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 |
|
---|
31 | package buildcol;
|
---|
32 |
|
---|
33 | BEGIN
|
---|
34 | {
|
---|
35 | die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
|
---|
36 | die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
|
---|
37 |
|
---|
38 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
|
---|
39 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
|
---|
40 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan/XML/XPath");
|
---|
41 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/plugins");
|
---|
42 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/classify");
|
---|
43 |
|
---|
44 | if (defined $ENV{'GSDLEXTS'})
|
---|
45 | {
|
---|
46 | my @extensions = split(/:/,$ENV{'GSDLEXTS'});
|
---|
47 | foreach my $e (@extensions)
|
---|
48 | {
|
---|
49 | my $ext_prefix = "$ENV{'GSDLHOME'}/ext/$e";
|
---|
50 | unshift (@INC, "$ext_prefix/perllib");
|
---|
51 | unshift (@INC, "$ext_prefix/perllib/cpan");
|
---|
52 | unshift (@INC, "$ext_prefix/perllib/plugins");
|
---|
53 | unshift (@INC, "$ext_prefix/perllib/classify");
|
---|
54 | }
|
---|
55 | }
|
---|
56 | if (defined $ENV{'GSDL3EXTS'})
|
---|
57 | {
|
---|
58 | my @extensions = split(/:/,$ENV{'GSDL3EXTS'});
|
---|
59 | foreach my $e (@extensions)
|
---|
60 | {
|
---|
61 | my $ext_prefix = "$ENV{'GSDL3SRCHOME'}/ext/$e";
|
---|
62 | unshift (@INC, "$ext_prefix/perllib");
|
---|
63 | unshift (@INC, "$ext_prefix/perllib/cpan");
|
---|
64 | unshift (@INC, "$ext_prefix/perllib/plugins");
|
---|
65 | unshift (@INC, "$ext_prefix/perllib/classify");
|
---|
66 | }
|
---|
67 | }
|
---|
68 | }
|
---|
69 |
|
---|
70 | use colcfg;
|
---|
71 | use dbutil;
|
---|
72 | use util;
|
---|
73 | use scriptutil;
|
---|
74 | use FileHandle;
|
---|
75 | use gsprintf;
|
---|
76 | use printusage;
|
---|
77 | use parse2;
|
---|
78 |
|
---|
79 | use strict;
|
---|
80 | no strict 'refs'; # allow filehandles to be variables and vice versa
|
---|
81 | no strict 'subs'; # allow barewords (eg STDERR) as function arguments
|
---|
82 |
|
---|
83 |
|
---|
84 | my $mode_list =
|
---|
85 | [ { 'name' => "all",
|
---|
86 | 'desc' => "{buildcol.mode.all}" },
|
---|
87 | { 'name' => "compress_text",
|
---|
88 | 'desc' => "{buildcol.mode.compress_text}" },
|
---|
89 | { 'name' => "build_index",
|
---|
90 | 'desc' => "{buildcol.mode.build_index}" },
|
---|
91 | { 'name' => "infodb",
|
---|
92 | 'desc' => "{buildcol.mode.infodb}" } ];
|
---|
93 |
|
---|
94 | my $sec_index_list =
|
---|
95 | [ {'name' => "never",
|
---|
96 | 'desc' => "{buildcol.sections_index_document_metadata.never}" },
|
---|
97 | {'name' => "always",
|
---|
98 | 'desc' => "{buildcol.sections_index_document_metadata.always}" },
|
---|
99 | {'name' => "unless_section_metadata_exists",
|
---|
100 | 'desc' => "{buildcol.sections_index_document_metadata.unless_section_metadata_exists}" }
|
---|
101 | ];
|
---|
102 |
|
---|
103 | my $arguments =
|
---|
104 | [ { 'name' => "remove_empty_classifications",
|
---|
105 | 'desc' => "{buildcol.remove_empty_classifications}",
|
---|
106 | 'type' => "flag",
|
---|
107 | 'reqd' => "no",
|
---|
108 | 'modegli' => "2" },
|
---|
109 | { 'name' => "archivedir",
|
---|
110 | 'desc' => "{buildcol.archivedir}",
|
---|
111 | 'type' => "string",
|
---|
112 | 'reqd' => "no",
|
---|
113 | 'hiddengli' => "yes" },
|
---|
114 | { 'name' => "builddir",
|
---|
115 | 'desc' => "{buildcol.builddir}",
|
---|
116 | 'type' => "string",
|
---|
117 | 'reqd' => "no",
|
---|
118 | 'hiddengli' => "yes" },
|
---|
119 | # { 'name' => "cachedir",
|
---|
120 | # 'desc' => "{buildcol.cachedir}",
|
---|
121 | # 'type' => "string",
|
---|
122 | # 'reqd' => "no" },
|
---|
123 | { 'name' => "collectdir",
|
---|
124 | 'desc' => "{buildcol.collectdir}",
|
---|
125 | 'type' => "string",
|
---|
126 | # parsearg left "" as default
|
---|
127 | #'deft' => &util::filename_cat ($ENV{'GSDLHOME'}, "collect"),
|
---|
128 | 'reqd' => "no",
|
---|
129 | 'hiddengli' => "yes" },
|
---|
130 | { 'name' => "site",
|
---|
131 | 'desc' => "{buildcol.site}",
|
---|
132 | 'type' => "string",
|
---|
133 | 'deft' => "",
|
---|
134 | 'reqd' => "no",
|
---|
