordPlug.check_toc}",
+    'type' => "flag" },
+      { 'name' => "toc_header",
+    'desc' => "{WordPlug.toc_header}",
+    'type' => "string" },
+      { 'name' => "tof_header",
+    'desc' => "{WordPlug.tof_header}",
+    'type' => "string" },
+      { 'name' => "level1_header",
+    'desc' => "{WordPlug.level1_header}",
+    'type' => "string" },
+      { 'name' => "level2_header",
+    'desc' => "{WordPlug.level2_header}",
+    'type' => "string" },     
+      { 'name' => "level3_header",
+    'desc' => "{WordPlug.level3_header}",
+    'type' => "string" }];
 
 my $options = { 'name'     => "WordPlug",
@@ -54 +77 @@
     push(@$pluginlist, $class);
 
+    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
+    my $ws_arg = { 'name' => "windows_scripting",
+               'desc' => "{WordPlug.windows_scripting}",
+               'type' => "flag" };
+    push(@$arguments,$ws_arg);
+    }
+
     if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
     if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
@@ -59 +89 @@
     my $self = (defined $hashArgOptLists)? new ConvertToPlug($pluginlist,$inputargs,$hashArgOptLists): new ConvertToPlug($pluginlist,$inputargs);
 
-     # wvWare will always produce html files encoded as utf-8
+    #this is passed through to gsConvert.pl by ConvertToPlug.pm
+    $self->{'convert_options'} = "-windows_scripting" if $self->{'windows_scripting'};
+    
+    my $secondary_plugin_options = $self->{'secondary_plugin_options'};
+    
+    if (!defined $secondary_plugin_options->{'HTMLPlug'}) {
+    $secondary_plugin_options->{'HTMLPlug'} = [];
+    }
+    if (!defined $secondary_plugin_options->{'TEXTPlug'}) {
+    $secondary_plugin_options->{'TEXTPlug'} = [];
+    }
+
+    my $html_options = $secondary_plugin_options->{'HTMLPlug'};
+    my $text_options = $secondary_plugin_options->{'TextPlug'};
+    
+    # wvWare will always produce html files encoded as utf-8
     if ($self->{'input_encoding'} eq "auto") {
     $self->{'input_encoding'} = "utf8";
     $self->{'extract_language'} = 1;
+    push(@$html_options,"-input_encoding", "utf8");
+    push(@$html_options,"-extract_language");
+    
+    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
+    # to extract these metadata fields from the HEAD META fields
+    push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
+    #push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
     }
+    
+    $self = bless $self, $class;
+    
+    $self->load_secondary_plugins($class,$secondary_plugin_options);
 
-    return bless $self, $class;
+    #return bless $self, $class;
+    return bless $self;
 }
 
@@ -74 +131 @@
 }
 
+sub convert_post_process
+{
+    my $self = shift (@_);
+    my ($conv_filename) = @_;
+
+    my $outhandle=$self->{'outhandle'};
+
+    my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
+
+    # read in file ($text will be in utf8)
+    my $text = "";
+    $self->read_file ($conv_filename, $encoding, $language, \$text);
+
+    # turn any high bytes that aren't valid utf-8 into utf-8.
+    unicode::ensure_utf8(\$text);
+
+    # Write it out again!
+}
+
+sub get_file_type {
+    my $self = shift (@_);
+    my $file_type = "Word";
+    return $file_type;
+}
+
+# Modified to cache HTML files for efficieny reasons rather
+# than delete all.  HTML is modified not to use IE's VML.
+# VML uses WML files, so these can be deleted.
+sub cleanup_tmp_area {
+    my ($self) = @_;
+    if (defined $self->{'files_dir'}) {
+    my $html_files_dir = $self->{'files_dir'};
+
+    if (opendir(DIN,$html_files_dir)) {
+        my @wmz_files = grep( /\.wmz$/, readdir(DIN));
+        foreach my $f (@wmz_files) {
+        my $full_f = &util::filename_cat($html_files_dir,$f);
+        &util::rm($full_f);
+        }
+        closedir(DIN);
+    }
+    else {
+        # if HTML file has no supporting images, then no _files dir made
+        # => do nothing
+    }
+    }
+}
+
 # do plugin specific processing of doc_obj for HTML type
 sub process {
     my $self = shift (@_);
-    #my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
-
-    my $outhandle = $self->{'outhandle'}; 
-    print $outhandle "WordPlug: passing $_[3] on to $self->{'converted_to'}Plug\n" 
-    if $self->{'verbosity'} > 1;
-    print STDERR "<Processing n='$_[3]' p='WordPlug'>\n" if ($_[6]);
-
-    return ConvertToPlug::process_type($self,"doc",@_);
+    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
+   
+    return $self->process_type("doc", $base_dir, $file, $doc_obj);
 }
 
 1;
+