source: main/branches/64_bit_Greenstone/greenstone2/perllib/plugins/BasePlugin.pm@ 23580

Last change on this file since 23580 was 23580, checked in by sjm84, 13 years ago

Merging the latest trunk changes into this branch
  • Property svn:keywords set to Author Date Id Revision
File size: 44.2 KB
Line 
1###########################################################################
2#
3# BasePlugin.pm -- base class for all the import plugins
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) 1999-2005 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
26package BasePlugin;
27
28use strict;
29no strict 'subs';
30no strict 'refs'; # allow filehandles to be variables and viceversa
31
32use File::Basename;
33use Encode;
34
35use encodings;
36use unicode;
37use doc;
38use ghtml;
39use gsprintf 'gsprintf';
40
41use PrintInfo;
42
43BEGIN {
44 @BasePlugin::ISA = ( 'PrintInfo' );
45}
46
47# the different methods that can be applied when renaming
48# imported documents and their associated files
49our $file_rename_method_list =
50 [ { 'name' => "url",
51 'desc' => "{BasePlugin.rename_method.url}" },
52 { 'name' => "base64",
53 'desc' => "{BasePlugin.rename_method.base64}" },
54 { 'name' => "none",
55 'desc' => "{BasePlugin.rename_method.none}",
56 'hiddengli' => "yes" } ];
57
58our $encoding_list =
59 [ { 'name' => "ascii",
60 'desc' => "{BasePlugin.encoding.ascii}" },
61 { 'name' => "utf8",
62 'desc' => "{BasePlugin.encoding.utf8}" },
63 { 'name' => "unicode",
64 'desc' => "{BasePlugin.encoding.unicode}" } ];
65
66
67my $e = $encodings::encodings;
68foreach my $enc (sort {$e->{$a}->{'name'} cmp $e->{$b}->{'name'}} keys (%$e))
69{
70 my $hashEncode =
71 {'name' => $enc,
72 'desc' => $e->{$enc}->{'name'}};
73
74 push(@{$encoding_list},$hashEncode);
75}
76
77our $encoding_plus_auto_list =
78 [ { 'name' => "auto",
79 'desc' => "{BasePlugin.filename_encoding.auto}" },
80 { 'name' => "auto-language-analysis",
81 'desc' => "{BasePlugin.filename_encoding.auto_language_analysis}" }, # textcat
82 { 'name' => "auto-filesystem-encoding",
83 'desc' => "{BasePlugin.filename_encoding.auto_filesystem_encoding}" }, # locale
84 { 'name' => "auto-fl",
85 'desc' => "{BasePlugin.filename_encoding.auto_fl}" }, # locale followed by textcat
86 { 'name' => "auto-lf",
87 'desc' => "{BasePlugin.filename_encoding.auto_lf}" } ]; # texcat followed by locale
88
89push(@{$encoding_plus_auto_list},@{$encoding_list});
90
91our $oidtype_list =
92 [ { 'name' => "auto",
93 'desc' => "{BasePlugin.OIDtype.auto}" },
94 { 'name' => "hash",
95 'desc' => "{import.OIDtype.hash}" },
96 { 'name' => "assigned",
97 'desc' => "{import.OIDtype.assigned}" },
98 { 'name' => "incremental",
99 'desc' => "{import.OIDtype.incremental}" },
100 { 'name' => "dirname",
101 'desc' => "{import.OIDtype.dirname}" } ];
102
103my $arguments =
104 [ { 'name' => "process_exp",
105 'desc' => "{BasePlugin.process_exp}",
106 'type' => "regexp",
107 'deft' => "",
108 'reqd' => "no" },
109 { 'name' => "no_blocking",
110 'desc' => "{BasePlugin.no_blocking}",
111 'type' => "flag",
112 'reqd' => "no"},
113 { 'name' => "block_exp",
114 'desc' => "{BasePlugin.block_exp}",
115 'type' => "regexp",
116 'deft' => "",
117 'reqd' => "no" },
118 { 'name' => "store_original_file",
119 'desc' => "{BasePlugin.store_original_file}",
120 'type' => "flag",
121 'reqd' => "no" },
122 { 'name' => "associate_ext",
123 'desc' => "{BasePlugin.associate_ext}",
124 'type' => "string",
125 'reqd' => "no" },
126 { 'name' => "associate_tail_re",
127 'desc' => "{BasePlugin.associate_tail_re}",
128 'type' => "string",
129 'reqd' => "no" },
130 { 'name' => "OIDtype",
131 'desc' => "{import.OIDtype}",
132 'type' => "enum",
133 'list' => $oidtype_list,
134 # leave default empty so we can tell if its been set or not - if not set will use option from import.pl
135 'deft' => "auto",
136 'reqd' => "no" },
137 { 'name' => "OIDmetadata",
138 'desc' => "{import.OIDmetadata}",
139 'type' => "metadata",
140 'deft' => "dc.Identifier",
141 'reqd' => "no" },
142# { 'name' => "use_as_doc_identifier",
143# 'desc' => "{BasePlugin.use_as_doc_identifier}",
144# 'type' => "string",
145# 'reqd' => "no" ,
146# 'deft' => "" } ,
147 { 'name' => "no_cover_image",
148 'desc' => "{BasePlugin.no_cover_image}",
149 'type' => "flag",
150 'reqd' => "no" },
151 { 'name' => "filename_encoding",
152 'desc' => "{BasePlugin.filename_encoding}",
153 'type' => "enum",
154 'deft' => "auto",
155 'list' => $encoding_plus_auto_list,
156 'reqd' => "no" },
157 { 'name' => "smart_block",
158 'desc' => "{common.deprecated}. {BasePlugin.smart_block}",
159 'type' => "flag",
160 'reqd' => "no",
161 'hiddengli' => "yes" }, # deprecated, but leave in for old collections
162 { 'name' => "file_rename_method",
163 'desc' => "{BasePlugin.file_rename_method}",
164 'type' => "enum",
165 'deft' => &get_default_file_rename_method(), # by default rename imported files and assoc files using this encoding
166 'list' => $file_rename_method_list,
167 'reqd' => "no"
168 }
169
170 ];
171
172
173my $options = { 'name' => "BasePlugin",
174 'desc' => "{BasePlugin.desc}",
175 'abstract' => "yes",
