Changeset 9958

Timestamp:

2005-05-25T17:31:13+12:00 (19 years ago)

Author:

davidb

Message:

Upgrading of OAIPlug to be based on XMLPlug. OAIPlug was originally written
before XMLPlug. The rewrite means it can now take advantage of the -xslt
option.

File:

• trunk/gsdl/perllib/plugins/OAIPlug.pm (modified) (8 diffs)

trunk/gsdl/perllib/plugins/OAIPlug.pm

@@ -32 +32 @@
 use parsargv;
 
+use XMLPlug;
+
 sub BEGIN {
-    @OAIPlug::ISA = ('BasPlug');
-}
+    @OAIPlug::ISA = ('XMLPlug');
+}
+
 
 my $arguments = 
@@ -52 +55 @@
 sub new {
     my $class = shift (@_);
-    my $self = new BasPlug ($class, @_);
+    my $self = new XMLPlug ($class, @_);
     $self->{'plugin_type'} = "OAIPlug";
     # 14-05-02 To allow for proper inheritance of arguments - John Thompson
@@ -75 +78 @@
 }
 
+sub xml_start_document {
+    $self->{'in_metadata_node'} = 0;
+    $self->{'rawxml'} = "";
+}
+
+sub xml_end_document {
+}
+
+sub xml_doctype {
+    my $self = shift(@_);
+
+    my ($expat, $name, $sysid, $pubid, $internal) = @_;
+
+    # allow the short-lived and badly named "GreenstoneArchive" files to be processed
+    # as well as the "Archive" files which should now be created by import.pl
+    die "" if ($name !~ /^OAI-PMH$/);
+
+    my $outhandle = $self->{'outhandle'};
+    print $outhandle "OAIPlug: processing $self->{'file'}\n" if $self->{'verbosity'} > 1;
+    print STDERR "<Processing n='$self->{'file'}' p='OAIPlug'>\n" if $self->{'gli'};
+
+}
+
+
+sub xml_start_tag {
+    my $self = shift(@_);
+    my ($expat,$element) = @_;
+
+    my %attr_hash = %_;
+
+    my $attr = "";
+    map { $attr .= " $_=$attr_hash{$_}"; } keys %attr_hash;
+
+    $self->{'rawxml'} .= "<$element$attr>";
+
+    if ($element eq "metadata") {
+    $self->{'in_metadata_node'} = 1;
+    $self->{'metadata_xml'} = "";
+    }
+
+    if ($self->{'in_metadata_node'}) {
+    $self->{'metadata_xml'} .= "<$element$attr>";
+    }
+}
+
+sub xml_end_tag {
+    my $self = shift(@_);
+    my ($expat, $element) = @_;
+
+    $self->{'rawxml'} .= "</$element>";
+
+    if ($self->{'in_metadata_node'}) {
+    $self->{'metadata_xml'} .= "</$element>";
+    }
+
+    if ($element eq "metadata") {
+    my $textref = \$self->{'metadata_xml'};
+    my $metadata = $self->{'metadata'};
+    $self->extract_oai_metadata($textref,$metadata);
+
+    $self->{'in_metadata_node'} = 0;    
+    }
+
+
+}
+
+sub xml_text {
+    my $self = shift(@_);
+    my ($expat) = @_;
+
+    $self->{'rawxml'} .= $_;
+
+    if ($self->{'in_metadata_node'}) {
+    $self->{'metadata_xml'} .= $_;
+    }
+}
+
+
+
 
 sub read {
@@ -87 +169 @@
 
     return 0 if ((-d $filename) && ($filename =~ m/srcdocs$/));
+
+    if ($self->SUPER::read(@_)) {
+
+    # Do encoding stuff
+    my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
     
