1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # docsave.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) 1999 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 | # This document processor saves a document in the
|
---|
27 | # archives directory of a collection (as xml)
|
---|
28 |
|
---|
29 | use strict;
|
---|
30 | no strict 'refs';
|
---|
31 |
|
---|
32 | package docsave;
|
---|
33 |
|
---|
34 | eval {require bytes};
|
---|
35 |
|
---|
36 | use arcinfo;
|
---|
37 | use docproc;
|
---|
38 | use util;
|
---|
39 |
|
---|
40 |
|
---|
41 | sub BEGIN {
|
---|
42 | @docsave::ISA = ('docproc');
|
---|
43 | }
|
---|
44 |
|
---|
45 | sub new {
|
---|
46 | my ($class, $collection, $archive_info, $verbosity,
|
---|
47 | $gzip, $groupsize, $outhandle) = @_;
|
---|
48 | my $self = new docproc ();
|
---|
49 |
|
---|
50 |
|
---|
51 | $groupsize=1 unless defined $groupsize;
|
---|
52 | $self->{'collection'} = $collection;
|
---|
53 | $self->{'archive_info'} = $archive_info;
|
---|
54 | $self->{'verbosity'} = $verbosity;
|
---|
55 | $self->{'gzip'} = $gzip;
|
---|
56 |
|
---|
57 | $self->{'groupsize'} = $groupsize;
|
---|
58 | $self->{'gs_count'} = 0;
|
---|
59 |
|
---|
60 | $self->{'outhandle'} = 'STDERR';
|
---|
61 | $self->{'outhandle'} = $outhandle if defined $outhandle;
|
---|
62 | # set a default for the archive directory
|
---|
63 | $self->{'archive_dir'} = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "archives");
|
---|
64 |
|
---|
65 | $self->{'sortmeta'} = undef;
|
---|
66 |
|
---|
67 | return bless $self, $class;
|
---|
68 | }
|
---|
69 |
|
---|
70 | sub setarchivedir {
|
---|
71 | my $self = shift (@_);
|
---|
72 | my ($archive_dir) = @_;
|
---|
73 |
|
---|
74 | &util::mk_all_dir ($archive_dir) unless -e $archive_dir;
|
---|
75 | $self->{'archive_dir'} = $archive_dir;
|
---|
76 | }
|
---|
77 |
|
---|
78 | sub set_sortmeta {
|
---|
79 | my $self = shift (@_);
|
---|
80 | my ($sortmeta) = @_;
|
---|
81 |
|
---|
82 | $self->{'sortmeta'} = $sortmeta;
|
---|
83 | }
|
---|
84 |
|
---|
85 | sub process {
|
---|
86 | my $self = shift (@_);
|
---|
87 | my ($doc_obj) = @_;
|
---|
88 |
|
---|
89 | my $outhandle = $self->{'outhandle'};
|
---|
90 |
|
---|
91 | if ($self->{'groupsize'} > 1) {
|
---|
92 | $self->group_process ($doc_obj);
|
---|
93 | return;
|
---|
94 | }
|
---|
95 |
|
---|
96 | # groupsize is 1 (i.e. one document per XML file) so sortmeta
|
---|
97 | # may be used
|
---|
98 |
|
---|
99 | my $OID = $doc_obj->get_OID();
|
---|
100 | $OID = "NULL" unless defined $OID;
|
---|
101 |
|
---|
102 | # get document's directory
|
---|
103 | my $doc_dir = $self->get_doc_dir ($OID);
|
---|
104 |
|
---|
105 |
|
---|
106 | # copy all the associated files, add this information as metadata
|
---|
107 | # to the document
|
---|
108 | $self->process_assoc_files ($doc_obj, $doc_dir);
|
---|
109 |
|
---|
110 | my $doc_file
|
---|
111 | = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.xml");
|
---|
112 |
|
---|
113 | #***define doctxt.xml file
|
---|
114 | my $doc_txt_file
|
---|
115 | = &util::filename_cat ($self->{'archive_dir'}, $doc_dir,"doctxt.xml");
|
---|
116 | my $working_dir
|
---|
117 | =&util::filename_cat ($self->{'archive_dir'}, $doc_dir);
|
---|
118 |
|
---|
119 | #***define docmets.xmlfile
|
---|
120 | my $doc_mets_file
|
---|
121 | = &util::filename_cat ($self->{'archive_dir'},$doc_dir, "docmets.xml");
|
---|
122 |
|
---|
123 | my $short_doc_file;
|
---|
124 | my $save_as = $self->{'saveas'} || "GA";
|
---|
125 | if ($save_as eq "GA") {
|
---|
126 | $short_doc_file = util::filename_cat ($doc_dir, "doc.xml");
|
---|
127 | } elsif ($save_as eq "METS") {
