1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # doc.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 redistr te 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 | # base class to hold documents
|
---|
27 |
|
---|
28 | package doc;
|
---|
29 | eval {require bytes};
|
---|
30 |
|
---|
31 | BEGIN {
|
---|
32 | die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
|
---|
33 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/dynamic/lib/site_perl/5.005/i686-linux");
|
---|
34 | }
|
---|
35 |
|
---|
36 | use unicode;
|
---|
37 | use util;
|
---|
38 | use ghtml;
|
---|
39 | use File::stat;
|
---|
40 | ##use hashdoc;
|
---|
41 | use docprint;
|
---|
42 |
|
---|
43 | # the document type may be indexed_doc, nonindexed_doc, or
|
---|
44 | # classification
|
---|
45 |
|
---|
46 | my $OIDcount = 0;
|
---|
47 |
|
---|
48 | sub new {
|
---|
49 | my $class = shift (@_);
|
---|
50 | my ($source_filename, $doc_type) = @_;
|
---|
51 |
|
---|
52 |
|
---|
53 | my $self = bless {'associated_files'=>[],
|
---|
54 | 'subsection_order'=>[],
|
---|
55 | 'next_subsection'=>1,
|
---|
56 | 'subsections'=>{},
|
---|
57 | 'metadata'=>[],
|
---|
58 | 'text'=>"",
|
---|
59 | 'OIDtype'=>"hash"}, $class;
|
---|
60 |
|
---|
61 | # used to set lastmodified here, but this can screw up the HASH ids, so
|
---|
62 | # the docsave processor now calls set_lastmodified
|
---|
63 |
|
---|
64 | $self->{'source_path'} = $source_filename;
|
---|
65 |
|
---|
66 | if (defined $source_filename) {
|
---|
67 | my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
|
---|
68 |
|
---|
69 | if (defined $collect_dir) {
|
---|
70 | my $dirsep = &util::get_dirsep();
|
---|
71 | if ($collect_dir !~ m/$dirsep$/) {
|
---|
72 | $collect_dir .= $dirsep;
|
---|
73 | }
|
---|
74 |
|
---|
75 | $collect_dir =~ s/\\/\\\\/g; # escape DOS style file separator
|
---|
76 |
|
---|
77 | # if from within GSDLCOLLECTDIR, then remove directory prefix
|
---|
78 | # so source_filename is realative to it. This is done to aid
|
---|
79 | # portability, i.e. the collection can be moved to somewhere
|
---|
80 | # else on the file system and the archives directory will still
|
---|
81 | # work. This is needed, for example in the applet version of
|
---|
82 | # GLI where GSDLHOME/collect on the server will be different to
|
---|
83 | # the collect directory of the remove user. Of course,
|
---|
84 | # GSDLCOLLECTDIR subsequently needs to be put back on to turn
|
---|
85 | # it back into a full pathname.
|
---|
86 |
|
---|
87 | if ($source_filename =~ /^$collect_dir(.*)$/) {
|
---|
88 | $source_filename = $1;
|
---|
89 | }
|
---|
90 | }
|
---|
91 |
|
---|
92 | $self->set_source_filename ($source_filename);
|
---|
93 | }
|
---|
94 |
|
---|
95 | $self->set_doc_type ($doc_type) if defined $doc_type;
|
---|
96 |
|
---|
97 | return $self;
|
---|
98 | }
|
---|
99 | # set lastmodified for OAI purposes, added by GRB, moved by kjdon
|
---|
100 | sub set_lastmodified {
|
---|
101 | my $self = shift (@_);
|
---|
102 |
|
---|
103 | my $source_path = $self->{'source_path'};
|
---|
104 |
|
---|
105 | if (defined $source_path && (-e $source_path)) {
|
---|
106 |
|
---|
107 | my $file_stat = stat($source_path);
|
---|
108 | my $mtime = $file_stat->mtime;
|
---|
109 | $self->add_utf8_metadata($self->get_top_section(), "lastmodified", $file_stat->mtime);
|
---|
110 | }
|
---|
111 | }
|
---|
112 |
|
---|
113 | # clone the $self object
|
---|
114 | sub duplicate {
|
---|
115 | my $self = shift (@_);
|
---|
116 |
|
---|
117 | my $newobj = {};
|
---|
118 |
|
---|
119 | foreach my $k (keys %$self) {
|
---|
120 | $newobj->{$k} = &clone ($self->{$k});
|
---|
121 | }
|
---|
122 |
|
---|
123 | bless $newobj, ref($self);
|
---|
124 | return $newobj;
|
---|
125 | }
|
---|
126 |
|
---|
127 | sub clone {
|
---|
128 | my ($from) = @_;
|
---|
129 | my $type = ref ($from);
|
---|
130 |
|
---|
131 | if ($type eq "HASH") {
|
---|
132 | my $to = {};
|
---|
133 | foreach my $key (keys %$from) {
|
---|
134 | $to->{$key} = &clone ($from->{$key});
|
---|
135 | }
|
---|
136 | return $to;
|
---|
137 | } elsif ($type eq "ARRAY") {
|
---|
138 | my $to = [];
|
---|
139 | foreach my $v (@$from) {
|
---|
140 | push (@$to, &clone ($v));
|
---|
141 | }
|
---|
142 | return $to;
|
---|
143 | } else {
|
---|
144 | return $from;
|
---|
145 | }
|
---|
146 | }
|
---|
147 |
|
---|
148 | sub set_OIDtype {
|
---|
149 | my $self = shift (@_);
|
---|
150 | my ($type, $metadata) = @_;
|
---|
151 |
|
---|
152 | if (defined $type && $type =~ /^(hash|incremental|dirname|assigned)$/) {
|
---|
