source: main/trunk/greenstone2/perllib/plugin.pm@ 23170

Last change on this file since 23170 was 23170, checked in by kjdon, 14 years ago

if infodbtype is gdbm-txtgz, we need to use gdbm for all archives dbs
  • Property svn:keywords set to Author Date Id Revision
File size: 13.9 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 gsprintf 'gsprintf';
35
36# mapping from old plugin names to new ones for backwards compatibility
37# can remove at sometime in future when we no longer want to support old xxPlug names in the config file
38my $plugin_name_map = {
39 'GAPlug' => 'GreenstoneXMLPlugin',
40 'ArcPlug' => 'ArchivesInfPlugin',
41 'RecPlug' => 'DirectoryPlugin',
42 'TEXTPlug' => 'TextPlugin',
43 'XMLPlug' => 'ReadXMLFile',
44 'EMAILPlug' => 'EmailPlugin',
45 'SRCPlug' => 'SourceCodePlugin',
46 'NULPlug' => 'NulPlugin',
47 'W3ImgPlug' => 'HTMLImagePlugin',
48 'PagedImgPlug' => 'PagedImagePlugin',
49 'METSPlug' => 'GreenstoneMETSPlugin',
50 'PPTPlug' => 'PowerPointPlugin',
51 'PSPlug' => 'PostScriptPlugin',
52 'DBPlug' => 'DatabasePlugin'
53 };
54
55# global variables
56my $stats = {'num_processed' => 0,
57 'num_blocked' => 0,
58 'num_not_processed' => 0,
59 'num_not_recognised' => 0,
60 'num_archives' => 0
61 };
62
63#globaloptions contains any options that should be passed to all plugins
64my ($verbosity, $outhandle, $failhandle, $globaloptions);
65
66sub get_valid_pluginname {
67 my ($pluginname) = @_;
68 my $valid_name = $pluginname;
69 if (defined $plugin_name_map->{$pluginname}) {
70 $valid_name = $plugin_name_map->{$pluginname};
71 } elsif ($pluginname =~ /Plug$/) {
72 $valid_name =~ s/Plug/Plugin/;
73
74 }
75 return $valid_name;
76}
77
78sub load_plugin_require
79{
80 my ($pluginname) = @_;
81
82 my @check_list = ();
83
84 # pp_plugname shorthand for 'perllib' 'plugin' '$pluginname.pm'
85 my $pp_plugname
86 = &util::filename_cat('perllib', 'plugins', "${pluginname}.pm");
87 my $collectdir = $ENV{'GSDLCOLLECTDIR'};
88
89 # find the plugin
90 if (defined($ENV{'GSDLCOLLECTION'}))
91 {
92 my $customplugname
93 = &util::filename_cat($collectdir, "custom",$ENV{'GSDLCOLLECTION'},
94 $pp_plugname);
95 push(@check_list,$customplugname);
96 }
97
98 my $colplugname = &util::filename_cat($collectdir, $pp_plugname);
99 push(@check_list,$colplugname);
100
101 if (defined $ENV{'GSDLEXTS'}) {
102
103 my $ext_prefix = &util::filename_cat($ENV{'GSDLHOME'}, "ext");
104
105 my @extensions = split(/:/,$ENV{'GSDLEXTS'});
106 foreach my $e (@extensions) {
107 my $extplugname = &util::filename_cat($ext_prefix, $e, $pp_plugname);
108 push(@check_list,$extplugname);
109
110 }
111 }
112 if (defined $ENV{'GSDL3EXTS'}) {
113
114 my $ext_prefix = &util::filename_cat($ENV{'GSDL3SRCHOME'}, "ext");
115
116 my @extensions = split(/:/,$ENV{'GSDL3EXTS'});
117 foreach my $e (@extensions) {
118 my $extplugname = &util::filename_cat($ext_prefix, $e, $pp_plugname);
119 push(@check_list,$extplugname);
120
121 }
122 }
123
124
125 my $mainplugname = &util::filename_cat($ENV{'GSDLHOME'}, $pp_plugname);
126 push(@check_list,$mainplugname);
127
128 my $success=0;
129 foreach my $plugname (@check_list) {
130 if (-e $plugname) {
131 require $plugname;
132 $success=1;
133 last;
134 }
135 }
136
137 if (!$success) {
138 &gsprintf(STDERR, "{plugin.could_not_find_plugin}\n",
139 $pluginname);
140 die "\n";
141 }
142}
143
144sub load_plugin_for_info {
145 my ($pluginname) = shift @_;
