1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # MARCXMLPlugin.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) 2001 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 | # Processes MARCXML documents. Note that this plugin does no
|
---|
27 | # syntax checking (though the XML::Parser module tests for
|
---|
28 | # well-formedness).
|
---|
29 |
|
---|
30 | package MARCXMLPlugin;
|
---|
31 |
|
---|
32 | use ReadXMLFile;
|
---|
33 | use ReadTextFile;
|
---|
34 | use marcmapping;
|
---|
35 |
|
---|
36 | use strict;
|
---|
37 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
---|
38 |
|
---|
39 | sub BEGIN {
|
---|
40 | @MARCXMLPlugin::ISA = ('ReadXMLFile', 'ReadTextFile');
|
---|
41 | }
|
---|
42 |
|
---|
43 | my $arguments = [{'name' => "metadata_mapping_file",
|
---|
44 | 'desc' => "{MARCXMLPlugin.metadata_mapping_file}",
|
---|
45 | 'type' => "string",
|
---|
46 | 'deft' => "marctodc.txt",
|
---|
47 | 'reqd' => "no" }];
|
---|
48 |
|
---|
49 | my $options = { 'name' => "MARCXMLPlugin",
|
---|
50 | 'desc' => "{MARCXMLPlugin.desc}",
|
---|
51 | 'abstract' => "no",
|
---|
52 | 'inherits' => "yes",
|
---|
53 | 'args' => $arguments
|
---|
54 | };
|
---|
55 |
|
---|
56 | sub new {
|
---|
57 | my ($class) = shift (@_);
|
---|
58 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
59 | push(@$pluginlist, $class);
|
---|
60 |
|
---|
61 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
62 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
63 |
|
---|
64 | # we want to be able to use the textcat methods from ReadTextFile
|
---|
65 | # to get the language and encoding
|
---|
66 | new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists, 1);
|
---|
67 |
|
---|
68 | my $self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
69 |
|
---|
70 | $self->{'content'} = "";
|
---|
71 | $self->{'xmlcontent'} = "";
|
---|
72 | $self->{'record_count'} = 1;
|
---|
73 | $self->{'language'} = "";
|
---|
74 | $self->{'encoding'} = "";
|
---|
75 | $self->{'marc_mapping'} = {};
|
---|
76 | $self->{'current_code'} = "";
|
---|
77 | $self->{'current_tag'} = "";
|
---|
78 | $self->{'current_element'} = "";
|
---|
79 | $self->{'metadata_mapping'} = undef;
|
---|
80 | $self->{'num_processed'} = 0;
|
---|
81 | $self->{'indent'} = 0;
|
---|
82 |
|
---|
83 | return bless $self, $class;
|
---|
84 | }
|
---|
85 |
|
---|
86 |
|
---|
87 |
|
---|
88 | sub get_doctype {
|
---|
89 | my $self = shift(@_);
|
---|
90 |
|
---|
91 | return "collection";
|
---|
92 | }
|
---|
93 |
|
---|
94 |
|
---|
95 | sub init {
|
---|
96 | my $self = shift (@_);
|
---|
97 | my ($verbosity, $outhandle, $failhandle) = @_;
|
---|
98 |
|
---|
99 | ## the mapping file has already been loaded
|
---|
100 | if (defined $self->{'metadata_mapping'} ){
|
---|
101 | $self->SUPER::init(@_);
|
---|
102 | return;
|
---|
103 | }
|
---|
104 |
|
---|
105 | # read in the metadata mapping files
|
---|
106 | my $mm_files = &util::locate_config_files($self->{'metadata_mapping_file'});
|
---|
107 |
|
---|
108 |
|
---|
109 | if (scalar(@$mm_files)==0)
|
---|
110 | {
|
---|
111 | my $msg = "MARCXMLPlugin ERROR: Can't locate mapping file \"" .
|
---|
112 | $self->{'metadata_mapping_file'} . "\".\n " .
|
---|
113 | " No metadata will be extracted from MARCXML files.\n";
|
---|
114 |
|
---|
115 | print $outhandle $msg;
|
---|
116 | print $failhandle $msg;
|
---|
117 | $self->{'metadata_mapping'} = undef;
|
---|
118 | # We pick up the error in process() if there is no $mm_file
|
---|
119 | # If we exit here, then pluginfo.pl will exit too!
