source: main/trunk/greenstone2/perllib/plugin.pm@ 32565

Last change on this file since 32565 was 32565, checked in by ak19, 2 years ago

I think this is a bugfix to plugin.pm::remove_some(): when processing files to delete on deletion/reindexing, it would return after the first successful remove_one(). This wasn't a noticeable problem in the past since no plugin had a real need for implementing remove_one(). But now GS SQLPlugin does remove_one() in earnest, we need all the docs marked for deletion to be processed.

  • Property svn:keywords set to Author Date Id Revision
File size: 15.7 KB
Line 
1###########################################################################
2#
3# plugin.pm -- functions to handle using 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 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 plugin;
27
28
29use strict; # to pick up typos and undeclared variables...
30no strict 'refs'; # ...but allow filehandles to be variables and vice versa
31no strict 'subs';
32
33require util;
34use FileUtils;
35use gsprintf 'gsprintf';
36
37# mapping from old plugin names to new ones for backwards compatibility
38# can remove at sometime in future when we no longer want to support old xxPlug names in the config file
39my $plugin_name_map = {
40 'GAPlug' => 'GreenstoneXMLPlugin',
41 'ArcPlug' => 'ArchivesInfPlugin',
42 'RecPlug' => 'DirectoryPlugin',
43 'TEXTPlug' => 'TextPlugin',
44 'XMLPlug' => 'ReadXMLFile',
45 'EMAILPlug' => 'EmailPlugin',
46 'SRCPlug' => 'SourceCodePlugin',
47 'NULPlug' => 'NulPlugin',
48 'W3ImgPlug' => 'HTMLImagePlugin',
49 'PagedImgPlug' => 'PagedImagePlugin',
50 'METSPlug' => 'GreenstoneMETSPlugin',
51 'PPTPlug' => 'PowerPointPlugin',
52 'PSPlug' => 'PostScriptPlugin',
53 'DBPlug' => 'DatabasePlugin'
54 };
55
56# global variables
57my $stats = {'num_processed' => 0,
58 'num_blocked' => 0,
59 'num_not_processed' => 0,
60 'num_not_recognised' => 0,
61 'num_archives' => 0
62 };
63
64#globaloptions contains any options that should be passed to all plugins
65my ($verbosity, $outhandle, $failhandle, $globaloptions);
66
67sub get_valid_pluginname {
68 my ($pluginname) = @_;
69 my $valid_name = $pluginname;
70 if (defined $plugin_name_map->{$pluginname}) {
71 $valid_name = $plugin_name_map->{$pluginname};
72 } elsif ($pluginname =~ /Plug$/) {
73 $valid_name =~ s/Plug/Plugin/;
74
75 }
76 return $valid_name;
77}
78
79sub load_plugin_require
80{
81 my ($pluginname) = @_;
82
83 my @check_list = ();
84
85 # pp_plugname shorthand for 'perllib' 'plugin' '$pluginname.pm'
86 my $pp_plugname
87 = &FileUtils::filenameConcatenate('perllib', 'plugins', "${pluginname}.pm");
88 my $collectdir = $ENV{'GSDLCOLLECTDIR'};
89
90 # find the plugin
91 if (defined($ENV{'GSDLCOLLECTION'}))
92 {
93 my $customplugname
94 = &FileUtils::filenameConcatenate($collectdir, "custom",$ENV{'GSDLCOLLECTION'},
95 $pp_plugname);
96 push(@check_list,$customplugname);
97 }
98
99 my $colplugname = &FileUtils::filenameConcatenate($collectdir, $pp_plugname);
100 push(@check_list,$colplugname);
101
102 if (defined $ENV{'GSDLEXTS'}) {
103
104 my $ext_prefix = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "ext");
105
106 my @extensions = split(/:/,$ENV{'GSDLEXTS'});
107 foreach my $e (@extensions) {
108 my $extplugname = &FileUtils::filenameConcatenate($ext_prefix, $e, $pp_plugname);
109 push(@check_list,$extplugname);
110
111 }
112 }
113 if (defined $ENV{'GSDL3EXTS'}) {
114
115 my $ext_prefix = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "ext");
116
117 my @extensions = split(/:/,$ENV{'GSDL3EXTS'});
118 foreach my $e (@extensions) {
119 my $extplugname = &FileUtils::filenameConcatenate($ext_prefix, $e, $pp_plugname);
120 push(@check_list,$extplugname);
121
122 }
123 }
124
125
126 my $mainplugname = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, $pp_plugname);
127 push(@check_list,$mainplugname);
128
129 my $success=0;
130 foreach my $plugname (@check_list) {
131 if (&FileUtils::fileExists($plugname)) {
132 # lets add perllib folder to INC
133 # check it isn't already there first [jmt12]
134 my ($perllibfolder) = $plugname =~ /^(.*[\/\\]perllib)[\/\\]plugins/;
135 if (&FileUtils::directoryExists($perllibfolder))
136 {
137 my $found_perllibfolder = 0;
138 foreach my $path (@INC)
139 {
140 if ($path eq $perllibfolder)
141 {
142 $found_perllibfolder = 1;
