source: gs2-extensions/parallel-building/trunk/src/perllib/plugins/OAIPlugin.pm@ 24626

Last change on this file since 24626 was 24626, checked in by jmt12, 13 years ago

An (almost) complete copy of the perllib directory from a (circa SEP2011) head checkout from Greenstone 2 trunk - in order to try and make merging in this extension a little easier later on (as there have been some major changes to buildcol.pl commited in the main trunk but not in the x64 branch)
File size: 16.3 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;
38use MetadataRead;
39
40# methods with identical signatures take precedence in the order given in the ISA list.
41sub BEGIN {
42 @OAIPlugin::ISA = ('MetadataRead', 'ReadXMLFile', 'ReadTextFile');
43}
44
45my $set_list =
46 [ { 'name' => "auto",
47 'desc' => "{OAIPlugin.metadata_set.auto}" },
48 { 'name' => "dc",
49 'desc' => "{OAIPlugin.metadata_set.dc}" }
50 ];
51
52my $arguments =
53 [ { 'name' => "process_exp",
54 'desc' => "{BasePlugin.process_exp}",
55 'type' => "regexp",
56 'reqd' => "no",
57 'deft' => &get_default_process_exp() },
58 { 'name' => "metadata_set",
59 'desc' => "{OAIPlugin.metadata_set}",
60 'type' => "enumstring",
61 'reqd' => "no",
62 'list' => $set_list,
63 'deft' => "dc" },
64 { 'name' => "document_field",
65 'desc' => "{OAIPlugin.document_field}",
66 'type' => "metadata",
67 'reqd' => "no",
68 'deft' => "gi.Sourcedoc" }
69 ];
70
71my $options = { 'name' => "OAIPlugin",
72 'desc' => "{OAIPlugin.desc}",
73 'abstract' => "no",
74 'inherits' => "yes",
75 'explodes' => "yes",
76 'args' => $arguments };
77
78
79sub new {
80 my ($class) = shift (@_);
81 my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
82 push(@$pluginlist, $class);
83
84 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
85 push(@{$hashArgOptLists->{"OptList"}},$options);
86
87 new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists,1);
88 my $self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
89
90 if ($self->{'info_only'}) {
91 # don't worry about modifying options
92 return bless $self, $class;
93 }
94 # trim any ex. from document field iff it's the only metadata namespace prefix
95 $self->{'document_field'} =~ s/^ex\.([^.]+)$/$1/;
96 return bless $self, $class;
97}
98
99sub get_default_process_exp {
100 my $self = shift (@_);
101
102 return q^(?i)(\.oai)$^;
103}
104
105sub get_doctype {
106 my $self = shift(@_);
107
108 return "OAI-PMH";
109}
110
111sub xml_start_document {
112 my $self = shift (@_);
113 $self->{'in_metadata_node'} = 0;
114 $self->{'rawxml'} = "";
115 $self->{'saved_metadata'} = {};
116}
117
118sub xml_end_document {
119}
120
121sub xml_doctype {
122 my $self = shift(@_);
123
124 my ($expat, $name, $sysid, $pubid, $internal) = @_;
125
126 ##die "" if ($name !~ /^OAI-PMH$/);
127
128 my $outhandle = $self->{'outhandle'};
129 print $outhandle "OAIPlugin: processing $self->{'file'}\n" if $self->{'verbosity'} > 1;
130 print STDERR "<Processing n='$self->{'file'}' p='OAIPlugin'>\n" if $self->{'gli'};
131
132}
133
134
135sub xml_start_tag {
136 my $self = shift(@_);
137 my ($expat,$element) = @_;
138
139 my %attr_hash = %_;
140
141 my $attr = "";
142 map { $attr .= " $_=$attr_hash{$_}"; } keys %attr_hash;
143
144 $self->{'rawxml'} .= "<$element$attr>";
145
146 if ($element eq "metadata") {
147 $self->{'in_metadata_node'} = 1;
148 $self->{'metadata_xml'} = "";
149 }
150
151 if ($self->{'in_metadata_node'}) {
152 $self->{'metadata_xml'} .= "<$element$attr>";
153 }
154}
155
156sub xml_end_tag {
157 my $self = shift(@_);
158 my ($expat, $element) = @_;
159
160 $self->{'rawxml'} .= "</$element>";
161
162 if ($self->{'in_metadata_node'}) {
163 $self->{'metadata_xml'} .= "</$element>";
164 }
165
166 if ($element eq "metadata") {
167 my $textref = \$self->{'metadata_xml'};
168 #my $metadata = $self->{'metadata'};
169 my $metadata = $self->{'saved_metadata'};
170 $self->extract_oai_metadata($textref,$metadata);
171
172 $self->{'in_metadata_node'} = 0;
173 }
174
175
176}
177
178sub xml_text {
179 my $self = shift(@_);
180 my ($expat) = @_;
181
182 $self->{'rawxml'} .= $_;
183
184 if ($self->{'in_metadata_node'}) {
185 $self->{'metadata_xml'} .= $_;
186 }
187}
188
189
190sub metadata_read {
191 my $self = shift (@_);
192
193 my ($pluginfo, $base_dir, $file, $block_hash,
194 $extrametakeys, $extrametadata, $extrametafile,
195 $processor, $gli, $aux) = @_;
196
197 # can we process this file??
