source: main/trunk/greenstone2/perllib/plugins/OAIPlugin.pm@ 23212

Last change on this file since 23212 was 23212, checked in by kjdon, 14 years ago

metadata_read no longer takes maxdocs args - metadata_read must process all docs, so that whatever few are actually processed by read will get their metadata
  • Property svn:keywords set to Author Date Id Revision
File size: 15.8 KB
Line 
1###########################################################################
2#
3# OAIPlug.pm -- basic Open Archives Initiative (OAI) plugin
4#
5# A component of the Greenstone digital library software
6# from the New Zealand Digital Library Project at the
7# University of Waikato, New Zealand.
8#
9# Copyright (C) 1999 New Zealand Digital Library Project
10#
11# This program is free software; you can redistribute it and/or modify
12# it under the terms of the GNU General Public License as published by
13# the Free Software Foundation; either version 2 of the License, or
14# (at your option) any later version.
15#
16# This program is distributed in the hope that it will be useful,
17# but WITHOUT ANY WARRANTY; without even the implied warranty of
18# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19# GNU General Public License for more details.
20#
21# You should have received a copy of the GNU General Public License
22# along with this program; if not, write to the Free Software
23# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
24#
25###########################################################################
26
27package OAIPlugin;
28
29use unicode;
30use util;
31
32use strict;
33no strict 'refs'; # allow filehandles to be variables and viceversa
34
35use ReadXMLFile;
36use ReadTextFile; # needed for subroutine textcat_get_language_encoding
37use metadatautil;
38
39sub BEGIN {
40 @OAIPlugin::ISA = ('ReadXMLFile', 'ReadTextFile');
41}
42
43my $set_list =
44 [ { 'name' => "auto",
45 'desc' => "{OAIPlugin.metadata_set.auto}" },
46 { 'name' => "dc",
47 'desc' => "{OAIPlugin.metadata_set.dc}" }
48 ];
49
50my $arguments =
51 [ { 'name' => "process_exp",
52 'desc' => "{BasePlugin.process_exp}",
53 'type' => "regexp",
54 'reqd' => "no",
55 'deft' => &get_default_process_exp() },
56 { 'name' => "metadata_set",
57 'desc' => "{OAIPlugin.metadata_set}",
58 'type' => "enumstring",
59 'reqd' => "no",
60 'list' => $set_list,
61 'deft' => "dc" },
62 { 'name' => "document_field",
63 'desc' => "{OAIPlugin.document_field}",
64 'type' => "metadata",
65 'reqd' => "no",
66 'deft' => "gi.Sourcedoc" }
67 ];
68
69my $options = { 'name' => "OAIPlugin",
70 'desc' => "{OAIPlugin.desc}",
71 'abstract' => "no",
72 'inherits' => "yes",
73 'explodes' => "yes",
74 'args' => $arguments };
75
76
77sub new {
78 my ($class) = shift (@_);
79 my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
80 push(@$pluginlist, $class);
81
82 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
83 push(@{$hashArgOptLists->{"OptList"}},$options);
84
85 new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists,1);
86 my $self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
87
88 if ($self->{'info_only'}) {
89 # don't worry about modifying options
90 return bless $self, $class;
91 }
92 # trim ex. from document field (if there)
93 $self->{'document_field'} =~ s/^ex\.//;
94 return bless $self, $class;
95}
96
97sub get_default_process_exp {
98 my $self = shift (@_);
99
100 return q^(?i)(\.oai)$^;
101}
102
103sub get_doctype {
104 my $self = shift(@_);
105
106 return "OAI-PMH";
107}
108
109sub xml_start_document {
110 my $self = shift (@_);
111 $self->{'in_metadata_node'} = 0;
112 $self->{'rawxml'} = "";
113 $self->{'saved_metadata'} = {};
114}
115
116sub xml_end_document {
117}
118
119sub xml_doctype {
120 my $self = shift(@_);
121
122 my ($expat, $name, $sysid, $pubid, $internal) = @_;
123
124 ##die "" if ($name !~ /^OAI-PMH$/);
125
126 my $outhandle = $self->{'outhandle'};
127 print $outhandle "OAIPlugin: processing $self->{'file'}\n" if $self->{'verbosity'} > 1;
128 print STDERR "<Processing n='$self->{'file'}' p='OAIPlugin'>\n" if $self->{'gli'};
129
130}
131
132
133sub xml_start_tag {
134 my $self = shift(@_);
135 my ($expat,$element) = @_;
136
137 my %attr_hash = %_;
138
139 my $attr = "";
140 map { $attr .= " $_=$attr_hash{$_}"; } keys %attr_hash;
141
142 $self->{'rawxml'} .= "<$element$attr>";
143
144 if ($element eq "metadata") {
145 $self->{'in_metadata_node'} = 1;
146 $self->{'metadata_xml'} = "";
147 }
148
149 if ($self->{'in_metadata_node'}) {
150 $self->{'metadata_xml'} .= "<$element$attr>";
151 }
152}
153
154sub xml_end_tag {
155 my $self = shift(@_);
156 my ($expat, $element) = @_;
157
158 $self->{'rawxml'} .= "</$element>";
159
160 if ($self->{'in_metadata_node'}) {
161 $self->{'metadata_xml'} .= "</$element>";
162 }
163
164 if ($element eq "metadata") {
165 my $textref = \$self->{'metadata_xml'};
166 #my $metadata = $self->{'metadata'};
167 my $metadata = $self->{'saved_metadata'};
168 $self->extract_oai_metadata($textref,$metadata);
169
170 $self->{'in_metadata_node'} = 0;
171 }
172
173
174}
175
176sub xml_text {
177 my $self = shift(@_);
178 my ($expat) = @_;
179
180 $self->{'rawxml'} .= $_;
181
182 if ($self->{'in_metadata_node'}) {
183 $self->{'metadata_xml'} .= $_;
184 }
185}
186
187
188sub metadata_read {
189 my $self = shift (@_);
190
191 my ($pluginfo, $base_dir, $file, $block_hash,
192 $extrametakeys, $extrametadata, $extrametafile,
193 $processor, $gli, $aux) = @_;
194
195 # can we process this file??
