1 | ##############################################################################
|
---|
2 | #
|
---|
3 | # buildcolutils.pm -- index and build the collection. The buildtime counterpart
|
---|
4 | # of inexport.pl
|
---|
5 | #
|
---|
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 | package buildcolutils;
|
---|
29 |
|
---|
30 | #use strict;
|
---|
31 | #no strict 'refs';
|
---|
32 |
|
---|
33 | use File::Basename;
|
---|
34 |
|
---|
35 | use colcfg;
|
---|
36 | use dbutil;
|
---|
37 | use util;
|
---|
38 | use FileUtils;
|
---|
39 | use scriptutil;
|
---|
40 | use servercontrol;
|
---|
41 | use gsprintf;
|
---|
42 | use printusage;
|
---|
43 | use parse2;
|
---|
44 |
|
---|
45 | ## @method new()
|
---|
46 | #
|
---|
47 | # Parses up and validates the arguments to the build process before creating
|
---|
48 | # the appropriate build process to do the actual work
|
---|
49 | #
|
---|
50 | # @note Added true incremental support - John Thompson, DL Consulting Ltd.
|
---|
51 | # @note There were several bugs regarding using directories other than
|
---|
52 | # "import" or "archives" during import and build quashed. - John
|
---|
53 | # Thompson, DL Consulting Ltd.
|
---|
54 | #
|
---|
55 | # @param $incremental If true indicates this build should not regenerate all
|
---|
56 | # the index and metadata files, and should instead just
|
---|
57 | # append the information found in the archives directory
|
---|
58 | # to the existing files. If this requires some complex
|
---|
59 | # work so as to correctly insert into a classifier so be
|
---|
60 | # it. Of course none of this is done here - instead the
|
---|
61 | # incremental argument is passed to the document
|
---|
62 | # processor.
|
---|
63 | #
|
---|
64 | sub new
|
---|
65 | {
|
---|
66 | my $class = shift(@_);
|
---|
67 | my ($argv, $options, $opt_listall_options) = @_;
|
---|
68 |
|
---|
69 | my $self = {'builddir' => undef,
|
---|
70 | 'buildtype' => undef,
|
---|
71 | 'close_faillog' => 0,
|
---|
72 | 'close_out' => 0,
|
---|
73 | 'mode' => '',
|
---|
74 | 'orthogonalbuildtypes' => undef,
|
---|
75 | 'realbuilddir' => undef,
|
---|
76 | 'textindex' => '',
|
---|
77 | 'xml' => 0
|
---|
78 | };
|
---|
79 |
|
---|
80 | # general options available to all plugins
|
---|
81 | my $arguments = $options->{'args'};
|
---|
82 | my $intArgLeftinAfterParsing = &parse2::parse($argv, $arguments, $self, "allow_extra_options");
|
---|
83 | # If parse returns -1 then something has gone wrong
|
---|
84 | if ($intArgLeftinAfterParsing == -1)
|
---|
85 | {
|
---|
86 | &PrintUsage::print_txt_usage($options, "{buildcol.params}",1);
|
---|
87 | print STDERR "Something went wrong during parsing the arguments. Scroll up for details.\n";
|
---|
88 | die "\n";
|
---|
89 | }
|
---|
90 |
|
---|
91 | # If $language has been specified, load the appropriate resource bundle
|
---|
92 | # (Otherwise, the default resource bundle will be loaded automatically)
|
---|
93 | if ($self->{'language'} && $self->{'language'} =~ /\S/)
|
---|
94 | {
|
---|
95 | &gsprintf::load_language_specific_resource_bundle($self->{'language'});
|
---|
96 | }
|
---|
97 |
|
---|
98 | # Do we need 'listall' support in buildcol? If so, copy code from inexport
|
---|
99 | # later [jmt12]
|
---|
100 |
|
---|
101 | # <insert explanation here>
|
---|
102 | if ($self->{'xml'})
|
---|
103 | {
|
---|
104 | &PrintUsage::print_xml_usage($options);
|
---|
105 | print "\n";
|
---|
106 | return bless($self, $class);
|
---|
107 | }
|
---|
108 |
|
---|
109 | # the gli wants strings to be in UTF-8
|
---|
110 | if ($gli)
|
---|
111 | {
|
---|
112 | &gsprintf::output_strings_in_UTF8;
|
---|
113 | }
|
---|
114 |
|
---|
115 | # If the user specified -h, then we output the usage
|
---|
116 | if (@$argv && $argv->[0] =~ /^\-+h/) {
|
---|
117 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
---|
118 | die "\n";
|
---|
119 | }
|
---|
120 |
|
---|
121 | # now check that we had exactly one leftover arg, which should be
|
---|
122 | # the collection name. We don't want to do this earlier, cos
|
---|
123 | # -xml arg doesn't need a collection name
|
---|
124 | if ($intArgLeftinAfterParsing != 1)
|
---|
125 | {
|
---|
126 | &PrintUsage::print_txt_usage($options, "{buildcol.params}", 1);
|
---|
127 | print STDERR "There should be one argument left after parsing the script args: the collection name.\n";
|
---|
128 | die "\n";
|
---|
129 | }
|
---|
130 |
|
---|
131 | my $out = $self->{'out'};
|
---|
132 | if ($out !~ /^(STDERR|STDOUT)$/i)
|
---|
133 | {
|
---|
134 | open (OUT, ">$out") || (&gsprintf::gsprintf(STDERR, "{common.cannot_open_output_file}\n", $out) && die);
|
---|
135 | $out = "buildcolutils::OUT";
|
---|
136 | $self->{'close_out'} = 1;
|
---|
137 | }
|
---|
138 | $out->autoflush(1);
|
---|
139 | $self->{'out'} = $out;
|
---|
140 |
|
---|
141 | # @ARGV should be only one item, the name of the collection
|
---|
142 | $self->{'collection'} = shift(@{$argv});
|
---|
143 |
|
---|
144 | return bless($self, $class);
|
---|
145 | }
|
---|
146 | # new()
|
---|
147 |
|
---|
148 | # newCGI()?