153 | $self->{'OIDtype'} = $type;
|
---|
154 | } else {
|
---|
155 | $self->{'OIDtype'} = "hash";
|
---|
156 | }
|
---|
157 | if ($type =~ /^assigned$/) {
|
---|
158 | if (defined $metadata) {
|
---|
159 | $self->{'OIDmetadata'} = $metadata;
|
---|
160 | } else {
|
---|
161 | $self->{'OIDmetadata'} = "dc.Identifier";
|
---|
162 | }
|
---|
163 | }
|
---|
164 | }
|
---|
165 |
|
---|
166 | sub set_source_filename {
|
---|
167 | my $self = shift (@_);
|
---|
168 | my ($source_filename) = @_;
|
---|
169 |
|
---|
170 | $self->set_metadata_element ($self->get_top_section(),
|
---|
171 | "gsdlsourcefilename",
|
---|
172 | $source_filename);
|
---|
173 | }
|
---|
174 |
|
---|
175 | sub set_converted_filename {
|
---|
176 | my $self = shift (@_);
|
---|
177 | my ($converted_filename) = @_;
|
---|
178 |
|
---|
179 | $self->set_metadata_element ($self->get_top_section(),
|
---|
180 | "gsdlconvertedfilename",
|
---|
181 | $converted_filename);
|
---|
182 | }
|
---|
183 |
|
---|
184 |
|
---|
185 | # returns the source_filename as it was provided
|
---|
186 | sub get_source_filename {
|
---|
187 | my $self = shift (@_);
|
---|
188 |
|
---|
189 | return $self->get_metadata_element ($self->get_top_section(), "gsdlsourcefilename");
|
---|
190 | }
|
---|
191 |
|
---|
192 | # returns converted filename if available else returns source filename
|
---|
193 | sub get_filename_for_hashing {
|
---|
194 | my $self = shift (@_);
|
---|
195 |
|
---|
196 | my $filename = $self->get_metadata_element ($self->get_top_section(), "gsdlconvertedfilename");
|
---|
197 |
|
---|
198 | if (!defined $filename) {
|
---|
199 | my $plugin_name = $self->get_metadata_element ($self->get_top_section(), "Plugin");
|
---|
200 | # if NULPlug processed file, then don't give a filename
|
---|
201 | if (defined $plugin_name && $plugin_name eq "NULPlug") {
|
---|
202 | $filename = undef;
|
---|
203 | } else {
|
---|
204 | $filename = $self->get_metadata_element ($self->get_top_section(), "gsdlsourcefilename");
|
---|
205 | }
|
---|
206 | }
|
---|
207 | return $filename;
|
---|
208 | }
|
---|
209 |
|
---|
210 | sub set_doc_type {
|
---|
211 | my $self = shift (@_);
|
---|
212 | my ($doc_type) = @_;
|
---|
213 |
|
---|
214 | $self->set_metadata_element ($self->get_top_section(),
|
---|
215 | "gsdldoctype",
|
---|
216 | $doc_type);
|
---|
217 | }
|
---|
218 |
|
---|
219 | # returns the gsdldoctype as it was provided
|
---|
220 | # the default of "indexed_doc" is used if no document
|
---|
221 | # type was provided
|
---|
222 | sub get_doc_type {
|
---|
223 | my $self = shift (@_);
|
---|
224 |
|
---|
225 | my $doc_type = $self->get_metadata_element ($self->get_top_section(), "gsdldoctype");
|
---|
226 | return $doc_type if (defined $doc_type);
|
---|
227 | return "indexed_doc";
|
---|
228 | }
|
---|
229 |
|
---|
230 |
|
---|
231 | # look up the reference to the a particular section
|
---|
232 | sub _lookup_section {
|
---|
233 | my $self = shift (@_);
|
---|
234 | my ($section) = @_;
|
---|
235 |
|
---|
236 | my ($num);
|
---|
237 | my $sectionref = $self;
|
---|
238 |
|
---|
239 | while (defined $section && $section ne "") {
|
---|
240 |
|
---|
241 | ($num, $section) = $section =~ /^\.?(\d+)(.*)$/;
|
---|
242 |
|
---|
243 | $num =~ s/^0+(\d)/$1/ if defined $num ; # remove leading 0s
|
---|
244 |
|
---|
245 | $section = "" unless defined $section;
|
---|
246 |
|
---|
247 |
|
---|
248 | if (defined $num && defined $sectionref->{'subsections'}->{$num}) {
|
---|
249 | $sectionref = $sectionref->{'subsections'}->{$num};
|
---|
250 | } else {
|
---|
251 | return undef;
|
---|
252 | }
|
---|
253 | }
|
---|
254 |
|
---|
255 | return $sectionref;
|
---|
256 | }
|
---|
257 |
|
---|
258 | # calculate OID by hashing the contents of the document
|
---|
259 | sub _calc_OID {
|
---|
260 | my $self = shift (@_);
|
---|
261 | my ($filename) = @_;
|
---|
262 |
|
---|
263 | my $osexe = &util::get_os_exe();
|
---|
264 |
|
---|
265 | my $hashfile_exe = &util::filename_cat($ENV{'GSDLHOME'},"bin",
|
---|
266 | $ENV{'GSDLOS'},"hashfile$osexe");
|
---|
267 |
|
---|
268 | my $result = "NULL";
|
---|
269 |
|
---|
270 | if (-e "$hashfile_exe") {
|
---|
271 | # $result = `\"$hashfile_exe\" \"$filename\"`;
|
---|
272 | $result = `hashfile$osexe \"$filename\"`;
|
---|
273 | ($result) = $result =~ /:\s*([0-9a-f]+)/i;
|
---|
274 |
|
---|
275 | } else {
|
---|
276 | print STDERR "doc::_calc_OID $hashfile_exe could not be found\n";
|
---|
277 | }
|
---|
278 | return "HASH$result";
|
---|
279 | }
|
---|
280 |
|
---|
281 | # methods dealing with OID, not groups of them.