135 | 'hiddengli' => "yes" },
|
---|
136 | { 'name' => "debug",
|
---|
137 | 'desc' => "{buildcol.debug}",
|
---|
138 | 'type' => "flag",
|
---|
139 | 'reqd' => "no",
|
---|
140 | 'hiddengli' => "yes" },
|
---|
141 | { 'name' => "faillog",
|
---|
142 | 'desc' => "{buildcol.faillog}",
|
---|
143 | 'type' => "string",
|
---|
144 | # parsearg left "" as default
|
---|
145 | #'deft' => &util::filename_cat("<collectdir>", "colname", "etc", "fail.log"),
|
---|
146 | 'reqd' => "no",
|
---|
147 | 'modegli' => "3" },
|
---|
148 | # Parallel Build Customization
|
---|
149 | # - this option was called 'index' but that doesn't get thru parsing [jmt12]
|
---|
150 | { 'name' => "indexname",
|
---|
151 | 'desc' => "{buildcol.index}",
|
---|
152 | 'type' => "string",
|
---|
153 | 'reqd' => "no",
|
---|
154 | 'modegli' => "3" },
|
---|
155 | # Parallel Build Customization
|
---|
156 | # - a new option to separate building of index levels [jmt12]
|
---|
157 | { 'name' => "indexlevel",
|
---|
158 | 'desc' => "{buildcol.indexlevel}",
|
---|
159 | 'type' => "string",
|
---|
160 | 'reqd' => "no",
|
---|
161 | 'modegli' => "3" },
|
---|
162 | { 'name' => "incremental",
|
---|
163 | 'desc' => "{buildcol.incremental}",
|
---|
164 | 'type' => "flag",
|
---|
165 | 'hiddengli' => "yes" },
|
---|
166 | { 'name' => "keepold",
|
---|
167 | 'desc' => "{buildcol.keepold}",
|
---|
168 | 'type' => "flag",
|
---|
169 | 'reqd' => "no",
|
---|
170 | #'modegli' => "3",
|
---|
171 | 'hiddengli' => "yes" },
|
---|
172 | { 'name' => "removeold",
|
---|
173 | 'desc' => "{buildcol.removeold}",
|
---|
174 | 'type' => "flag",
|
---|
175 | 'reqd' => "no",
|
---|
176 | #'modegli' => "3",
|
---|
177 | 'hiddengli' => "yes" },
|
---|
178 | { 'name' => "language",
|
---|
179 | 'desc' => "{scripts.language}",
|
---|
180 | 'type' => "string",
|
---|
181 | 'reqd' => "no",
|
---|
182 | 'modegli' => "3" },
|
---|
183 | { 'name' => "maxdocs",
|
---|
184 | 'desc' => "{buildcol.maxdocs}",
|
---|
185 | 'type' => "int",
|
---|
186 | 'reqd' => "no",
|
---|
187 | 'hiddengli' => "yes" },
|
---|
188 | { 'name' => "maxnumeric",
|
---|
189 | 'desc' => "{buildcol.maxnumeric}",
|
---|
190 | 'type' => "int",
|
---|
191 | 'reqd' => "no",
|
---|
192 | 'deft' => "4",
|
---|
193 | 'range' => "4,512",
|
---|
194 | 'modegli' => "3" },
|
---|
195 | { 'name' => "mode",
|
---|
196 | 'desc' => "{buildcol.mode}",
|
---|
197 | 'type' => "enum",
|
---|
198 | 'list' => $mode_list,
|
---|
199 | # parsearg left "" as default
|
---|
200 | # 'deft' => "all",
|
---|
201 | 'reqd' => "no",
|
---|
202 | 'modegli' => "3" },
|
---|
203 | { 'name' => "no_strip_html",
|
---|
204 | 'desc' => "{buildcol.no_strip_html}",
|
---|
205 | 'type' => "flag",
|
---|
206 | 'reqd' => "no",
|
---|
207 | 'modegli' => "3" },
|
---|
208 | { 'name' => "no_text",
|
---|
209 | 'desc' => "{buildcol.no_text}",
|
---|
210 | 'type' => "flag",
|
---|
211 | 'reqd' => "no",
|
---|
212 | 'modegli' => "2" },
|
---|
213 | { 'name' => "sections_index_document_metadata",
|
---|
214 | 'desc' => "{buildcol.sections_index_document_metadata}",
|
---|
215 | 'type' => "enum",
|
---|
216 | 'list' => $sec_index_list,
|
---|
217 | 'reqd' => "no",
|
---|
218 | 'modegli' => "2" },
|
---|
219 | { 'name' => "out",
|
---|
220 | 'desc' => "{buildcol.out}",
|
---|
221 | 'type' => "string",
|
---|
222 | 'deft' => "STDERR",
|
---|
223 | 'reqd' => "no",
|
---|
224 | 'hiddengli' => "yes" },
|
---|
225 | { 'name' => "verbosity",
|
---|
226 | 'desc' => "{buildcol.verbosity}",
|
---|
227 | 'type' => "int",
|
---|
228 | # parsearg left "" as default
|
---|
229 | #'deft' => "2",
|
---|
230 | 'reqd' => "no",
|
---|
231 | 'modegli' => "3" },
|
---|
232 | { 'name' => "gli",
|
---|
233 | 'desc' => "",
|
---|
234 | 'type' => "flag",
|
---|
235 | 'reqd' => "no",
|
---|
236 | 'hiddengli' => "yes" },
|
---|
237 | { 'name' => "xml",
|
---|
238 | 'desc' => "{scripts.xml}",
|
---|
239 | 'type' => "flag",
|
---|
240 | 'reqd' => "no",
|
---|
241 | 'hiddengli' => "yes" },
|
---|
242 | { 'name' => "parallel",
|
---|
243 | 'desc' => "{scripts.parallel}",
|
---|
244 | 'type' => "flag",
|
---|
245 | 'reqd' => "no",
|
---|
246 | 'hiddengli' => "yes" },
|
---|