|
---|
128 | #my $short_txt_doc_file=&util::filename_cat ($doc_dir, "doctxt.xml");
|
---|
129 | $short_doc_file=&util::filename_cat ($doc_dir, "docmets.xml");
|
---|
130 | } else {
|
---|
131 | return;
|
---|
132 | }
|
---|
133 | # save for later (for close_file_output())
|
---|
134 | $self->{'short_doc_file'}=$short_doc_file;
|
---|
135 |
|
---|
136 | if ($save_as eq "GA") {
|
---|
137 | if (!open (OUTDOC, ">$doc_file")) {
|
---|
138 | print $outhandle "docsave::process could not write to file $doc_file\n";
|
---|
139 | return;
|
---|
140 | }
|
---|
141 |
|
---|
142 | # save this document
|
---|
143 | $self->output_xml_header('docsave::OUTDOC');
|
---|
144 | $doc_obj->output_section('docsave::OUTDOC',
|
---|
145 | $doc_obj->get_top_section());
|
---|
146 | $self->output_xml_footer('docsave::OUTDOC');
|
---|
147 |
|
---|
148 | close OUTDOC;
|
---|
149 | } elsif ($save_as eq "METS") {
|
---|
150 | # save the document without metadata:doctxt.xml
|
---|
151 |
|
---|
152 | if (!open(OUTDOC_TXT, ">$doc_txt_file")){
|
---|
153 | print $outhandle "docsave::process could not write to file $doc_mets_file\n";
|
---|
154 | return;
|
---|
155 | }
|
---|
156 |
|
---|
157 | $self->output_txt_xml_header('docsave::OUTDOC_TXT');
|
---|
158 | $doc_obj->output_txt_section('docsave::OUTDOC_TXT', $doc_obj->get_top_section());
|
---|
159 | #$self->output_txt_xml_footer('docsave::OUTDOC_TXT');
|
---|
160 |
|
---|
161 | # Convert doctxt.xml file to docmets.xml
|
---|
162 | if (!open(OUTDOC_METS,">$doc_mets_file")){
|
---|
163 | print $outhandle "docsave::process could not write to file $doc_mets_file\n";
|
---|
164 | return;
|
---|
165 | }
|
---|
166 |
|
---|
167 | $self->output_mets_xml_header('docsave::OUTDOC_METS', $OID);
|
---|
168 | $doc_obj->output_mets_section('docsave::OUTDOC_METS',
|
---|
169 | $doc_obj->get_top_section(),
|
---|
170 | $working_dir);
|
---|
171 | $self->output_mets_xml_footer('docsave::OUTDOC_METS');
|
---|
172 |
|
---|
173 | close OUTDOC_TXT;
|
---|
174 | close OUTDOC_METS;
|
---|
175 | } else { # save_as isn't GA or METS
|
---|
176 | print $outhandle "docsave::process unrecognised saveas type, $save_as\n";
|
---|
177 | return;
|
---|
178 | }
|
---|
179 |
|
---|
180 | if ($self->{'gzip'}) {
|
---|
181 | my $doc_file = $self->{'gs_filename'};
|
---|
182 | `gzip $doc_file`;
|
---|
183 | $doc_file .= ".gz";
|
---|
184 | $short_doc_file .= ".gz";
|
---|
185 | if (!-e $doc_file) {
|
---|
186 | print $outhandle "error while gzipping: $doc_file doesn't exist\n";
|
---|
187 | return 0;
|
---|
188 | }
|
---|
189 | }
|
---|
190 |
|
---|
191 | # do the sortmeta thing
|
---|
192 | my ($metadata); if (defined ($self->{'sortmeta'})) {
|
---|
193 | $metadata = $doc_obj->get_metadata_element($doc_obj->get_top_section(),
|
---|
194 | $self->{'sortmeta'});
|
---|
195 | }
|
---|
196 |
|
---|
197 | # store reference in the archive_info
|
---|
198 | $self->{'archive_info'}->add_info($OID, $short_doc_file, $metadata);
|
---|
199 | }
|
---|
200 |
|
---|
201 |
|
---|
202 | sub group_process {
|
---|
203 | my $self = shift (@_);
|
---|
204 | my ($doc_obj) = @_;
|
---|
205 |
|
---|
206 | my $outhandle = $self->{'outhandle'};
|
---|
207 |
|
---|
208 | my $OID = $doc_obj->get_OID();
|
---|
209 | $OID = "NULL" unless defined $OID;
|
---|
210 |
|
---|
211 | my $groupsize = $self->{'groupsize'};
|
---|
212 | my $gs_count = $self->{'gs_count'};
|
---|
213 | my $open_new_file = (($gs_count % $groupsize)==0);
|
---|
214 |
|
---|
215 | # opening a new file, or document has assoicated files => directory needed
|
---|
216 | if (($open_new_file) || (scalar(@{$doc_obj->get_assoc_files()})>0)) {
|