196 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
197 return undef unless $self->can_process_this_file($filename_full_path);
198
199 if (!$self->parse_file($filename_full_path, $file, $gli)) {
200 $self->{'saved_metadata'} = undef;
201 return undef;
202 }
203
204 my $new_metadata = $self->{'saved_metadata'};
205 $self->{'saved_metadata'} = undef;
206
207 # add the pretty metadata table as metadata
208 my $ppmd_table = $self->{'ppmd_table'};
209 $new_metadata->{'prettymd'} = $ppmd_table;
210 $self->{'ppmd_table'} = undef;
211
212 my $document_metadata_field = $self->{'document_field'};
213 my $url_array = $new_metadata->{$document_metadata_field};
214 if (!defined $url_array) {
215 # try ex.
216 $url_array = $new_metadata->{"ex.$document_metadata_field"};
217 }
218 my $num_urls = (defined $url_array) ? scalar(@$url_array) : 0;
219 ##print STDERR "$num_urls urls for $file\n";
220 my $srcdoc_exists = 0;
221 my $srcdoc_pos = 0;
222 my $filename_dir = &util::filename_head($filename_full_path);
223
224 # filenames in extrametadata must be relative to current dir, as
225 # DirectoryPlugin adds path info on itself
226 my ($filename_for_metadata) = $file =~ /([^\\\/]+)$/; # this assumes there will only be one record per oai file - is this always the case??
227 for (my $i=0; $i<$num_urls; $i++) {
228
229 if ($url_array->[$i] !~ m/^(https?|ftp):/) {
230
231 my $src_filename = &util::filename_cat($filename_dir, $url_array->[$i]);
232 if (-e $src_filename) {
233 $srcdoc_pos = $i;
234 $srcdoc_exists = 1;
235 # get the slashes the right way, use filename_cat
236 $filename_for_metadata = &util::filename_cat($url_array->[$i]);
237 last;
238 }
239 }
240 }
241
242 if ($srcdoc_exists) {
243 $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 1;
244 }
245 else {
246 # save the rawxml for the source document
247 $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 0;
248 $self->{'oai-files'}->{$file}->{'rawxml'} = $self->{'rawxml'};
249 $self->{'rawxml'} = "";
250 }
251
252 # return all the metadata we have extracted to the caller.
253 # Directory plug will pass it back in at read time, so we don't need to extract it again.