247 | ];
|
---|
248 |
|
---|
249 | my $options = { 'name' => "buildcol.pl",
|
---|
250 | 'desc' => "{buildcol.desc}",
|
---|
251 | 'args' => $arguments };
|
---|
252 |
|
---|
253 |
|
---|
254 | # globals
|
---|
255 | my $collection;
|
---|
256 | my $configfilename;
|
---|
257 | my $out;
|
---|
258 |
|
---|
259 | # used to signify "gs2"(default) or "gs3"
|
---|
260 | my $gs_mode = "gs2";
|
---|
261 |
|
---|
262 | ## @method gsprintf()
|
---|
263 | # Print a string to the screen after looking it up from a locale dependant
|
---|
264 | # strings file. This function is losely based on the idea of resource
|
---|
265 | # bundles as used in Java.
|
---|
266 | #
|
---|
267 | # @param $error The STDERR stream.
|
---|
268 | # @param $text The string containing GS keys that should be replaced with
|
---|
269 | # their locale dependant equivilents.
|
---|
270 | # @param $out The output stream.
|
---|
271 | # @return The locale-based string to output.
|
---|
272 | #
|
---|
273 | sub gsprintf()
|
---|
274 | {
|
---|
275 | return &gsprintf::gsprintf(@_);
|
---|
276 | }
|
---|
277 | ## gsprintf() ##
|
---|
278 |
|
---|
279 | &main();
|
---|
280 |
|
---|
281 | ## @method main()
|
---|
282 | #
|
---|
283 | # [Parses up and validates the arguments to the build process before creating
|
---|
284 | # the appropriate build process to do the actual work - John]
|
---|
285 | #
|
---|
286 | # @note Added true incremental support - John Thompson, DL Consulting Ltd.
|
---|
287 | # @note There were several bugs regarding using directories other than
|
---|
288 | # "import" or "archives" during import and build quashed. - John
|
---|
289 | # Thompson, DL Consulting Ltd.
|
---|
290 | #
|
---|
291 | # @param $incremental If true indicates this build should not regenerate all
|
---|
292 | # the index and metadata files, and should instead just
|
---|
293 | # append the information found in the archives directory
|
---|
294 | # to the existing files. If this requires some complex
|
---|
295 | # work so as to correctly insert into a classifier so be
|
---|
296 | # it. Of course none of this is done here - instead the
|
---|
297 | # incremental argument is passed to the document
|
---|
298 | # processor.
|
---|
299 | #
|
---|
300 | sub main
|
---|
301 | {
|
---|
302 | # command line args
|
---|
303 | # - new argument to allow control of index level [jmt12]
|
---|
304 | my ($verbosity, $archivedir, $cachedir, $builddir, $site, $maxdocs,
|
---|
305 | $debug, $mode, $indexname, $removeold, $keepold,
|
---|
306 | $incremental, $incremental_mode,
|
---|
307 | $remove_empty_classifications,
|
---|
308 | $collectdir, $build, $type, $textindex,
|
---|
309 | $no_strip_html, $no_text, $faillog, $gli, $index, $language,
|
---|
310 | $sections_index_document_metadata, $maxnumeric, $indexlevel,
|
---|
311 | $parallel);
|
---|
312 |
|
---|
313 | my $xml = 0;
|
---|
314 | my $hashParsingResult = {};
|
---|
315 | # general options available to all plugins
|
---|
316 | my $intArgLeftinAfterParsing = parse2::parse(\@ARGV,$arguments,$hashParsingResult,"allow_extra_options");
|
---|
317 |
|
---|
318 | # If parse returns -1 then something has gone wrong
|
---|
319 | if ($intArgLeftinAfterParsing == -1)
|
---|
320 | {
|
---|
321 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
---|
322 | die "\n";
|
---|
323 | }
|
---|
324 |
|
---|
325 | foreach my $strVariable (keys %$hashParsingResult)
|
---|
326 | {
|
---|
327 | eval "\$$strVariable = \$hashParsingResult->{\"\$strVariable\"}";
|
---|
328 | }
|
---|
329 |
|
---|
330 | # If $language has been specified, load the appropriate resource bundle
|
---|
331 | # (Otherwise, the default resource bundle will be loaded automatically)
|
---|
332 | if ($language && $language =~ /\S/) {
|
---|
333 | &gsprintf::load_language_specific_resource_bundle($language);
|
---|
334 | }
|
---|
335 |
|
---|
336 | if ($xml) {
|
---|
337 | &PrintUsage::print_xml_usage($options);
|
---|
338 | print "\n";
|
---|
339 | return;
|
---|
340 | }
|
---|
341 |
|
---|
342 | if ($gli) { # the gli wants strings to be in UTF-8
|
---|
343 | &gsprintf::output_strings_in_UTF8;
|
---|
344 | }
|
---|
345 |
|
---|
