Changeset 15871 for gsdl/trunk/perllib/plugins/ReadXMLFile.pm

Timestamp:

2008-06-05T09:26:56+12:00 (16 years ago)

Author:

kjdon

Message:

plugin overhaul: Split plug renamed to SplitTextFile, XMLPlug renamed to ReadXMLFile, ConvertToPlug renamed to ConvertBinaryFile. With the exception of BasePlugin, only 'real' plugins (top level ones) are named xxPlugin.

File:

• gsdl/trunk/perllib/plugins/ReadXMLFile.pm (modified) (18 diffs)

gsdl/trunk/perllib/plugins/ReadXMLFile.pm

@@ -1 +1 @@
 ###########################################################################
 #
-# XMLPlug.pm -- base class for XML plugins
+# ReadXMLFile.pm -- base class for XML plugins
 # A component of the Greenstone digital library software
 # from the New Zealand Digital Library Project at the 
@@ -24 +24 @@
 ###########################################################################
 
-package XMLPlug;
-
-use BasPlug;
+package ReadXMLFile;
+
+use BasePlugin;
 use doc;
 use strict;
@@ -32 +32 @@
 
 sub BEGIN {
-    @XMLPlug::ISA = ('BasPlug');
+    @ReadXMLFile::ISA = ('BasePlugin');
     unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
 }
@@ -40 +40 @@
 my $arguments =
     [ { 'name' => "process_exp",
-    'desc' => "{BasPlug.process_exp}",
+    'desc' => "{BasePlugin.process_exp}",
     'type' => "regexp",
     'deft' => &get_default_process_exp(),
     'reqd' => "no" },
       { 'name' => "xslt",
-    'desc' => "{XMLPlug.xslt}",
+    'desc' => "{ReadXMLFile.xslt}",
     'type' => "string",
     'deft' => "",
     'reqd' => "no" } ];
 
-my $options = { 'name'     => "XMLPlug",
-        'desc'     => "{XMLPlug.desc}",
+my $options = { 'name'     => "ReadXMLFile",
+        'desc'     => "{ReadXMLFile.desc}",
         'abstract' => "yes",
         'inherits' => "yes",
@@ -61 +61 @@
     push(@$pluginlist, $class);
 
-    if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
-    if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
-    
-    # $self is global for use within subroutines called by XML::Parser
-    my $self = new BasPlug($pluginlist, $inputargs, $hashArgOptLists);
+    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
+    push(@{$hashArgOptLists->{"OptList"}},$options);
+    
+    my $self = new BasePlugin($pluginlist, $inputargs, $hashArgOptLists);
 
     if ($self->{'info_only'}) {
-    # don't worry about any options etc
+    # don't worry about creating the XML parser as all we want is the 
+    # list of plugin options
     return bless $self, $class;
     }
 
     my $parser = new XML::Parser('Style' => 'Stream',
-                                 'Pkg' => 'XMLPlug',
+                                 'Pkg' => 'ReadXMLFile',
                                  'PluginObj' => $self,
                  'Handlers' => {'Char' => \&Char,
@@ -198 +198 @@
     if (defined $result) {
     # we think we are processing this, but check that we actually are
-    my $filename = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
+    my $filename = $self->get_full_filename($base_dir, $file);
 
     if ($self->check_doctype($filename)) {
@@ -207 +207 @@
 }
 
+# we need to implement read cos we are not just using process_exp to determine
+# whether to process this or not.
 sub read {
     my $self = shift (@_);  
@@ -213 +215 @@
 
     # Make sure we're processing the correct file, do blocking etc
-    my ($block_status,$filename) = $self->read_block(@_);    
+    my ($block_status,$filename_full_path) = $self->read_block(@_);    
     return $block_status if ((!defined $block_status) || ($block_status==0));
 
     ## check the doctype to see whether we really want to process the file
-    if (!$self->check_doctype($filename)) {
+    if (!$self->check_doctype($filename_full_path)) {
     # this file is not for us
     return undef;
@@ -225 +227 @@
     $self->{'base_dir'} = $base_dir;
     $self->{'file'} = $file;
-    $self->{'filename'} = $filename;
+    $self->{'filename'} = $filename_full_path;
     $self->{'processor'} = $processor;
     $self->{'metadata'} = $metadata;
@@ -233 +235 @@
     if (defined $xslt && ($xslt ne "")) {
         # perform xslt
-        my $transformed_xml = $self->apply_xslt($xslt,$filename);
+        my $transformed_xml = $self->apply_xslt($xslt,$filename_full_path);
 
         # feed transformed file (now in memory as string) into XML parser
@@ -239 +241 @@
     }
     else {
-        $self->{'parser'}->parsefile($filename);
+        $self->{'parser'}->parsefile($filename_full_path);
     }
     };
@@ -246 +248 @@
 
     # parsefile may either croak somewhere in XML::Parser (e.g. because
-    # the document is not well formed) or die somewhere in XMLPlug or a
+    # the document is not well formed) or die somewhere in ReadXMLFile or a
     # derived plugin (e.g. because we're attempting to process a
     # document whose DOCTYPE is not meant for this plugin). For the
@@ -271 +273 @@
 }
 
-# the following two methods are for if you want to do the parsing from a
-# plugin that inherits from this. it seems that you can't call the parse 
-# methods directly. WHY???
-#
-# [Stefan 27/5/07] These two methods may not be necessary any more as I've
-# fixed XMLPlug so $self is no longer required to be a global variable
-# (that was why inheritance wasn't working quite right with XMLPlug I
-# think). I don't really know what other plugins rely on these methods
-# though so have left them here for now.
-sub parse_file {
-    my $self = shift (@_); 
-    my ($filename) = @_;
-    $self->{'parser'}->parsefile($filename);
-}
-
-sub parse_string {
-    my $self = shift (@_); 
-    my ($xml_string) = @_;
-    $self->{'parser'}->parse($xml_string);
-}
 
 sub get_default_process_exp {
@@ -344 +326 @@
 
     my ($expat, $name, $sysid, $pubid, $internal) = @_;
-    die "XMLPlug Cannot process XML document with DOCTYPE of $name";
+    die "ReadXMLFile Cannot process XML document with DOCTYPE of $name";
 }
 
@@ -395 +377 @@
     $self->{'doc_obj'} = new doc ($self->{'filename'}, "indexed_doc");
     $self->{'doc_obj'}->set_OIDtype ($self->{'processor'}->{'OIDtype'}, $self->{'processor'}->{'OIDmetadata'});
+    $self->{'doc_obj'}->add_utf8_metadata($self->{'doc_obj'}->get_top_section(), "Plugin", "$self->{'plugin_type'}");
+
+    # do we want other auto metadata here (see BasePlugin.read_into_doc_obj)
 }
 
@@ -400 +385 @@
     my $self = shift(@_);
     my $doc_obj = $self->{'doc_obj'};
+
+    # do we want other auto stuff here, see BasePlugin.read_into_doc_obj
+
     # include any metadata passed in from previous plugins 
     # note that this metadata is associated with the top level section
@@ -410 +398 @@
    
     # add an OID
-    $doc_obj->set_OID();
+    $self->add_OID();
     
     $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
@@ -419 +407 @@
     
     $self->{'num_processed'} ++;
+    undef $self->{'doc_obj'};
+    undef $doc_obj; # is this the same as above??
 }