|
---|
282 |
|
---|
283 | # if $OID is not provided one is calculated
|
---|
284 | sub set_OID {
|
---|
285 | my $self = shift (@_);
|
---|
286 | my ($OID) = @_;
|
---|
287 |
|
---|
288 | my $use_hash_oid = 0;
|
---|
289 | # if an OID wasn't provided claculate one
|
---|
290 | if (!defined $OID) {
|
---|
291 | $OID = "NULL";
|
---|
292 | if ($self->{'OIDtype'} eq "hash") {
|
---|
293 | $use_hash_oid = 1;
|
---|
294 | } elsif ($self->{'OIDtype'} eq "incremental") {
|
---|
295 | $OID = "D" . $OIDcount;
|
---|
296 | $OIDcount ++;
|
---|
297 |
|
---|
298 | } elsif ($self->{'OIDtype'} eq "dirname") {
|
---|
299 | $OID = 'J';
|
---|
300 | my $filename = $self->get_source_filename();
|
---|
301 | if (defined($filename)) { # && -e $filename) {
|
---|
302 | $OID = &File::Basename::dirname($filename);
|
---|
303 | if (defined $OID) {
|
---|
304 | $OID = 'J'.&File::Basename::basename($OID);
|
---|
305 | $OID =~ s/\.//; #remove any periods
|
---|
306 | } else {
|
---|
307 | print STDERR "Failed to find base for filename ($filename)...generating hash id\n";
|
---|
308 | $use_hash_oid = 1;
|
---|
309 | }
|
---|
310 | } else {
|
---|
311 | print STDERR "Failed to find filename, generating hash id\n";
|
---|
312 | $use_hash_oid = 1;
|
---|
313 | }
|
---|
314 |
|
---|
315 | } elsif ($self->{'OIDtype'} eq "assigned") {
|
---|
316 | my $identifier = $self->get_metadata_element ($self->get_top_section(), $self->{'OIDmetadata'});
|
---|
317 | if (defined $identifier && $identifier ne "") {
|
---|
318 | $OID = "D" . $identifier;
|
---|
319 | $OID =~ s/\.//; #remove any periods
|
---|
320 | } else {
|
---|
321 | # need a hash id
|
---|
322 | print STDERR "no $self->{'OIDmetadata'} metadata found, generating hash id\n";
|
---|
323 | $use_hash_oid = 1;
|
---|
324 | }
|
---|
325 |
|
---|
326 | } else {
|
---|
327 | $use_hash_oid = 1;
|
---|
328 | }
|
---|
329 |
|
---|
330 | if ($use_hash_oid) {
|
---|
331 |
|
---|
332 | # "hash" OID - feed file to hashfile.exe
|
---|
333 | #my $filename = $self->get_source_filename();
|
---|
334 | # we want to use the converted file for hashing if available
|
---|
335 | # cos its quicker
|
---|
336 | my $filename = $self->get_filename_for_hashing();
|
---|
337 | # -z: don't want to hash on the file if it is zero size
|
---|
338 | if (defined($filename) && -e $filename && !-z $filename) {
|
---|
339 | $OID = $self->_calc_OID ($filename);
|
---|
340 | } else {
|
---|
341 | $filename = &util::get_tmp_filename();
|
---|
342 | if (!open (OUTFILE, ">$filename")) {
|
---|
343 | print STDERR "doc::set_OID could not write to $filename\n";
|
---|
344 | } else {
|
---|
345 | my $doc_text = &docprint::get_section_xml($self, $self->get_top_section());
|
---|
346 | print OUTFILE $doc_text;
|
---|
347 | close (OUTFILE);
|
---|
348 | }
|
---|
349 | $OID = $self->_calc_OID ($filename);
|
---|
350 | &util::rm ($filename);
|
---|
351 | }
|
---|
352 | }
|
---|
353 | }
|
---|
354 | $self->set_metadata_element ($self->get_top_section(), "Identifier", $OID);
|
---|
355 | }
|
---|
356 |
|
---|
357 | # this uses hashdoc (embedded c thingy) which is faster but still
|
---|
358 | # needs a little work to be suffiently stable
|
---|
359 | sub ___set_OID {
|
---|
360 | my $self = shift (@_);
|
---|
361 | my ($OID) = @_;
|
---|
362 |
|
---|
363 | # if an OID wasn't provided then calculate hash value based on document
|
---|
364 | if (!defined $OID)
|
---|
365 | {
|
---|
366 | my $hash_text = &docprint::get_section_xml($self, $self->get_top_section());
|
---|
367 | my $hash_len = length($hash_text);
|
---|
368 |
|
---|
369 | $OID = &hashdoc::buffer($hash_text,$hash_len);
|
---|
370 | }
|
---|
371 |
|
---|
372 | $self->set_metadata_element ($self->get_top_section(), "Identifier", $OID);
|
---|
373 | }
|
---|
374 |
|
---|
375 | # returns the OID for this document
|
---|
376 | sub get_OID {
|
---|
377 | my $self = shift (@_);
|
---|
378 | my $OID = $self->get_metadata_element ($self->get_top_section(), "Identifier");
|
---|
379 | return $OID if (defined $OID);
|
---|
380 | return "NULL";
|
---|
381 | }
|
---|
382 |
|
---|
383 | sub delete_OID {
|
---|
384 | my $self = shift (@_);
|
---|
385 |
|
---|
386 | $self->set_metadata_element ($self->get_top_section(), "Identifier", "NULL");
|
---|
387 | }
|
---|
388 |
|
---|
389 |
|
---|
390 | # methods for manipulating section names
|
---|
391 |
|
---|
392 | # returns the name of the top-most section (the top
|
---|
393 | # level of the document
|
---|
394 | sub get_top_section {
|
---|
395 | my $self = shift (@_);
|
---|
396 |
|
---|
397 | return "";
|
---|
398 | }
|
---|
399 |
|
---|
400 | # returns a section
|
---|
401 | sub get_parent_section {
|
---|
402 | my $self = shift (@_);
|
---|
403 | my ($section) = @_;
|
---|
404 |
|
---|
405 | $section =~ s/(^|\.)\d+$//;
|
---|
406 |
|
---|
407 | return $section;
|
---|