|
---|
120 | }
|
---|
121 | else {
|
---|
122 | $self->{'metadata_mapping'} = &marcmapping::parse_marc_metadata_mapping($mm_files, $outhandle);
|
---|
123 | }
|
---|
124 |
|
---|
125 |
|
---|
126 | ##map { print STDERR $_."=>".$self->{'metadata_mapping'}->{$_}."\n"; } keys %{$self->{'metadata_mapping'}};
|
---|
127 |
|
---|
128 | $self->SUPER::init(@_);
|
---|
129 | }
|
---|
130 |
|
---|
131 | # Called for DOCTYPE declarations - use die to bail out if this doctype
|
---|
132 | # is not meant for this plugin
|
---|
133 | sub xml_doctype {
|
---|
134 | my $self = shift(@_);
|
---|
135 |
|
---|
136 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
137 | return;
|
---|
138 |
|
---|
139 | }
|
---|
140 |
|
---|
141 |
|
---|
142 | sub xml_start_document {
|
---|
143 | my $self = shift(@_);
|
---|
144 |
|
---|
145 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
146 |
|
---|
147 |
|
---|
148 | my $file = $self->{'file'};
|
---|
149 | my $filename = $self->{'filename'};
|
---|
150 |
|
---|
151 | my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
|
---|
152 |
|
---|
153 | $self->{'language'} = $language;
|
---|
154 | $self->{'encoding'} = $encoding;
|
---|
155 | $self->{'element_count'} = 1;
|
---|
156 | $self->{'indent'} = 0;
|
---|
157 | my $outhandle = $self->{'outhandle'};
|
---|
158 | print $outhandle "MARCXMLPlugin: processing $self->{'file'}\n" if $self->{'verbosity'} > 1;
|
---|
159 | print STDERR "<Processing n='$self->{'file'}' p='MARCXMLPlugin'>\n" if $self->{'gli'};
|
---|
160 |
|
---|
161 | }
|
---|
162 |
|
---|
163 | sub xml_end_document {
|
---|
164 |
|
---|
165 | }
|
---|
166 |
|
---|
167 | sub xml_start_tag {
|
---|
168 | my $self = shift;
|
---|
169 | my $expat = shift;
|
---|
170 | my $element = shift;
|
---|
171 |
|
---|
172 | my $text = $_;
|
---|
173 | my $escaped_text = $self->escape_text($_);
|
---|
174 |
|
---|
175 | $self->{'current_element'} = $element;
|
---|
176 |
|
---|
177 | ##get all atributes of this element and store it in a map name=>value
|
---|
178 | my %attr_map = ();
|
---|
179 | my $attrstring = $_;
|
---|
180 | while ($attrstring =~ /(\w+)=\"(\w+)\"/){
|
---|
181 | $attr_map{$1}=$2;
|
---|
182 | $attrstring = $'; #'
|
---|
183 | }
|
---|
184 |
|
---|
185 |
|
---|
186 | my $processor = $self->{'processor'};
|
---|
187 |
|
---|
188 | ##create a new document for each record
|
---|
189 | if ($element eq "record") {
|
---|
190 | my $filename = $self->{'filename'};
|
---|
191 | my $language = $self->{'language'};
|
---|
192 | my $encoding = $self->{'encoding'};
|
---|
193 | my $file = $self->{'file'};
|
---|
194 | my $doc_obj = new doc($filename);
|
---|
195 | $doc_obj->set_OIDtype ($processor->{'OIDtype'}, $processor->{'OIDmetadata'});
|
---|
196 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Language", $language);
|
---|
197 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Encoding", $encoding);
|
---|
198 | my ($filemeta) = $file =~ /([^\\\/]+)$/;
|
---|
199 | $self->set_Source_metadata($doc_obj, $filemeta, $encoding);
|
---|
200 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "SourceSegment", "$self->{'record_count'}");
|
---|
201 | if ($self->{'cover_image'}) {
|
---|
202 | $self->associate_cover_image($doc_obj, $filename);
|
---|
203 | }
|
---|