|
---|
149 |
|
---|
150 | # @function get_collection
|
---|
151 | #
|
---|
152 | sub get_collection
|
---|
153 | {
|
---|
154 | my $self = shift @_;
|
---|
155 | return $self->{'collection'};
|
---|
156 | }
|
---|
157 | # get_collection()
|
---|
158 |
|
---|
159 | # @function read_collection_cfg
|
---|
160 | #
|
---|
161 | sub read_collection_cfg
|
---|
162 | {
|
---|
163 | my $self = shift(@_);
|
---|
164 | my ($collection, $options) = @_;
|
---|
165 |
|
---|
166 | my $collectdir = $self->{'collectdir'};
|
---|
167 | my $site = $self->{'site'};
|
---|
168 | my $out = $self->{'out'};
|
---|
169 |
|
---|
170 | # get and check the collection
|
---|
171 | if (($collection = &colcfg::use_collection($site, $collection, $collectdir)) eq "")
|
---|
172 | {
|
---|
173 | #&PrintUsage::print_txt_usage($options, "{buildcol.params}", 1);
|
---|
174 | die "\n";
|
---|
175 | }
|
---|
176 |
|
---|
177 | # set gs_version 2/3
|
---|
178 | $self->{'gs_version'} = "2";
|
---|
179 | if ((defined $site) && ($site ne ""))
|
---|
180 | {
|
---|
181 | # gs3
|
---|
182 | $self->{'gs_version'} = "3";
|
---|
183 | }
|
---|
184 |
|
---|
185 | # add collection's perllib dir into include path in case we have collection
|
---|
186 | # specific modules
|
---|
187 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib'));
|
---|
188 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib', 'classify'));
|
---|
189 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib', 'plugins'));
|
---|
190 |
|
---|
191 | # check that we can open the faillog
|
---|
192 | my $faillog = $self->{'faillog'};
|
---|
193 | if ($faillog eq "")
|
---|
194 | {
|
---|
195 | $faillog = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
|
---|
196 | }
|
---|
197 | # note that we're appending to the faillog here (import.pl clears it each time)
|
---|
198 | # this could potentially create a situation where the faillog keeps being added
|
---|
199 | # to over multiple builds (if the import process is being skipped)
|
---|
200 | open (FAILLOG, ">>$faillog") || (&gsprintf::gsprintf(STDERR, "{common.cannot_open_fail_log}\n", $faillog) && die);
|
---|
201 | $faillog = 'buildcolutils::FAILLOG';
|
---|
202 | $faillog->autoflush(1);
|
---|
203 | $self->{'faillog'} = $faillog;
|
---|
204 | $self->{'faillogname'} = $faillog;
|
---|
205 | $self->{'close_faillog'} = 1;
|
---|
206 |
|
---|
207 | # Read in the collection configuration file.
|
---|
208 | my $gs_mode = "gs".$self->{'gs_version'}; #gs2 or gs3
|
---|
209 | my $config_filename = &colcfg::get_collect_cfg_name($out, $gs_mode);
|
---|
210 | my $collect_cfg = &colcfg::read_collection_cfg($config_filename, $gs_mode);
|
---|
211 |
|
---|
212 | return ($config_filename, $collect_cfg);
|
---|
213 | }
|
---|
214 | # read_collection_cfg()
|
---|
215 |
|
---|
216 | # @function set_collection_options
|
---|
217 | # This function copies across values for arguments from the collection
|
---|
218 | # configuration file if they are not already provided by the user, then
|
---|
219 | # sets reasonable defaults for any required arguments that remains without
|
---|
220 | # a value.