143 last;
144 }
145 }
146 if (!$found_perllibfolder)
147 {
148 unshift (@INC, $perllibfolder);
149 }
150 }
151 require $plugname;
152 $success=1;
153 last;
154 }
155 }
156
157 if (!$success) {
158 &gsprintf(STDERR, "{plugin.could_not_find_plugin}\n",
159 $pluginname);
160 die "\n";
161 }
162}
163
164sub load_plugin_for_info {
165 my ($pluginname, $gs_version) = (@_);
166 $pluginname = &get_valid_pluginname($pluginname);
167 load_plugin_require($pluginname);
168
169 # create a plugin object
170 my ($plugobj);
171 my $options = "-gsdlinfo,-gs_version,$gs_version";
172
173 eval ("\$plugobj = new \$pluginname([],[$options])");
174 die "$@" if $@;
175
176 return $plugobj;
177}
178
179sub load_plugins {
180 my ($plugin_list) = shift @_;
181 my ($incremental_mode, $gs_version, $site);
182 ($verbosity, $outhandle, $failhandle, $globaloptions, $incremental_mode, $gs_version, $site) = @_; # globals
183 my @plugin_objects = ();
184 $verbosity = 2 unless defined $verbosity;
185 $outhandle = 'STDERR' unless defined $outhandle;
186 $failhandle = 'STDERR' unless defined $failhandle;
187
188 # before pushing collection perl and plugin directories onto INC, test that
189 # they aren't already there [jmt12]
190 &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},'perllib'));
191 &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},'perllib','plugins'));
192
193 map { $_ = "\"$_\""; } @$globaloptions;
194 my $globals = join (",", @$globaloptions);
195
196 foreach my $pluginoptions (@$plugin_list) {
197 my $pluginname = shift @$pluginoptions;
198 next unless defined $pluginname;
199 $pluginname = &get_valid_pluginname($pluginname);
200 load_plugin_require($pluginname);
201
202 # create a plugin object
203 my ($plugobj);
204 # put quotes around each option to the plugin, unless the option is already quoted
205 map { $_ = "\"$_\"" unless ($_ =~ m/^\s*\".*\"\s*$/) ; } @$pluginoptions;
206 my $site_option = $site ? "\"-site_name\",\"$site\"," : "";
207 my $options = "$site_option"."-gs_version,$gs_version,".join (",", @$pluginoptions);
208 if ($globals) {
209 if (@$pluginoptions) {
210 $options .= ",";
211 }
212 $options .= "$globals";
213 }
214 # need to escape backslash before putting in to the eval
215 # but watch out for any \" (which shouldn't be further escaped)
216 $options =~ s/\\([^"])/\\\\$1/g; #"
217 $options =~ s/\$/\\\$/g;
218
219 eval ("\$plugobj = new \$pluginname([],[$options])");
220 die "$@" if $@;
221
222 # initialize plugin
223 $plugobj->init($verbosity, $outhandle, $failhandle);
224
225 $plugobj->set_incremental($incremental_mode);
226
227 # add this object to the list
228 push (@plugin_objects, $plugobj);
229 }
230
231 return \@plugin_objects;
232}
233
234
235sub begin {
236 my ($pluginfo, $base_dir, $processor, $maxdocs, $gli) = @_;
237
238 map { $_->{'gli'} = $gli; } @$pluginfo;
239 map { $_->begin($pluginfo, $base_dir, $processor, $maxdocs); } @$pluginfo;
240}
241
242 sub remove_all {
243 my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
244
245 map { $_->remove_all($pluginfo, $base_dir, $processor, $maxdocs); } @$pluginfo;
246}
247
248sub remove_some {
249 my ($pluginfo, $infodbtype, $archivedir, $deleted_files) = @_;
250 return if (scalar(@$deleted_files)==0);
251 $infodbtype = "gdbm" if $infodbtype eq "gdbm-txtgz";
252 my $arcinfo_src_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-src", $archivedir);
253
254 foreach my $file (@$deleted_files) {
255 # use 'archiveinf-src' info database to look up all the OIDs
256 # that this file is used in (note in most cases, it's just one OID)
257
258 my $processed_file = 0; # set to 1 if a plugin could process the file and did so successfully
259
260 my $file_with_placeholders = &util::abspath_to_placeholders($file);
261 my $src_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_src_filename, $file_with_placeholders);
262 my $oids = $src_rec->{'oid'};
263 my $rv;
264 foreach my $plugobj (@$pluginfo) {
265
266 $rv = $plugobj->remove_one($file, $oids, $archivedir);
267 if (defined $rv && $rv != -1) {
268 #return $rv;
269 $processed_file = 1;
270 last; # break and continue with outer for loop, to process other deleted files
271 } # else undefined (was not recognised by the plugin) or there was an error, try the next one
272 }
273 #return 0;
274
275 if (!$processed_file) { # no plugin could recognise file.