176 'inherits' => "no",
177 'args' => $arguments };
178
179sub new {
180
181 my ($class) = shift (@_);
182 my ($pluginlist,$inputargs,$hashArgOptLists,$auxiliary) = @_;
183 push(@$pluginlist, $class);
184
185 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
186 push(@{$hashArgOptLists->{"OptList"}},$options);
187
188 my $self = new PrintInfo($pluginlist, $inputargs, $hashArgOptLists,$auxiliary);
189
190 if ($self->{'info_only'}) {
191 # don't worry about any options etc
192 return bless $self, $class;
193 }
194
195 if ($self->{'smart_block'}) {
196 print STDERR "WARNING: -smart_block option has been deprecated and is no longer useful\n";
197 }
198 $self->{'smart_block'} = undef;
199
200 my $plugin_name = (defined $pluginlist->[0]) ? $pluginlist->[0] : $class;
201 $self->{'plugin_type'} = $plugin_name;
202
203 # remove ex. from OIDmetadata
204 $self->{'OIDmetadata'} =~ s/^ex\.// if defined $self->{'OIDmetadata'};
205 $self->{'num_processed'} = 0;
206 $self->{'num_not_processed'} = 0;
207 $self->{'num_blocked'} = 0;
208 $self->{'num_archives'} = 0;
209 $self->{'cover_image'} = 1; # cover image is on by default
210 $self->{'cover_image'} = 0 if ($self->{'no_cover_image'});
211 $self->{'can_process_directories'} = 0;
212 #$self->{'option_list'} = $hashArgOptLists->{"OptList"};
213
214 my $associate_ext = $self->{'associate_ext'};
215 if ((defined $associate_ext) && ($associate_ext ne "")) {
216
217 my $associate_tail_re = $self->{'associate_tail_re'};
218 if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
219 my $outhandle = $self->{'outhandle'};
220 print $outhandle "Warning: can only specify 'associate_ext' or 'associate_tail_re'\n";
221 print $outhandle " defaulting to 'associate_tail_re'\n";
222 }
223 else {
224 my @exts = split(/,/,$associate_ext);
225
226 my @exts_bracketed = map { $_ = "(?:\\.$_)" } @exts;
227 my $associate_tail_re = join("|",@exts_bracketed);
228 $self->{'associate_tail_re'} = $associate_tail_re;
229 }
230
231 delete $self->{'associate_ext'};
232 }
233
234 return bless $self, $class;
235
236}
237
238sub merge_inheritance
239{
240 my $self = {};
241 my @child_selfs = @_;
242
243 foreach my $child_self (@child_selfs) {
244 foreach my $key (keys %$child_self) {
245 if (defined $self->{$key}) {
246 if ($self->{$key} ne $child_self->{$key}) {
247# print STDERR "Warning: Conflicting value in multiple inheritance for '$key'\n";
248# print STDERR "Existing stored value = $self->{$key}\n";
249# print STDERR "New (child) value = $child_self->{$key}\n";
250# print STDERR "Keeping existing value\n";
251 # Existing value seems to be option specified in collect.cfg
252
253 ### $self->{$key} = $child_self->{$key};
254
255 }
256 else {
257## print STDERR "****Info: Value $self->{$key} for $key already defined through multiple inheritance as the same value\n";
258 }
259
260 }
261 else {
262 $self->{$key} = $child_self->{$key};
263 }
264 }
265 }
266
267 return $self;
268}
269
270# initialize BasePlugin options
271# if init() is overridden in a sub-class, remember to call BasePlugin::init()
272sub init {
273 my $self = shift (@_);
274 my ($verbosity, $outhandle, $failhandle) = @_;
275
276 # verbosity is passed through from the processor
277 $self->{'verbosity'} = $verbosity;
278
279 # as are the outhandle and failhandle
280 $self->{'outhandle'} = $outhandle if defined $outhandle;
281 $self->{'failhandle'} = $failhandle;
282# $self->SUPER::init(@_);
283
284 # set process_exp and block_exp to defaults unless they were
285 # explicitly set
286
287 if ((!$self->is_recursive()) and
288 (!defined $self->{'process_exp'}) || ($self->{'process_exp'} eq "")) {
289
290 $self->{'process_exp'} = $self->get_default_process_exp ();
291 if ($self->{'process_exp'} eq "") {
292 warn ref($self) . " Warning: Non-recursive plugin has no process_exp\n";
293 }
294 }
295
296 if ((!defined $self->{'block_exp'}) || ($self->{'block_exp'} eq "")) {
297 $self->{'block_exp'} = $self->get_default_block_exp ();
298 }
299
300}
301
302sub begin {
303 my $self = shift (@_);
304 my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
305
306 if ($self->{'OIDtype'} eq "auto") {
307 # hasn't been set in the plugin, use the processor values
308 $self->{'OIDtype'} = $processor->{'OIDtype'};
309 $self->{'OIDmetadata'} = $processor->{'OIDmetadata'};
310 }
311 if ($self->{'OIDtype'} eq "hash") {
312 # should we hash on the file or on the doc xml??
313 $self->{'OIDtype'} = $self->get_oid_hash_type();
314 if ($self->{'OIDtype'} !~ /^(hash_on_file|hash_on_ga_xml)$/) {
315 $self->{'OIDtype'} = "hash_on_file";
316 }
317 }
318}
319
320# This is called once if removeold is set with import.pl. Most plugins will do
321# nothing but if a plugin does any stuff outside of creating doc obj, then
322# it may need to clear something.
323sub remove_all {
324 my $self = shift (@_);
325 my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
326}
327
328# This is called per document for docs that have been deleted from the
329# collection. Most plugins will do nothing
330# but if a plugin does any stuff outside of creating doc obj, then it may need
331# to clear something.