|
---|
221 | sub set_collection_options
|
---|
222 | {
|
---|
223 | my $self = shift @_;
|
---|
224 | my ($collectcfg) = @_;
|
---|
225 | my ($buildtype, $orthogonalbuildtypes);
|
---|
226 |
|
---|
227 | # If the infodbtype value wasn't defined in the collect.cfg file, use the default
|
---|
228 | if (!defined($collectcfg->{'infodbtype'}))
|
---|
229 | {
|
---|
230 | $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
|
---|
231 | }
|
---|
232 | # - just so I don't have to pass collectcfg around as well
|
---|
233 | $self->{'infodbtype'} = $collectcfg->{'infodbtype'};
|
---|
234 |
|
---|
235 | if ($self->{'verbosity'} !~ /\d+/)
|
---|
236 | {
|
---|
237 | if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/)
|
---|
238 | {
|
---|
239 | $self->{'verbosity'} = $collectcfg->{'verbosity'};
|
---|
240 | }
|
---|
241 | else
|
---|
242 | {
|
---|
243 | $self->{'verbosity'} = 2; # the default
|
---|
244 | }
|
---|
245 | }
|
---|
246 |
|
---|
247 | # we use searchtype for determining buildtype, but for old versions, use buildtype
|
---|
248 | if (defined $collectcfg->{'buildtype'})
|
---|
249 | {
|
---|
250 | $self->{'buildtype'} = $collectcfg->{'buildtype'};
|
---|
251 | }
|
---|
252 | elsif (defined $collectcfg->{'searchtypes'} || defined $collectcfg->{'searchtype'})
|
---|
253 | {
|
---|
254 | $self->{'buildtype'} = "mgpp";
|
---|
255 | }
|
---|
256 | else
|
---|
257 | {
|
---|
258 | $self->{'buildtype'} = "mg"; #mg is the default
|
---|
259 | }
|
---|
260 |
|
---|
261 | if ($self->{'buildtype'} eq "mgpp" && defined $collectcfg->{'textcompress'})
|
---|
262 | {
|
---|
263 | $self->{'textindex'} = $collectcfg->{'textcompress'};
|
---|
264 | }
|
---|
265 |
|
---|
266 | # is it okay to always clobber or possible remain undefined? [jmt12]
|
---|
267 | if (defined $collectcfg->{'orthogonalbuildtypes'})
|
---|
268 | {
|
---|
269 | $self->{'orthogonalbuildtypes'} = $collectcfg->{'orthogonalbuildtypes'};
|
---|
270 | }
|
---|
271 |
|
---|
272 | # - resolve (and possibly set to default) builddir
|
---|
273 | if (defined $collectcfg->{'archivedir'} && $self->{'archivedir'} eq "")
|
---|
274 | {
|
---|
275 | $self->{'archivedir'} = $collectcfg->{'archivedir'};
|
---|
276 | }
|
---|
277 | # Modified so that the archivedir, if provided as an argument, is made
|
---|
278 | # absolute if it isn't already
|
---|
279 | if ($self->{'archivedir'} eq "")
|
---|
280 | {
|
---|
281 | $self->{'archivedir'} = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "archives");
|
---|
282 | }
|
---|
283 | else
|
---|
284 | {
|
---|
285 | $self->{'archivedir'} = &util::make_absolute($ENV{'GSDLCOLLECTDIR'}, $self->{'archivedir'});
|
---|
286 | }
|
---|
287 | # End Mod
|
---|
288 | $self->{'archivedir'} = &FileUtils::sanitizePath($self->{'archivedir'});
|
---|
289 | #$self->{'archivedir'} =~ s/[\\\/]+/\//g;
|
---|
290 | #$self->{'archivedir'} =~ s/\/$//;
|
---|
291 |
|
---|
292 | # - resolve (and possibly set to default) builddir
|
---|
293 | if (defined $collectcfg->{'builddir'} && $self->{'builddir'} eq "")
|
---|
294 | {
|
---|
295 | $self->{'builddir'} = $collectcfg->{'builddir'};
|
---|
296 | }
|
---|
297 | if ($self->{'builddir'} eq "")
|
---|
298 | {
|
---|
299 | $self->{'builddir'} = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'building');
|
---|
300 | if ($incremental)
|
---|
301 | {
|
---|
302 | &gsprintf::gsprintf($out, "{buildcol.incremental_default_builddir}\n");
|
---|
303 | }
|
---|
304 | } else {
|
---|
305 | # make absolute if not already
|
---|
306 | $self->{'builddir'} = &util::make_absolute($ENV{'GSDLCOLLECTDIR'}, $self->{'builddir'});
|
---|
307 | }
|
---|
308 |
|
---|
309 | $self->{'builddir'} = &FileUtils::sanitizePath($self->{'builddir'});
|
---|
310 | #$self->{'builddir'} =~ s/[\\\/]+/\//g;
|
---|
311 | #$self->{'builddir'} =~ s/\/$//;
|
---|
312 |
|
---|
313 | if (defined $collectcfg->{'cachedir'} && $self->{'cachedir'} eq "")
|
---|
314 | {
|
---|
315 | $self->{'cachedir'} = $collectcfg->{'cachedir'};
|
---|
316 | }
|
---|
317 |
|
---|
318 | if ($self->{'maxdocs'} !~ /\-?\d+/)
|
---|
319 | {
|
---|
320 | if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/)
|
---|
321 | {
|
---|
322 | $self->{'maxdocs'} = $collectcfg->{'maxdocs'};
|
---|
323 | }
|
---|
324 | else
|
---|
325 | {
|
---|
326 | $self->{'maxdocs'} = -1; # the default
|
---|
327 | }
|
---|
328 | }
|
---|
329 |
|
---|
330 | # always clobbers? [jmt12]