204 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
|
---|
205 | $doc_obj->add_metadata($doc_obj->get_top_section(), "FileFormat", "MARCXML");
|
---|
206 |
|
---|
207 | my $outhandle = $self->{'outhandle'};
|
---|
208 | print $outhandle "Record $self->{'record_count'}\n" if $self->{'verbosity'} > 1;
|
---|
209 |
|
---|
210 | $self->{'record_count'}++;
|
---|
211 | $self->{'doc_obj'} = $doc_obj;
|
---|
212 | $self->{'num_processed'}++;
|
---|
213 |
|
---|
214 | }
|
---|
215 |
|
---|
216 | ## get the marc code, for example 520
|
---|
217 | if ($element eq "datafield") {
|
---|
218 | if (defined $attr_map{'tag'} and $attr_map{'tag'} ne ""){
|
---|
219 | $self->{'current_tag'} = $attr_map{tag};
|
---|
220 | }
|
---|
221 | }
|
---|
222 |
|
---|
223 |
|
---|
224 | ## append the subcode to the marc code for example 520a or 520b
|
---|
225 | if ($element eq "subfield"){
|
---|
226 | if (defined $attr_map{'code'} and $attr_map{'code'} ne "" and $self->{'current_tag'} ne ""){
|
---|
227 | $self->{'current_code'} = $attr_map{'code'};
|
---|
228 | }
|
---|
229 | }
|
---|
230 |
|
---|
231 | if ($element eq "record"){
|
---|
232 | $self->{'indent'} = 0;
|
---|
233 | $self->{'content'} = "";
|
---|
234 | $self->{'xmlcontent'} = "";
|
---|
235 | }
|
---|
236 | else {
|
---|
237 | if ($element ne "subfield"){
|
---|
238 | $self->{'indent'} = 1;
|
---|
239 | }
|
---|
240 | else{
|
---|
241 | $self->{'indent'} = 2;
|
---|
242 | }
|
---|
243 | }
|
---|
244 |
|
---|
245 |
|
---|
246 | if ($element eq "collection") {
|
---|
247 | # remember the full start tag for <collection ...>
|
---|
248 | # This is needed to wrap around each <record> when generating its associate MARCXML file
|
---|
249 |
|
---|
250 | $self->{'xmlcollectiontag'} = $text;
|
---|
251 | }
|
---|
252 | else {
|
---|
253 | $self->{'content'} .= "<br/>" if ($element ne "record");
|
---|
254 | $self->{'content'} .= $self->calculate_indent($self->{'indent'}).$escaped_text;
|
---|
255 | $self->{'xmlcontent'} .= $text;
|
---|
256 | }
|
---|
257 |
|
---|
258 | }
|
---|
259 |
|
---|
260 |
|
---|
261 |
|
---|
262 | sub xml_end_tag {
|
---|
263 | my $self = shift(@_);
|
---|
264 | my ($expat, $element) = @_;
|
---|
265 |
|
---|
266 | my $text = $_;
|
---|
267 | my $escaped_text = $self->escape_text($_);
|
---|
268 |
|
---|
269 | if ($element eq "record" and defined $self->{'doc_obj'}) {
|
---|
270 | # process the document
|
---|
271 | my $processor = $self->{'processor'};
|
---|
272 | my $doc_obj = $self->{'doc_obj'};
|
---|
273 | $self->{'content'} .= "<br/>".$escaped_text;
|
---|
274 | $self->{'xmlcontent'} .= $text;
|
---|
275 |
|
---|
276 |
|
---|
277 | my $top_section = $doc_obj->get_top_section();
|
---|
278 |
|
---|
279 | my $tmp_marcxml_filename = &util::get_tmp_filename("xml");
|
---|
280 | if (open (XMLOUT,">$tmp_marcxml_filename")) {
|
---|
281 |
|
---|
282 | print XMLOUT "<?xml-stylesheet type=\"text/xsl\" href=\"MARC21slim2English.xsl\"?>\n";
|
---|
283 | my $xml_content = $self->{'xmlcontent'};
|
---|
284 |
|
---|
285 | $xml_content = $self->{'xmlcollectiontag'}.$xml_content."</collection>";
|
---|
286 |
|
---|
287 | print XMLOUT $xml_content;
|
---|
288 |
|
---|
289 | close(XMLOUT);
|
---|
290 |
|
---|
291 | $doc_obj->associate_file($tmp_marcxml_filename,"marcxml.xml","text/xml", $top_section);