408 | }
|
---|
409 |
|
---|
410 | # returns the first child section (or the end child
|
---|
411 | # if there isn't any)
|
---|
412 | sub get_begin_child {
|
---|
413 | my $self = shift (@_);
|
---|
414 | my ($section) = @_;
|
---|
415 |
|
---|
416 | my $section_ptr = $self->_lookup_section($section);
|
---|
417 | return "" unless defined $section_ptr;
|
---|
418 |
|
---|
419 | if (defined $section_ptr->{'subsection_order'}->[0]) {
|
---|
420 | return "$section.$section_ptr->{'subsection_order'}->[0]";
|
---|
421 | }
|
---|
422 |
|
---|
423 | return $self->get_end_child ($section);
|
---|
424 | }
|
---|
425 |
|
---|
426 | # returns the next child of a parent section
|
---|
427 | sub get_next_child {
|
---|
428 | my $self = shift (@_);
|
---|
429 | my ($section) = @_;
|
---|
430 |
|
---|
431 | my $parent_section = $self->get_parent_section($section);
|
---|
432 | my $parent_section_ptr = $self->_lookup_section($parent_section);
|
---|
433 | return undef unless defined $parent_section_ptr;
|
---|
434 |
|
---|
435 | my ($section_num) = $section =~ /(\d+)$/;
|
---|
436 | return undef unless defined $section_num;
|
---|
437 |
|
---|
438 | my $i = 0;
|
---|
439 | my $section_order = $parent_section_ptr->{'subsection_order'};
|
---|
440 | while ($i < scalar(@$section_order)) {
|
---|
441 | last if $section_order->[$i] eq $section_num;
|
---|
442 | $i++;
|
---|
443 | }
|
---|
444 |
|
---|
445 | $i++; # the next child
|
---|
446 | if ($i < scalar(@$section_order)) {
|
---|
447 | return $section_order->[$i] if $parent_section eq "";
|
---|
448 | return "$parent_section.$section_order->[$i]";
|
---|
449 | }
|
---|
450 |
|
---|
451 | # no more sections in this level
|
---|
452 | return undef;
|
---|
453 | }
|
---|
454 |
|
---|
455 | # returns a reference to a list of children
|
---|
456 | sub get_children {
|
---|
457 | my $self = shift (@_);
|
---|
458 | my ($section) = @_;
|
---|
459 |
|
---|
460 | my $section_ptr = $self->_lookup_section($section);
|
---|
461 | return [] unless defined $section_ptr;
|
---|
462 |
|
---|
463 | my @children = @{$section_ptr->{'subsection_order'}};
|
---|
464 |
|
---|
465 | map {$_ = "$section.$_"; $_ =~ s/^\.+//;} @children;
|
---|
466 | return \@children;
|
---|
467 | }
|
---|
468 |
|
---|
469 | # returns the child section one past the last one (which
|
---|
470 | # is coded as "0")
|
---|
471 | sub get_end_child {
|
---|
472 | my $self = shift (@_);
|
---|
473 | my ($section) = @_;
|
---|
474 |
|
---|
475 | return $section . ".0" unless $section eq "";
|
---|
476 | return "0";
|
---|
477 | }
|
---|
478 |
|
---|
479 | # returns the next section in book order
|
---|
480 | sub get_next_section {
|
---|
481 | my $self = shift (@_);
|
---|
482 | my ($section) = @_;
|
---|
483 |
|
---|
484 | return undef unless defined $section;
|
---|
485 |
|
---|
486 | my $section_ptr = $self->_lookup_section($section);
|
---|
487 | return undef unless defined $section_ptr;
|
---|
488 |
|
---|
489 | # first try to find first child
|
---|
490 | if (defined $section_ptr->{'subsection_order'}->[0]) {
|
---|
491 | return $section_ptr->{'subsection_order'}->[0] if ($section eq "");
|
---|
492 | return "$section.$section_ptr->{'subsection_order'}->[0]";
|
---|
493 | }
|
---|
494 |
|
---|
495 | do {
|
---|
496 | # try to find sibling
|
---|
497 | my $next_child = $self->get_next_child ($section);
|
---|
498 | return $next_child if (defined $next_child);
|
---|
499 |
|
---|
500 | # move up one level
|
---|
501 | $section = $self->get_parent_section ($section);
|
---|
502 | } while $section =~ /\d/;
|
---|
503 |
|
---|
504 | return undef;
|
---|
505 | }
|
---|
506 |
|
---|
507 | sub is_leaf_section {
|
---|
508 | my $self = shift (@_);
|
---|
509 | my ($section) = @_;
|
---|
510 |
|
---|
511 | my $section_ptr = $self->_lookup_section($section);
|
---|
512 | return 1 unless defined $section_ptr;
|
---|
513 |
|
---|
514 | return (scalar (@{$section_ptr->{'subsection_order'}}) == 0);
|
---|
515 | }
|
---|
516 |
|
---|
517 | # methods for dealing with sections
|
---|
518 |
|
---|
519 | # returns the name of the inserted section
|
---|
520 | sub insert_section {
|
---|
521 | my $self = shift (@_);
|
---|
522 | my ($before_section) = @_;
|
---|
523 |
|
---|
524 | # get the child to insert before and its parent section
|
---|
525 | my $parent_section = "";
|
---|
526 | my $before_child = "0";
|
---|
527 | my @before_section = split (/\./, $before_section);
|
---|
528 | if (scalar(@before_section) > 0) {
|
---|
529 | $before_child = pop (@before_section);
|
---|
530 | $parent_section = join (".", @before_section);
|
---|
531 | }
|
---|
532 |
|
---|
533 | my $parent_section_ptr = $self->_lookup_section($parent_section);
|
---|
534 | if (!defined $parent_section_ptr) {
|
---|
535 | print STDERR "doc::insert_section couldn't find parent section " .
