source: main/trunk/greenstone2/bin/script/buildcol.pl@ 22743

Last change on this file since 22743 was 22743, checked in by mdewsnip, 11 years ago

Consistency changes to copyright statement.

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 19.5 KB
Line 
1#!/usr/bin/perl -w
2
3###########################################################################
4#
5# buildcol.pl --
6# A component of the Greenstone digital library software
7# from the New Zealand Digital Library Project at the
8# University of Waikato, New Zealand.
9#
10# Copyright (C) 1999 New Zealand Digital Library Project
11#
12# This program is free software; you can redistribute it and/or modify
13# it under the terms of the GNU General Public License as published by
14# the Free Software Foundation; either version 2 of the License, or
15# (at your option) any later version.
16#
17# This program is distributed in the hope that it will be useful,
18# but WITHOUT ANY WARRANTY; without even the implied warranty of
19# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20# GNU General Public License for more details.
21#
22# You should have received a copy of the GNU General Public License
23# along with this program; if not, write to the Free Software
24# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
25#
26###########################################################################
27
28
29# This program will build a particular collection.
30
31package buildcol;
32
33BEGIN {
34 die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
35 die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
36 unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
37 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
38 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan/XML/XPath");
39 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/plugins");
40 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/classify");
41
42 if (defined $ENV{'GSDLEXTS'}) {
43 my @extensions = split(/:/,$ENV{'GSDLEXTS'});
44 foreach my $e (@extensions) {
45 my $ext_prefix = "$ENV{'GSDLHOME'}/ext/$e";
46
47 unshift (@INC, "$ext_prefix/perllib");
48 unshift (@INC, "$ext_prefix/perllib/cpan");
49 unshift (@INC, "$ext_prefix/perllib/plugins");
50 unshift (@INC, "$ext_prefix/perllib/classify");
51 }
52 }
53 if (defined $ENV{'GSDL3EXTS'}) {
54 my @extensions = split(/:/,$ENV{'GSDL3EXTS'});
55 foreach my $e (@extensions) {
56 my $ext_prefix = "$ENV{'GSDL3SRCHOME'}/ext/$e";
57
58 unshift (@INC, "$ext_prefix/perllib");
59 unshift (@INC, "$ext_prefix/perllib/cpan");
60 unshift (@INC, "$ext_prefix/perllib/plugins");
61 unshift (@INC, "$ext_prefix/perllib/classify");
62 }
63 }
64
65}
66
67use colcfg;
68use dbutil;
69use util;
70use scriptutil;
71use FileHandle;
72use gsprintf;
73use printusage;
74use parse2;
75
76use strict;
77no strict 'refs'; # allow filehandles to be variables and vice versa
78no strict 'subs'; # allow barewords (eg STDERR) as function arguments
79
80
81my $mode_list =
82 [ { 'name' => "all",
83 'desc' => "{buildcol.mode.all}" },
84 { 'name' => "compress_text",
85 'desc' => "{buildcol.mode.compress_text}" },
86 { 'name' => "build_index",
87 'desc' => "{buildcol.mode.build_index}" },
88 { 'name' => "infodb",
89 'desc' => "{buildcol.mode.infodb}" } ];
90
91my $sec_index_list =
92 [ {'name' => "never",
93 'desc' => "{buildcol.sections_index_document_metadata.never}" },
94 {'name' => "always",
95 'desc' => "{buildcol.sections_index_document_metadata.always}" },
96 {'name' => "unless_section_metadata_exists",
97 'desc' => "{buildcol.sections_index_document_metadata.unless_section_metadata_exists}" }
98 ];
99
100my $arguments =
101 [ { 'name' => "remove_empty_classifications",
102 'desc' => "{buildcol.remove_empty_classifications}",
103 'type' => "flag",
104 'reqd' => "no",
105 'modegli' => "2" },
106 { 'name' => "archivedir",
107 'desc' => "{buildcol.archivedir}",
108 'type' => "string",
109 'reqd' => "no",
110 'hiddengli' => "yes" },
111 { 'name' => "builddir",