146 $pluginname = &get_valid_pluginname($pluginname);
147 load_plugin_require($pluginname);
148
149 # create a plugin object
150 my ($plugobj);
151 my $options = "-gsdlinfo";
152
153 eval ("\$plugobj = new \$pluginname([],[$options])");
154 die "$@" if $@;
155
156 return $plugobj;
157}
158
159sub load_plugins {
160 my ($plugin_list) = shift @_;
161 my $incremental_mode;
162 ($verbosity, $outhandle, $failhandle, $globaloptions, $incremental_mode) = @_; # globals
163 my @plugin_objects = ();
164 $verbosity = 2 unless defined $verbosity;
165 $outhandle = 'STDERR' unless defined $outhandle;
166 $failhandle = 'STDERR' unless defined $failhandle;
167
168 my $colplugindir = &util::filename_cat($ENV{'GSDLCOLLECTDIR'},"perllib/plugins");
169 unshift (@INC, $colplugindir);
170
171 map { $_ = "\"$_\""; } @$globaloptions;
172 my $globals = join (",", @$globaloptions);
173
174 foreach my $pluginoptions (@$plugin_list) {
175 my $pluginname = shift @$pluginoptions;
176 next unless defined $pluginname;
177 $pluginname = &get_valid_pluginname($pluginname);
178 load_plugin_require($pluginname);
179
180 # create a plugin object
181 my ($plugobj);
182 map { $_ = "\"$_\""; } @$pluginoptions;
183 my $options = join (",", @$pluginoptions);
184 if ($globals) {
185 if (@$pluginoptions) {
186 $options .= ",";
187 }
188 $options .= "$globals";
189 }
190 # need to escape backslash before putting in to the eval
191 # but watch out for any \" (which shouldn't be further escaped)
192 $options =~ s/\\([^"])/\\\\$1/g; #"
193 $options =~ s/\$/\\\$/g;
194
195 eval ("\$plugobj = new \$pluginname([],[$options])");
196 die "$@" if $@;
197
198 # initialize plugin
199 $plugobj->init($verbosity, $outhandle, $failhandle);
200
201 $plugobj->set_incremental($incremental_mode);
202
203 # add this object to the list
204 push (@plugin_objects, $plugobj);
205 }
206
207 return \@plugin_objects;
208}
209
210
211sub begin {
212 my ($pluginfo, $base_dir, $processor, $maxdocs, $gli) = @_;
213
214 map { $_->{'gli'} = $gli; } @$pluginfo;
215 map { $_->begin($pluginfo, $base_dir, $processor, $maxdocs); } @$pluginfo;
216}
217
218 sub remove_all {
219 my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
220
221 map { $_->remove_all($pluginfo, $base_dir, $processor, $maxdocs); } @$pluginfo;
222}
223
224sub remove_some {
225 my ($pluginfo, $infodbtype, $archivedir, $deleted_files) = @_;
226 return if (scalar(@$deleted_files)==0);
227 $infodbtype = "gdbm" if $infodbtype eq "gdbm-txtgz";
228 my $arcinfo_src_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-src", $archivedir);
229
230 foreach my $file (@$deleted_files) {
231 # use 'archiveinf-src' info database to look up all the OIDs
232 # that this file is used in (note in most cases, it's just one OID)
233
234 my $src_rec_string = &dbutil::read_infodb_entry($infodbtype, $arcinfo_src_filename, $file);
235 my $src_rec = &dbutil::convert_infodb_string_to_hash($src_rec_string);
236 my $oids = $src_rec->{'oid'};
237 my $rv;
238 foreach my $plugobj (@$pluginfo) {
239
240 $rv = $plugobj->remove_one($file, $oids, $archivedir);
241 if (defined $rv && $rv != -1) {
242 return $rv;
243 } # else undefined (was not recognised by the plugin) or there was an error, try the next one
244 }
245 return 0;
246 }
247
248}
249sub file_block_read {
250 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
251
252
253 $gli = 0 unless defined $gli;
254
255 my $rv = 0;
256 my $glifile = $file;
257