|
---|
331 | if (defined $collectcfg->{'maxnumeric'} && $collectcfg->{'maxnumeric'} =~ /\d+/)
|
---|
332 | {
|
---|
333 | $self->{'maxnumeric'} = $collectcfg->{'maxnumeric'};
|
---|
334 | }
|
---|
335 | if ($self->{'maxnumeric'} < 4 || $self->{'maxnumeric'} > 512)
|
---|
336 | {
|
---|
337 | $self->{'maxnumeric'} = 4;
|
---|
338 | }
|
---|
339 |
|
---|
340 | if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i)
|
---|
341 | {
|
---|
342 | $self->{'debug'} = 1;
|
---|
343 | }
|
---|
344 |
|
---|
345 | if ($self->{'mode'} !~ /^(all|compress_text|build_index|infodb|extra)$/)
|
---|
346 | {
|
---|
347 | if (defined $collectcfg->{'mode'} && $collectcfg->{'mode'} =~ /^(all|compress_text|build_index|infodb|extra)$/)
|
---|
348 | {
|
---|
349 | $self->{'mode'} = $collectcfg->{'mode'};
|
---|
350 | }
|
---|
351 | else
|
---|
352 | {
|
---|
353 | $self->{'mode'} = "all"; # the default
|
---|
354 | }
|
---|
355 | }
|
---|
356 |
|
---|
357 | # Presumably 'index' from the collect.cfg still works [jmt12]
|
---|
358 | if (defined $collectcfg->{'index'} && $self->{'indexname'} eq "")
|
---|
359 | {
|
---|
360 | $self->{'indexname'} = $collectcfg->{'index'};
|
---|
361 | }
|
---|
362 | # - 'index' from the command line doesn't make it through parsing so I
|
---|
363 | # renamed this option 'indexname' [jmt12]
|
---|
364 | if (defined $collectcfg->{'indexname'} && $self->{'indexname'} eq "")
|
---|
365 | {
|
---|
366 | $self->{'indexname'} = $collectcfg->{'indexname'};
|
---|
367 | }
|
---|
368 | # - we may also define the index level to build now [jmt12]
|
---|
369 | if (defined $collectcfg->{'indexlevel'} && $self->{'indexlevel'} eq "")
|
---|
370 | {
|
---|
371 | $self->{'indexlevel'} = $collectcfg->{'indexlevel'};
|
---|
372 | }
|
---|
373 |
|
---|
374 | if (defined $collectcfg->{'no_text'} && $self->{'no_text'} == 0)
|
---|
375 | {
|
---|
376 | if ($collectcfg->{'no_text'} =~ /^true$/i)
|
---|
377 | {
|
---|
378 | $self->{'no_text'} = 1;
|
---|
379 | }
|
---|
380 | }
|
---|
381 |
|
---|
382 | if (defined $collectcfg->{'no_strip_html'} && $self->{'no_strip_html'} == 0)
|
---|
383 | {
|
---|
384 | if ($collectcfg->{'no_strip_html'} =~ /^true$/i)
|
---|
385 | {
|
---|
386 | $self->{'no_strip_html'} = 1;
|
---|
387 | }
|
---|
388 | }
|
---|
389 |
|
---|
390 | if (defined $collectcfg->{'store_metadata_coverage'} && $self->{'store_metadata_coverage'} == 0)
|
---|
391 | {
|
---|
392 | if ($collectcfg->{'store_metadata_coverage'} =~ /^true$/i)
|
---|
393 | {
|
---|
394 | $self->{'store_metadata_coverage'} = 1;
|
---|
395 | }
|
---|
396 | }
|
---|
397 |
|
---|
398 | if (defined $collectcfg->{'remove_empty_classifications'} && $self->{'remove_empty_classifications'} == 0)
|
---|
399 | {
|
---|
400 | if ($collectcfg->{'remove_empty_classifications'} =~ /^true$/i)
|
---|
401 | {
|
---|
402 | $self->{'remove_empty_classifications'} = 1;
|
---|
403 | }
|
---|
404 | }
|
---|
405 |
|
---|
406 | if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i)
|
---|
407 | {
|
---|
408 | $self->{'gli'} = 1;
|
---|
409 | }
|
---|
410 | if (!defined $self->{'gli'})
|
---|
411 | {
|
---|
412 | $self->{'gli'} = 0;
|
---|
413 | }
|
---|
414 |
|
---|
415 | if ($self->{'sections_index_document_metadata'} !~ /\S/ && defined $collectcfg->{'sections_index_document_metadata'})
|
---|
416 | {
|
---|
417 | $self->{'sections_index_document_metadata'} = $collectcfg->{'sections_index_document_metadata'};
|
---|
418 | }
|
---|
419 |
|
---|
420 | if ($self->{'sections_index_document_metadata'} !~ /^(never|always|unless_section_metadata_exists)$/) {
|
---|
421 | $self->{'sections_index_document_metadata'} = 'never';
|
---|
422 | }
|
---|
423 |
|
---|
424 | if ($self->{'sections_sort_on_document_metadata'} !~ /\S/ && defined $collectcfg->{'sections_sort_on_document_metadata'})
|
---|
425 | {
|
---|
426 | $self->{'sections_sort_on_document_metadata'} = $collectcfg->{'sections_sort_on_document_metadata'};
|
---|
427 | }
|
---|
428 |
|
---|
429 | if ($self->{'sections_sort_on_document_metadata'} !~ /^(never|always|unless_section_metadata_exists)$/) {
|
---|
430 | $self->{'sections_sort_on_document_metadata'} = 'never';
|
---|
431 | }
|
---|
432 |
|
---|
433 | my ($removeold, $keepold, $incremental, $incremental_mode)
|
---|
434 | = &scriptutil::check_removeold_and_keepold($self->{'removeold'}, $self->{'keepold'},
|
---|
435 | $self->{'incremental'}, 'building',