112 'desc' => "{buildcol.builddir}",
113 'type' => "string",
114 'reqd' => "no",
115 'hiddengli' => "yes" },
116# { 'name' => "cachedir",
117# 'desc' => "{buildcol.cachedir}",
118# 'type' => "string",
119# 'reqd' => "no" },
120 { 'name' => "collectdir",
121 'desc' => "{buildcol.collectdir}",
122 'type' => "string",
123 # parsearg left "" as default
124 #'deft' => &util::filename_cat ($ENV{'GSDLHOME'}, "collect"),
125 'reqd' => "no",
126 'hiddengli' => "yes" },
127 { 'name' => "site",
128 'desc' => "{buildcol.site}",
129 'type' => "string",
130 'deft' => "",
131 'reqd' => "no",
132 'hiddengli' => "yes" },
133 { 'name' => "debug",
134 'desc' => "{buildcol.debug}",
135 'type' => "flag",
136 'reqd' => "no",
137 'hiddengli' => "yes" },
138 { 'name' => "faillog",
139 'desc' => "{buildcol.faillog}",
140 'type' => "string",
141 # parsearg left "" as default
142 #'deft' => &util::filename_cat("<collectdir>", "colname", "etc", "fail.log"),
143 'reqd' => "no",
144 'modegli' => "3" },
145 { 'name' => "index",
146 'desc' => "{buildcol.index}",
147 'type' => "string",
148 'reqd' => "no",
149 'modegli' => "3" },
150 { 'name' => "incremental",
151 'desc' => "{buildcol.incremental}",
152 'type' => "flag",
153 'hiddengli' => "yes" },
154 { 'name' => "keepold",
155 'desc' => "{buildcol.keepold}",
156 'type' => "flag",
157 'reqd' => "no",
158 #'modegli' => "3",
159 'hiddengli' => "yes" },
160 { 'name' => "removeold",
161 'desc' => "{buildcol.removeold}",
162 'type' => "flag",
163 'reqd' => "no",
164 #'modegli' => "3",
165 'hiddengli' => "yes" },
166 { 'name' => "language",
167 'desc' => "{scripts.language}",
168 'type' => "string",
169 'reqd' => "no",
170 'modegli' => "3" },
171 { 'name' => "maxdocs",
172 'desc' => "{buildcol.maxdocs}",
173 'type' => "int",
174 'reqd' => "no",
175 'hiddengli' => "yes" },
176 { 'name' => "maxnumeric",
177 'desc' => "{buildcol.maxnumeric}",
178 'type' => "int",
179 'reqd' => "no",
180 'deft' => "4",
181 'range' => "4,512",
182 'modegli' => "3" },
183 { 'name' => "mode",
184 'desc' => "{buildcol.mode}",
185 'type' => "enum",
186 'list' => $mode_list,
187 # parsearg left "" as default
188# 'deft' => "all",
189 'reqd' => "no",
190 'modegli' => "3" },
191 { 'name' => "no_strip_html",
192 'desc' => "{buildcol.no_strip_html}",
193 'type' => "flag",
194 'reqd' => "no",
195 'modegli' => "3" },
196 { 'name' => "no_text",
197 'desc' => "{buildcol.no_text}",
198 'type' => "flag",
199 'reqd' => "no",
200 'modegli' => "2" },
201 { 'name' => "sections_index_document_metadata",
202 'desc' => "{buildcol.sections_index_document_metadata}",
203 'type' => "enum",
204 'list' => $sec_index_list,
205 'reqd' => "no",
206 'modegli' => "2" },
207 { 'name' => "out",
208 'desc' => "{buildcol.out}",
209 'type' => "string",
210 'deft' => "STDERR",
211 'reqd' => "no",
212 'hiddengli' => "yes" },
213 { 'name' => "verbosity",
214 'desc' => "{buildcol.verbosity}",
215 'type' => "int",
216 # parsearg left "" as default
217 #'deft' => "2",
218 'reqd' => "no",
219 'modegli' => "3" },
220 { 'name' => "gli",
221 'desc' => "",
222 'type' => "flag",
223 'reqd' => "no",
224 'hiddengli' => "yes" },
225 { 'name' => "xml",
226 'desc' => "{scripts.xml}",
227 'type' => "flag",
228 'reqd' => "no",
229 'hiddengli' => "yes" },
230 ];
231
232my $options = { 'name' => "buildcol.pl",
233 'desc' => "{buildcol.desc}",
234 'args' => $arguments };
235
236
237# globals
238my $collection;
239my $configfilename;
240my $out;
241
242# used to signify "gs2"(default) or "gs3"
243my $gs_mode = "gs2";
244
245## @method gsprintf()
246# Print a string to the screen after looking it up from a locale dependant
247# strings file. This function is losely based on the idea of resource
248# bundles as used in Java.
249#
250# @param $error The STDERR stream.
251# @param $text The string containing GS keys that should be replaced with
252# their locale dependant equivilents.
253# @param $out The output stream.
254# @return The locale-based string to output.