|
---|
292 |
|
---|
293 | # assicate xsl style file for presentation as HTML
|
---|
294 | my $xsl_filename = &util::filename_cat($ENV{'GSDLHOME'},"etc","MARC21slim2English.xsl");
|
---|
295 | $doc_obj->associate_file($xsl_filename,"MARC21slim2English.xsl","text/xml", $top_section);
|
---|
296 |
|
---|
297 | }
|
---|
298 | else {
|
---|
299 | my $outhandle = $self->{'outhandle'};
|
---|
300 | print $outhandle "Warning: Unable for write out associated MARCXML file $tmp_marcxml_filename\n";
|
---|
301 | }
|
---|
302 |
|
---|
303 | # include any metadata passed in from previous plugins
|
---|
304 | # note that this metadata is associated with the top level section
|
---|
305 |
|
---|
306 | $self->extra_metadata ($doc_obj,
|
---|
307 | $doc_obj->get_top_section(),
|
---|
308 | $self->{'metadata'});
|
---|
309 |
|
---|
310 |
|
---|
311 | $self->add_OID($doc_obj, $self->{'record_count'});
|
---|
312 |
|
---|
313 | $doc_obj->add_utf8_text($doc_obj->get_top_section(),$self->{'content'});
|
---|
314 | $processor->process($doc_obj);
|
---|
315 |
|
---|
316 | ##clean up
|
---|
317 | $self->{'content'} = "";
|
---|
318 | $self->{'xmlcontent'} = "";
|
---|
319 | $self->{'doc_obj'} = undef;
|
---|
320 | return;
|
---|
321 | }
|
---|
322 |
|
---|
323 | ## map the xmlmarc to gsdl metadata
|
---|
324 | if ($element eq "datafield" and defined $self->{'doc_obj'} and defined $self->{'marc_mapping'} and defined $self->{'metadata_mapping'}){
|
---|
325 | my $metadata_mapping = $self->{'metadata_mapping'};
|
---|
326 | my $marc_mapping = $self->{'marc_mapping'};
|
---|
327 | my $doc_obj = $self->{'doc_obj'};
|
---|
328 |
|
---|
329 | ## print STDERR "**** Marc Record\n";
|
---|
330 | ## map { print STDERR $_."=>".$marc_mapping->{$_}."\n"; } keys %$marc_mapping;
|
---|
331 | ## print STDERR "**** Metadata Mapping\n";
|
---|
332 | ## map { print STDERR $_."=>".$metadata_mapping->{$_}."\n"; } keys %$metadata_mapping;
|
---|
333 |
|
---|
334 |
|
---|
335 | foreach my $marc_field (keys %$metadata_mapping){
|
---|
336 |
|
---|
337 | ## test whether this field has subfield
|
---|
338 | my $subfield = undef;
|
---|
339 | if ($marc_field =~ /(\d\d\d)(?:\$|\^)?(\w)/){
|
---|
340 | $marc_field = $1;
|
---|
341 | $subfield = $2;
|
---|
342 | }
|
---|
343 |
|
---|
344 | my $matched_field = $marc_mapping->{$marc_field};
|
---|
345 |
|
---|
346 | if (defined $matched_field) {
|
---|
347 |
|
---|
348 | my $meta_name = undef;
|
---|
349 | my $meta_value = undef;
|
---|
350 |
|
---|
351 | if (defined $subfield){
|
---|
352 | $meta_name = $metadata_mapping->{$marc_field."\$".$subfield};
|
---|
353 |
|
---|
354 | $meta_value = $matched_field->{$subfield};
|
---|
355 |
|
---|
356 | if (!defined $meta_value) {
|
---|
357 | # record read in does not have the specified subfield
|
---|
358 | next;
|
---|
359 | }
|
---|
360 | }
|
---|
361 | else {
|
---|
362 | $meta_name = $metadata_mapping->{$marc_field};
|
---|
363 |
|
---|
364 | # no subfield => get all the values
|
---|
365 | foreach my $value (sort keys %{$matched_field}) {
|
---|
366 | $meta_value .= $matched_field->{$value} ." ";
|
---|
367 | }
|
---|
368 |
|
---|
369 | }
|
---|
370 |
|
---|
371 | ## escape [ and ]
|
---|
372 | $meta_value =~ s/\[/\\\[/g;
|
---|
373 | $meta_value =~ s/\]/\\\]/g;
|
---|
