root/main/trunk/greenstone2/perllib/plugins/MetadataXMLPlugin.pm @ 22853

Revision 22853, 11.1 KB (checked in by kjdon, 10 years ago)

print parse errors to failhandle and GLI xml as well as to outhandle

  • Property svn:keywords set to Author Date Id Revision
Line 
1###########################################################################
2#
3# MetadataXMLPlugin.pm --
4# A component of the Greenstone digital library software
5# from the New Zealand Digital Library Project at the
6# University of Waikato, New Zealand.
7#
8# Copyright (C) 2006 New Zealand Digital Library Project
9#
10# This program is free software; you can redistribute it and/or modify
11# it under the terms of the GNU General Public License as published by
12# the Free Software Foundation; either version 2 of the License, or
13# (at your option) any later version.
14#
15# This program is distributed in the hope that it will be useful,
16# but WITHOUT ANY WARRANTY; without even the implied warranty of
17# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18# GNU General Public License for more details.
19#
20# You should have received a copy of the GNU General Public License
21# along with this program; if not, write to the Free Software
22# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
23#
24###########################################################################
25
26# MetadataXMLPlugin process metadata.xml files in a collection
27
28# Here's an example of a metadata file that uses three FileSet structures
29# (ignore the # characters):
30
31#<?xml version="1.0" encoding="UTF-8" standalone="no"?>
32#<!DOCTYPE DirectoryMetadata SYSTEM "http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd">
33#<DirectoryMetadata>
34#  <FileSet>
35#    <FileName>nugget.*</FileName>
36#    <Description>
37#      <Metadata name="Title">Nugget Point, The Catlins</Metadata>
38#      <Metadata name="Place" mode="accumulate">Nugget Point</Metadata>
39#    </Description>
40#  </FileSet>
41#  <FileSet>
42#    <FileName>nugget-point-1.jpg</FileName>
43#    <Description>
44#      <Metadata name="Title">Nugget Point Lighthouse, The Catlins</Metadata>
45#      <Metadata name="Subject">Lighthouse</Metadata>
46#    </Description>
47#  </FileSet>
48#  <FileSet>
49#    <FileName>kaka-point-dir</FileName>
50#    <Description>
51#      <Metadata name="Title">Kaka Point, The Catlins</Metadata>
52#    </Description>
53#  </FileSet>
54#</DirectoryMetadata>
55
56# Metadata elements are read and applied to files in the order they appear
57# in the file.
58#
59# The FileName element describes the subfiles in the directory that the
60# metadata applies to as a perl regular expression (a FileSet group may
61# contain multiple FileName elements). So, <FileName>nugget.*</FileName>
62# indicates that the metadata records in the following Description block
63# apply to every subfile that starts with "nugget".  For these files, a
64# Title metadata element is set, overriding any old value that the Title
65# might have had.
66#
67# Occasionally, we want to have multiple metadata values applied to a
68# document; in this case we use the "mode=accumulate" attribute of the
69# particular Metadata element.  In the second metadata element of the first
70# FileSet above, the "Place" metadata is accumulating, and may therefore be
71# given several values.  If we wanted to override these values and use a
72# single metadata element again, we could set the mode attribute to
73# "override" instead.  Remember: every element is assumed to be in override
74# mode unless you specify otherwise, so if you want to accumulate metadata
75# for some field, every occurance must have "mode=accumulate" specified.
76#
77# The second FileSet element above applies to a specific file, called
78# nugget-point-1.jpg.  This element overrides the Title metadata set in the
79# first FileSet, and adds a "Subject" metadata field.
80#
81# The third and final FileSet sets metadata for a subdirectory rather than
82# a file.  The metadata specified (a Title) will be passed into the
83# subdirectory and applied to every file that occurs in the subdirectory
84# (and to every subsubdirectory and its contents, and so on) unless the
85# metadata is explictly overridden later in the import.