|
---|
536 | "$parent_section\n";
|
---|
537 | return;
|
---|
538 | }
|
---|
539 |
|
---|
540 | # get the next section number
|
---|
541 | my $section_num = $parent_section_ptr->{'next_subsection'}++;
|
---|
542 |
|
---|
543 | my $i = 0;
|
---|
544 | while ($i < scalar(@{$parent_section_ptr->{'subsection_order'}}) &&
|
---|
545 | $parent_section_ptr->{'subsection_order'}->[$i] ne $before_child) {
|
---|
546 | $i++;
|
---|
547 | }
|
---|
548 |
|
---|
549 | # insert the section number into the order list
|
---|
550 | splice (@{$parent_section_ptr->{'subsection_order'}}, $i, 0, $section_num);
|
---|
551 |
|
---|
552 | # add this section to the parent section
|
---|
553 | my $section_ptr = {'subsection_order'=>[],
|
---|
554 | 'next_subsection'=>1,
|
---|
555 | 'subsections'=>{},
|
---|
556 | 'metadata'=>[],
|
---|
557 | 'text'=>""};
|
---|
558 | $parent_section_ptr->{'subsections'}->{$section_num} = $section_ptr;
|
---|
559 |
|
---|
560 | # work out the full section number
|
---|
561 | my $section = $parent_section;
|
---|
562 | $section .= "." unless $section eq "";
|
---|
563 | $section .= $section_num;
|
---|
564 |
|
---|
565 | return $section;
|
---|
566 | }
|
---|
567 |
|
---|
568 | # creates a pre-named section
|
---|
569 | sub create_named_section {
|
---|
570 | my $self = shift (@_);
|
---|
571 | my ($mastersection) = @_;
|
---|
572 |
|
---|
573 | my ($num);
|
---|
574 | my $section = $mastersection;
|
---|
575 | my $sectionref = $self;
|
---|
576 |
|
---|
577 | while ($section ne "") {
|
---|
578 | ($num, $section) = $section =~ /^\.?(\d+)(.*)$/;
|
---|
579 | $num =~ s/^0+(\d)/$1/; # remove leading 0s
|
---|
580 | $section = "" unless defined $section;
|
---|
581 |
|
---|
582 | if (defined $num) {
|
---|
583 | if (!defined $sectionref->{'subsections'}->{$num}) {
|
---|
584 | push (@{$sectionref->{'subsection_order'}}, $num);
|
---|
585 | $sectionref->{'subsections'}->{$num} = {'subsection_order'=>[],
|
---|
586 | 'next_subsection'=>1,
|
---|
587 | 'subsections'=>{},
|
---|
588 | 'metadata'=>[],
|
---|
589 | 'text'=>""};
|
---|
590 | if ($num >= $sectionref->{'next_subsection'}) {
|
---|
591 | $sectionref->{'next_subsection'} = $num + 1;
|
---|
592 | }
|
---|
593 | }
|
---|
594 | $sectionref = $sectionref->{'subsections'}->{$num};
|
---|
595 |
|
---|
596 | } else {
|
---|
597 | print STDERR "doc::create_named_section couldn't create section ";
|
---|
598 | print STDERR "$mastersection\n";
|
---|
599 | last;
|
---|
600 | }
|
---|
601 | }
|
---|
602 | }
|
---|
603 |
|
---|
604 | # returns a reference to a list of subsections
|
---|
605 | sub list_subsections {
|
---|
606 | my $self = shift (@_);
|
---|
607 | my ($section) = @_;
|
---|
608 |
|
---|
609 | my $section_ptr = $self->_lookup_section ($section);
|
---|
610 | if (!defined $section_ptr) {
|
---|
611 | print STDERR "doc::list_subsections couldn't find section $section\n";
|
---|
612 | return [];
|
---|
613 | }
|
---|
614 |
|
---|
615 | return [@{$section_ptr->{'subsection_order'}}];
|
---|
616 | }
|
---|
617 |
|
---|
618 | sub delete_section {
|
---|
619 | my $self = shift (@_);
|
---|
620 | my ($section) = @_;
|
---|
621 |
|
---|
622 | # my $section_ptr = {'subsection_order'=>[],
|
---|
623 | # 'next_subsection'=>1,
|
---|
624 | # 'subsections'=>{},
|
---|
625 | # 'metadata'=>[],
|
---|
626 | # 'text'=>""};
|
---|
627 |
|
---|
628 | # if this is the top section reset everything
|
---|
629 | if ($section eq "") {
|
---|
630 | $self->{'subsection_order'} = [];
|
---|
631 | $self->{'subsections'} = {};
|
---|
632 | $self->{'metadata'} = [];
|
---|
633 | $self->{'text'} = "";
|
---|
634 | return;
|
---|
635 | }
|
---|
636 |
|
---|
637 | # find the parent of the section to delete
|
---|
638 | my $parent_section = "";
|
---|
639 | my $child = "0";
|
---|
640 | my @section = split (/\./, $section);
|
---|
641 | if (scalar(@section) > 0) {
|
---|
642 | $child = pop (@section);
|
---|
643 | $parent_section = join (".", @section);
|
---|
644 | }
|
---|
645 |
|
---|
646 | my $parent_section_ptr = $self->_lookup_section($parent_section);
|
---|
647 | if (!defined $parent_section_ptr) {
|
---|
648 | print STDERR "doc::delete_section couldn't find parent section " .