254 # extrametadata keys should be regular expressions
255 $filename_for_metadata = &util::filename_to_regex($filename_for_metadata);
256 $extrametadata->{$filename_for_metadata} = $new_metadata;
257 push(@$extrametakeys, $filename_for_metadata);
258 if ($srcdoc_exists) {
259 if (!defined $extrametafile->{$filename_for_metadata}) {
260 $extrametafile->{$filename_for_metadata} = {};
261 }
262 #maps the file to full path
263 $extrametafile->{$filename_for_metadata}->{$file} = $filename_full_path;
264 }
265 return 1;
266
267}
268
269
270sub read {
271 my $self = shift (@_);
272
273 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
274
275 if (!defined $self->{'oai-files'}->{$file}) {
276 return undef;
277 }
278
279 my $srcdoc_exists = $self->{'oai-files'}->{$file}->{'srcdoc_exists'};
280 if ($srcdoc_exists) {
281 # do nothing more - all the metadata has been extracted and associated with the srcdoc
282 # no more need to access details of this $file => tidy up as you go
283 delete $self->{'oai-files'}->{$file};
284 return 0; # not processed here, but don't pass on to rest of plugins
285 }
286
287 my $filename = $file;
288 $filename = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
289
290 # Do encoding stuff on metadata
291 my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
292
293 # create a new document
294 my $doc_obj = new doc ($filename, "indexed_doc", $self->{'file_rename_method'});
295 my $top_section = $doc_obj->get_top_section;
296 my $plugin_type = $self->{'plugin_type'};
297
298 my ($filemeta) = $file =~ /([^\\\/]+)$/;
299 $self->set_Source_metadata($doc_obj, $filemeta, $encoding);
300 $doc_obj->add_utf8_metadata($top_section, "Language", $language);
301 $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
302 $doc_obj->add_utf8_metadata($top_section, "Plugin", $plugin_type);
303 $doc_obj->add_metadata($top_section, "FileFormat", "OAI");
304 $doc_obj->add_metadata($top_section, "FileSize", (-s $filename));
305
306 # include any metadata passed in from previous plugins
307 # note that this metadata is associated with the top level section
308 # this will include all the metadata from the oai file that we extracted
309 # during metadata_read
310 $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
311
312 # do plugin specific processing of doc_obj
313 my $text = $self->{'oai-files'}->{$file}->{'rawxml'};
314 delete $self->{'oai-files'}->{$file};
315
316 unless (defined ($self->process(\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj))) {
317 print STDERR "<ProcessingError n='$file'>\n" if ($gli);
318 return -1;
319 }
320
321 # do any automatic metadata extraction
322 $self->auto_extract_metadata ($doc_obj);
323
324 # add an OID
325 $self->add_OID($doc_obj);
326
327 # process the document
328 $processor->process($doc_obj);
329
330 $self->{'num_processed'} ++;
331
332 return 1; # processed the file
333}
334
335
336# do plugin specific processing of doc_obj
337sub process {
338 my $self = shift (@_);
339 my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
340 my $outhandle = $self->{'outhandle'};
341
342 print STDERR "<Processing n='$file' p='OAIPlugin'>\n" if ($gli);
343 print $outhandle "OAIPlugin: processing $file\n"
344 if $self->{'verbosity'} > 1;
345
346 my $cursection = $doc_obj->get_top_section();
347
348## $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection);
349
350 # add text to document object
351
352# $$textref =~ s/<(.*?)>/$1 /g;
353 $$textref =~ s/</&lt;/g;
354 $$textref =~ s/>/&gt;/g;
355 $$textref =~ s/\[/&#91;/g;
356 $$textref =~ s/\]/&#93;/g;
357
358 $doc_obj->add_utf8_text($cursection, $$textref);
359
360 return 1;
361}
362
363
364# Improvement is to merge this with newer version in MetadataPass
365
366sub open_prettyprint_metadata_table
367{
368 my $self = shift(@_);
369
370 my $att = "width=100% cellspacing=2";
371 my $style = "style=\'border-bottom: 4px solid #000080\'";
372
373 $self->{'ppmd_table'} = "\n<table $att $style>";
374}
375
376sub add_prettyprint_metadata_line
377{
378 my $self = shift(@_);
379 my ($metaname, $metavalue_utf8) = @_;
380
381 $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
382
383 $self->{'ppmd_table'} .= " <tr bgcolor=#b5d3cd>\n";
384 $self->{'ppmd_table'} .= " <td width=30%>\n";
385 $self->{'ppmd_table'} .= " $metaname\n";
386 $self->{'ppmd_table'} .= " </td>\n";
387 $self->{'ppmd_table'} .= " <td>\n";
388 $self->{'ppmd_table'} .= " $metavalue_utf8\n";
389 $self->{'ppmd_table'} .= " </td>\n";
390 $self->{'ppmd_table'} .= " </tr>\n";
391
392}
393
394sub close_prettyprint_metadata_table
395{
396 my $self = shift(@_);
397
398 $self->{'ppmd_table'} .= "</table>\n";
399}
400
401my $qualified_dc_mapping = {