---|
217 |
|
---|
218 | # get document's directory
|
---|
219 | my $doc_dir = $self->get_doc_dir ($OID);
|
---|
220 |
|
---|
221 | # copy all the associated files, add this information as metadata
|
---|
222 | # to the document
|
---|
223 | $self->process_assoc_files ($doc_obj, $doc_dir);
|
---|
224 |
|
---|
225 |
|
---|
226 | if ($open_new_file) {
|
---|
227 | # only if opening new file
|
---|
228 | my $doc_file
|
---|
229 | = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.xml");
|
---|
230 | my $short_doc_file = &util::filename_cat ($doc_dir, "doc.xml");
|
---|
231 |
|
---|
232 | if ($gs_count>0)
|
---|
233 | {
|
---|
234 | return if (!$self->close_file_output());
|
---|
235 | }
|
---|
236 |
|
---|
237 | if (!open (OUTDOC, ">$doc_file")) {
|
---|
238 | print $outhandle "docsave::group_process could not write to file $doc_file\n";
|
---|
239 | return;
|
---|
240 | }
|
---|
241 | $self->{'gs_filename'} = $doc_file;
|
---|
242 | $self->{'gs_short_filename'} = $short_doc_file;
|
---|
243 | $self->{'gs_OID'} = $OID;
|
---|
244 |
|
---|
245 | $self->output_xml_header('docsave::OUTDOC');
|
---|
246 | }
|
---|
247 | }
|
---|
248 |
|
---|
249 | # save this document
|
---|
250 | $doc_obj->output_section('docsave::OUTDOC', $doc_obj->get_top_section());
|
---|
251 |
|
---|
252 | $self->{'gs_count'}++;
|
---|
253 | }
|
---|
254 |
|
---|
255 |
|
---|
256 | sub get_doc_dir {
|
---|
257 | my $self = shift (@_);
|
---|
258 | my ($OID) = @_;
|
---|
259 |
|
---|
260 | my $doc_info = $self->{'archive_info'}->get_info($OID);
|
---|
261 | my $doc_dir = "";
|
---|
262 | if (defined $doc_info && scalar(@$doc_info) >= 1) {
|
---|
263 | # this OID already has an assigned directory, use the
|
---|
264 | # same one.
|
---|
265 | $doc_dir = $doc_info->[0];
|
---|
266 | $doc_dir =~ s/\/?doc\.xml(\.gz)?$//;
|
---|
267 | } else {
|
---|
268 | # have to get a new document directory
|
---|
269 | my $doc_dir_rest = $OID;
|
---|
270 | my $doc_dir_num = 0;
|
---|
271 | do {
|
---|
272 | $doc_dir .= "/" if $doc_dir_num > 0;
|
---|
273 | if ($doc_dir_rest =~ s/^(.{1,8})//) {
|
---|
274 | $doc_dir .= $1;
|
---|
275 | $doc_dir_num++;
|
---|
276 | }
|
---|
277 | } while ($doc_dir_rest ne "" &&
|
---|
278 | ((-d &util::filename_cat ($self->{'archive_dir'}, "$doc_dir.dir")) ||
|
---|
279 | ($self->{'archive_info'}->size() >= 1024 && $doc_dir_num < 2)));
|
---|
280 | $doc_dir .= ".dir";
|
---|
281 |
|
---|
282 | }
|
---|
283 |
|
---|
284 | &util::mk_all_dir (&util::filename_cat ($self->{'archive_dir'}, $doc_dir));
|
---|
285 |
|
---|
286 | return $doc_dir;
|
---|
287 | }
|
---|
288 |
|
---|
289 |
|
---|
290 | sub process_assoc_files {
|
---|
291 | my $self = shift (@_);
|
---|
292 | my ($doc_obj, $doc_dir) = @_;
|
---|
293 |
|
---|
294 | my $outhandle = $self->{'outhandle'};
|
---|
295 |
|
---|
296 | my @assoc_files = ();
|
---|
297 | foreach my $assoc_file (@{$doc_obj->get_assoc_files()}) {
|
---|
298 | my ($dir, $afile) = $assoc_file->[1] =~ /^(.*?)([^\/\\]+)$/;
|
---|
299 | $dir = "" unless defined $dir;
|
---|
300 | if (-e $assoc_file->[0]) {
|
---|
301 | my $filepath = &util::filename_cat($self->{'archive_dir'}, $doc_dir, $afile);
|
---|
302 | &util::hard_link ($assoc_file->[0], $filepath);
|
---|
303 | $doc_obj->add_utf8_metadata ($doc_obj->get_top_section(),
|
---|
304 | "gsdlassocfile",
|
---|
305 | "$afile:$assoc_file->[2]:$dir");
|
---|
306 | $doc_obj->set_utf8_metadata_element ($doc_obj->get_top_section(),
|
---|
307 | "assocfilepath",
|
---|
308 | "$doc_dir");
|
---|
309 | } elsif ($self->{'verbosity'} > 2) {
|
---|
310 | print $outhandle "docsave::process couldn't copy the associated file " .