|
---|
649 | "$parent_section\n";
|
---|
650 | return;
|
---|
651 | }
|
---|
652 |
|
---|
653 | # remove this section from the subsection_order list
|
---|
654 | my $i = 0;
|
---|
655 | while ($i < scalar (@{$parent_section_ptr->{'subsection_order'}})) {
|
---|
656 | if ($parent_section_ptr->{'subsection_order'}->[$i] eq $child) {
|
---|
657 | splice (@{$parent_section_ptr->{'subsection_order'}}, $i, 1);
|
---|
658 | last;
|
---|
659 | }
|
---|
660 | $i++;
|
---|
661 | }
|
---|
662 |
|
---|
663 | # remove this section from the subsection hash
|
---|
664 | if (defined ($parent_section_ptr->{'subsections'}->{$child})) {
|
---|
665 | undef $parent_section_ptr->{'subsections'}->{$child};
|
---|
666 | }
|
---|
667 | }
|
---|
668 |
|
---|
669 | #--
|
---|
670 | # methods for dealing with metadata
|
---|
671 |
|
---|
672 | # set_metadata_element and get_metadata_element are for metadata
|
---|
673 | # which should only have one value. add_meta_data and get_metadata
|
---|
674 | # are for metadata which can have more than one value.
|
---|
675 |
|
---|
676 | # returns the first metadata value which matches field
|
---|
677 |
|
---|
678 | # This version of get metadata element works much like the one above,
|
---|
679 | # except it allows for the namespace portion of a metadata element to
|
---|
680 | # be ignored, thus if you are searching for dc.Title, the first piece
|
---|
681 | # of matching metadata ending with the name Title (once any namespace
|
---|
682 | # is removed) would be returned.
|
---|
683 | # 28-11-2003 John Thompson
|
---|
684 | sub get_metadata_element {
|
---|
685 | my $self = shift (@_);
|
---|
686 | my ($section, $field, $ignore_namespace) = @_;
|
---|
687 | my ($data);
|
---|
688 |
|
---|
689 | $ignore_namespace = 0 unless defined $ignore_namespace;
|
---|
690 |
|
---|
691 | my $section_ptr = $self->_lookup_section($section);
|
---|
692 | if (!defined $section_ptr) {
|
---|
693 | print STDERR "doc::get_metadata_element couldn't find section ", $section, "\n";
|
---|
694 | return;
|
---|
695 | }
|
---|
696 |
|
---|
697 | # Remove the any namespace if we are being told to ignore them
|
---|
698 | if($ignore_namespace) {
|
---|
699 | $field =~ s/^\w*\.//;
|
---|
700 | }
|
---|
701 |
|
---|
702 | foreach $data (@{$section_ptr->{'metadata'}}) {
|
---|
703 |
|
---|
704 | my $data_name = $data->[0];
|
---|
705 | # Remove the any namespace if we are being told to ignore them
|
---|
706 | if($ignore_namespace) {
|
---|
707 | $data_name =~ s/^\w*\.//;
|
---|
708 | }
|
---|
709 |
|
---|
710 | return $data->[1] if (scalar(@$data) >= 2 && $data_name eq $field);
|
---|
711 | }
|
---|
712 |
|
---|
713 | return undef; # was not found
|
---|
714 | }
|
---|
715 |
|
---|
716 | # returns a list of the form [value1, value2, ...]
|
---|
717 | sub get_metadata {
|
---|
718 | my $self = shift (@_);
|
---|
719 | my ($section, $field, $ignore_namespace) = @_;
|
---|
720 | my ($data);
|
---|
721 |
|
---|
722 | $ignore_namespace = 0 unless defined $ignore_namespace;
|
---|
723 |
|
---|
724 | my $section_ptr = $self->_lookup_section($section);
|
---|
725 | if (!defined $section_ptr) {
|
---|
726 | print STDERR "doc::get_metadata couldn't find section ",
|
---|
727 | $section, "\n";
|
---|
728 | return;
|
---|
729 | }
|
---|
730 |
|
---|
731 | # Remove the any namespace if we are being told to ignore them
|
---|
732 | if($ignore_namespace) {
|
---|
733 | $field =~ s/^\w*\.//;
|
---|
734 | }
|
---|
735 |
|
---|
736 | my @metadata = ();
|
---|
737 | foreach $data (@{$section_ptr->{'metadata'}}) {
|
---|
738 |
|
---|
739 | my $data_name = $data->[0];
|
---|
740 | # Remove the any namespace if we are being told to ignore them
|
---|
741 | if($ignore_namespace) {
|
---|
742 | $data_name =~ s/^\w*\.//;
|
---|
743 | }
|
---|
744 |
|
---|
745 | push (@metadata, $data->[1]) if ($data_name eq $field);
|
---|
746 | }
|
---|
747 |
|
---|
748 | return \@metadata;
|
---|
749 | }
|
---|
750 |
|
---|
751 | # returns a list of the form [[field,value],[field,value],...]