-    if ($self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/) {
-    $self->{'num_blocked'} ++;
-    return 0;
-    }
-    if ($filename !~ /$self->{'process_exp'}/ || !-f $filename) {
+    my $url_array = $metadata->{'URL'};
+    my $num_urls = scalar(@$url_array);
+    
+    my $srcdoc_exists = 0;
+    my $srcdoc_pos = 0;
+    my $filename_dir = &util::filename_head($filename);
+    
+    for (my $i=0; $i<$num_urls; $i++) {
+        
+        if ($url_array->[$i] !~ m/^(http|ftp):/) {
+        
+        my $src_filename = &util::filename_cat($filename_dir, $url_array->[$i]);
+        
+        if (-e $src_filename) {
+            $srcdoc_pos = $i;
+            $srcdoc_exists = 1;
+        }
+        }
+    }
+    
+    if ($srcdoc_exists)
+    {
+        print $outhandle "OAIPlug: passing metadata on to $url_array->[0]\n"
+        if ($self->{'verbosity'}>1);
+        
+        
+        # Make pretty print metadata table stick with src filename
+        my $ppmd_table = $self->{'ppmd_table'};
+        $metadata->{'prettymd'} = [ $ppmd_table ];
+        $self->{'ppmd_table'} = undef;
+        
+        return &plugin::read ($pluginfo, $filename_dir, $url_array->[0],
+                  $metadata, $processor, $maxdocs, $total_count, $gli);
+    }
+    else
+    {
+        # create a new document
+        my $doc_obj = new doc ($filename, "indexed_doc");
+        my $top_section = $doc_obj->get_top_section;
+        my $plugin_type = $self->{'plugin_type'};
+        
+        $doc_obj->add_utf8_metadata($top_section, "Language", $language);
+        $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
+        $doc_obj->add_utf8_metadata($top_section, "Plugin", $plugin_type);
+        $doc_obj->add_metadata($top_section, "FileFormat", "OAI");
+        $doc_obj->add_metadata($top_section, "FileSize", (-s $filename));
+        
+        # include any metadata passed in from previous plugins 
+        # note that this metadata is associated with the top level section
+        $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
+        
+        # do plugin specific processing of doc_obj
+        my $textref = \$self->{'rawxml'};
+        unless (defined ($self->process($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj))) {
+        print STDERR "<ProcessingError n='$file'>\n" if ($gli);
+        return -1;
+        }
+        
+        # do any automatic metadata extraction
+        $self->auto_extract_metadata ($doc_obj);
+        
+        # add an OID
+        $doc_obj->set_OID();
+        
+        my $ppmd_table = $self->{'ppmd_table'};
+        $doc_obj->add_utf8_metadata($top_section,"prettymd",$ppmd_table);
+        $self->{'ppmd_table'} = undef;
+        
+        # process the document
+        $processor->process($doc_obj);
+        
+        $self->{'num_processed'} ++;
+        
+        return 1; # processed the file
+    }
+    }
+    else {
     return undef;
-    }
-    $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
-
-    # Do encoding stuff
-    my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
-
-    ####
-    # Above code exactly the same as in BasPlug
-    # => consider making supporting function?
-    ###
-    
-    # read in file ($text will be in utf8)
-    my $text = "";
-    $self->read_file ($filename, $encoding, $language, \$text);
-
-    if (!length ($text)) {
-    print $outhandle "$plugin_name: ERROR: $file contains no text\n" if $self->{'verbosity'};
-    return 0;
-    }
-
-    print STDERR "<Processing n='$file' p='OAIPlug'>\n" if ($gli);
-    print $outhandle "OAIPlug: extracting metadata from $file\n"
-    if ($self->{'verbosity'}>1);
-
-    $self->extract_oai_metadata(\$text,$metadata);
-
-    my $url_array = $metadata->{'URL'};
-
-    if (defined $url_array && ($url_array->[0] !~ m/^http:/))
-    {
-    ## my $source_file =  &util::filename_cat($base_dir, $file);
-
-    my $url_base_dir = &util::filename_head($filename);
-
-##  print STDERR "*** url base dir = $url_base_dir/$url_array->[0]\n";
-    print $outhandle "OAIPlug: passing metadata on to $url_array->[0]\n"
-        if ($self->{'verbosity'}>1);
-
-    return &plugin::read ($pluginfo, $url_base_dir, $url_array->[0],
-                  $metadata, $processor, $maxdocs, $total_count, $gli);
-    }
-    else
-    {
-    # create a new document
-    my $doc_obj = new doc ($filename, "indexed_doc");
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Language", $language);
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Encoding", $encoding);
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
-    $doc_obj->add_metadata($doc_obj->get_top_section(), "FileFormat", "OAI");
-    $doc_obj->add_metadata($doc_obj->get_top_section(), "FileSize", (-s $filename));
-
-
-    # include any metadata passed in from previous plugins 
-    # note that this metadata is associated with the top level section
-    $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
-
-
-    # do plugin specific processing of doc_obj
-    unless (defined ($self->process(\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj))) {
-        print STDERR "<ProcessingError n='$file'>\n" if ($gli);
-        return -1;
-    }
-
-    # do any automatic metadata extraction
-    $self->auto_extract_metadata ($doc_obj);
-
-    # add an OID
-    $doc_obj->set_OID();
-
-    # process the document
-    $processor->process($doc_obj);
-
-    return 1; # processed the file
     }
 }
@@ -199 +285 @@
 
 
+# Improvement is to merge this with newer version in MetadataPass
+
+sub open_prettyprint_metadata_table
+{
+    my $self = shift(@_);
+
+    my $att   = "width=100% cellspacing=2";
+    my $style = "style=\'border-bottom: 4px solid #000080\'";
+
+    $self->{'ppmd_table'} = "\n<table $att $style>";
+}
+
+sub add_prettyprint_metadata_line 
+{
+    my $self = shift(@_);
+    my ($metaname, $metavalue_utf8) = @_;
+
+    $metavalue_utf8 =~ s/hdl\.handle\.net/mcgonagall.cs.waikato.ac.nz:8080\/dspace\/handle/;
+    $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
+
+    $self->{'ppmd_table'} .= "  <tr bgcolor=#b5d3cd>\n";
+    $self->{'ppmd_table'} .= "    <td width=30%>\n";
+    $self->{'ppmd_table'} .= "      $metaname\n";
+    $self->{'ppmd_table'} .= "    </td>\n";
+    $self->{'ppmd_table'} .= "    <td>\n";
+    $self->{'ppmd_table'} .= "      $metavalue_utf8\n";
+    $self->{'ppmd_table'} .= "    </td>\n";
+    $self->{'ppmd_table'} .= "  </tr>\n";
+
+}
+
+sub close_prettyprint_metadata_table
+{
+    my $self = shift(@_);
+
+    $self->{'ppmd_table'} .= "</table>\n";
+}
+
+
+
 
 sub extract_oai_metadata {
@@ -205 +331 @@
     my $outhandle = $self->{'outhandle'};
 
-    
-    if ($$textref =~ m/<metadata>(.*?)<\/metadata>/s)
+    # Only handles DC metadata
+
+    $self->open_prettyprint_metadata_table();
+
+    if ($$textref =~ m/<metadata\s*>(.*?)<\/metadata\s*>/s)
     {
     $metadata_text = $1;
@@ -215 +344 @@
         # if URL given for document as identifier metadata, store it ...
         # $doc_obj->add_utf8_metadata($cursection, "URL", $web_url);
+
         my $metaname = $1;
         my $metavalue = $2;
@@ -237 +367 @@
         }
 
-
+        $self->add_prettyprint_metadata_line($metaname, $metavalue);
+        
     }
     }
+
+    $self->close_prettyprint_metadata_table();
 }