86
87package MetadataXMLPlugin;
88
89use strict;
90no strict 'refs';
91use BasePlugin;
92use util;
93use metadatautil;
94
95sub BEGIN {
96    @MetadataXMLPlugin::ISA = ('BasePlugin');
97    unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
98}
99
100use XMLParser;
101
102my $arguments = [
103      { 'name' => "process_exp",
104    'desc' => "{BasePlugin.process_exp}",
105    'type' => "regexp",
106    'reqd' => "no",
107    'deft' => &get_default_process_exp() }
108
109];
110
111my $options = { 'name'     => "MetadataXMLPlugin",
112        'desc'     => "{MetadataXMLPlugin.desc}",
113        'abstract' => "no",
114        'inherits' => "yes",
115        'args'     => $arguments };
116
117sub new {
118    my ($class) = shift (@_);
119    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
120    push(@$pluginlist, $class);
121
122    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
123    push(@{$hashArgOptLists->{"OptList"}},$options);
124
125    my $self = new BasePlugin($pluginlist, $inputargs, $hashArgOptLists);
126
127    if ($self->{'info_only'}) {
128    # don't worry about any options or initialisations etc
129    return bless $self, $class;
130    }
131   
132    # create XML::Parser object for parsing metadata.xml files
133    my $parser;
134    if ($]<5.008) {
135    # Perl 5.6
136    $parser = new XML::Parser('Style' => 'Stream',
137                                  'PluginObj' => $self,
138                  'Handlers' => {'Char' => \&Char,
139                         'Doctype' => \&Doctype
140                         });
141    }
142    else {
143    # Perl 5.8 or greater
144    $parser = new XML::Parser('Style' => 'Stream',
145                                  'PluginObj' => $self,
146                  'ProtocolEncoding' => 'ISO-8859-1',
147                  'Handlers' => {'Char' => \&Char,
148                         'Doctype' => \&Doctype
149                         });
150    }
151
152    $self->{'parser'} = $parser;
153    $self->{'in_filename'} = 0;
154   
155   
156    return bless $self, $class;
157}
158
159
160sub get_default_process_exp
161{
162    return q^metadata\.xml$^;
163}
164
165sub get_doctype {
166    my $self = shift(@_);
167   
168    return "(Greenstone)?DirectoryMetadata"
169}
170
171sub can_process_this_file {
172    my $self = shift(@_);
173    my ($filename) = @_;
174
175    if (-f $filename && $self->SUPER::can_process_this_file($filename) && $self->check_doctype($filename)) {
176    return 1; # its a file for us
177    }
178    return 0;
179}
180
181sub check_doctype {
182    my $self = shift (@_);
183   
184    my ($filename) = @_;
185
186    if (open(XMLIN,"<$filename")) {
187    my $doctype = $self->get_doctype();
188    ## check whether the doctype has the same name as the root element tag
189    while (defined (my $line = <XMLIN>)) {
190        ## find the root element
191        if ($line =~ /<([\w\d:]+)[\s>]/){
192        my $root = $1;
193        if ($root !~ $doctype){
194            close(XMLIN);
195            return 0;
196        }
197        else {
198            close(XMLIN);
199            return 1;
200        }
201        }
202    }
203    close(XMLIN);
204    }
205   
206    return undef; # haven't found a valid line
207   
208}
209
210sub file_block_read {
211    my $self = shift (@_);
212    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
213
214    my $filename_full_path = &util::filename_cat($base_dir, $file);
215    return undef unless $self->can_process_this_file($filename_full_path);   
216
217    $block_hash->{'metadata_files'}->{$filename_full_path} = 1;
218
219    return 1;
220}
221
222sub metadata_read
223{
224    my $self = shift (@_);
225    my ($pluginfo, $base_dir, $file, $block_hash,
226    $extrametakeys, $extrametadata,$extrametafile,
227    $processor, $maxdocs, $gli) = @_;
228
229    my $filename = &util::filename_cat($base_dir, $file);
230    return undef unless $self->can_process_this_file($filename);   
231
232    $self->{'metadata-file'} = $file;
233    $self->{'metadata-filename'} = $filename;
234
235    my $outhandle = $self->{'outhandle'};
236   
237    print STDERR "\n<Processing n='$file' p='MetadataXMLPlugin'>\n" if ($gli);
238    print $outhandle "MetadataXMLPlugin: processing $file\n" if ($self->{'verbosity'})> 1;
239    # add the file to the block list so that it won't be processed in read, as we will do all we can with it here
240    $block_hash->{'file_blocks'}->{$filename} = 1;
241
242    $self->{'metadataref'} = $extrametadata;
243    $self->{'metafileref'} = $extrametafile;
244    $self->{'metakeysref'} = $extrametakeys;
245   
246    eval {
247    $self->{'parser'}->parsefile($filename);
248    };
249
250    if ($@) {
251    my $plugin_name = ref ($self);
252    my $failhandle = $self->{'failhandle'};
253    print $outhandle "$plugin_name failed to process $file ($@)\n";
254    print $failhandle "$plugin_name failed to process $file ($@)\n";
255    print STDERR "<ProcessingError n='$file'>\n" if ($gli);
256    return -1; #error
257    }
258
259    return 1;
260
261}
262
263
264# Updated by Jeffrey 2010/04/16 @ DL Consulting Ltd.