|
---|
752 | sub get_all_metadata {
|
---|
753 | my $self = shift (@_);
|
---|
754 | my ($section) = @_;
|
---|
755 |
|
---|
756 | my $section_ptr = $self->_lookup_section($section);
|
---|
757 | if (!defined $section_ptr) {
|
---|
758 | print STDERR "doc::get_all_metadata couldn't find section ", $section, "\n";
|
---|
759 | return;
|
---|
760 | }
|
---|
761 |
|
---|
762 | return $section_ptr->{'metadata'};
|
---|
763 | }
|
---|
764 |
|
---|
765 | # $value is optional
|
---|
766 | sub delete_metadata {
|
---|
767 | my $self = shift (@_);
|
---|
768 | my ($section, $field, $value) = @_;
|
---|
769 |
|
---|
770 | my $section_ptr = $self->_lookup_section($section);
|
---|
771 | if (!defined $section_ptr) {
|
---|
772 | print STDERR "doc::delete_metadata couldn't find section ", $section, "\n";
|
---|
773 | return;
|
---|
774 | }
|
---|
775 |
|
---|
776 | my $i = 0;
|
---|
777 | while ($i < scalar (@{$section_ptr->{'metadata'}})) {
|
---|
778 | if (($section_ptr->{'metadata'}->[$i]->[0] eq $field) &&
|
---|
779 | (!defined $value || $section_ptr->{'metadata'}->[$i]->[1] eq $value)) {
|
---|
780 | splice (@{$section_ptr->{'metadata'}}, $i, 1);
|
---|
781 | } else {
|
---|
782 | $i++;
|
---|
783 | }
|
---|
784 | }
|
---|
785 | }
|
---|
786 |
|
---|
787 | sub delete_all_metadata {
|
---|
788 | my $self = shift (@_);
|
---|
789 | my ($section) = @_;
|
---|
790 |
|
---|
791 | my $section_ptr = $self->_lookup_section($section);
|
---|
792 | if (!defined $section_ptr) {
|
---|
793 | print STDERR "doc::delete_all_metadata couldn't find section ", $section, "\n";
|
---|
794 | return;
|
---|
795 | }
|
---|
796 |
|
---|
797 | $section_ptr->{'metadata'} = [];
|
---|
798 | }
|
---|
799 |
|
---|
800 | sub set_metadata_element {
|
---|
801 | my $self = shift (@_);
|
---|
802 | my ($section, $field, $value) = @_;
|
---|
803 |
|
---|
804 | $self->set_utf8_metadata_element ($section, $field,
|
---|
805 | &unicode::ascii2utf8(\$value));
|
---|
806 | }
|
---|
807 |
|
---|
808 | # set_utf8_metadata_element assumes the text has already been
|
---|
809 | # converted to the UTF-8 encoding.
|
---|
810 | sub set_utf8_metadata_element {
|
---|
811 | my $self = shift (@_);
|
---|
812 | my ($section, $field, $value) = @_;
|
---|
813 |
|
---|
814 | $self->delete_metadata ($section, $field);
|
---|
815 | $self->add_utf8_metadata ($section, $field, $value);
|
---|
816 | }
|
---|
817 |
|
---|
818 |
|
---|
819 | # add_metadata assumes the text is in (extended) ascii form. For
|
---|
820 | # text which has already been converted to the UTF-8 format use
|
---|
821 | # add_utf8_metadata.
|
---|
822 | sub add_metadata {
|
---|
823 | my $self = shift (@_);
|
---|
824 | my ($section, $field, $value) = @_;
|
---|
825 |
|
---|
826 | $self->add_utf8_metadata ($section, $field,
|
---|
827 | &unicode::ascii2utf8(\$value));
|
---|
828 | }
|
---|
829 |
|
---|
830 | sub add_utf8_metadata {
|
---|
831 | my $self = shift (@_);
|
---|
832 | my ($section, $field, $value) = @_;
|
---|
833 |
|
---|
834 | my $section_ptr = $self->_lookup_section($section);
|
---|
835 | if (!defined $section_ptr) {
|
---|
836 | print STDERR "doc::add_utf8_metadata couldn't find section ", $section, "\n";
|
---|
837 | return;
|
---|
838 | }
|
---|
839 | if (!defined $value) {
|
---|
840 | print STDERR "doc::add_utf8_metadata undefined value for $field\n";
|
---|
841 | return;
|
---|
842 | }
|
---|
843 | if (!defined $field) {
|
---|
844 | print STDERR "doc::add_utf8_metadata undefined metadata type \n";
|
---|
845 | return;
|
---|
846 | }
|
---|
847 |
|
---|
848 | #print STDERR "###$field=$value\n";
|
---|
849 | # double check that the value is utf-8
|
---|
850 | if (unicode::ensure_utf8(\$value)) {
|
---|
851 | print STDERR "doc::add_utf8_metadata: warning: '$field' wasn't utf8\n";
|
---|
852 | }
|
---|
853 |
|
---|
854 | push (@{$section_ptr->{'metadata'}}, [$field, $value]);
|
---|
855 | }
|
---|
856 |
|
---|
857 |
|
---|
858 | # methods for dealing with text
|
---|
859 |
|
---|
860 | # returns the text for a section
|
---|
861 | sub get_text {
|
---|
862 | my $self = shift (@_);
|
---|
863 | my ($section) = @_;
|
---|
864 |
|
---|
865 | my $section_ptr = $self->_lookup_section($section);
|
---|
866 | if (!defined $section_ptr) {
|
---|
867 | print STDERR "doc::get_text couldn't find section " .