402 "alternative" => "dc.title",
403 "tableOfContents" => "dc.description",
404 "abstract" => "dc.description",
405 "created" => "dc.date",
406 "valid" => "dc.date",
407 "available" => "dc.date",
408 "issued" => "dc.date",
409 "modified" => "dc.date",
410 "dateAccepted" => "dc.date",
411 "dateCopyrighted" => "dc.date",
412 "dateSubmitted" => "dc.date",
413 "extent" => "dc.format",
414 "medium" => "dc.format",
415 "isVersionOf" => "dc.relation",
416 "hasVersion" => "dc.relation",
417 "isReplacedBy" => "dc.relation",
418 "replaces" => "dc.relation",
419 "isRequiredBy" => "dc.relation",
420 "requires" => "dc.relation",
421 "isPartOf" => "dc.relation",
422 "hasPart" => "dc.relation",
423 "isReferencedBy" => "dc.relation",
424 "references" => "dc.relation",
425 "isFormatOf" => "dc.relation",
426 "hasFormat" => "dc.relation",
427 "conformsTo" => "dc.relation",
428 "spatial" => "dc.coverage",
429 "temporal" => "dc.coverage",
430# these are now top level elements in our qualified dc metadata set
431# "audience" => "dc.any",
432# "accrualMethod" => "dc.any",
433# "accrualPeriodicity" => "dc.any",
434# "accrualPolicy" => "dc.any",
435# "instructionalMethod" => "dc.any",
436# "provenance" => "dc.any",
437# "rightsHolder" => "dc.any",
438 "mediator" => "dc.audience",
439 "educationLevel" => "dc.audience",
440 "accessRights" => "dc.rights",
441 "license" => "dc.rights",
442 "bibliographicCitation" => "dc.identifier"
443 };
444
445sub remap_dc_metadata
446{
447 my $self = shift(@_);
448
449 my ($metaname) = @_;
450
451 my ($prefix,$name) = ($metaname =~ m/^(.*?)\.(.*?)$/);
452
453 if (defined $qualified_dc_mapping->{$name}) {
454
455 return $qualified_dc_mapping->{$name}."^".$name;
456 }
457
458
459 return $metaname; # didn't get a match, return param passed in unchanged
460}
461
462
463sub extract_oai_metadata {
464 my $self = shift (@_);
465 my ($textref, $metadata) = @_;
466 my $outhandle = $self->{'outhandle'};
467
468 $self->open_prettyprint_metadata_table();
469
470 if ($$textref =~ m/<metadata\s*>(.*?)<\/metadata\s*>/s)
471 {
472 my $metadata_text = $1;
473
474 # locate and remove outermost tag (ignoring any attribute information in top-level tag)
475 my ($outer_tagname,$inner_metadata_text) = ($metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>/s);
476 # split tag into namespace and tag name
477 my($namespace,$top_level_prefix) = ($outer_tagname =~ m/^(.*?):(.*?)$/);
478 # sometimes, the dc namespace is not specified as the prefix in each element (like <dc:title>)
479 # but is rather defined in the wrapper element containing the various dc meta elements,
480 # like <dc><title></title><creator></creator></dc>.
481 # In such a case, we use this wrapper element as the top_level_prefix
482
483 # if there was no prefix, then the tag itself becomes the top_level_prefix
484 if(!defined $top_level_prefix && defined $outer_tagname) {
485 $top_level_prefix = $outer_tagname;
486 }
487
488 #process each element one by one
489 while ($inner_metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>(.*)/s)
490 {
491
492 my $metaname = $1;
493 my $metavalue = $2;
494 $inner_metadata_text = $3;
495
496 # greenstone uses . for namespace, while oai uses :
497 $metaname =~ s/:/\./;
498 # if there is no namespace, then we use the outer tag name or
499 # namespace for this element
500 if ($metaname !~ m/\./)
501 {
502 $metaname = "$top_level_prefix.$metaname";
503 }
504
505 # if metadata set is auto, leave as is, otherwise convert to
506 # specified namespace
507 if ($self->{'metadata_set'} ne "auto") {
508 if ($metaname !~ /^gi\./) { # hack to not overwrite gi metadata
509 $metaname =~ s/^([^\.]*)\./$self->{'metadata_set'}\./;
510 if ($self->{'metadata_set'} eq "dc") {
511 # convert qualified dc terms to gs version, e.g.
512 # spatial becomes coverage^spatial
513 $metaname = $self->remap_dc_metadata($metaname);
514 }
515 }
516 }
517
518 # uppercase the first char of the name
519 $metaname =~ s/\.(.)/\.\u$1/;
520 $metavalue =~ s/\[/&#91;/g;
521 $metavalue =~ s/\]/&#93;/g;
522
523 # so that GLI can see this metadata, store here as ex.dc.Title etc
524 my $ex_metaname = "ex.$metaname";
525
526 if (defined $metadata->{$ex_metaname})
527 {
528 push(@{$metadata->{$ex_metaname}},$metavalue);
529
530 }
531 else
532 {
533 $metadata->{$ex_metaname} = [ $metavalue ];
534 }
535
536 # but don't add ex to the pretty print line
537 $self->add_prettyprint_metadata_line($metaname, $metavalue);
538
539 }
540 }
541
542 $self->close_prettyprint_metadata_table();
543}
544
545## we know from the file extension, so doesn't need to check the doctype
546sub check_doctype {
547
548 return 1;
549}
550
5511;
Note: See TracBrowser for help on using the repository browser.