374 | ##print STDERR "$meta_name=$meta_value\n";
|
---|
375 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(),$meta_name, $meta_value);
|
---|
376 |
|
---|
377 | }
|
---|
378 |
|
---|
379 | }
|
---|
380 |
|
---|
381 | ##clean up
|
---|
382 | $self->{'marc_mapping'} = undef;
|
---|
383 | $self->{'current_tag'} = "";
|
---|
384 | }
|
---|
385 |
|
---|
386 | if ($element eq "datafield"){
|
---|
387 | $self->{'indent'} = 1;
|
---|
388 | $self->{'content'} .= "<br/>".$self->calculate_indent($self->{'indent'}).$escaped_text;
|
---|
389 | $self->{'xmlcontent'} .= $text;
|
---|
390 | }
|
---|
391 | else{
|
---|
392 | $self->{'content'} .= $escaped_text;
|
---|
393 | $self->{'xmlcontent'} .= $text;
|
---|
394 | }
|
---|
395 |
|
---|
396 | }
|
---|
397 |
|
---|
398 |
|
---|
399 | sub set_OID {
|
---|
400 | my $self = shift (@_);
|
---|
401 | my ($doc_obj, $record_number) = @_;
|
---|
402 |
|
---|
403 | # first set it to generate hash value
|
---|
404 | $doc_obj->set_OID();
|
---|
405 |
|
---|
406 | # then top it up with an "r" + record-number suffix
|
---|
407 | my $id = $doc_obj->get_OID();
|
---|
408 | $doc_obj->set_OID($id . "r" . $record_number);
|
---|
409 | }
|
---|
410 |
|
---|
411 | sub xml_text {
|
---|
412 | my $self = shift(@_);
|
---|
413 | my ($expat) = @_;
|
---|
414 |
|
---|
415 | my $text = $_;
|
---|
416 | my $escaped_text = $self->escape_text($_);
|
---|
417 |
|
---|
418 | # protect against & in raw text file
|
---|
419 | $text =~ s/&/&/g; # can't have & in raw form, even in 'raw' xml text
|
---|
420 |
|
---|
421 | ## store the text of a marc code, for exapmle 520a=>A poem about....
|
---|
422 | if ($self->{'current_element'} eq "subfield" and $self->{'current_code'} ne "" and $_ ne "" ){
|
---|
423 | ##stored it in the marc_mapping
|
---|
424 |
|
---|
425 | my $current_tag = $self->{'current_tag'};
|
---|
426 | my $current_code = $self->{'current_code'};
|
---|
427 |
|
---|
428 | $self->{'marc_mapping'}->{$current_tag}->{$current_code} .= $_;
|
---|
429 |
|
---|
430 | $self->{'current_code'} = "";
|
---|
431 | }
|
---|
432 |
|
---|
433 | $self->{'content'} .= $escaped_text;
|
---|
434 | $self->{'xmlcontent'} .= $text;
|
---|
435 |
|
---|
436 | }
|
---|
437 |
|
---|
438 | sub calculate_indent{
|
---|
439 | my ($self,$num) = @_;
|
---|
440 |
|
---|
441 | my $indent ="";
|
---|
442 |
|
---|
443 | for (my $i=0; $i<$num;$i++){
|
---|
444 | $indent .= " ";
|
---|
445 | }
|
---|
446 |
|
---|
447 | return $indent;
|
---|
448 |
|
---|
449 | }
|
---|
450 |
|
---|
451 | sub escape_text {
|
---|
452 | my ($self,$text) = @_;
|
---|
453 | # special characters in the xml encoding
|
---|
454 | $text =~ s/&/&/g; # this has to be first...
|
---|
455 | $text =~ s/</</g;
|
---|
456 | $text =~ s/>/>/g;
|
---|
457 | $text =~ s/\"/"/g;
|
---|
458 |
|
---|
459 | return $text;
|
---|
460 | }
|
---|
461 |
|
---|
462 |
|
---|
463 | sub unescape_text {
|
---|
464 | my ($self,$text) = @_;
|
---|
465 | # special characters in the xml encoding
|
---|
466 | $text =~ s/</</g;
|
---|
467 | $text =~ s/>/>/g;
|
---|
468 | $text =~ s/"/\"/g;
|
---|
469 |
|
---|
470 | $text =~ s/&/&/g; # can't have & in raw form, even in unescaped xml!
|
---|
471 |
|
---|
472 | return $text;
|
---|
473 | }
|
---|
474 |
|
---|
475 |
|
---|
476 | 1;
|
---|
477 |
|
---|
478 |
|
---|