332sub remove_one {
333 my $self = shift (@_);
334
335 my ($file, $oids, $archivedir) = @_;
336 return 0 if $self->can_process_this_file($file);
337 return undef;
338}
339
340sub end {
341 # potentially called at the end of each plugin pass
342 # import.pl only has one plugin pass, but buildcol.pl has multiple ones
343
344 my ($self) = shift (@_);
345}
346
347sub deinit {
348 # called only once, after all plugin passes have been done
349
350 my ($self) = @_;
351}
352
353# default hashing type is to hash on the original file (or converted file)
354# override this to return hash_on_ga_xml for filetypes where hashing on the
355# file is no good eg video
356sub get_oid_hash_type {
357
358 my $self = shift (@_);
359
360 return "hash_on_file";
361}
362
363
364# this function should be overridden to return 1
365# in recursive plugins
366sub is_recursive {
367 my $self = shift (@_);
368
369 return 0;
370}
371
372sub get_default_block_exp {
373 my $self = shift (@_);
374
375 return "";
376}
377
378sub get_default_process_exp {
379 my $self = shift (@_);
380
381 return "";
382}
383
384
385# rename imported files and assoc files using URL encoding by default
386# as this will work for most plugins and give more legible filenames
387sub get_default_file_rename_method() {
388 my $self = shift (@_);
389 return "url";
390}
391
392# returns this plugin's active (possibly user-selected) file_rename_method
393sub get_file_rename_method() {
394 my $self = shift (@_);
395 my $rename_method = $self->{'file_rename_method'};
396 if($rename_method) {
397 return $rename_method;
398 } else {
399 return $self->get_default_file_rename_method();
400 }
401}
402
403# default implementation is to do nothing
404sub store_block_files {
405
406 my $self =shift (@_);
407 my ($filename_full_path, $block_hash) = @_;
408
409}
410
411# put files to block into hash
412sub use_block_expressions {
413
414 my $self =shift (@_);
415 my ($filename_full_path, $block_hash) = @_;
416
417 $filename_full_path = &util::upgrade_if_dos_filename($filename_full_path);
418
419 if ($self->{'block_exp'} ne "" && $filename_full_path =~ /$self->{'block_exp'}/) {
420 &util::block_filename($block_hash,$filename_full_path);
421 }
422
423}
424
425#default implementation is to block a file with same name as this, but extension jpg or JPG, if cover_images is on.
426sub block_cover_image
427{
428 my $self =shift;
429 my ($filename, $block_hash) = @_;
430
431 $filename = &util::upgrade_if_dos_filename($filename);
432
433 if ($self->{'cover_image'}) {
434 my $coverfile = $filename;
435 $coverfile =~ s/\.[^\\\/\.]+$/\.jpg/;
436 if (!&util::fd_exists($coverfile)) {
437 $coverfile =~ s/jpg$/JPG/;
438 }
439 if (&util::fd_exists($coverfile)) {
440 &util::block_filename($block_hash,$coverfile);
441 }
442 }
443
444 return;
445}
446
447
448# discover all the files that should be blocked by this plugin
449# check the args ...
450sub file_block_read {
451
452 my $self = shift (@_);
453 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
454 # Keep track of filenames with same root but different extensions
455 # Used to support -associate_ext and the more generalised
456 # -associate_tail_re
457 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
458
459 if (!-d $filename_full_path) {
460 $block_hash->{'all_files'}->{$file} = 1;
461 }
462
463 my $associate_tail_re = $self->{'associate_tail_re'};
464 if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
465 my ($file_prefix,$file_ext)
466 = &util::get_prefix_and_tail_by_regex($filename_full_path,$associate_tail_re);
467 if ((defined $file_prefix) && (defined $file_ext)) {
468 my $shared_fileroot = $block_hash->{'shared_fileroot'};
469 if (!defined $shared_fileroot->{$file_prefix}) {
470 my $file_prefix_rec = { 'tie_to' => undef,
471 'exts' => {} };
472 $shared_fileroot->{$file_prefix} = $file_prefix_rec;
473 }
474
475 my $file_prefix_rec = $shared_fileroot->{$file_prefix};
476
477 if ($self->can_process_this_file($filename_full_path)) {
478 # This is the document the others should be tied to
479 $file_prefix_rec->{'tie_to'} = $file_ext;
480 }
481 else {
482 if ($file_ext =~ m/$associate_tail_re$/) {
483 # this file should be associated to the main one
484 $file_prefix_rec->{'exts'}->{$file_ext} = 1;
485 }
486 }
487
488 }
489 }
490
491 # check block expressions
492 $self->use_block_expressions($filename_full_path, $block_hash) unless $self->{'no_blocking'};
493
494 # now check whether we are actually processing this
495 if (!-f $filename_full_path || !$self->can_process_this_file($filename_full_path)) {
496 return undef; # can't recognise
497 }
498
499 # if we have a block_exp, then this overrides the normal 'smart' blocking
500 $self->store_block_files($filename_full_path, $block_hash) unless ($self->{'no_blocking'} || $self->{'block_exp'} ne "");
501
502 # block the cover image if there is one
503 if ($self->{'cover_image'}) {
504 $self->block_cover_image($filename_full_path, $block_hash);
505 }
506
507 return 1;
508}
509
510# plugins that rely on more than process_exp (eg XML plugins) can override this method
511sub can_process_this_file {
512 my $self = shift(@_);
513 my ($filename) = @_;
514
515 if (-d $filename && !$self->{'can_process_directories'}) {
516 return 0;
517 }
518
519 if ($self->{'process_exp'} ne "" && $filename =~ /$self->{'process_exp'}/) {
520 return 1;
521 }
522 return 0;
523
524}
525
526# just converts path as is to utf8.
527sub filepath_to_utf8 {
528 my $self = shift (@_);
529 my ($file, $file_encoding) = @_;
530 my $filemeta = $file;
531
532 my $filename_encoding = $self->{'filename_encoding'}; # filename encoding setting
533
534 # Whenever filename-encoding is set to any of the auto settings, we
535 # check if the filename is already in UTF8. If it is, then we're done.