|
---|
311 | "$assoc_file->[0] to $afile\n";
|
---|
312 | }
|
---|
313 | }
|
---|
314 | }
|
---|
315 |
|
---|
316 |
|
---|
317 | sub close_file_output
|
---|
318 | {
|
---|
319 | my ($self) = @_;
|
---|
320 |
|
---|
321 | # make sure that the handle has been opened - it won't be if we failed
|
---|
322 | # to import any documents...
|
---|
323 | if (defined(fileno(docsave::OUTDOC))) {
|
---|
324 | $self->output_xml_footer('docsave::OUTDOC');
|
---|
325 | close OUTDOC;
|
---|
326 | }
|
---|
327 |
|
---|
328 | my $OID = $self->{'gs_OID'};
|
---|
329 | my $short_doc_file;
|
---|
330 | # can we use 'short_doc_file' for GA too?
|
---|
331 | if (exists($self->{'saveas'}) && $self->{'saveas'} eq "METS") {
|
---|
332 | $short_doc_file=$self->{'short_doc_file'};
|
---|
333 | } else { # "GA"
|
---|
334 | $short_doc_file=$self->{'gs_short_filename'};
|
---|
335 | }
|
---|
336 |
|
---|
337 | if ($self->{'gzip'}) {
|
---|
338 | my $doc_file = $self->{'gs_filename'};
|
---|
339 | `gzip $doc_file`;
|
---|
340 | $doc_file .= ".gz";
|
---|
341 | $short_doc_file .= ".gz";
|
---|
342 | if (!-e $doc_file) {
|
---|
343 | my $outhandle = $self->{'outhandle'};
|
---|
344 | print $outhandle "error while gzipping: $doc_file doesn't exist\n";
|
---|
345 | return 0;
|
---|
346 | }
|
---|
347 | }
|
---|
348 |
|
---|
349 | # store reference in the archive_info
|
---|
350 | $self->{'archive_info'}->add_info($OID, $short_doc_file);
|
---|
351 |
|
---|
352 | return 1;
|
---|
353 | }
|
---|
354 |
|
---|
355 | sub output_xml_header {
|
---|
356 | my $self = shift (@_);
|
---|
357 | my ($handle) = @_;
|
---|
358 |
|
---|
359 | print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
|
---|
360 |
|
---|
361 | print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
|
---|
362 | print $handle "<Archive>\n";
|
---|
363 | }
|
---|
364 |
|
---|
365 | sub output_xml_footer {
|
---|
366 | my $self = shift (@_);
|
---|
367 | my ($handle) = @_;
|
---|
368 |
|
---|
369 | print $handle "</Archive>\n";
|
---|
370 | }
|
---|
371 |
|
---|
372 | sub output_txt_xml_header{
|
---|
373 | my $self = shift (@_);
|
---|
374 | my ($handle) = @_;
|
---|
375 | print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
|
---|
376 | print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
|
---|
377 | }
|
---|
378 |
|
---|
379 | sub output_txt_xml_footer{
|
---|
380 | my $self = shift(@_);
|
---|
381 | my ($handle) = @_;
|
---|
382 | print $handle "<the end of the file>\n";
|
---|
383 | }
|
---|
384 |
|
---|
385 | sub output_mets_xml_header(){
|
---|
386 | my $self = shift(@_);
|
---|
387 | my ($handle, $OID) = @_;
|
---|
388 | print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
|
---|
389 | print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
|
---|
390 | print $handle '<mets:mets OBJID="'. $OID. ':2">' . "\n";
|
---|
391 | }
|
---|
392 |
|
---|
393 | sub output_mets_xml_footer() {
|
---|
394 | my $self = shift(@_);
|
---|
395 | my ($handle) = @_;
|
---|
396 | print $handle '</mets:mets>' . "\n";
|
---|
397 | }
|
---|
398 |
|
---|
399 | 1;
|
---|
400 |
|
---|
401 |
|
---|
402 |
|
---|