198 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
199 return undef unless $self->can_process_this_file_for_metadata($filename_full_path);
200
201 if (!$self->parse_file($filename_full_path, $file, $gli)) {
202 $self->{'saved_metadata'} = undef;
203 return undef;
204 }
205
206 my $new_metadata = $self->{'saved_metadata'};
207 $self->{'saved_metadata'} = undef;
208
209 # add the pretty metadata table as metadata
210 my $ppmd_table = $self->{'ppmd_table'};
211 $new_metadata->{'prettymd'} = $ppmd_table;
212 $self->{'ppmd_table'} = undef;
213
214 my $document_metadata_field = $self->{'document_field'};
215 my $url_array = $new_metadata->{$document_metadata_field};
216 if (!defined $url_array) {
217 # try ex.
218 $url_array = $new_metadata->{"ex.$document_metadata_field"};
219 }
220 my $num_urls = (defined $url_array) ? scalar(@$url_array) : 0;
221 ##print STDERR "$num_urls urls for $file\n";
222 my $srcdoc_exists = 0;
223 my $srcdoc_pos = 0;
224 my $filename_dir = &util::filename_head($filename_full_path);
225
226 # filenames in extrametadata must be relative to current dir, as
227 # DirectoryPlugin adds path info on itself
228 my ($filename_for_metadata) = $file =~ /([^\\\/]+)$/; # this assumes there will only be one record per oai file - is this always the case??
229 for (my $i=0; $i<$num_urls; $i++) {
230
231 if ($url_array->[$i] !~ m/^(https?|ftp):/) {
232
233 my $src_filename = &util::filename_cat($filename_dir, $url_array->[$i]);
234 if (-e $src_filename) {
235 $srcdoc_pos = $i;
236 $srcdoc_exists = 1;
237 # get the slashes the right way, use filename_cat
238 $filename_for_metadata = &util::filename_cat($url_array->[$i]);
239 last;
240 }
241 }
242 }
243
244 if ($srcdoc_exists) {
245 $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 1;
246 }
247 else {
248 # save the rawxml for the source document
249 $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 0;
250 $self->{'oai-files'}->{$file}->{'rawxml'} = $self->{'rawxml'};
251 $self->{'rawxml'} = "";
252 }
253
254 # return all the metadata we have extracted to the caller.
255 # Directory plug will pass it back in at read time, so we don't need to extract it again.
256 # extrametadata keys should be regular expressions
257 $filename_for_metadata = &util::filename_to_regex($filename_for_metadata);
258 $extrametadata->{$filename_for_metadata} = $new_metadata;
259 push(@$extrametakeys, $filename_for_metadata);
260 if ($srcdoc_exists) {
261 if (!defined $extrametafile->{$filename_for_metadata}) {
262 $extrametafile->{$filename_for_metadata} = {};
263 }
264 #maps the file to full path
265 $extrametafile->{$filename_for_metadata}->{$file} = $filename_full_path;
266 }
267 return 1;
268
269}
270
271
272sub read {
273 my $self = shift (@_);
274
275 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
276
277 if (!defined $self->{'oai-files'}->{$file}) {
278 return undef;
279 }
280
281 my $srcdoc_exists = $self->{'oai-files'}->{$file}->{'srcdoc_exists'};
282 if ($srcdoc_exists) {
283 # do nothing more - all the metadata has been extracted and associated with the srcdoc
284 # no more need to access details of this $file => tidy up as you go
285 delete $self->{'oai-files'}->{$file};
286 return 0; # not processed here, but don't pass on to rest of plugins
287 }
288
289 my $filename = $file;
290 $filename = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
291
292 # Do encoding stuff on metadata
293 my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
294
295 # create a new document
296 my $doc_obj = new doc ($filename, "indexed_doc", $self->{'file_rename_method'});
297 my $top_section = $doc_obj->get_top_section;
298 my $plugin_type = $self->{'plugin_type'};
299
300 my ($filemeta) = $file =~ /([^\\\/]+)$/;
301 my $plugin_filename_encoding = $self->{'filename_encoding'};
302 my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
303 $self->set_Source_metadata($doc_obj, $filename, $filename_encoding);
304