258 $glifile =~ s/^[\/\\]+//; # file sometimes starts with a / so get rid of it
259
260 # Announce to GLI that we are handling a file
261 print STDERR "<File n='$glifile'>\n" if $gli;
262
263 # the .kill file is a handy (if not very elegant) way of aborting
264 # an import.pl or buildcol.pl process
265 if (-e &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, ".kill")) {
266 gsprintf($outhandle, "{plugin.kill_file}\n");
267 die "\n";
268 }
269
270 foreach my $plugobj (@$pluginfo) {
271
272 $rv = $plugobj->file_block_read($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli);
273 #last if (defined $rv && $rv==1); # stop this file once we have found something to 'process' it
274 }
275
276}
277
278
279sub metadata_read {
280 my ($pluginfo, $base_dir, $file, $block_hash,
281 $extrametakeys, $extrametadata, $extrametafile,
282 $processor, $maxdocs, $gli, $aux) = @_;
283
284 $maxdocs = -1 unless defined $maxdocs && $maxdocs =~ /\d/;
285 $gli = 0 unless defined $gli;
286
287 my $rv = 0;
288 my $glifile = $file;
289
290 $glifile =~ s/^[\/\\]+//; # file sometimes starts with a / so get rid of it
291
292 # Announce to GLI that we are handling a file
293 print STDERR "<File n='$glifile'>\n" if $gli;
294
295 # the .kill file is a handy (if not very elegant) way of aborting
296 # an import.pl or buildcol.pl process
297 if (-e &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, ".kill")) {
298 gsprintf($outhandle, "{plugin.kill_file}\n");
299 die "\n";
300 }
301
302 my $had_error = 0;
303 # pass this file by each of the plugins in turn until one
304 # is found which will process it
305 # read must return:
306 # undef - could not recognise
307 # -1 - tried but error
308 # 0 - blocked
309 # anything else for successful processing
310
311 foreach my $plugobj (@$pluginfo) {
312
313 $rv = $plugobj->metadata_read($pluginfo, $base_dir, $file, $block_hash,
314 $extrametakeys, $extrametadata, $extrametafile,
315 $processor, $maxdocs, $gli, $aux);
316
317 if (defined $rv) {
318 if ($rv == -1) {
319 # an error has occurred
320 $had_error = 1;
321 print STDERR "<ProcessingError n='$glifile'>\n" if $gli;
322 } else {
323 return $rv;
324 }
325 } # else undefined - was not recognised by the plugin
326 }
327
328 return 0;
329}
330
331sub read {
332 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli, $aux) = @_;
333
334 $maxdocs = -1 unless defined $maxdocs && $maxdocs =~ /\d/;
335 $total_count = 0 unless defined $total_count && $total_count =~ /\d/;
336 $gli = 0 unless defined $gli;
337
338 my $rv = 0;
339 my $glifile = $file;
340
341 $glifile =~ s/^[\/\\]+//; # file sometimes starts with a / so get rid of it
342
343 # Announce to GLI that we are handling a file
344 print STDERR "<File n='$glifile'>\n" if $gli;
345
346 # the .kill file is a handy (if not very elegant) way of aborting
347 # an import.pl or buildcol.pl process
348 if (-e &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, ".kill")) {
349 gsprintf($outhandle, "{plugin.kill_file}\n");
350 die "\n";
351 }
352
353 my $had_error = 0;
354 # pass this file by each of the plugins in turn until one
355 # is found which will process it
356 # read must return:
357 # undef - could not recognise
358 # -1 - tried but error
359 # 0 - blocked
360 # anything else for successful processing
361
362 foreach my $plugobj (@$pluginfo) {
363
364 $rv = $plugobj->read($pluginfo, $base_dir, $file,
365 $block_hash, $metadata, $processor, $maxdocs,