255#
256sub gsprintf()
257{
258 return &gsprintf::gsprintf(@_);
259}
260## gsprintf() ##
261
262&main();
263
264## @method main()
265#
266# [Parses up and validates the arguments to the build process before creating
267# the appropriate build process to do the actual work - John]
268#
269# @note Added true incremental support - John Thompson, DL Consulting Ltd.
270# @note There were several bugs regarding using directories other than
271# "import" or "archives" during import and build quashed. - John
272# Thompson, DL Consulting Ltd.
273#
274# @param $incremental If true indicates this build should not regenerate all
275# the index and metadata files, and should instead just
276# append the information found in the archives directory
277# to the existing files. If this requires some complex
278# work so as to correctly insert into a classifier so be
279# it. Of course none of this is done here - instead the
280# incremental argument is passed to the document
281# processor.
282#
283sub main
284{
285 # command line args
286 my ($verbosity, $archivedir, $cachedir, $builddir, $site, $maxdocs,
287 $debug, $mode, $indexname, $removeold, $keepold,
288 $incremental, $incremental_mode,
289 $remove_empty_classifications,
290 $collectdir, $build, $type, $textindex,
291 $no_strip_html, $no_text, $faillog, $gli, $index, $language,
292 $sections_index_document_metadata, $maxnumeric);
293
294 my $xml = 0;
295 my $hashParsingResult = {};
296 # general options available to all plugins
297 my $intArgLeftinAfterParsing = parse2::parse(\@ARGV,$arguments,$hashParsingResult,"allow_extra_options");
298
299 # If parse returns -1 then something has gone wrong
300 if ($intArgLeftinAfterParsing == -1)
301 {
302 &PrintUsage::print_txt_usage($options, "{buildcol.params}");
303 die "\n";
304 }
305
306 foreach my $strVariable (keys %$hashParsingResult)
307 {
308 eval "\$$strVariable = \$hashParsingResult->{\"\$strVariable\"}";
309 }
310
311 # If $language has been specified, load the appropriate resource bundle
312 # (Otherwise, the default resource bundle will be loaded automatically)
313 if ($language && $language =~ /\S/) {
314 &gsprintf::load_language_specific_resource_bundle($language);
315 }
316
317 if ($xml) {
318 &PrintUsage::print_xml_usage($options);
319 print "\n";
320 return;
321 }
322
323 if ($gli) { # the gli wants strings to be in UTF-8
324 &gsprintf::output_strings_in_UTF8;
325 }
326
327 # now check that we had exactly one leftover arg, which should be
328 # the collection name. We don't want to do this earlier, cos
329 # -xml arg doesn't need a collection name
330 # Or if the user specified -h, then we output the usage also
331 if ($intArgLeftinAfterParsing != 1 || (@ARGV && $ARGV[0] =~ /^\-+h/))
332 {
333 &PrintUsage::print_txt_usage($options, "{buildcol.params}");
334 die "\n";
335 }
336
337 $textindex = "";
338 my $close_out = 0;
339 if ($out !~ /^(STDERR|STDOUT)$/i) {
340 open (OUT, ">$out") ||
341 (&gsprintf(STDERR, "{common.cannot_open_output_file}\n", $out) && die);
342 $out = "buildcol::OUT";
343 $close_out = 1;
344 }
345 $out->autoflush(1);
346
347 # get and check the collection
348 if (($collection = &colcfg::use_collection($site, @ARGV, $collectdir)) eq "") {
349 &PrintUsage::print_txt_usage($options, "{buildcol.params}");
350 die "\n";
351 }
352
353 if ($faillog eq "") {
354 $faillog = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
355 }
356 # note that we're appending to the faillog here (import.pl clears it each time)
357 # this could potentially create a situation where the faillog keeps being added
358 # to over multiple builds (if the import process is being skipped)
359 open (FAILLOG, ">>$faillog") ||
360 (&gsprintf(STDERR, "{common.cannot_open_fail_log}\n", $faillog) && die);
361 $faillog = 'buildcol::FAILLOG';
362 $faillog->autoflush(1);
363
364 unshift (@INC, "$ENV{'GSDLCOLLECTDIR'}/perllib");
365 # Don't know why this didn't already happen, but now collection specific
366 # classify and plugins directory also added to include path
367 unshift (@INC, "$ENV{'GSDLCOLLECTDIR'}/perllib/classify");
368 unshift (@INC, "$ENV{'GSDLCOLLECTDIR'}/perllib/plugins");
369
370 # Read in the collection configuration file.