305 $doc_obj->add_utf8_metadata($top_section, "Language", $language);
306 $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
307 $doc_obj->add_utf8_metadata($top_section, "Plugin", $plugin_type);
308 $doc_obj->add_metadata($top_section, "FileFormat", "OAI");
309 $doc_obj->add_metadata($top_section, "FileSize", (-s $filename));
310
311 # include any metadata passed in from previous plugins
312 # note that this metadata is associated with the top level section
313 # this will include all the metadata from the oai file that we extracted
314 # during metadata_read
315 $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
316
317 # do plugin specific processing of doc_obj
318 my $text = $self->{'oai-files'}->{$file}->{'rawxml'};
319 delete $self->{'oai-files'}->{$file};
320
321 unless (defined ($self->process(\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj))) {
322 print STDERR "<ProcessingError n='$file'>\n" if ($gli);
323 return -1;
324 }
325
326 # do any automatic metadata extraction
327 $self->auto_extract_metadata ($doc_obj);
328
329 # add an OID
330 $self->add_OID($doc_obj);
331
332 # process the document
333 $processor->process($doc_obj);
334
335 $self->{'num_processed'} ++;
336
337 return 1; # processed the file
338}
339
340
341# do plugin specific processing of doc_obj
342sub process {
343 my $self = shift (@_);
344 my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
345 my $outhandle = $self->{'outhandle'};
346
347 print STDERR "<Processing n='$file' p='OAIPlugin'>\n" if ($gli);
348 print $outhandle "OAIPlugin: processing $file\n"
349 if $self->{'verbosity'} > 1;
350
351 my $cursection = $doc_obj->get_top_section();
352
353## $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection);
354
355 # add text to document object
356
357# $$textref =~ s/<(.*?)>/$1 /g;
358 $$textref =~ s/</&lt;/g;
359 $$textref =~ s/>/&gt;/g;
360 $$textref =~ s/\[/&#91;/g;
361 $$textref =~ s/\]/&#93;/g;
362
363 $doc_obj->add_utf8_text($cursection, $$textref);
364
365 return 1;
366}
367
368
369# Improvement is to merge this with newer version in MetadataPass
370
371sub open_prettyprint_metadata_table
372{
373 my $self = shift(@_);
374
375 my $att = "width=100% cellspacing=2";
376 my $style = "style=\'border-bottom: 4px solid #000080\'";
377
378 $self->{'ppmd_table'} = "\n<table $att $style>";
379}
380
381sub add_prettyprint_metadata_line
382{
383 my $self = shift(@_);
384 my ($metaname, $metavalue_utf8) = @_;
385
386 $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
387
388 $self->{'ppmd_table'} .= " <tr bgcolor=#b5d3cd>\n";
389 $self->{'ppmd_table'} .= " <td width=30%>\n";
390 $self->{'ppmd_table'} .= " $metaname\n";
391 $self->{'ppmd_table'} .= " </td>\n";
392 $self->{'ppmd_table'} .= " <td>\n";
393 $self->{'ppmd_table'} .= " $metavalue_utf8\n";
394 $self->{'ppmd_table'} .= " </td>\n";
395 $self->{'ppmd_table'} .= " </tr>\n";
396
397}
398
399sub close_prettyprint_metadata_table
400{
401 my $self = shift(@_);
402
403 $self->{'ppmd_table'} .= "</table>\n";
404}
405
406my $qualified_dc_mapping = {
407 "alternative" => "dc.title",
408 "tableOfContents" => "dc.description",
409 "abstract" => "dc.description",
410 "created" => "dc.date",
411 "valid" => "dc.date",
412 "available" => "dc.date",
413 "issued" => "dc.date",
414 "modified" => "dc.date",
415 "dateAccepted" => "dc.date",
416 "dateCopyrighted" => "dc.date",
417 "dateSubmitted" => "dc.date",
418 "extent" => "dc.format",
419 "medium" => "dc.format",
420 "isVersionOf" => "dc.relation",
421 "hasVersion" => "dc.relation",
422 "isReplacedBy" => "dc.relation",
423 "replaces" => "dc.relation",
424 "isRequiredBy" => "dc.relation",
425 "requires" => "dc.relation",
426 "isPartOf" => "dc.relation",
427 "hasPart" => "dc.relation",
428 "isReferencedBy" => "dc.relation",
429 "references" => "dc.relation",
430 "isFormatOf" => "dc.relation",
431 "hasFormat" => "dc.relation",
432 "conformsTo" => "dc.relation",
433 "spatial" => "dc.coverage",