536 if($filename_encoding =~ m/auto/) {
537 if(&unicode::check_is_utf8($filemeta))
538 {
539 $filename_encoding = "utf8";
540 return $filemeta;
541 }
542 }
543
544 # Auto setting, but filename is not utf8
545 if ($filename_encoding eq "auto")
546 {
547 # try textcat
548 $filename_encoding = $self->textcat_encoding($filemeta);
549
550 # check the locale next
551 $filename_encoding = $self->locale_encoding() if $filename_encoding eq "undefined";
552
553
554 # now try the encoding of the document, if available
555 if ($filename_encoding eq "undefined" && defined $file_encoding) {
556 $filename_encoding = $file_encoding;
557 }
558
559 }
560
561 elsif ($filename_encoding eq "auto-language-analysis")
562 {
563 $filename_encoding = $self->textcat_encoding($filemeta);
564
565 # now try the encoding of the document, if available
566 if ($filename_encoding eq "undefined" && defined $file_encoding) {
567 $filename_encoding = $file_encoding;
568 }
569 }
570
571 elsif ($filename_encoding eq "auto-filesystem-encoding")
572 {
573 # try locale
574 $filename_encoding = $self->locale_encoding();
575 }
576
577 elsif ($filename_encoding eq "auto-fl")
578 {
579 # filesystem-encoding (locale) then language-analysis (textcat)
580 $filename_encoding = $self->locale_encoding();
581
582 # try textcat
583 $filename_encoding = $self->textcat_encoding($filemeta) if $filename_encoding eq "undefined";
584
585 # else assume filename encoding is encoding of file content, if that's available
586 if ($filename_encoding eq "undefined" && defined $file_encoding) {
587 $filename_encoding = $file_encoding;
588 }
589 }
590
591 elsif ($filename_encoding eq "auto-lf")
592 {
593 # language-analysis (textcat) then filesystem-encoding (locale)
594 $filename_encoding = $self->textcat_encoding($filemeta);
595
596 # guess filename encoding from encoding of file content, if available
597 if ($filename_encoding eq "undefined" && defined $file_encoding) {
598 $filename_encoding = $file_encoding;
599 }
600
601 # try locale
602 $filename_encoding = $self->locale_encoding() if $filename_encoding eq "undefined";
603 }
604
605 # if still undefined, use utf8 as fallback
606 if ($filename_encoding eq "undefined") {
607 $filename_encoding = "utf8";
608 }
609
610 #print STDERR "**** UTF8 encoding the filename $filemeta ";
611
612 # if the filename encoding is set to utf8 but it isn't utf8 already--such as when
613 # 1. the utf8 fallback is used, or 2. if the system locale is used and happens to
614 # be always utf8 (in which case the filename's encoding is also set as utf8 even
615 # though the filename need not be if it originates from another system)--in such
616 # cases attempt to make the filename utf8 to match.
617 if($filename_encoding eq "utf8" && !&unicode::check_is_utf8($filemeta)) {
618 &unicode::ensure_utf8(\$filemeta);
619 }
620
621 # convert non-unicode encodings to utf8
622 if ($filename_encoding !~ m/(?:ascii|utf8|unicode)/) {
623 $filemeta = &unicode::unicode2utf8(
624 &unicode::convert2unicode($filename_encoding, \$filemeta)
625 );
626 }
627
628 #print STDERR " from encoding $filename_encoding -> $filemeta\n";
629 return $filemeta;
630}
631
632# gets the filename with no path, converts to utf8, and then dm safes it.
633# filename_encoding set by user
634sub filename_to_utf8_metadata
635{
636 my $self = shift (@_);
637 my ($file, $file_encoding) = @_;
638
639 my $outhandle = $self->{'outhandle'};
640
641 print $outhandle "****!!!!**** BasePlugin::filename_to_utf8_metadata now deprecated\n";
642 my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(0);
643 print $outhandle "Calling method: $cfilename:$cline $cpackage->$csubr\n";
644
645
646 my ($filemeta) = $file =~ /([^\\\/]+)$/; # getting the tail of the filepath (skips all string parts containing slashes upto the end)
647 $filemeta = $self->filepath_to_utf8($filemeta, $file_encoding);
648
649 return $filemeta;
650}
651
652sub locale_encoding {
653 my $self = shift(@_);
654
655 if (!defined $self->{'filesystem_encoding'}) {
656 $self->{'filesystem_encoding'} = $self->get_filesystem_encoding();
657 }
658
659 #print STDERR "*** filename encoding determined based on locale: " . $self->{'filesystem_encoding'} . "\n";
660 return $self->{'filesystem_encoding'}; # can be the string "undefined"
661}
662
663sub textcat_encoding {
664 my $self = shift(@_);
665 my ($filemeta) = @_;
666
667 # analyse filenames without extensions and digits (and trimmed of
668 # surrounding whitespace), so that irrelevant chars don't confuse
669 # textcat
670 my $strictfilemeta = $filemeta;
671 $strictfilemeta =~ s/\.[^\.]+$//g;
672 $strictfilemeta =~ s/\d//g;
673 $strictfilemeta =~ s/^\s*//g;
674 $strictfilemeta =~ s/\s*$//g;
675
676 my $filename_encoding = $self->encoding_from_language_analysis($strictfilemeta);
677 if(!defined $filename_encoding) {
678 $filename_encoding = "undefined";
679 }
680
681 return $filename_encoding; # can be the string "undefined"
682}
683
684# performs textcat
685sub encoding_from_language_analysis {
686 my $self = shift(@_);
687 my ($text) = @_;
688
689 my $outhandle = $self->{'outhandle'};
690 my $best_encoding = undef;
691
692 # get the language/encoding of the textstring using textcat
693 require textcat; # Only load the textcat module if it is required
694 $self->{'textcat'} = new textcat() unless defined($self->{'textcat'});
695 my $results = $self->{'textcat'}->classify_cached_filename(\$text);
696
697
698 if (scalar @$results < 0) {
699 return undef;
700 }
701
702 # We have some results, we choose the first
703 my ($language, $encoding) = $results->[0] =~ /^([^-]*)(?:-(.*))?$/;
704
705 $best_encoding = $encoding;
706 if (!defined $best_encoding) {
707 return undef;
708 }
709
710 if (defined $best_encoding && $best_encoding =~ m/^iso_8859/ && &unicode::check_is_utf8($text)) {
711 # the text is valid utf8, so assume that's the real encoding (since textcat is based on probabilities)
712 $best_encoding = 'utf8';
713 }
714
715
716 # check for equivalents where textcat doesn't have some encodings...
717 # eg MS versions of standard encodings
718 if (defined $best_encoding && $best_encoding =~ /^iso_8859_(\d+)/) {
719 my $iso = $1; # which variant of the iso standard?