346 | # now check that we had exactly one leftover arg, which should be
|
---|
347 | # the collection name. We don't want to do this earlier, cos
|
---|
348 | # -xml arg doesn't need a collection name
|
---|
349 | # Or if the user specified -h, then we output the usage also
|
---|
350 | if ($intArgLeftinAfterParsing != 1 || (@ARGV && $ARGV[0] =~ /^\-+h/))
|
---|
351 | {
|
---|
352 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
---|
353 | die "\n";
|
---|
354 | }
|
---|
355 |
|
---|
356 | $textindex = "";
|
---|
357 | my $close_out = 0;
|
---|
358 | if ($out !~ /^(STDERR|STDOUT)$/i) {
|
---|
359 | open (OUT, ">$out") ||
|
---|
360 | (&gsprintf(STDERR, "{common.cannot_open_output_file}\n", $out) && die);
|
---|
361 | $out = "buildcol::OUT";
|
---|
362 | $close_out = 1;
|
---|
363 | }
|
---|
364 | $out->autoflush(1);
|
---|
365 |
|
---|
366 | # get and check the collection
|
---|
367 | if (($collection = &colcfg::use_collection($site, @ARGV, $collectdir)) eq "") {
|
---|
368 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
---|
369 | die "\n";
|
---|
370 | }
|
---|
371 |
|
---|
372 | if ($faillog eq "") {
|
---|
373 | $faillog = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
|
---|
374 | }
|
---|
375 | # note that we're appending to the faillog here (import.pl clears it each time)
|
---|
376 | # this could potentially create a situation where the faillog keeps being added
|
---|
377 | # to over multiple builds (if the import process is being skipped)
|
---|
378 | open (FAILLOG, ">>$faillog") ||
|
---|
379 | (&gsprintf(STDERR, "{common.cannot_open_fail_log}\n", $faillog) && die);
|
---|
380 | $faillog = 'buildcol::FAILLOG';
|
---|
381 | $faillog->autoflush(1);
|
---|
382 |
|
---|
383 | # Don't know why this didn't already happen, but now collection specific
|
---|
384 | # classify and plugins directory also added to include path
|
---|
385 | unshift (@INC, "$ENV{'GSDLCOLLECTDIR'}/perllib"); # [jmt12]
|
---|
386 | unshift (@INC, "$ENV{'GSDLCOLLECTDIR'}/perllib/classify"); # [jmt12]
|
---|
387 | unshift (@INC, "$ENV{'GSDLCOLLECTDIR'}/perllib/plugins"); # [jmt12]
|
---|
388 |
|
---|
389 | # Read in the collection configuration file.
|
---|
390 | my ($collectcfg, $buildtype, $orthogonalbuildtypes);
|
---|
391 | ($configfilename, $gs_mode) = &colcfg::get_collect_cfg_name($out);
|
---|
392 | $collectcfg = &colcfg::read_collection_cfg ($configfilename, $gs_mode);
|
---|
393 |
|
---|
394 | # If the infodbtype value wasn't defined in the collect.cfg file, use the default
|
---|
395 | if (!defined($collectcfg->{'infodbtype'}))
|
---|
396 | {
|
---|
397 | $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
|
---|
398 | }
|
---|
399 | # sanity check - you currently can't have SQLite as the infodb while
|
---|
400 | # asking for a parallel build
|
---|
401 | elsif ($collectcfg->{'infodbtype'} eq 'sqlite' && $parallel)
|
---|
402 | {
|
---|
403 | print STDERR "WARNING: Parallel builds not current supported by SQLite - reverting to serial build\n";
|
---|
404 | $parallel = 0;
|
---|
405 | }
|
---|
406 |
|
---|
407 | if ($verbosity !~ /\d+/) {
|
---|
408 | if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/) {
|
---|
409 | $verbosity = $collectcfg->{'verbosity'};
|
---|
410 | } else {
|
---|
411 | $verbosity = 2; # the default
|
---|
412 | }
|
---|
413 | }
|
---|
414 | # we use searchtype for determining buildtype, but for old versions, use buildtype
|
---|
415 | if (defined $collectcfg->{'buildtype'}) {
|
---|
416 | $buildtype = $collectcfg->{'buildtype'};
|
---|
417 | } elsif (defined $collectcfg->{'searchtypes'} || defined $collectcfg->{'searchtype'}) {
|
---|
418 | $buildtype = "mgpp";
|
---|
419 | } else {
|
---|
420 | $buildtype = "mg"; #mg is the default
|
---|
421 | }
|
---|
422 |
|
---|
423 | if (defined $collectcfg->{'orthogonalbuildtypes'}) {
|
---|
424 | $orthogonalbuildtypes = $collectcfg->{'orthogonalbuildtypes'};
|
---|
425 | }
|
---|
426 |
|
---|
427 | if (defined $collectcfg->{'archivedir'} && $archivedir eq "") {
|
---|
428 | $archivedir = $collectcfg->{'archivedir'};
|
---|
429 | }
|
---|
430 | if (defined $collectcfg->{'cachedir'} && $cachedir eq "") {
|
---|