434 "temporal" => "dc.coverage",
435# these are now top level elements in our qualified dc metadata set
436# "audience" => "dc.any",
437# "accrualMethod" => "dc.any",
438# "accrualPeriodicity" => "dc.any",
439# "accrualPolicy" => "dc.any",
440# "instructionalMethod" => "dc.any",
441# "provenance" => "dc.any",
442# "rightsHolder" => "dc.any",
443 "mediator" => "dc.audience",
444 "educationLevel" => "dc.audience",
445 "accessRights" => "dc.rights",
446 "license" => "dc.rights",
447 "bibliographicCitation" => "dc.identifier"
448 };
449
450sub remap_dc_metadata
451{
452 my $self = shift(@_);
453
454 my ($metaname) = @_;
455
456 my ($prefix,$name) = ($metaname =~ m/^(.*?)\.(.*?)$/);
457
458 if (defined $qualified_dc_mapping->{$name}) {
459
460 return $qualified_dc_mapping->{$name}."^".$name;
461 }
462
463
464 return $metaname; # didn't get a match, return param passed in unchanged
465}
466
467
468sub extract_oai_metadata {
469 my $self = shift (@_);
470 my ($textref, $metadata) = @_;
471 my $outhandle = $self->{'outhandle'};
472
473 $self->open_prettyprint_metadata_table();
474
475 if ($$textref =~ m/<metadata\s*>(.*?)<\/metadata\s*>/s)
476 {
477 my $metadata_text = $1;
478
479 # locate and remove outermost tag (ignoring any attribute information in top-level tag)
480 my ($outer_tagname,$inner_metadata_text) = ($metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>/s);
481 # split tag into namespace and tag name
482 my($namespace,$top_level_prefix) = ($outer_tagname =~ m/^(.*?):(.*?)$/);
483 # sometimes, the dc namespace is not specified as the prefix in each element (like <dc:title>)
484 # but is rather defined in the wrapper element containing the various dc meta elements,
485 # like <dc><title></title><creator></creator></dc>.
486 # In such a case, we use this wrapper element as the top_level_prefix
487
488 # if there was no prefix, then the tag itself becomes the top_level_prefix
489 if(!defined $top_level_prefix && defined $outer_tagname) {
490 $top_level_prefix = $outer_tagname;
491 }
492
493 #process each element one by one
494 while ($inner_metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>(.*)/s)
495 {
496
497 my $metaname = $1;
498 my $metavalue = $2;
499 $inner_metadata_text = $3;
500
501 # greenstone uses . for namespace, while oai uses :
502 $metaname =~ s/:/\./;
503 # if there is no namespace, then we use the outer tag name or
504 # namespace for this element
505 if ($metaname !~ m/\./)
506 {
507 $metaname = "$top_level_prefix.$metaname";
508 }
509
510 # if metadata set is auto, leave as is, otherwise convert to
511 # specified namespace
512 if ($self->{'metadata_set'} ne "auto") {
513 if ($metaname !~ /^gi\./) { # hack to not overwrite gi metadata
514 $metaname =~ s/^([^\.]*)\./$self->{'metadata_set'}\./;
515 if ($self->{'metadata_set'} eq "dc") {
516 # convert qualified dc terms to gs version, e.g.
517 # spatial becomes coverage^spatial
518 $metaname = $self->remap_dc_metadata($metaname);
519 }
520 }
521 }
522
523 # uppercase the first char of the name
524 $metaname =~ s/\.(.)/\.\u$1/;
525 $metavalue =~ s/\[/&#91;/g;
526 $metavalue =~ s/\]/&#93;/g;
527
528 # so that GLI can see this metadata, store here as ex.dc.Title etc
529 my $ex_metaname = $metaname;
530 $ex_metaname =~ s/^ex\.//; # remove any pre-existing ex. prefix
531 $ex_metaname = "ex.$ex_metaname"; # at last can prefix ex.
532
533 if (defined $metadata->{$ex_metaname})
534 {
535 push(@{$metadata->{$ex_metaname}},$metavalue);
536
537 }
538 else
539 {
540 $metadata->{$ex_metaname} = [ $metavalue ];
541 }
542
543 # but don't add ex to the pretty print line
544 $self->add_prettyprint_metadata_line($metaname, $metavalue);
545
546 }
547 }
548
549 $self->close_prettyprint_metadata_table();
550}
551
552## we know from the file extension, so doesn't need to check the doctype
553sub check_doctype {
554
555 return 1;
556}
557
5581;
Note: See TracBrowser for help on using the repository browser.