720 # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
721 if ($text =~ /[\x80-\x9f]/) {
722 # Western Europe
723 if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
724 elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
725 elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
726 elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
727 elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
728 elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
729 elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
730 }
731 }
732
733 if (defined $best_encoding && $best_encoding !~ /^(ascii|utf8|unicode)$/ &&
734 !defined $encodings::encodings->{$best_encoding})
735 {
736 if ($self->{'verbosity'}) {
737 gsprintf($outhandle, "BasePlugin: {ReadTextFile.unsupported_encoding}\n", $text, $best_encoding, "undef");
738 }
739 $best_encoding = undef;
740 }
741
742 return $best_encoding;
743}
744
745# uses locale
746sub get_filesystem_encoding
747{
748
749 my $self = shift(@_);
750
751 my $outhandle = $self->{'outhandle'};
752 my $filesystem_encoding = undef;
753
754 eval {
755 # Works for Windows as well, returning the DOS code page in use
756 use POSIX qw(locale_h);
757
758 # With only one parameter, setlocale retrieves the
759 # current value
760 my $current_locale = setlocale(LC_CTYPE);
761
762 my $char_encoding = undef;
763 if ($current_locale =~ m/\./) {
764 ($char_encoding) = ($current_locale =~ m/^.*\.(.*?)$/);
765 $char_encoding = lc($char_encoding);
766 }
767 else {
768 if ($current_locale =~ m/^(posix|c)$/i) {
769 $char_encoding = "ascii";
770 }
771 }
772
773 if (defined $char_encoding) {
774 if ($char_encoding =~ m/^(iso)(8859)(\d{1,2})$/) {
775 $char_encoding = "$1\_$2\_$3";
776 }
777
778 $char_encoding =~ s/-/_/g;
779 $char_encoding =~ s/^utf_8$/utf8/;
780
781 if ($char_encoding =~ m/^\d+$/) {
782 if (defined $encodings::encodings->{"windows_$char_encoding"}) {
783 $char_encoding = "windows_$char_encoding";
784 }
785 elsif (defined $encodings::encodings->{"dos_$char_encoding"}) {
786 $char_encoding = "dos_$char_encoding";
787 }
788 }
789
790 if (($char_encoding =~ m/(?:ascii|utf8|unicode)/)
791 || (defined $encodings::encodings->{$char_encoding})) {
792 $filesystem_encoding = $char_encoding;
793 }
794 else {
795 print $outhandle "Warning: Unsupported character encoding '$char_encoding' from locale '$current_locale'\n";
796 }
797 }
798
799
800 };
801 if ($@) {
802 print $outhandle "$@\n";
803 print $outhandle "Warning: Unable to establish locale. Will assume filesystem is UTF-8\n";
804
805 }
806
807 return $filesystem_encoding;
808}
809
810
811
812sub deduce_filename_encoding
813{
814 my $self = shift (@_);
815 my ($file,$metadata,$plugin_filename_encoding) = @_;
816
817 my $gs_filename_encoding = $metadata->{"gs.filenameEncoding"};
818 my $deduced_filename_encoding = undef;
819
820 # Start by looking for manually assigned metadata
821 if (defined $gs_filename_encoding) {
822 if (ref ($gs_filename_encoding) eq "ARRAY") {
823 my $outhandle = $self->{'outhandle'};
824
825 $deduced_filename_encoding = $gs_filename_encoding->[0];
826
827 my $num_vals = scalar(@$gs_filename_encoding);
828 if ($num_vals>1) {
829 print $outhandle "Warning: gs.filenameEncoding multiply defined for $file\n";
830 print $outhandle " Selecting first value: $deduced_filename_encoding\n";
831 }
832 }
833 else {
834 $deduced_filename_encoding = $gs_filename_encoding;
835 }
836 }
837
838 if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
839 # Look to see if plugin specifies this value
840
841 if (defined $plugin_filename_encoding) {
842 # First look to see if we're using any of the "older" (i.e. deprecated auto-... plugin options)
843 if ($plugin_filename_encoding =~ m/^auto-.*$/) {
844 my $outhandle = $self->{'outhandle'};
845 print $outhandle "Warning: $plugin_filename_encoding is no longer supported\n";
846 print $outhandle " default to 'auto'\n";
847 $self->{'filename_encoding'} = $plugin_filename_encoding = "auto";
848 }
849
850 if ($plugin_filename_encoding ne "auto") {
851 # We've been given a specific filenamne encoding
852 # => so use it!
853 $deduced_filename_encoding = $plugin_filename_encoding;
854 }
855 }
856 }
857
858 if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
859
860 # Look to file system to provide a character encoding
861
862 # If Windows NTFS, then -- assuming we work with long file names got through
863 # Win32::GetLongFilePath() -- then the underlying file system is UTF16
864
865 if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
866 # Can do better than working with the DOS character encoding returned by locale
867 $deduced_filename_encoding = "unicode";
868 }
869 else {
870 # Unix of some form or other
871
872 # See if we can determine the file system encoding through locale
873 $deduced_filename_encoding = $self->locale_encoding();
874
875 # if locale shows us filesystem is utf8, check to see filename is consistent
876 # => if not, then we have an "alien" filename on our hands
877
878 if (defined $deduced_filename_encoding && $deduced_filename_encoding =~ m/^utf-?8$/i) {
879 if (!&unicode::check_is_utf8($file)) {
880 # "alien" filename, so revert
881 $deduced_filename_encoding = undef;
882 }
883 }
884 }
885 }
886
887# if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
888# # Last chance, apply textcat to deduce filename encoding
889# $deduced_filename_encoding = $self->textcat_encoding($file);
890# }
891
892 if ($self->{'verbosity'}>3) {
893 my $outhandle = $self->{'outhandle'};
894
895 if (defined $deduced_filename_encoding) {
896 print $outhandle " Deduced filename encoding as: $deduced_filename_encoding\n";
897 }
898 else {
899 print $outhandle " No filename encoding deduced\n";
900 }
901 }
902
903 return $deduced_filename_encoding;
904}
905
906
907
908
909# Notionally written to be called once for each document, it is however safe to
910# call multiple times (as in the case of ImagePlugin) which calls this later on
911# after the original image has potentially been converted to a *new* source image
912# format (e.g. TIFF to PNG)
913
914sub set_Source_metadata {
915 my $self = shift (@_);
916 my ($doc_obj, $raw_filename, $filename_encoding, $section) = @_;
917
918 # 1. Sets the filename (Source) for display encoded as Unicode if possible,
919 # and (as a fallback) using %xx if not for non-ascii chars
920 # 2. Sets the url ref (SourceFile) to the URL encoded version
921 # of filename for generated files
922
923 my ($unused_full_rf, $raw_file) = &util::get_full_filenames("", $raw_filename);
924