431 | $cachedir = $collectcfg->{'cachedir'};
|
---|
432 | }
|
---|
433 | if (defined $collectcfg->{'builddir'} && $builddir eq "") {
|
---|
434 | $builddir = $collectcfg->{'builddir'};
|
---|
435 | }
|
---|
436 | if ($maxdocs !~ /\-?\d+/) {
|
---|
437 | if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/) {
|
---|
438 | $maxdocs = $collectcfg->{'maxdocs'};
|
---|
439 | } else {
|
---|
440 | $maxdocs = -1; # the default
|
---|
441 | }
|
---|
442 | }
|
---|
443 | if (defined $collectcfg->{'maxnumeric'} && $collectcfg->{'maxnumeric'} =~ /\d+/) {
|
---|
444 | $maxnumeric = $collectcfg->{'maxnumeric'};
|
---|
445 | }
|
---|
446 |
|
---|
447 | if ($maxnumeric < 4 || $maxnumeric > 512) {
|
---|
448 | $maxnumeric = 4;
|
---|
449 | }
|
---|
450 |
|
---|
451 | if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
|
---|
452 | $debug = 1;
|
---|
453 | }
|
---|
454 | if ($mode !~ /^(all|compress_text|build_index|infodb)$/) {
|
---|
455 | if (defined $collectcfg->{'mode'} && $collectcfg->{'mode'} =~ /^(all|compress_text|build_index|infodb)$/) {
|
---|
456 | $mode = $collectcfg->{'mode'};
|
---|
457 | } else {
|
---|
458 | $mode = "all"; # the default
|
---|
459 | }
|
---|
460 | }
|
---|
461 | # - 'index' doesn't make it through parsing so I renamed this option
|
---|
462 | # 'indexname' [jmt12]
|
---|
463 | if (defined $collectcfg->{'indexname'} && $indexname eq "")
|
---|
464 | {
|
---|
465 | $indexname = $collectcfg->{'indexname'};
|
---|
466 | }
|
---|
467 | # - we may also define the index level to build now
|
---|
468 | if (defined $collectcfg->{'indexlevel'} && $indexlevel eq "")
|
---|
469 | {
|
---|
470 | $indexlevel = $collectcfg->{'indexlevel'};
|
---|
471 | }
|
---|
472 | if (defined $collectcfg->{'no_text'} && $no_text == 0) {
|
---|
473 | if ($collectcfg->{'no_text'} =~ /^true$/i) {
|
---|
474 | $no_text = 1;
|
---|
475 | }
|
---|
476 | }
|
---|
477 | if (defined $collectcfg->{'no_strip_html'} && $no_strip_html == 0) {
|
---|
478 | if ($collectcfg->{'no_strip_html'} =~ /^true$/i) {
|
---|
479 | $no_strip_html = 1;
|
---|
480 | }
|
---|
481 | }
|
---|
482 | if (defined $collectcfg->{'remove_empty_classifications'} && $remove_empty_classifications == 0) {
|
---|
483 | if ($collectcfg->{'remove_empty_classifications'} =~ /^true$/i) {
|
---|
484 | $remove_empty_classifications = 1;
|
---|
485 | }
|
---|
486 | }
|
---|
487 |
|
---|
488 | if ($buildtype eq "mgpp" && defined $collectcfg->{'textcompress'}) {
|
---|
489 | $textindex = $collectcfg->{'textcompress'};
|
---|
490 | }
|
---|
491 | if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i) {
|
---|
492 | $gli = 1;
|
---|
493 | }
|
---|
494 |
|
---|
495 | if ($sections_index_document_metadata !~ /\S/ && defined $collectcfg->{'sections_index_document_metadata'}) {
|
---|
496 | $sections_index_document_metadata = $collectcfg->{'sections_index_document_metadata'};
|
---|
497 | }
|
---|
498 |
|
---|
499 | if ($sections_index_document_metadata !~ /^(never|always|unless_section_metadata_exists)$/) {
|
---|
500 | $sections_index_document_metadata = "never";
|
---|
501 | }
|
---|
502 |
|
---|
503 | ($removeold, $keepold, $incremental, $incremental_mode)
|
---|
504 | = &scriptutil::check_removeold_and_keepold($removeold, $keepold,
|
---|
505 | $incremental, "building",
|
---|
506 | $collectcfg);
|
---|
507 |
|
---|
508 | $gli = 0 unless defined $gli;
|
---|
509 |
|
---|
510 | # New argument to track whether build is incremental
|
---|
511 | $incremental = 0 unless defined $incremental;
|
---|
512 |
|
---|
513 | print STDERR "<Build>\n" if $gli;
|
---|
514 |
|
---|
515 | #set the text index
|
---|
516 | if (($buildtype eq "mgpp") || ($buildtype eq "lucene")) {
|
---|
517 | if ($textindex eq "") {
|
---|
518 | $textindex = "text";
|
---|
519 | }
|
---|
520 | }
|
---|
521 | else {
|
---|
522 | $textindex = "section:text";
|
---|
523 | }
|
---|
524 |
|
---|
525 | # fill in the default archives and building directories if none
|
---|
526 | # were supplied, turn all \ into / and remove trailing /
|
---|
527 |
|
---|
528 | my ($realarchivedir, $realbuilddir);
|
---|
529 | # Modified so that the archivedir, if provided as an argument, is made
|