|
---|
868 | "$section\n";
|
---|
869 | return "";
|
---|
870 | }
|
---|
871 |
|
---|
872 | return $section_ptr->{'text'};
|
---|
873 | }
|
---|
874 |
|
---|
875 | # returns the (utf-8 encoded) length of the text for a section
|
---|
876 | sub get_text_length {
|
---|
877 | my $self = shift (@_);
|
---|
878 | my ($section) = @_;
|
---|
879 |
|
---|
880 | my $section_ptr = $self->_lookup_section($section);
|
---|
881 | if (!defined $section_ptr) {
|
---|
882 | print STDERR "doc::get_text_length couldn't find section " .
|
---|
883 | "$section\n";
|
---|
884 | return 0;
|
---|
885 | }
|
---|
886 |
|
---|
887 | return length ($section_ptr->{'text'});
|
---|
888 | }
|
---|
889 |
|
---|
890 | sub delete_text {
|
---|
891 | my $self = shift (@_);
|
---|
892 | my ($section) = @_;
|
---|
893 |
|
---|
894 | my $section_ptr = $self->_lookup_section($section);
|
---|
895 | if (!defined $section_ptr) {
|
---|
896 | print STDERR "doc::delete_text couldn't find section " .
|
---|
897 | "$section\n";
|
---|
898 | return;
|
---|
899 | }
|
---|
900 |
|
---|
901 | $section_ptr->{'text'} = "";
|
---|
902 | }
|
---|
903 |
|
---|
904 | # add_text assumes the text is in (extended) ascii form. For
|
---|
905 | # text which has been already converted to the UTF-8 format
|
---|
906 | # use add_utf8_text.
|
---|
907 | sub add_text {
|
---|
908 | my $self = shift (@_);
|
---|
909 | my ($section, $text) = @_;
|
---|
910 |
|
---|
911 | # convert the text to UTF-8 encoded unicode characters
|
---|
912 | # and add the text
|
---|
913 | $self->add_utf8_text($section, &unicode::ascii2utf8(\$text));
|
---|
914 | }
|
---|
915 |
|
---|
916 |
|
---|
917 | # add_utf8_text assumes the text to be added has already
|
---|
918 | # been converted to the UTF-8 encoding. For ascii text use
|
---|
919 | # add_text
|
---|
920 | sub add_utf8_text {
|
---|
921 | my $self = shift (@_);
|
---|
922 | my ($section, $text) = @_;
|
---|
923 |
|
---|
924 | my $section_ptr = $self->_lookup_section($section);
|
---|
925 | if (!defined $section_ptr) {
|
---|
926 | print STDERR "doc::add_utf8_text couldn't find section " .
|
---|
927 | "$section\n";
|
---|
928 | return;
|
---|
929 | }
|
---|
930 |
|
---|
931 | $section_ptr->{'text'} .= $text;
|
---|
932 | }
|
---|
933 |
|
---|
934 |
|
---|
935 | # methods for dealing with associated files
|
---|
936 |
|
---|
937 | # a file is associated with a document, NOT a section.
|
---|
938 | # if section is defined it is noted in the data structure
|
---|
939 | # only so that files associated from a particular section
|
---|
940 | # may be removed later (using delete_section_assoc_files)
|
---|
941 | sub associate_file {
|
---|
942 | my $self = shift (@_);
|
---|
943 | my ($real_filename, $assoc_filename, $mime_type, $section) = @_;
|
---|
944 | $mime_type = &ghtml::guess_mime_type ($real_filename) unless defined $mime_type;
|
---|
945 |
|
---|
946 | # remove all associated files with the same name
|
---|
947 | $self->delete_assoc_file ($assoc_filename);
|
---|
948 |
|
---|
949 | push (@{$self->{'associated_files'}},
|
---|
950 | [$real_filename, $assoc_filename, $mime_type, $section]);
|
---|
951 | }
|
---|
952 |
|
---|
953 | # returns a list of associated files in the form
|
---|
954 | # [[real_filename, assoc_filename, mimetype], ...]
|
---|
955 | sub get_assoc_files {
|
---|
956 | my $self = shift (@_);
|
---|
957 |
|
---|
958 | return $self->{'associated_files'};
|
---|
959 | }
|
---|
960 |
|
---|
961 | sub delete_section_assoc_files {
|
---|
962 | my $self = shift (@_);
|
---|
963 | my ($section) = @_;
|
---|
964 |
|
---|
965 | my $i=0;
|
---|
966 | while ($i < scalar (@{$self->{'associated_files'}})) {
|
---|
967 | if (defined $self->{'associated_files'}->[$i]->[3] &&
|
---|
968 | $self->{'associated_files'}->[$i]->[3] eq $section) {
|
---|
969 | splice (@{$self->{'associated_files'}}, $i, 1);
|
---|
970 | } else {
|
---|
971 | $i++;
|
---|
972 | }
|
---|
973 | }
|
---|
974 | }
|
---|
975 |
|
---|
976 | sub delete_assoc_file {
|
---|
977 | my $self = shift (@_);
|
---|
978 | my ($assoc_filename) = @_;
|
---|
979 |
|
---|
980 | my $i=0;
|
---|
981 | while ($i < scalar (@{$self->{'associated_files'}})) {
|
---|
982 | if ($self->{'associated_files'}->[$i]->[1] eq $assoc_filename) {
|
---|
983 | splice (@{$self->{'associated_files'}}, $i, 1);
|
---|
984 | } else {
|
---|
985 | $i++;
|
---|
986 | }
|
---|
987 | }
|
---|
988 | }
|
---|
989 |
|
---|
990 | sub reset_nextsection_ptr {
|
---|
991 | my $self = shift (@_);
|
---|
992 | my ($section) = @_;
|
---|
993 |
|
---|
994 | my $section_ptr = $self->_lookup_section($section);
|
---|
995 | $section_ptr->{'next_subsection'} = 1;
|
---|
996 | }
|
---|
997 |
|
---|
998 | 1;
|
---|