925 my $this_section = (defined $section)? $section : $doc_obj->get_top_section();
926
927 my $octet_file = $raw_file;
928
929 # UTF-8 version of filename
930# if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
931# print STDERR "**** Setting Source Metadata given: $octet_file\n";
932# }
933
934 # Deal with (on Windows) raw filenames that are in their
935 # abbreviated DOS form
936
937 if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
938 if ((defined $filename_encoding) && ($filename_encoding eq "unicode")) {
939 if (-e $raw_filename) {
940 my $unicode_filename = Win32::GetLongPathName($raw_filename);
941
942 my $unused_full_uf;
943 ($unused_full_uf, $octet_file) = &util::get_full_filenames("", $unicode_filename);
944 }
945 }
946 }
947
948 my $url_encoded_filename;
949 if ((defined $filename_encoding) && ($filename_encoding ne "ascii")) {
950 # => Generate a pretty print version of filename that is mapped to Unicode
951
952 # Use filename_encoding to map raw filename to a Perl unicode-aware string
953 $url_encoded_filename = decode($filename_encoding,$octet_file);
954 }
955 else {
956 # otherwise generate %xx encoded version of filename for char > 127
957 $url_encoded_filename = &unicode::raw_filename_to_url_encoded($octet_file);
958 }
959
960# if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
961# print STDERR "****** saving Source as: $url_encoded_filename\n";
962# }
963
964
965 # Source is the UTF8 display name - not necessarily the name of the file on the system
966 $doc_obj->set_utf8_metadata_element($this_section, "Source", $url_encoded_filename);
967
968 my $renamed_raw_file = &util::rename_file($raw_file, $self->{'file_rename_method'});
969 # If using URL encoding, then SourceFile is the url-reference to url-encoded
970 # renamed_raw_url: it's a url that refers to the actual file on the system
971 my $renamed_raw_url = &unicode::filename_to_url($renamed_raw_file);
972
973 $doc_obj->set_utf8_metadata_element($this_section, "SourceFile",
974 $renamed_raw_url);
975
976# if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
977# print STDERR "****** saving SourceFile as: $renamed_raw_url\n";
978# }
979}
980
981# this should be called by all plugins to set the oid of the doc obj, rather
982# than calling doc_obj->set_OID directly
983sub add_OID {
984 my $self = shift (@_);
985 my ($doc_obj) = @_;
986
987 $doc_obj->set_OIDtype($self->{'OIDtype'}, $self->{'OIDmetadata'});
988
989 # see if there is a plugin specific set_OID function
990 if (defined ($self->can('set_OID'))) {
991 $self->set_OID(@_); # pass through doc_obj and any extra arguments
992 }
993 else {
994 # use the default set_OID() in doc.pm
995 $doc_obj->set_OID();
996 }
997
998}
999
1000# The BasePlugin read_into_doc_obj() function. This function does all the
1001# right things to make general options work for a given plugin. It doesn't do anything with the file other than setting reads in
1002# a file and sets up a slew of metadata all saved in doc_obj, which
1003# it then returns as part of a tuple (process_status,doc_obj)
1004#
1005# Much of this functionality used to reside in read, but it was broken
1006# down into a supporting routine to make the code more flexible.
1007#
1008# recursive plugins (e.g. RecPlug) and specialized plugins like those
1009# capable of processing many documents within a single file (e.g.
1010# GMLPlug) will normally want to implement their own version of
1011# read_into_doc_obj()
1012#
1013# Note that $base_dir might be "" and that $file might
1014# include directories
1015
1016# currently blocking has been done before it gets here - does this affect secondary plugin stuff??
1017sub read_into_doc_obj {
1018 my $self = shift (@_);
1019 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
1020
1021 my $outhandle = $self->{'outhandle'};
1022
1023 # should we move this to read? What about secondary plugins?
1024 my $pp_file = &util::prettyprint_file($base_dir,$file,$gli);
1025 print STDERR "<Processing n='$file' p='$self->{'plugin_type'}'>\n" if ($gli);
1026 print $outhandle "$self->{'plugin_type'} processing $pp_file\n"
1027 if $self->{'verbosity'} > 1;
1028
1029 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
1030
1031 # create a new document
1032 my $doc_obj = new doc ($filename_full_path, "indexed_doc", $self->{'file_rename_method'});
1033 my $top_section = $doc_obj->get_top_section();
1034
1035 $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
1036 $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename_full_path));
1037
1038
1039 my $plugin_filename_encoding = $self->{'filename_encoding'};
1040 my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
1041 $self->set_Source_metadata($doc_obj,$filename_full_path,$filename_encoding,$top_section);
1042
1043 # plugin specific stuff - what args do we need here??
1044 unless (defined ($self->process($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli))) {
1045 print STDERR "<ProcessingError n='$file'>\n" if ($gli);
1046 return -1;
1047 }
1048
1049 # include any metadata passed in from previous plugins
1050 # note that this metadata is associated with the top level section
1051 my $section = $doc_obj->get_top_section();
1052 # can we merge these two methods??
1053 $self->add_associated_files($doc_obj, $filename_full_path);
1054 $self->extra_metadata ($doc_obj, $section, $metadata);
1055 $self->auto_extract_metadata($doc_obj);
1056
1057 # if we haven't found any Title so far, assign one
1058 # this was shifted to here from inside read()
1059 $self->title_fallback($doc_obj,$section,$filename_no_path);
1060
1061 $self->add_OID($doc_obj);
1062
1063 $self->post_process_doc_obj($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli);
1064 return (1,$doc_obj);
1065}
1066
1067sub post_process_doc_obj {
1068 my $self = shift (@_);
1069 my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
1070
1071 return 1;
1072}
1073
1074sub add_dummy_text {
1075 my $self = shift(@_);
1076 my ($doc_obj, $section) = @_;
1077
1078 # add NoText metadata so we can hide this dummy text in format statements
1079 $doc_obj->add_metadata($section, "NoText", "1");
1080 $doc_obj->add_text($section, &gsprintf::lookup_string("{BasePlugin.dummy_text}",1));
1081
1082}
1083
1084# does nothing. Can be overridden by subclass
1085sub auto_extract_metadata {
1086 my $self = shift(@_);
1087 my ($doc_obj) = @_;
1088}
1089
1090# adds cover image, associate_file options stuff. Should be called by sub class
1091# read_into_doc_obj
1092sub add_associated_files {
1093 my $self = shift(@_);
1094 # whatis filename??