371 my ($collectcfg, $buildtype);
372 ($configfilename, $gs_mode) = &colcfg::get_collect_cfg_name($out);
373 $collectcfg = &colcfg::read_collection_cfg ($configfilename, $gs_mode);
374
375 # If the infodbtype value wasn't defined in the collect.cfg file, use the default
376 if (!defined($collectcfg->{'infodbtype'}))
377 {
378 $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
379 }
380
381 if ($verbosity !~ /\d+/) {
382 if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/) {
383 $verbosity = $collectcfg->{'verbosity'};
384 } else {
385 $verbosity = 2; # the default
386 }
387 }
388 # we use searchtype for determining buildtype, but for old versions, use buildtype
389 if (defined $collectcfg->{'buildtype'}) {
390 $buildtype = $collectcfg->{'buildtype'};
391 } elsif (defined $collectcfg->{'searchtypes'} || defined $collectcfg->{'searchtype'}) {
392 $buildtype = "mgpp";
393 } else {
394 $buildtype = "mg"; #mg is the default
395 }
396 if (defined $collectcfg->{'archivedir'} && $archivedir eq "") {
397 $archivedir = $collectcfg->{'archivedir'};
398 }
399 if (defined $collectcfg->{'cachedir'} && $cachedir eq "") {
400 $cachedir = $collectcfg->{'cachedir'};
401 }
402 if (defined $collectcfg->{'builddir'} && $builddir eq "") {
403 $builddir = $collectcfg->{'builddir'};
404 }
405 if ($maxdocs !~ /\-?\d+/) {
406 if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/) {
407 $maxdocs = $collectcfg->{'maxdocs'};
408 } else {
409 $maxdocs = -1; # the default
410 }
411 }
412 if (defined $collectcfg->{'maxnumeric'} && $collectcfg->{'maxnumeric'} =~ /\d+/) {
413 $maxnumeric = $collectcfg->{'maxnumeric'};
414 }
415
416 if ($maxnumeric < 4 || $maxnumeric > 512) {
417 $maxnumeric = 4;
418 }
419
420 if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
421 $debug = 1;
422 }
423 if ($mode !~ /^(all|compress_text|build_index|infodb)$/) {
424 if (defined $collectcfg->{'mode'} && $collectcfg->{'mode'} =~ /^(all|compress_text|build_index|infodb)$/) {
425 $mode = $collectcfg->{'mode'};
426 } else {
427 $mode = "all"; # the default
428 }
429 }
430 if (defined $collectcfg->{'index'} && $indexname eq "") {
431 $indexname = $collectcfg->{'index'};
432 }
433 if (defined $collectcfg->{'no_text'} && $no_text == 0) {
434 if ($collectcfg->{'no_text'} =~ /^true$/i) {
435 $no_text = 1;
436 }
437 }
438 if (defined $collectcfg->{'no_strip_html'} && $no_strip_html == 0) {
439 if ($collectcfg->{'no_strip_html'} =~ /^true$/i) {
440 $no_strip_html = 1;
441 }
442 }
443 if (defined $collectcfg->{'remove_empty_classifications'} && $remove_empty_classifications == 0) {
444 if ($collectcfg->{'remove_empty_classifications'} =~ /^true$/i) {
445 $remove_empty_classifications = 1;
446 }
447 }
448
449 if ($buildtype eq "mgpp" && defined $collectcfg->{'textcompress'}) {
450 $textindex = $collectcfg->{'textcompress'};
451 }
452 if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i) {
453 $gli = 1;
454 }
455
456 if ($sections_index_document_metadata !~ /\S/ && defined $collectcfg->{'sections_index_document_metadata'}) {
457 $sections_index_document_metadata = $collectcfg->{'sections_index_document_metadata'};
458 }
459
460 if ($sections_index_document_metadata !~ /^(never|always|unless_section_metadata_exists)$/) {
461 $sections_index_document_metadata = "never";
462 }
463
464 ($removeold, $keepold, $incremental, $incremental_mode)
465 = &scriptutil::check_removeold_and_keepold($removeold, $keepold,
466 $incremental, "building",
467 $collectcfg);
468
469 $gli = 0 unless defined $gli;
470
471 # New argument to track whether build is incremental
472 $incremental = 0 unless defined $incremental;
473
474 print STDERR "<Build>\n" if $gli;
475
476 #set the text index
477 if (($buildtype eq "mgpp") || ($buildtype eq "lucene")) {
478 if ($textindex eq "") {
479 $textindex = "text";
480 }
481 }
482 else {
483 $textindex = "section:text";
484 }
485
486 # fill in the default archives and building directories if none
487 # were supplied, turn all \ into / and remove trailing /
488
489 my ($realarchivedir, $realbuilddir);