|
---|
436 | $collectcfg);
|
---|
437 | $self->{'removeold'} = $removeold;
|
---|
438 | $self->{'keepold'} = $keepold;
|
---|
439 | $self->{'incremental'} = $incremental;
|
---|
440 | $self->{'incremental_mode'} = $incremental_mode;
|
---|
441 |
|
---|
442 | # New argument to track whether build is incremental
|
---|
443 | if (!defined $self->{'incremental'})
|
---|
444 | {
|
---|
445 | $self->{'incremental'} = 0;
|
---|
446 | }
|
---|
447 |
|
---|
448 | #set the text index
|
---|
449 | if (($self->{'buildtype'} eq 'mgpp') || ($self->{'buildtype'} eq 'lucene') || ($self->{'buildtype'} eq 'solr'))
|
---|
450 | {
|
---|
451 | if ($self->{'textindex'} eq '')
|
---|
452 | {
|
---|
453 | $self->{'textindex'} = 'text';
|
---|
454 | }
|
---|
455 | }
|
---|
456 | else
|
---|
457 | {
|
---|
458 | $self->{'textindex'} = 'section:text';
|
---|
459 | }
|
---|
460 | }
|
---|
461 | # set_collection_options()
|
---|
462 |
|
---|
463 | # @function prepare_builders
|
---|
464 | #
|
---|
465 | sub prepare_builders
|
---|
466 | {
|
---|
467 | my $self = shift @_;
|
---|
468 | my ($config_filename,$collectcfg) = @_;
|
---|
469 |
|
---|
470 | my $archivedir = $self->{'archivedir'};
|
---|
471 | my $builddir = $self->{'builddir'};
|
---|
472 | my $buildtype = $self->{'buildtype'};
|
---|
473 | my $cachedir = $self->{'cachedir'};
|
---|
474 | my $collectdir = $self->{'collectdir'};
|
---|
475 | my $collection = $self->{'collection'};
|
---|
476 | my $debug = $self->{'debug'};
|
---|
477 | my $faillog = $self->{'faillog'};
|
---|
478 | my $gli = $self->{'gli'};
|
---|
479 | my $incremental = $self->{'incremental'};
|
---|
480 | my $incremental_mode = $self->{'incremental_mode'};
|
---|
481 | my $keepold = $self->{'keepold'};
|
---|
482 | my $maxdocs = $self->{'maxdocs'};
|
---|
483 | my $maxnumeric = $self->{'maxnumeric'};
|
---|
484 | my $no_strip_html = $self->{'no_strip_html'};
|
---|
485 | my $no_text = $self->{'no_text'};
|
---|
486 | my $orthogonalbuildtypes = $self->{'orthogonalbuildtypes'};
|
---|
487 | my $out = $self->{'out'};
|
---|
488 | my $remove_empty_classifications = $self->{'remove_empty_classifications'};
|
---|
489 | my $sections_index_document_metadata = $self->{'sections_index_document_metadata'};
|
---|
490 | my $sections_sort_on_document_metadata = $self->{'sections_sort_on_document_metadata'};
|
---|
491 | my $site = $self->{'site'};
|
---|
492 | my $store_metadata_coverage = $self->{'store_metadata_coverage'};
|
---|
493 | my $verbosity = $self->{'verbosity'};
|
---|
494 |
|
---|
495 | if ($gli)
|
---|
496 | {
|
---|
497 | print STDERR "<Build>\n";
|
---|
498 | }
|
---|
499 |
|
---|
500 | # fill in the default archives and building directories if none
|
---|
501 | # were supplied, turn all \ into / and remove trailing /
|
---|
502 |
|
---|
503 | my ($realarchivedir, $realbuilddir);
|
---|
504 | # update the archive cache if needed
|
---|
505 | if ($cachedir)
|
---|
506 | {
|
---|
507 | if ($verbosity >= 1)
|
---|
508 | {
|
---|
509 | &gsprintf::gsprintf($out, "{buildcol.updating_archive_cache}\n")
|
---|
510 | }
|
---|
511 |
|
---|
512 | $cachedir =~ s/[\\\/]+$//;
|
---|
513 | if ($cachedir !~ /collect[\/\\]$collection/)
|
---|
514 | {
|
---|
515 | $cachedir = &FileUtils::filenameConcatenate($cachedir, 'collect', $collection);
|
---|
516 | }
|
---|
517 |
|
---|
518 | $realarchivedir = &FileUtils::filenameConcatenate($cachedir, 'archives');
|
---|
519 | $realbuilddir = &FileUtils::filenameConcatenate($cachedir, 'building');
|
---|
520 | &FileUtils::makeAllDirectories($realarchivedir);
|
---|
521 | &FileUtils::makeAllDirectories($realbuilddir);
|
---|
522 | &FileUtils::synchronizeDirectory($archivedir, $realarchivedir, $verbosity);
|
---|
523 | }
|
---|
524 | else
|
---|
525 | {
|
---|
526 | $realarchivedir = $archivedir;
|
---|
527 | $realbuilddir = $builddir;
|
---|
528 | }
|
---|
529 | $self->{'realarchivedir'} = $realarchivedir;
|
---|
530 | $self->{'realbuilddir'} = $realbuilddir;
|
---|
531 |
|
---|
532 | # build it in realbuilddir
|
---|
533 | &FileUtils::makeAllDirectories($realbuilddir);
|
---|
534 |
|
---|
535 | my ($buildertype, $builderdir, $builder);
|
---|
536 | # if a builder class has been created for this collection, use it
|
---|
537 | # otherwise, use the mg or mgpp builder
|
---|
538 | if (-e "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib/custombuilder.pm")
|