1095 my ($doc_obj, $filename) = @_;
1096
1097 # add in the cover image
1098 if ($self->{'cover_image'}) {
1099 $self->associate_cover_image($doc_obj, $filename);
1100 }
1101 # store the original (used for eg TextPlugin to store the original for OAI)
1102 if ($self->{'store_original_file'}) {
1103 $self->associate_source_file($doc_obj, $filename);
1104 }
1105
1106
1107}
1108
1109# implement this if you are extracting metadata for other documents
1110sub metadata_read {
1111 my $self = shift (@_);
1112 my ($pluginfo, $base_dir, $file, $block_hash,
1113 $extrametakeys, $extrametadata, $extrametafile,
1114 $processor, $gli, $aux) = @_;
1115
1116 # can we process this file??
1117 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
1118 return undef unless $self->can_process_this_file($filename_full_path);
1119
1120 return 1; # we recognise the file, but don't actually do anything with it
1121}
1122
1123
1124# The BasePlugin read() function. This function calls read_into_doc_obj()
1125# to ensure all the right things to make general options work for a
1126# given plugin are done. It then calls the process() function which
1127# does all the work specific to a plugin (like the old read functions
1128# used to do). Most plugins should define their own process() function
1129# and let this read() function keep control.
1130#
1131# recursive plugins (e.g. RecPlug) and specialized plugins like those
1132# capable of processing many documents within a single file (e.g.
1133# GMLPlug) might want to implement their own version of read(), but
1134# more likely need to implement their own version of read_into_doc_obj()
1135#
1136# Return number of files processed, undef if can't recognise, -1 if can't
1137# process
1138
1139sub read {
1140 my $self = shift (@_);
1141 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
1142
1143 # can we process this file??
1144 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
1145
1146 return undef unless $self->can_process_this_file($filename_full_path);
1147
1148 my ($process_status,$doc_obj) = $self->read_into_doc_obj(@_);
1149
1150 if ((defined $process_status) && ($process_status == 1)) {
1151
1152 # process the document
1153 $processor->process($doc_obj);
1154
1155 $self->{'num_processed'} ++;
1156 undef $doc_obj;
1157 }
1158 # delete any temp files that we may have created
1159 $self->clean_up_after_doc_obj_processing();
1160
1161
1162 # if process_status == 1, then the file has been processed.
1163 return $process_status;
1164
1165}
1166
1167# returns undef if file is rejected by the plugin
1168sub process {
1169 my $self = shift (@_);
1170 my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
1171
1172 gsprintf(STDERR, "BasePlugin::process {common.must_be_implemented}\n");
1173
1174 my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
1175 print STDERR "Calling method: $cfilename:$cline $cpackage->$csubr\n";
1176
1177 die "\n";
1178
1179 return undef; # never gets here
1180}
1181
1182# overwrite this method to delete any temp files that we have created
1183sub clean_up_after_doc_obj_processing {
1184 my $self = shift(@_);
1185
1186}
1187
1188# write_file -- used by ConvertToPlug, for example in post processing
1189#
1190# where should this go, is here the best place??
1191sub utf8_write_file {
1192 my $self = shift (@_);
1193 my ($textref, $filename) = @_;
1194
1195 if (!open (FILE, ">:utf8", $filename)) {
1196 gsprintf(STDERR, "ConvertToPlug::write_file {ConvertToPlug.could_not_open_for_writing} ($!)\n", $filename);
1197 die "\n";
1198 }
1199 print FILE $$textref;
1200
1201 close FILE;
1202}
1203
1204
1205sub filename_based_title
1206{
1207 my $self = shift (@_);
1208 my ($file) = @_;
1209
1210 my $file_derived_title = $file;
1211 $file_derived_title =~ s/_/ /g;
1212 $file_derived_title =~ s/\.[^.]+$//;
1213
1214 return $file_derived_title;
1215}
1216
1217
1218sub title_fallback
1219{
1220 my $self = shift (@_);
1221 my ($doc_obj,$section,$file) = @_;
1222
1223 if (!defined $doc_obj->get_metadata_element ($section, "Title")
1224 || $doc_obj->get_metadata_element($section, "Title") eq "") {
1225
1226 my $source_file = $doc_obj->get_metadata_element($section, "Source");
1227 my $file_derived_title;
1228 if (defined $source_file) {
1229 $file_derived_title = $self->filename_based_title($source_file);
1230 }
1231 else {
1232 # pp = pretty print
1233 my $pp_file = (defined $source_file) ? $source_file : $file;
1234
1235 my $raw_title = $self->filename_based_title($file);
1236 my $file_derived_title = &unicode::raw_filename_to_url_encoded($raw_title);
1237 }
1238
1239
1240 if (!defined $doc_obj->get_metadata_element ($section, "Title")) {
1241 $doc_obj->add_utf8_metadata ($section, "Title", $file_derived_title);
1242 }
1243 else {
1244 $doc_obj->set_utf8_metadata_element ($section, "Title", $file_derived_title);
1245 }
1246 }
1247
1248}
1249
1250# add any extra metadata that's been passed around from one
1251# plugin to another.
1252# extra_metadata uses add_utf8_metadata so it expects metadata values
1253# to already be in utf8
1254sub extra_metadata {
1255 my $self = shift (@_);
1256 my ($doc_obj, $cursection, $metadata) = @_;
1257
1258 my $associate_tail_re = $self->{'associate_tail_re'};
1259
1260 foreach my $field (keys(%$metadata)) {
1261 # $metadata->{$field} may be an array reference
1262 if ($field eq "gsdlassocfile_tobe") {
1263 # 'gsdlassocfile_tobe' is artificially introduced metadata
1264 # that is used to signal that certain additional files should
1265 # be tied to this document. Useful in situations where a
1266 # metadata pass in the plugin pipeline works out some files
1267 # need to be associated with a document, but the document hasn't
1268 # been formed yet.