---|
530 | # absolute if it isn't already
|
---|
531 | if ($archivedir eq "")
|
---|
532 | {
|
---|
533 | $archivedir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "archives");
|
---|
534 | }
|
---|
535 | else
|
---|
536 | {
|
---|
537 | $archivedir = &util::make_absolute($ENV{'GSDLCOLLECTDIR'}, $archivedir);
|
---|
538 | }
|
---|
539 | # End Mod
|
---|
540 | $archivedir =~ s/[\\\/]+/\//g;
|
---|
541 | $archivedir =~ s/\/$//;
|
---|
542 |
|
---|
543 | if ($builddir eq "") {
|
---|
544 | $builddir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "building");
|
---|
545 | if ($incremental) {
|
---|
546 | &gsprintf($out, "{buildcol.incremental_default_builddir}\n");
|
---|
547 | }
|
---|
548 | }
|
---|
549 | $builddir =~ s/[\\\/]+/\//g;
|
---|
550 | $builddir =~ s/\/$//;
|
---|
551 |
|
---|
552 | # update the archive cache if needed
|
---|
553 | if ($cachedir) {
|
---|
554 | &gsprintf($out, "{buildcol.updating_archive_cache}\n")
|
---|
555 | if ($verbosity >= 1);
|
---|
556 |
|
---|
557 | $cachedir =~ s/[\\\/]+$//;
|
---|
558 | $cachedir .= "/collect/$collection" unless
|
---|
559 | $cachedir =~ /collect\/$collection/;
|
---|
560 |
|
---|
561 | $realarchivedir = "$cachedir/archives";
|
---|
562 | $realbuilddir = "$cachedir/building";
|
---|
563 | &util::mk_all_dir ($realarchivedir);
|
---|
564 | &util::mk_all_dir ($realbuilddir);
|
---|
565 | &util::cachedir ($archivedir, $realarchivedir, $verbosity);
|
---|
566 |
|
---|
567 | } else {
|
---|
568 | $realarchivedir = $archivedir;
|
---|
569 | $realbuilddir = $builddir;
|
---|
570 | }
|
---|
571 |
|
---|
572 | # build it in realbuilddir
|
---|
573 | &util::mk_all_dir ($realbuilddir);
|
---|
574 |
|
---|
575 | my ($buildertype, $builderdir, $builder);
|
---|
576 | # if a builder class has been created for this collection, use it
|
---|
577 | # otherwise, use the mg or mgpp builder
|
---|
578 | if (-e "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib/custombuilder.pm") {
|
---|
579 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib";
|
---|
580 | $buildertype = "custombuilder";
|
---|
581 | } elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/custombuilder.pm") {
|
---|
582 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
|
---|
583 | $buildertype = "custombuilder";
|
---|
584 | } elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/${collection}builder.pm") {
|
---|
585 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
|
---|
586 | $buildertype = "${collection}builder";
|
---|
587 | } else {
|
---|
588 |
|
---|
589 | $builderdir = undef;
|
---|
590 | if ($buildtype ne "") {
|
---|
591 | # caters for extension-based build types, such as 'solr'
|
---|
592 | $buildertype = $buildtype."builder";
|
---|
593 | }
|
---|
594 | else {
|
---|
595 | # Default to mgpp
|
---|
596 | $buildertype = "mgppbuilder";
|
---|
597 | }
|
---|
598 | }
|
---|
599 | # check for extension specific builders
|
---|
600 | # (that will then be run after main builder.pm
|
---|
601 |
|
---|
602 | my @builderdir_list = ($builderdir);
|
---|
603 | my @buildertype_list = ($buildertype);
|
---|
604 |
|
---|
605 | if (defined $orthogonalbuildtypes) {
|
---|
606 | foreach my $obt (@$orthogonalbuildtypes) {
|
---|
607 |
|
---|
608 | push(@builderdir_list,undef); # rely on @INC to find it
|
---|
609 | push(@buildertype_list,$obt."Builder");
|
---|
610 | }
|
---|
611 | }
|
---|
612 |
|
---|
613 | # Set up array of the main builder.pm, followed by any ones
|
---|
614 | # from the extension folders
|
---|
615 |
|
---|
616 | my $num_builders = scalar(@buildertype_list);
|
---|
617 | my @builders = ();
|
---|
618 |
|
---|
619 | for (my $i=0; $i<$num_builders; $i++) {
|
---|
620 | my $this_builder;
|
---|
621 | my $this_buildertype = $buildertype_list[$i];
|
---|
622 | my $this_builderdir = $builderdir_list[$i];
|
---|
623 |
|
---|
624 | if ((defined $this_builderdir) && ($this_builderdir ne "")) {
|
---|
625 | require "$this_builderdir/$this_buildertype.pm";
|
---|
626 | }
|
---|
627 | else {
|
---|
628 | require "$this_buildertype.pm";
|
---|
629 | }
|
---|
630 |
|
---|
631 | eval("\$this_builder = new $this_buildertype(\$site, \$collection, " .
