Changeset 10275 for trunk/gsdl/perllib/plugins/PPTPlug.pm

Timestamp:

2005-07-25T13:52:52+12:00 (19 years ago)

Author:

chi

Message:

A modification to allow a secondary plugin setting through ConvertToPlug

File:

• trunk/gsdl/perllib/plugins/PPTPlug.pm (modified) (3 diffs)

trunk/gsdl/perllib/plugins/PPTPlug.pm

@@ -41 +41 @@
     'type' => "regexp",
     'reqd' => "no",
-    'deft' => &get_default_process_exp() },
+    'deft' => &get_default_process_exp()}
       ];
 
@@ -55 +55 @@
     push(@$pluginlist, $class);
 
+    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
+    my $ws_arg = { 'name' => "windows_scripting",
+               'desc' => "PPTPlug.windows_scripting}",
+               'type' => "flag",
+               'reqd' => "no" };
+    push(@$arguments,$ws_arg);
+    }
+    
     if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
     if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
 
     my $self = (defined $hashArgOptLists)? new ConvertToPlug($pluginlist,$inputargs,$hashArgOptLists): new ConvertToPlug($pluginlist,$inputargs);
+ 
+    # ppthtml outputs utf-8 already.
+    #these are passed through to gsConvert.pl by ConvertToPlug.pm
+    $self->{'convert_options'} = "-windows_scripting" if $self->{'window_scripting'};
+    my $secondary_plugin_options = $self->{'secondary_plugin_options'};
 
-    # ppthtml outputs utf-8 already.
+    if ($self->{'windows_scripting'} && ($self->{'convert_to'} eq "PagedImg")) {
+        $secondary_plugin_options->{'PagedImgPlug'} = [];
+    } else {
+    $secondary_plugin_options->{'HTMLPlug'} = [];
+    }
+    my $html_options = $secondary_plugin_options->{'HTMLPlug'};
+    my $pageimg_options = $secondary_plugin_options->{'PagedImgPlug'};
+
     if ($self->{'input_encoding'} eq "auto") {
     $self->{'input_encoding'} = "utf8";
+    $self->{'extract_language'} = 1;
+    if (defined $secondary_plugin_options->{'HTMLPlug'}){
+        push(@$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+)?');
+    } 
+    if (defined $secondary_plugin_options->{'PagedImgPlug'}){
+        push(@$pageimg_options,"-input_encoding", "utf8");
+        push(@$pageimg_options,"-extract_language");
+    }
     }
 
-    return bless $self, $class;
+    $self = bless $self, $class;
+
+    $self->load_secondary_plugins($class,$secondary_plugin_options);
+    return $self;
 }
 
@@ -72 +109 @@
     return q^(?i)\.ppt$^;
 }
-    
+
+sub get_file_type {
+    my $self = shift (@_);
+    my $file_type = "PPT";
+    return $file_type;
+}
+
+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!
+    $self->utf8_write_file (\$text, $conv_filename);
+}
+
 sub process {
     my $self = shift (@_);
+    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
 
-    my $outhandle = $self->{'outhandle'};
-    print $outhandle "PPTPlug: passing $_[3] on to $self->{'converted_to'}Plug\n"
-    if $self->{'verbosity'} > 1;
-    print STDERR "<Processing n='$_[3]' p='PPTPlug'>\n" if ($_[6]);
-
-    return ConvertToPlug::process_type($self,"ppt",@_);
+    return $self->process_type("ppt",$base_dir,$file,$doc_obj);
 }
 
 1;
+