---|
539 | {
|
---|
540 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib";
|
---|
541 | $buildertype = "custombuilder";
|
---|
542 | }
|
---|
543 | elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/custombuilder.pm")
|
---|
544 | {
|
---|
545 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
|
---|
546 | $buildertype = "custombuilder";
|
---|
547 | }
|
---|
548 | elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/${collection}builder.pm")
|
---|
549 | {
|
---|
550 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
|
---|
551 | $buildertype = $collection . 'builder';
|
---|
552 | }
|
---|
553 | else
|
---|
554 | {
|
---|
555 | $builderdir = undef;
|
---|
556 | if ($buildtype ne '')
|
---|
557 | {
|
---|
558 | # caters for extension-based build types, such as 'solr'
|
---|
559 | $buildertype = $buildtype . 'builder';
|
---|
560 | }
|
---|
561 | else
|
---|
562 | {
|
---|
563 | # Default to mgpp
|
---|
564 | $buildertype = 'mgppbuilder';
|
---|
565 | }
|
---|
566 | }
|
---|
567 | # check for extension specific builders
|
---|
568 | # (that will then be run after main builder.pm
|
---|
569 | my @builderdir_list = ($builderdir);
|
---|
570 | my @buildertype_list = ($buildertype);
|
---|
571 |
|
---|
572 | my $mode = $self->{'mode'};
|
---|
573 |
|
---|
574 | if ($mode eq "extra") {
|
---|
575 | # knock out the main builder type, by reseting the lists to be empty
|
---|
576 | @builderdir_list = ();
|
---|
577 | @buildertype_list = ();
|
---|
578 | }
|
---|
579 |
|
---|
580 | if (defined $orthogonalbuildtypes)
|
---|
581 | {
|
---|
582 | foreach my $obt (@$orthogonalbuildtypes)
|
---|
583 | {
|
---|
584 | push(@builderdir_list,undef); # rely on @INC to find it
|
---|
585 | push(@buildertype_list,$obt."Builder");
|
---|
586 | }
|
---|
587 | }
|
---|
588 |
|
---|
589 | # Set up array of the main builder.pm, followed by any ones
|
---|
590 | # from the extension folders
|
---|
591 |
|
---|
592 | my $num_builders = scalar(@buildertype_list);
|
---|
593 | my @builders = ();
|
---|
594 |
|
---|
595 | for (my $i=0; $i<$num_builders; $i++)
|
---|
596 | {
|
---|
597 | my $this_builder;
|
---|
598 | my $this_buildertype = $buildertype_list[$i];
|
---|
599 | my $this_builderdir = $builderdir_list[$i];
|
---|
600 |
|
---|
601 | if ((defined $this_builderdir) && ($this_builderdir ne ""))
|
---|
602 | {
|
---|
603 | require "$this_builderdir/$this_buildertype.pm";
|
---|
604 | }
|
---|
605 | else
|
---|
606 | {
|
---|
607 | require "$this_buildertype.pm";
|
---|
608 | }
|
---|
609 |
|
---|
610 | eval("\$this_builder = new $this_buildertype(\$site, \$collection, " .
|
---|
611 | "\$realarchivedir, \$realbuilddir, \$verbosity, " .
|
---|
612 | "\$maxdocs, \$debug, \$keepold, \$incremental, \$incremental_mode, " .
|
---|
613 | "\$remove_empty_classifications, " .
|
---|
614 | "\$out, \$no_text, \$faillog, \$gli)");
|
---|
615 | die "$@" if $@;
|
---|
616 |
|
---|
617 | push(@builders,$this_builder);
|
---|
618 | }
|
---|
619 |
|
---|
620 | # Init phase for builders
|
---|
621 | for (my $i=0; $i<$num_builders; $i++)
|
---|
622 | {
|
---|
623 | my $this_buildertype = $buildertype_list[$i];
|
---|
624 | my $this_builderdir = $builderdir_list[$i];
|
---|
625 | my $this_builder = $builders[$i];
|
---|
626 |
|
---|
627 | $this_builder->init();
|
---|
628 | $this_builder->set_maxnumeric($maxnumeric);
|
---|
629 |
|
---|
630 | if (($this_buildertype eq "mgppbuilder") && $no_strip_html)
|
---|
631 | {
|
---|
632 | $this_builder->set_strip_html(0);
|
---|
633 | }
|
---|
634 |
|
---|
635 | if ($sections_index_document_metadata ne "never")
|
---|
636 | {
|
---|
637 | $this_builder->set_sections_index_document_metadata($sections_index_document_metadata);
|
---|
638 | }
|
---|
639 | if (($this_buildertype eq "lucenebuilder" || $this_buildertype eq "solrbuilder") && $sections_sort_on_document_metadata ne "never")
|
---|
640 | {
|
---|
641 | $this_builder->set_sections_sort_on_document_metadata($sections_sort_on_document_metadata);
|
---|
642 | }
|
---|
643 |
|
---|
644 | if ($store_metadata_coverage)
|
---|
645 | {
|
---|
646 | $this_builder->set_store_metadata_coverage(1);
|
---|
647 | }
|
---|
648 | }
|
---|
649 | return \@builders;
|
---|
650 | }
|
---|
651 |
|
---|
652 | sub build_collection
|
---|
653 | {
|
---|
654 | my $self = shift(@_);
|
---|
655 | my @builders = @{shift(@_)};
|
---|
656 |
|
---|