490 # Modified so that the archivedir, if provided as an argument, is made
491 # absolute if it isn't already
492 if ($archivedir eq "")
493 {
494 $archivedir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "archives");
495 }
496 else
497 {
498 $archivedir = &util::make_absolute($ENV{'GSDLCOLLECTDIR'}, $archivedir);
499 }
500 # End Mod
501 $archivedir =~ s/[\\\/]+/\//g;
502 $archivedir =~ s/\/$//;
503
504 if ($builddir eq "") {
505 $builddir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "building");
506 if ($incremental) {
507 &gsprintf($out, "{buildcol.incremental_default_builddir}\n");
508 }
509 }
510 $builddir =~ s/[\\\/]+/\//g;
511 $builddir =~ s/\/$//;
512
513 # update the archive cache if needed
514 if ($cachedir) {
515 &gsprintf($out, "{buildcol.updating_archive_cache}\n")
516 if ($verbosity >= 1);
517
518 $cachedir =~ s/[\\\/]+$//;
519 $cachedir .= "/collect/$collection" unless
520 $cachedir =~ /collect\/$collection/;
521
522 $realarchivedir = "$cachedir/archives";
523 $realbuilddir = "$cachedir/building";
524 &util::mk_all_dir ($realarchivedir);
525 &util::mk_all_dir ($realbuilddir);
526 &util::cachedir ($archivedir, $realarchivedir, $verbosity);
527
528 } else {
529 $realarchivedir = $archivedir;
530 $realbuilddir = $builddir;
531 }
532
533 # build it in realbuilddir
534 &util::mk_all_dir ($realbuilddir);
535
536 my ($buildertype, $builderdir, $builder);
537 # if a builder class has been created for this collection, use it
538 # otherwise, use the mg or mgpp builder
539 if (-e "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib/custombuilder.pm") {
540 $builderdir = "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib";
541 $buildertype = "custombuilder";
542 } elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/custombuilder.pm") {
543 $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
544 $buildertype = "custombuilder";
545 } elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/${collection}builder.pm") {
546 $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
547 $buildertype = "${collection}builder";
548 } else {
549 $builderdir = "$ENV{'GSDLHOME'}/perllib";
550 if ($buildtype eq "lucene") {
551 $buildertype = "lucenebuilder";
552 }
553 elsif ($buildtype eq "mgpp") {
554 $buildertype = "mgppbuilder";
555 }
556 else {
557 $buildertype = "mgbuilder";
558 }
559 }
560
561 require "$builderdir/$buildertype.pm";
562
563 eval("\$builder = new $buildertype(\$collection, " .
564 "\$realarchivedir, \$realbuilddir, \$verbosity, " .
565 "\$maxdocs, \$debug, \$keepold, \$incremental, \$incremental_mode, " .
566 "\$remove_empty_classifications, " .
567 "\$out, \$no_text, \$faillog, \$gli)");
568 die "$@" if $@;
569
570 $builder->init();
571 $builder->set_maxnumeric($maxnumeric);
572
573 if (($buildertype eq "mgppbuilder") && $no_strip_html) {
574 $builder->set_strip_html(0);
575 }
576 if ($sections_index_document_metadata ne "never") {
577 $builder->set_sections_index_document_metadata($sections_index_document_metadata);
578 }
579
580 if ($mode =~ /^all$/i) {
581 $builder->compress_text($textindex);
582 $builder->build_indexes($indexname);
583 $builder->make_infodatabase();
584 $builder->collect_specific();
585 } elsif ($mode =~ /^compress_text$/i) {
586 $builder->compress_text($textindex);
587 } elsif ($mode =~ /^build_index$/i) {
588 $builder->build_indexes($indexname);
589 } elsif ($mode =~ /^infodb$/i) {
590 $builder->make_infodatabase();
591 } else {
592 (&gsprintf(STDERR, "{buildcol.unknown_mode}\n", $mode) && die);
593 }
594
595 $builder->make_auxiliary_files() if !$debug;
596 $builder->deinit();
597
598 if (($realbuilddir ne $builddir) && !$debug) {
599 &gsprintf($out, "{buildcol.copying_back_cached_build}\n")
600 if ($verbosity >= 1);
601 &util::rm_r ($builddir);
602 &util::cp_r ($realbuilddir, $builddir);
603 }
604
605 close OUT if $close_out;
606 close FAILLOG;
607
608 print STDERR "</Build>\n" if $gli;
609}
610## main() ##
611
612
Note: See TracBrowser for help on using the repository browser.