|
---|
632 | "\$realarchivedir, \$realbuilddir, \$verbosity, " .
|
---|
633 | "\$maxdocs, \$debug, \$keepold, \$incremental, \$incremental_mode, " .
|
---|
634 | "\$remove_empty_classifications, " .
|
---|
635 | "\$out, \$no_text, \$faillog, \$gli)");
|
---|
636 | die "$@" if $@;
|
---|
637 |
|
---|
638 | push(@builders,$this_builder);
|
---|
639 | }
|
---|
640 |
|
---|
641 | # Init phase for builders
|
---|
642 | for (my $i=0; $i<$num_builders; $i++) {
|
---|
643 | my $this_buildertype = $buildertype_list[$i];
|
---|
644 | my $this_builderdir = $builderdir_list[$i];
|
---|
645 | my $this_builder = $builders[$i];
|
---|
646 |
|
---|
647 | $this_builder->init();
|
---|
648 | $this_builder->set_maxnumeric($maxnumeric);
|
---|
649 |
|
---|
650 | if (($this_buildertype eq "mgppbuilder") && $no_strip_html) {
|
---|
651 | $this_builder->set_strip_html(0);
|
---|
652 | }
|
---|
653 | if ($sections_index_document_metadata ne "never") {
|
---|
654 | $this_builder->set_sections_index_document_metadata($sections_index_document_metadata);
|
---|
655 | }
|
---|
656 | }
|
---|
657 |
|
---|
658 | # Run the requested passes
|
---|
659 |
|
---|
660 | # Parallel Building Support
|
---|
661 | # - if parallel building is requested then we subvert the normal 'all' mode
|
---|
662 | # process, insert attempting to create an XML 'recipe' for building this
|
---|
663 | # collection. We then pass this recipe to an Open MPI augmented compiled
|
---|
664 | # executable (which will in turn make multiple calls back to buildcol.pl
|
---|
665 | # according to the instructions in the recipe)!
|
---|
666 | if ($parallel)
|
---|
667 | {
|
---|
668 | print $out "*** parallel building\n";
|
---|
669 | # Some infodb modes (namely GDBMServer at the moment) need to open the
|
---|
670 | # connection to the database in such a way that it persists over the
|
---|
671 | # child threads. We do this by adding a dummy call to build the file path
|
---|
672 | # to archiveinf-doc as it is the database in question. The '1' at the end
|
---|
673 | # means launch the server... it will then persist until this block passes
|
---|
674 | # out of scope (presumably after all the child mpi processes are done)
|
---|
675 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($collectcfg->{'infodbtype'}, "archiveinf-doc", $archivedir, 1);
|
---|
676 |
|
---|
677 | # we initially create the recipe as a datastructure to make it easier for
|
---|
678 | # each builder to determine what has already been defined
|
---|
679 | # - each step of the recipe will have a command as a string and a (possibly
|
---|
680 | # empty) array of steps that depend on this step (possibly recursive)
|
---|
681 | print $out "Generating build 'recipe'\n";
|
---|
682 | my $recipe = [];
|
---|
683 | # pass to each builder to have it populated with appropriate commands
|
---|
684 | map { local $_=$_; $_->prepare_build_recipe($collection, $recipe); } @builders;
|
---|
685 | # now write the recipe to an XML file, resolving any path macros
|
---|
686 | my $max_parallel_tasks = scalar(@{$recipe});
|
---|
687 | my $xml_lines = ();
|
---|
688 | push(@{$xml_lines},'<?xml version="1.0" standalone="no" ?>');
|
---|
689 | push(@{$xml_lines},'<Recipe>');
|
---|
690 | foreach my $item (@{$recipe})
|
---|
691 | {
|
---|
692 | my $max_parallel_child_tasks = &print_recipe($xml_lines, $item);
|
---|
693 | if ($max_parallel_child_tasks > $max_parallel_tasks)
|
---|
694 | {
|
---|
695 | $max_parallel_tasks = $max_parallel_child_tasks;
|
---|
696 | }
|
---|
697 | }
|
---|
698 | push(@{$xml_lines}, '</Recipe>');
|
---|
699 | my $recipe_path = &util::get_tmp_filename('.xml');
|
---|
700 | open(XMLOUT, ">:utf8", $recipe_path) or die("Error! Failed to open recipe file for writing: " . $recipe_path . "\nReason: " . $!);
|
---|
701 | print XMLOUT join("\n", @{$xml_lines});
|
---|
702 | close(XMLOUT);
|
---|
703 | # determine the 'optimal' number of threads (based on number of processor
|
---|
704 | # cores and number of indexes)
|
---|
705 | my $number_of_threads = $max_parallel_tasks + 1; # any more is waste
|
---|
706 | my $number_of_cores = `grep "processor" /proc/cpuinfo | wc -l`;
|
---|
707 | $number_of_cores =~ s/\r?\n//g;
|
---|
708 | print $out "Calculating optimal threads => Max parallel tasks: $max_parallel_tasks, Number of cores: $number_of_cores\n";
|
---|