366 $total_count, $gli, $aux);
367
368 if (defined $rv) {
369 if ($rv == -1) {
370 # an error has occurred
371 $had_error = 1;
372 } else {
373 return $rv;
374 }
375 } # else undefined - was not recognised by the plugin
376 }
377
378 if ($had_error) {
379 # was recognised but couldn't be processed
380 if ($verbosity >= 2) {
381 gsprintf($outhandle, "{plugin.no_plugin_could_process}\n", $file);
382 }
383 # tell the GLI that it was not processed
384 print STDERR "<NonProcessedFile n='$glifile'>\n" if $gli;
385
386 gsprintf($failhandle, "$file: {plugin.no_plugin_could_process_this_file}\n");
387 $stats->{'num_not_processed'} ++;
388 } else {
389 # was not recognised
390 if ($verbosity >= 2) {
391 gsprintf($outhandle, "{plugin.no_plugin_could_recognise}\n",$file);
392 }
393 # tell the GLI that it was not processed
394 print STDERR "<NonRecognisedFile n='$glifile'>\n" if $gli;
395
396 gsprintf($failhandle, "$file: {plugin.no_plugin_could_recognise_this_file}\n");
397 $stats->{'num_not_recognised'} ++;
398 }
399 return 0;
400}
401
402# write out some general stats that the plugins have compiled - note that
403# the buildcol.pl process doesn't currently call this process so the stats
404# are only output after import.pl -
405sub write_stats {
406 my ($pluginfo, $statshandle, $faillog, $gli) = @_;
407
408 $gli = 0 unless defined $gli;
409
410 foreach my $plugobj (@$pluginfo) {
411 $plugobj->compile_stats($stats);
412 }
413
414 my $total = $stats->{'num_processed'} + $stats->{'num_blocked'} +
415 $stats->{'num_not_processed'} + $stats->{'num_not_recognised'};
416
417 print STDERR "<ImportComplete considered='$total' processed='$stats->{'num_processed'}' blocked='$stats->{'num_blocked'}' ignored='$stats->{'num_not_recognised'}' failed='$stats->{'num_not_processed'}'>\n" if $gli;
418
419 if ($total == 1) {
420 gsprintf($statshandle, "* {plugin.one_considered}\n");
421 } else {
422 gsprintf($statshandle, "* {plugin.n_considered}\n", $total);
423 }
424 if ($stats->{'num_archives'}) {
425 if ($stats->{'num_archives'} == 1) {
426 gsprintf($statshandle, " ({plugin.including_archive})\n");
427 }
428 else {
429 gsprintf($statshandle, " ({plugin.including_archives})\n",
430 $stats->{'num_archives'});
431 }
432 }
433 if ($stats->{'num_processed'} == 1) {
434 gsprintf($statshandle, "* {plugin.one_included}\n");
435 } else {
436 gsprintf($statshandle, "* {plugin.n_included}\n", $stats->{'num_processed'});
437 }
438 if ($stats->{'num_not_recognised'}) {
439 if ($stats->{'num_not_recognised'} == 1) {
440 gsprintf($statshandle, "* {plugin.one_unrecognised}\n");
441 } else {
442 gsprintf($statshandle, "* {plugin.n_unrecognised}\n",
443 $stats->{'num_not_recognised'});
444 }
445
446 }
447 if ($stats->{'num_not_processed'}) {
448 if ($stats->{'num_not_processed'} == 1) {
449 gsprintf($statshandle, "* {plugin.one_rejected}\n");
450 } else {
451 gsprintf($statshandle, "* {plugin.n_rejected}\n",
452 $stats->{'num_not_processed'});
453 }
454 }
455 if ($stats->{'num_not_processed'} || $stats->{'num_not_recognised'}) {
456 gsprintf($statshandle, " {plugin.see_faillog}\n", $faillog);
457 }
458}
459
460sub end {
461 my ($pluginfo, $processor) = @_;
462 map { $_->end($processor); } @$pluginfo;
463}
464
465sub deinit {
466 my ($pluginfo, $processor) = @_;
467
468
469 map { $_->deinit($processor); } @$pluginfo;
470}
471
4721;
Note: See TracBrowser for help on using the repository browser.