657 | my $indexlevel = $self->{'indexlevel'};
|
---|
658 | my $indexname = $self->{'indexname'};
|
---|
659 | my $mode = $self->{'mode'};
|
---|
660 | my $textindex = $self->{'textindex'};
|
---|
661 |
|
---|
662 | # Run the requested passes
|
---|
663 | if ($mode =~ /^(all|extra)$/i)
|
---|
664 | {
|
---|
665 | # 'map' modifies the elements of the original array, so calling
|
---|
666 | # methods -- as done below -- will cause (by default) @builders
|
---|
667 | # to be changed to whatever these functions return (which is *not*
|
---|
668 | # what we want -- we want to leave the values unchanged)
|
---|
669 | # => Use 'local' (dynamic scoping) to give each 'map' call its
|
---|
670 | # own local copy This could also be done with:
|
---|
671 | # (my $new =$_)->method(); $new
|
---|
672 | # but is a bit more cumbersome to write
|
---|
673 | map { local $_=$_; $_->compress_text($textindex); } @builders;
|
---|
674 | # - we pass the required indexname and indexlevel (if specified) to the
|
---|
675 | # processor [jmt12]
|
---|
676 | map { local $_=$_; $_->build_indexes($indexname, $indexlevel); } @builders;
|
---|
677 |
|
---|
678 | # If incremental, need to deactivate the collection for collections whose db don't support concurrent R+W
|
---|
679 | # All except the collection (1st parameter) can be empty. For GS3, also set the site parameter
|
---|
680 | my $gsserver = new servercontrol( $self->get_collection(), $self->{'site'}, $self->{'verbosity'}, $self->{'builddir'}, $self->{'indexdir'}, $self->{'collectdir'}, $self->{'library_url'}, $self->{'library_name'});
|
---|
681 |
|
---|
682 | # when *incrementally* rebuilding a collection using any db that *doesn't* support concurrent
|
---|
683 | # read and write (e.g. gdbm), need to deactivate the collection before make_infodatabase()
|
---|
684 | map {
|
---|
685 | local $_=$_;
|
---|
686 |
|
---|
687 | if($_->supports_make_infodatabase()) {
|
---|
688 | my $infodbtype = $_->{'infodbtype'};
|
---|
689 | my $dbSupportsConcurrentRW = &dbutil::supportsConcurrentReadAndWrite($infodbtype);
|
---|
690 |
|
---|
691 | if(!$dbSupportsConcurrentRW && $self->{'incremental'}) {
|
---|
692 | $gsserver->print_task_msg("About to deactivate collection ".$self->get_collection());
|
---|
693 | $gsserver->do_deactivate();
|
---|
694 | }
|
---|
695 | $_->make_infodatabase();
|
---|
696 | }
|
---|
697 |
|
---|
698 | } @builders;
|
---|
699 |
|
---|
700 | map { local $_=$_; $_->collect_specific(); } @builders;
|
---|
701 | }
|
---|
702 | elsif ($mode =~ /^compress_text$/i)
|
---|
703 | {
|
---|
704 | map { local $_=$_; $_->compress_text($textindex); } @builders;
|
---|
705 | }
|
---|
706 | elsif ($mode =~ /^build_index$/i)
|
---|
707 | {
|
---|
708 | map { local $_=$_; $_->build_indexes($indexname, $indexlevel); } @builders;
|
---|
709 | }
|
---|
710 | elsif ($mode =~ /^infodb$/i)
|
---|
711 | {
|
---|
712 | map {
|
---|
713 | local $_=$_;
|
---|
714 |
|
---|
715 | # when *incrementally* rebuilding a collection using any db that *doesn't* support concurrent
|
---|
716 | # read and write (e.g. gdbm), need to deactivate the collection before make_infodatabase()
|
---|
717 |
|
---|
718 | if($_->supports_make_infodatabase()) {
|
---|
719 | my $infodbtype = $_->{'infodbtype'};
|
---|
720 | my $dbSupportsConcurrentRW = &dbutil::supportsConcurrentReadAndWrite($infodbtype);
|
---|
721 |
|
---|
722 | if(!$dbSupportsConcurrentRW && $self->{'incremental'}) {
|
---|
723 | $gsserver->print_task_msg("About to deactivate collection ".$self->get_collection());
|
---|
724 | $gsserver->do_deactivate();
|
---|
725 | }
|
---|
726 | $_->make_infodatabase();
|
---|
727 | }
|
---|
728 | } @builders;
|
---|
729 | }
|
---|
730 | else
|
---|
731 | {
|
---|
732 | (&gsprintf::gsprintf(STDERR, "{buildcol.unknown_mode}\n", $mode) && die);
|
---|
733 | }
|
---|
734 | }
|
---|
735 | # build_collection()
|
---|
736 |
|
---|
737 | # @function build_auxiliary_files
|
---|
738 | #
|
---|
739 | sub build_auxiliary_files
|
---|
740 | {
|
---|
741 | my $self = shift(@_);
|
---|
742 | my @builders = @{shift(@_)};
|
---|
743 | if (!$self->{'debug'})
|
---|
744 | {
|
---|
745 | map {local $_=$_; $_->make_auxiliary_files(); } @builders;
|
---|
746 | }
|
---|
747 | }
|
---|
748 | # build_auxiliary_files()
|
---|
749 |
|
---|
750 | # @function complete_builders
|
---|
751 | #
|
---|
752 | sub complete_builders
|
---|
753 | {
|
---|