709 | if ($number_of_cores =~ /^\d+$/ && $max_parallel_tasks > $number_of_cores)
|
---|
710 | {
|
---|
711 | # optimal threads for processor bound load (even though it's probably
|
---|
712 | # IO load that has us bound)
|
---|
713 | $number_of_threads = $number_of_cores + 1;
|
---|
714 | }
|
---|
715 | print $out "Optimal threads: " . $number_of_threads . "\n";
|
---|
716 | # call mpibuildcol executable using mpirun and passing path to recipe
|
---|
717 | #rint "The path *in* Perl: " . $ENV{'PATH'} . "\n";
|
---|
718 | my $mpirun_cmd = 'mpirun -n ' . $number_of_threads . ' mpibuildcol "' . $recipe_path . '"';
|
---|
719 | #$mpirun_cmd = 'echo $PATH; echo $PERLLIB;';
|
---|
720 | print $out "Running command: " . $mpirun_cmd . "\n";
|
---|
721 | print `$mpirun_cmd`;
|
---|
722 | # clean up recipe
|
---|
723 | #unlink($recipe_path);
|
---|
724 | }
|
---|
725 | # Normal mode - run compress text, followed by index building, followed by
|
---|
726 | # infodb
|
---|
727 | elsif ($mode =~ /^all$/i) {
|
---|
728 |
|
---|
729 | # 'map' modifies the elements of the original array, so calling
|
---|
730 | # methods -- as done below -- will cause (by default) @builders
|
---|
731 | # to be changed to whatever these functions return (which is *not*
|
---|
732 | # what we want -- we want to leave the values unchanged)
|
---|
733 | # => Use 'local' (dynamic scoping) to give each 'map' call its
|
---|
734 | # own local copy This could also be done with:
|
---|
735 | # (my $new =$_)->method(); $new
|
---|
736 | # but is a bit more cumbersome to write
|
---|
737 |
|
---|
738 | map { local $_=$_; $_->compress_text($textindex); } @builders;
|
---|
739 | # - note we pass the required indexname and indexlevel (if specified)
|
---|
740 | # to the processor
|
---|
741 | map { local $_=$_; $_->build_indexes($indexname, $indexlevel); } @builders;
|
---|
742 | map { local $_=$_; $_->make_infodatabase(); } @builders;
|
---|
743 | map { local $_=$_; $_->collect_specific(); } @builders;
|
---|
744 | } elsif ($mode =~ /^compress_text$/i) {
|
---|
745 | map { local $_=$_; $_->compress_text($textindex); } @builders;
|
---|
746 | } elsif ($mode =~ /^build_index$/i) {
|
---|
747 | map { local $_=$_; $_->build_indexes($indexname, $indexlevel); } @builders;
|
---|
748 | } elsif ($mode =~ /^infodb$/i) {
|
---|
749 | map { local $_=$_; $_->make_infodatabase(); } @builders;
|
---|
750 | } else {
|
---|
751 | (&gsprintf(STDERR, "{buildcol.unknown_mode}\n", $mode) && die);
|
---|
752 | }
|
---|
753 |
|
---|
754 | if (!$debug && !$parallel) {
|
---|
755 | map {local $_=$_; $_->make_auxiliary_files(); } @builders;
|
---|
756 | }
|
---|
757 | map {local $_=$_; $_->deinit(); } @builders;
|
---|
758 |
|
---|
759 | if (($realbuilddir ne $builddir) && !$debug) {
|
---|
760 | &gsprintf($out, "{buildcol.copying_back_cached_build}\n")
|
---|
761 | if ($verbosity >= 1);
|
---|
762 | &util::rm_r ($builddir);
|
---|
763 | &util::cp_r ($realbuilddir, $builddir);
|
---|
764 | }
|
---|
765 |
|
---|
766 | close OUT if $close_out;
|
---|
767 | close FAILLOG;
|
---|
768 |
|
---|
769 | print STDERR "</Build>\n" if $gli;
|
---|
770 | }
|
---|
771 |
|
---|
772 | sub print_recipe
|
---|
773 | {
|
---|
774 | my ($xml_lines, $item) = @_;
|
---|
775 | my $max_parallel_tasks = 0;
|
---|
776 |
|
---|
777 | # start building up the command in our xml buffer
|
---|
778 | push(@{$xml_lines}, '<Task>');
|
---|
779 | my $command = $item->{'command'};
|
---|
780 | $command =~ s/&/&/g;
|
---|
781 | $command =~ s/</</g;
|
---|
782 | $command =~ s/>/>/g;
|
---|
783 | push(@{$xml_lines}, '<Command>' . $command . '</Command>');
|
---|
784 |
|
---|
785 | # - print children before closing task
|
---|
786 | if (defined $item->{'children'})
|
---|
787 | {
|
---|
788 | $max_parallel_tasks = scalar(@{$item->{'children'}});
|
---|
789 |
|
---|
790 | foreach my $child_item (@{$item->{'children'}})
|
---|
791 | {
|
---|
792 | my $max_parallel_child_tasks = &print_recipe($xml_lines, $child_item);
|
---|
793 | if ($max_parallel_child_tasks > $max_parallel_tasks)
|
---|
794 | {
|
---|
795 | $max_parallel_tasks = $max_parallel_child_tasks;
|
---|
796 | }
|
---|
797 | }
|
---|
798 | }
|
---|
799 | # - now we can close the task having printed nested children
|
---|
800 | push(@{$xml_lines},'</Task>');
|
---|
801 | # done
|
---|
802 | return $max_parallel_tasks;
|
---|
803 | }
|
---|