1269 my $equiv_form = "";
1270 foreach my $gaf (@{$metadata->{$field}}) {
1271 my ($full_filename,$mimetype) = ($gaf =~ m/^(.*):(.*):$/);
1272 my ($tail_filename) = ($full_filename =~ /^.*[\/\\](.+?)$/);
1273
1274 # we need to make sure the filename is valid utf-8 - we do
1275 # this by url or base64 encoding it
1276 # $tail_filename is the name that we store the file as
1277 $tail_filename = &util::rename_file($tail_filename, $self->{'file_rename_method'});
1278 $doc_obj->associate_file($full_filename,$tail_filename,$mimetype);
1279 $doc_obj->associate_source_file($full_filename);
1280 # If the filename is url_encoded, we need to encode the % signs
1281 # in the filename, so that it works in a url
1282 my $url_tail_filename = &unicode::filename_to_url($tail_filename);
1283 # work out extended tail extension (i.e. matching tail re)
1284
1285 my ($file_prefix,$file_extended_ext)
1286 = &util::get_prefix_and_tail_by_regex($tail_filename,$associate_tail_re);
1287 my ($pre_doc_ext) = ($file_extended_ext =~ m/^(.*)\..*$/);
1288 my ($doc_ext) = ($tail_filename =~ m/^.*\.(.*)$/);
1289 my $start_doclink = "<a href=\"_httpprefix_/collect/[collection]/index/assoc/{Or}{[parent(Top):assocfilepath],[assocfilepath]}/$url_tail_filename\">";
1290 my $srcicon = "_icon".$doc_ext."_";
1291 my $end_doclink = "</a>";
1292
1293 my $assoc_form = "$start_doclink\{If\}{$srcicon,$srcicon,$doc_ext\}$end_doclink";
1294
1295 if (defined $pre_doc_ext && $pre_doc_ext ne "") {
1296 # for metadata such as [mp3._edited] [mp3._full] ...
1297 $doc_obj->add_utf8_metadata ($cursection, "$doc_ext.$pre_doc_ext", $assoc_form);
1298 }
1299
1300 # for multiple metadata such as [mp3.assoclink]
1301 $doc_obj->add_utf8_metadata ($cursection, "$doc_ext.assoclink", $assoc_form);
1302
1303 $equiv_form .= " $assoc_form";
1304 }
1305 $doc_obj->add_utf8_metadata ($cursection, "equivlink", $equiv_form);
1306 }
1307 elsif ($field eq "gsdlzipfilename") {
1308 # special case for when files have come out of a zip. source_path
1309 # (used for archives dbs and keeping track for incremental import)
1310 # must be set to the zip file name
1311 my $zip_filename = $metadata->{$field};
1312 # overwrite the source_path
1313 $doc_obj->set_source_path($zip_filename);
1314 # and set the metadata
1315 $zip_filename = &util::filename_within_collection($zip_filename);
1316 $zip_filename = $doc_obj->encode_filename($zip_filename, $self->{'file_rename_method'});
1317 $doc_obj->add_utf8_metadata ($cursection, $field, $zip_filename);
1318 }
1319 elsif (ref ($metadata->{$field}) eq "ARRAY") {
1320 map {
1321 $doc_obj->add_utf8_metadata ($cursection, $field, $_);
1322 } @{$metadata->{$field}};
1323 } else {
1324 $doc_obj->add_utf8_metadata ($cursection, $field, $metadata->{$field});
1325 }
1326 }
1327}
1328
1329
1330sub compile_stats {
1331 my $self = shift(@_);
1332 my ($stats) = @_;
1333
1334 $stats->{'num_processed'} += $self->{'num_processed'};
1335 $stats->{'num_not_processed'} += $self->{'num_not_processed'};
1336 $stats->{'num_archives'} += $self->{'num_archives'};
1337
1338}
1339sub associate_source_file {
1340 my $self = shift(@_);
1341
1342 my ($doc_obj, $filename) = @_;
1343 my $cursection = $doc_obj->get_top_section();
1344 my $assocfilename = $doc_obj->get_assocfile_from_sourcefile();
1345
1346 $doc_obj->associate_file($filename, $assocfilename, undef, $cursection);
1347 $doc_obj->add_utf8_metadata ($cursection, "srclink_file", $doc_obj->get_sourcefile());
1348}
1349
1350sub associate_cover_image {
1351 my $self = shift(@_);
1352 my ($doc_obj, $filename) = @_;
1353
1354 my $upgraded_filename = &util::upgrade_if_dos_filename($filename);
1355
1356 $filename =~ s/\.[^\\\/\.]+$/\.jpg/;
1357 $upgraded_filename =~ s/\.[^\\\/\.]+$/\.jpg/;
1358
1359 if (exists $self->{'covers_missing_cache'}->{$upgraded_filename}) {
1360 # don't stat() for existence e.g. for multiple document input files
1361 # (eg SplitPlug)
1362 return;
1363 }
1364
1365 my $top_section=$doc_obj->get_top_section();
1366
1367 if (&util::fd_exists($upgraded_filename)) {
1368 $doc_obj->associate_source_file($filename);
1369 $doc_obj->associate_file($filename, "cover.jpg", "image/jpeg");
1370 $doc_obj->add_utf8_metadata($top_section, "hascover", 1);
1371 } else {
1372 my $upper_filename = $filename;
1373 my $upgraded_upper_filename = $upgraded_filename;
1374
1375 $upper_filename =~ s/jpg$/JPG/;
1376 $upgraded_upper_filename =~ s/jpg$/JPG/;
1377
1378 if (&util::fd_exists($upgraded_upper_filename)) {
1379 $doc_obj->associate_source_file($upper_filename);
1380 $doc_obj->associate_file($upper_filename, "cover.jpg",
1381 "image/jpeg");
1382 $doc_obj->add_utf8_metadata($top_section, "hascover", 1);
1383 } else {
1384 # file doesn't exist, so record the fact that it's missing so
1385 # we don't stat() again (stat is slow)
1386 $self->{'covers_missing_cache'}->{$upgraded_filename} = 1;
1387 }
1388 }
1389
1390}
1391
1392
1393# Overridden by exploding plugins (eg. ISISPlug)
1394sub clean_up_after_exploding
1395{
1396 my $self = shift(@_);
1397}
1398
1399
1400
14011;
Note: See TracBrowser for help on using the repository browser.