276 # Should we continue processing other deleted files or not?
277 print STDERR "WARNING: plugin::remove_some() failed to process $file with oid(s) ". join(",", $oids) . "\n";
278 return 0;
279 } # else some plugin processed the current deleted file
280 # continue to process next deleted file
281
282 }
283 return 1; # if we got here, all deleted files got processed successfully
284}
285
286sub file_block_read {
287 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
288
289
290 $gli = 0 unless defined $gli;
291
292 my $rv = 0;
293 my $glifile = $file;
294
295 $glifile =~ s/^[\/\\]+//; # file sometimes starts with a / so get rid of it
296
297 # Announce to GLI that we are handling a file
298 print STDERR "<File n='$glifile'>\n" if $gli;
299
300 # the .kill file is a handy (if not very elegant) way of aborting
301 # an import.pl or buildcol.pl process
302 if (&FileUtils::fileExists(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, ".kill"))) {
303 gsprintf($outhandle, "{plugin.kill_file}\n");
304 die "\n";
305 }
306
307 foreach my $plugobj (@$pluginfo) {
308
309 $rv = $plugobj->file_block_read($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli);
310 #last if (defined $rv && $rv==1); # stop this file once we have found something to 'process' it
311 }
312
313}
314
315
316sub metadata_read {
317 my ($pluginfo, $base_dir, $file, $block_hash,
318 $extrametakeys, $extrametadata, $extrametafile,
319 $processor, $gli, $aux) = @_;
320
321 $gli = 0 unless defined $gli;
322
323 my $rv = 0;
324 my $glifile = $file;
325
326 $glifile =~ s/^[\/\\]+//; # file sometimes starts with a / so get rid of it
327
328 # Announce to GLI that we are handling a file
329 print STDERR "<File n='$glifile'>\n" if $gli;
330
331 # the .kill file is a handy (if not very elegant) way of aborting
332 # an import.pl or buildcol.pl process
333 if (&FileUtils::fileExists(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, ".kill"))) {
334 gsprintf($outhandle, "{plugin.kill_file}\n");
335 die "\n";
336 }
337
338 my $had_error = 0;
339 # pass this file by each of the plugins in turn until one
340 # is found which will process it
341 # read must return:
342 # undef - could not recognise
343 # -1 - tried but error
344 # 0 - blocked
345 # anything else for successful processing
346
347 foreach my $plugobj (@$pluginfo) {
348
349 $rv = $plugobj->metadata_read($pluginfo, $base_dir, $file, $block_hash,
350 $extrametakeys, $extrametadata, $extrametafile,
351 $processor, $gli, $aux);
352
353 if (defined $rv) {
354 if ($rv == -1) {
355 # an error has occurred
356 $had_error = 1;
357 print STDERR "<ProcessingError n='$glifile'>\n" if $gli;
358 } else {
359 return $rv;
360 }
361 } # else undefined - was not recognised by the plugin
362 }
363
364 return 0;
365}
366
367sub read {
368 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli, $aux) = @_;
369
370 $maxdocs = -1 unless defined $maxdocs && $maxdocs =~ /\d/;
371 $total_count = 0 unless defined $total_count && $total_count =~ /\d/;
372 $gli = 0 unless defined $gli;
373
374 my $rv = 0;
375 my $glifile = $file;
376
377 $glifile =~ s/^[\/\\]+//; # file sometimes starts with a / so get rid of it
378
379 # Announce to GLI that we are handling a file
380 print STDERR "<File n='$glifile'>\n" if $gli;
381
382 # the .kill file is a handy (if not very elegant) way of aborting
383 # an import.pl or buildcol.pl process
384 if (&FileUtils::fileExists(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, ".kill"))) {
385 gsprintf($outhandle, "{plugin.kill_file}\n");
386 die "\n";
387 }
388
389 my $had_error = 0;
390 # pass this file by each of the plugins in turn until one
391 # is found which will process it
392 # read must return:
393 # undef - could not recognise
394 # -1 - tried but error
395 # 0 - blocked
396 # anything else for successful processing
397
398 foreach my $plugobj (@$pluginfo) {
399
400 $rv = $plugobj->read($pluginfo, $base_dir, $file,
401 $block_hash, $metadata, $processor, $maxdocs,
402 $total_count, $gli, $aux);
403
404 if (defined $rv) {
405 if ($rv == -1) {
406 # an error has occurred
407 $had_error = 1;
408 } else {
409 return $rv;
410 }
411 } # else undefined - was not recognised by the plugin
412 }
413
414 if ($had_error) {
415 # was recognised but couldn't be processed
416 if ($verbosity >= 2) {
417 gsprintf($outhandle, "{plugin.no_plugin_could_process}\n", $file);
418 }
419 # tell the GLI that it was not processed
420 print STDERR "<NonProcessedFile n='$glifile'>\n" if $gli;
421
422 gsprintf($failhandle, "$file: {plugin.no_plugin_could_process_this_file}\n");
423 $stats->{'num_not_processed'} ++;
424 } else {
425 # was not recognised
426 if ($verbosity >= 2) {
427 gsprintf($outhandle, "{plugin.no_plugin_could_recognise}\n",$file);
428 }
429 # tell the GLI that it was not processed
430 print STDERR "<NonRecognisedFile n='$glifile'>\n" if $gli;
431
432 gsprintf($failhandle, "$file: {plugin.no_plugin_could_recognise_this_file}\n");
433 $stats->{'num_not_recognised'} ++;
434 }
435 return 0;
436}
437
438# write out some general stats that the plugins have compiled - note that
439# the buildcol.pl process doesn't currently call this process so the stats
440# are only output after import.pl -
441sub write_stats {
442 my ($pluginfo, $statshandle, $faillog, $gli) = @_;
443
444 $gli = 0 unless defined $gli;
445
446 foreach my $plugobj (@$pluginfo) {
447 $plugobj->compile_stats($stats);
448 }
449
450 my $total = $stats->{'num_processed'} + $stats->{'num_blocked'} +
451 $stats->{'num_not_processed'} + $stats->{'num_not_recognised'};
452
453 print STDERR "<ImportComplete considered='$total' processed='$stats->{'num_processed'}' blocked='$stats->{'num_blocked'}' ignored='$stats->{'num_not_recognised'}' failed='$stats->{'num_not_processed'}'>\n" if $gli;
454
455 if ($total == 1) {
456 gsprintf($statshandle, "* {plugin.one_considered}\n");
457 } else {
458 gsprintf($statshandle, "* {plugin.n_considered}\n", $total);
459 }
460 if ($stats->{'num_archives'}) {
461 if ($stats->{'num_archives'} == 1) {
462 gsprintf($statshandle, " ({plugin.including_archive})\n");
463 }
464 else {
465 gsprintf($statshandle, " ({plugin.including_archives})\n",
466 $stats->{'num_archives'});
467 }
468 }
469 if ($stats->{'num_processed'} == 1) {
470 gsprintf($statshandle, "* {plugin.one_included}\n");
471 } else {
472 gsprintf($statshandle, "* {plugin.n_included}\n", $stats->{'num_processed'});
473 }
474 if ($stats->{'num_not_recognised'}) {
475 if ($stats->{'num_not_recognised'} == 1) {
476 gsprintf($statshandle, "* {plugin.one_unrecognised}\n");
477 } else {
478 gsprintf($statshandle, "* {plugin.n_unrecognised}\n",
479 $stats->{'num_not_recognised'});
480 }
481
482 }
483 if ($stats->{'num_not_processed'}) {
484 if ($stats->{'num_not_processed'} == 1) {
485 gsprintf($statshandle, "* {plugin.one_rejected}\n");
486 } else {
487 gsprintf($statshandle, "* {plugin.n_rejected}\n",
488 $stats->{'num_not_processed'});
489 }
490 }
491 if ($stats->{'num_not_processed'} || $stats->{'num_not_recognised'}) {
492 gsprintf($statshandle, " {plugin.see_faillog}\n", $faillog);
493 }
494}
495
496sub end {
497 my ($pluginfo, $processor) = @_;
498 map { $_->end($processor); } @$pluginfo;
499}
500
501sub deinit {
502 my ($pluginfo, $processor) = @_;
503
504
505 map { $_->deinit($processor); } @$pluginfo;
506}
507
5081;
Note: See TracBrowser for help on using the repository browser.