265# Get rid off the global $self as it cause problems when there are 2+ MetadataXMLPlugin in your collect.cfg...
266# For example when you have an OAIMetadataXMLPlugin that is a child of MetadataXMLPlugin
267sub Doctype {$_[0]->{'PluginObj'}->xml_doctype(@_);}
268sub StartTag {$_[0]->{'PluginObj'}->xml_start_tag(@_);}
269sub EndTag {$_[0]->{'PluginObj'}->xml_end_tag(@_);}
270sub Text {$_[0]->{'PluginObj'}->xml_text(@_);}
271
272
273sub xml_doctype {
274    my $self = shift(@_);
275    my ($expat, $name, $sysid, $pubid, $internal) = @_;
276
277    # allow the short-lived and badly named "GreenstoneDirectoryMetadata" files
278    # to be processed as well as the "DirectoryMetadata" files which should now
279    # be created by import.pl
280    die if ($name !~ /^(Greenstone)?DirectoryMetadata$/);
281}
282
283sub xml_start_tag {
284    my $self = shift(@_);
285    my ($expat, $element) = @_;
286
287    if ($element eq "FileSet") {
288    $self->{'saved_targets'} = [];
289    $self->{'saved_metadata'} = {};
290    }
291    elsif ($element eq "FileName") {
292    $self->{'in_filename'} = 1;
293    }
294    elsif ($element eq "Metadata") {
295    $self->{'metadata_name'} = $_{'name'};
296    $self->{'metadata_value'} = "";
297    if ((defined $_{'mode'}) && ($_{'mode'} eq "accumulate")) {
298        $self->{'metadata_accumulate'} = 1;
299    } else {
300        $self->{'metadata_accumulate'} = 0;
301    }
302    }
303}
304
305sub xml_end_tag {
306    my $self = shift(@_);
307    my ($expat, $element) = @_;
308
309    if ($element eq "FileSet") {
310    foreach my $target (@{$self->{'saved_targets'}}) {
311        my $file_metadata = $self->{'metadataref'}->{$target};
312        my $saved_metadata = $self->{'saved_metadata'};
313
314        if (!defined $file_metadata) {
315        $self->{'metadataref'}->{$target} = $saved_metadata;
316
317        # not had target before
318        push (@{$self->{'metakeysref'}}, $target);
319        }
320        else {
321        &metadatautil::combine_metadata_structures($file_metadata,$saved_metadata);
322        }
323
324       
325        # now record which metadata.xml file it came from
326
327        my $file = $self->{'metadata-file'};
328        my $filename = $self->{'metadata-filename'};
329
330        if (!defined $self->{'metafileref'}->{$target}) {
331            $self->{'metafileref'}->{$target} = {};
332        }
333
334        $self->{'metafileref'}->{$target}->{$file} = $filename
335    }
336    }
337    elsif ($element eq "FileName") {
338    $self->{'in_filename'} = 0;
339    }
340    elsif ($element eq "Metadata") {
341    &metadatautil::store_saved_metadata($self,$self->{'metadata_name'}, $self->{'metadata_value'}, $self->{'metadata_accumulate'});
342    $self->{'metadata_name'} = "";
343    }
344
345}
346
347sub xml_text {
348    my $self = shift(@_);
349
350    if ($self->{'in_filename'}) {
351    # $_ == FileName content
352    push (@{$self->{'saved_targets'}}, $_);
353    }
354    elsif (defined ($self->{'metadata_name'}) && $self->{'metadata_name'} ne "") {
355    # $_ == Metadata content
356    $self->{'metadata_value'} = $_;
357    }
358}
359
360# This Char function overrides the one in XML::Parser::Stream to overcome a
361# problem where $expat->{Text} is treated as the return value, slowing
362# things down significantly in some cases.
363sub Char {
364    if ($]<5.008) {
365    use bytes;  # Necessary to prevent encoding issues with XML::Parser 2.31+ and Perl 5.6
366    }
367    $_[0]->{'Text'} .= $_[1];
368    return undef;
369}
370
371
372
3731;
Note: See TracBrowser for help on using the browser.