754 | my $self = shift(@_);
|
---|
755 | my @builders = @{shift(@_)};
|
---|
756 |
|
---|
757 | map {local $_=$_; $_->deinit(); } @builders;
|
---|
758 |
|
---|
759 | if (($self->{'realbuilddir'} ne $self->{'builddir'}) && !$self->{'debug'})
|
---|
760 | {
|
---|
761 | if ($self->{'verbosity'} >= 1)
|
---|
762 | {
|
---|
763 | &gsprintf::gsprintf($out, "{buildcol.copying_back_cached_build}\n");
|
---|
764 | }
|
---|
765 | &FileUtils::removeFilesRecursive($self->{'builddir'});
|
---|
766 | &FileUtils::copyFilesRecursive($self->{'realbuilddir'}, $self->{'builddir'});
|
---|
767 | }
|
---|
768 |
|
---|
769 | # for RSS support: Need rss-items.rdf file in index folder
|
---|
770 | # check if a file called rss-items.rdf exists in archives, then copy it into the building folder
|
---|
771 | # so that when building is moved to index, this file will then also be in index as desired
|
---|
772 | my $collection_dir = &util::resolve_collection_dir($self->{'collectdir'},
|
---|
773 | $self->{'collection'},
|
---|
774 | $self->{'site'});
|
---|
775 | my $rss_items_rdf_file = &FileUtils::filenameConcatenate($self->{'archivedir'}, 'rss-items.rdf');
|
---|
776 | # @todo FileUtils
|
---|
777 | if(defined $self->{'builddir'} && &FileUtils::directoryExists($self->{'builddir'}) && &FileUtils::fileExists($rss_items_rdf_file))
|
---|
778 | {
|
---|
779 | if ($self->{'verbosity'} >= 1)
|
---|
780 | {
|
---|
781 | my $archivedir_tail = "'".basename($self->{'archivedir'})."'";
|
---|
782 | my $builddir_tail = "'".basename($self->{'builddir'})."'";
|
---|
783 |
|
---|
784 | &gsprintf::gsprintf($self->{'out'}, "{buildcol.copying_rss_items_rdf}\n", $archivedir_tail, $builddir_tail);
|
---|
785 | }
|
---|
786 | &FileUtils::copyFiles($rss_items_rdf_file, $self->{'builddir'});
|
---|
787 | }
|
---|
788 |
|
---|
789 | if ($self->{'gli'})
|
---|
790 | {
|
---|
791 | print STDERR "</Build>\n";
|
---|
792 | }
|
---|
793 | }
|
---|
794 | # complete_builders()
|
---|
795 |
|
---|
796 | # @function activate_collection
|
---|
797 | #
|
---|
798 | sub activate_collection
|
---|
799 | {
|
---|
800 | my $self = shift(@_);
|
---|
801 |
|
---|
802 | # if buildcol.pl was run with -activate, need to run activate.pl
|
---|
803 | # now that building's complete
|
---|
804 | if ($self->{'activate'})
|
---|
805 | {
|
---|
806 | #my $quoted_argv = join(" ", map { "\"$_\"" } @ARGV);
|
---|
807 | my @activate_argv = ();
|
---|
808 | push(@activate_argv, '-library_url', $self->{'library_url'}) if ($self->{'library_url'});
|
---|
809 | push(@activate_argv, '-library_name', $self->{'library_name'}) if ($self->{'library_name'});
|
---|
810 | push(@activate_argv, '-collectdir', $self->{'collectdir'}) if ($self->{'collectdir'});
|
---|
811 | push(@activate_argv, '-builddir', $self->{'builddir'}) if ($self->{'builddir'});
|
---|
812 | push(@activate_argv, '-indexdir', $self->{'indexdir'}) if ($self->{'indexdir'});
|
---|
813 | push(@activate_argv, '-site', $self->{'site'}) if ($self->{'site'});
|
---|
814 | push(@activate_argv, '-verbosity', $self->{'verbosity'}) if ($self->{'verbosity'});
|
---|
815 | push(@activate_argv, '-removeold') if ($self->{'removeold'});
|
---|
816 | push(@activate_argv, '-keepold') if ($self->{'keepold'});
|
---|
817 | push(@activate_argv, '-incremental') if ($self->{'incremental'});
|
---|
818 | push(@activate_argv, '-skipactivation', $self->{'skipactivation'}) if ($self->{'skipactivation'});
|
---|
819 |
|
---|
820 | my $quoted_argv = join(' ', map { "\"$_\"" } @activate_argv);
|
---|
821 | my $activatecol_cmd = '"' . &util::get_perl_exec(). '" -S activate.pl ' . $quoted_argv . ' "' . $self->get_collection() . '"';
|
---|
822 | my $activatecol_status = system($activatecol_cmd)/256;
|
---|
823 |
|
---|
824 | if ($activatecol_status != 0)
|
---|
825 | {
|
---|
826 | print STDERR "Error: Failed to run: $activatecol_cmd\n";
|
---|
827 | print STDERR " $!\n" if ($! ne '');
|
---|
828 | exit(-1);
|
---|
829 | }
|
---|
830 | }
|
---|
831 | }
|
---|
832 |
|
---|
833 | # @function deinit()
|
---|
834 | #
|
---|
835 | sub deinit
|
---|
836 | {
|
---|
837 | my $self = shift(@_);
|
---|
838 |
|
---|
839 | if ($self->{'close_out'})
|
---|
840 | {
|
---|
841 | close OUT;
|
---|
842 | }
|
---|
843 | if ($self->{'close_faillog'})
|
---|
844 | {
|
---|
845 | close FAILLOG;
|
---|
846 | }
|
---|
847 | }
|
---|
848 | # deinit()
|
---|
849 |
|
---|
850 | 1;
|
---|