source: gs3-extensions/solr/trunk/src/perllib/solrbuilder.pm@ 24501

Last change on this file since 24501 was 24501, checked in by davidb, 9 years ago

Relocation of files to make solr.solr.home more natural. Plus, more carefully control the order in which the build_dir/index_dir folder is deleted in. For solr we need to do this earlier than lucene

File size: 16.7 KB
Line 
1###########################################################################
2#
3# solrbuilder.pm -- perl wrapper for building index with Solr
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
26
27package solrbuilder;
28
29use strict;
30no strict 'refs';
31
32use lucenebuilder;
33use solrserver;
34use Config; # for getting the perlpath in the recommended way
35
36sub BEGIN {
37 @solrbuilder::ISA = ('lucenebuilder');
38}
39
40
41sub new {
42 my $class = shift(@_);
43 my $self = new lucenebuilder (@_);
44 $self = bless $self, $class;
45
46 $self->{'buildtype'} = "solr";
47
48 my $solr_passes_script = "solr_passes.pl";
49
50 $self->{'solr_passes'} = "$solr_passes_script";
51 # Tack perl on the beginning to ensure execution
52 $self->{'solr_passes_exe'} = "\"$Config{perlpath}\" -S \"$solr_passes_script\"";
53 return $self;
54}
55
56
57sub default_buildproc {
58 my $self = shift (@_);
59
60 return "solrbuildproc";
61}
62
63# This writes a nice version of the text docs
64#
65# Essentially the same as the lucenebuilder.pm version, only using solr_passes
66# => refactor and make better use of inheritence
67#
68sub compress_text
69{
70 my $self = shift (@_);
71 # do nothing if we don't want compressed text
72 return if $self->{'no_text'};
73
74 my ($textindex) = @_;
75
76 # workaround to avoid hard-coding "solr" check into buildcol.pl
77 $textindex =~ s/^section://;
78
79 my $outhandle = $self->{'outhandle'};
80
81 # the text directory
82 my $text_dir = &util::filename_cat($self->{'build_dir'}, "text");
83 my $build_dir = &util::filename_cat($self->{'build_dir'},"");
84 &util::mk_all_dir ($text_dir);
85
86 my $osextra = "";
87 if ($ENV{'GSDLOS'} =~ /^windows$/i)
88 {
89 $text_dir =~ s@/@\\@g;
90 }
91 else
92 {
93 if ($outhandle ne "STDERR")
94 {
95 # so solr_passes doesn't print to stderr if we redirect output
96 $osextra .= " 2>/dev/null";
97 }
98 }
99
100 # Find the perl script to call to run solr
101 my $solr_passes = $self->{'solr_passes'};
102 my $solr_passes_exe = $self->{'solr_passes_exe'};
103
104 my $solr_passes_sections = "Doc";
105
106 my ($handle);
107
108 if ($self->{'debug'})
109 {
110 $handle = *STDOUT;
111 }
112 else
113 {
114 my $site = $self->{'site'};
115 my $collect = $self->{'collection'};
116 my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
117 my $core = $core_prefix; # unused in this call to solr_passes
118
119 print STDERR "Executable: $solr_passes_exe\n";
120 print STDERR "Sections: $solr_passes_sections\n";
121 print STDERR "Build Dir: $build_dir\n";
122 print STDERR "Cmd: $solr_passes_exe $core text \"$build_dir\" \"dummy\" $osextra\n";
123 if (!open($handle, "| $solr_passes_exe $core text \"$build_dir\" \"dummy\" $osextra"))
124 {
125 print STDERR "<FatalError name='NoRunSolrPasses'/>\n</Stage>\n" if $self->{'gli'};
126 die "solrbuilder::build_index - couldn't run $solr_passes_exe\n$!\n";
127 }
128 }
129
130 # stored text is always Doc and Sec levels
131 my $levels = { 'document' => 1, 'section' => 1 };
132 # always do database at section level
133 my $db_level = "section";
134
135 # set up the document processr
136 $self->{'buildproc'}->set_output_handle ($handle);
137 $self->{'buildproc'}->set_mode ('text');
138 $self->{'buildproc'}->set_index ($textindex);
139 $self->{'buildproc'}->set_indexing_text (0);
140 #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
141 $self->{'buildproc'}->set_levels ($levels);
142 $self->{'buildproc'}->set_db_level ($db_level);
143 $self->{'buildproc'}->reset();
144
145 &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
146 $self->{'buildproc'}, $self->{'maxdocs'});
147 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
148 "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
149 &plugin::end($self->{'pluginfo'});
150
151 close ($handle) unless $self->{'debug'};
152 $self->print_stats();
153
154 print STDERR "</Stage>\n" if $self->{'gli'};
155}
156
157#----
158
159
160
161sub filter_in_out_file
162{
163 my ($in_filename,$out_filename,$replace_rules) = @_;
164
165 if (open(SIN,"<$in_filename")) {
166
167 if (open(SOUT,">$out_filename")) {
168
169 my $line;
170 while (defined ($line=<SIN>)) {
171 chomp $line;
172
173 my $done_insert = 0;
174 foreach my $rule (@$replace_rules) {
175 my $line_re = $rule->{'regexp'};
176 my $insert = $rule->{'insert'};
177
178 if ($line =~ m/$line_re/) {
179 print SOUT $insert;
180 $done_insert = 1;
181 last;
182 }
183 }
184 if (!$done_insert) {
185 print SOUT "$line\n";;
186 }
187 }
188
189 close(SOUT);
190 }
191 else {
192 print STDERR "Error: Failed to open $out_filename\n";
193 print STDERR " $!\n";
194 }
195
196 close(SIN);
197 }
198 else {
199 print STDERR "Error: Failed to open $in_filename\n";
200 print STDERR " $!\n";
201 }
202
203}
204
205# Generate solr schema.xml file based on indexmapfield and other associated
206# config files
207#
208# Unlike make_auxiliary_files(), this needs to be done up-front (rather
209# than at the end) so the data-types in schema.xml are correctly set up
210# prior to document content being pumped through solr_passes.pl
211
212
213sub premake_solr_auxiliary_files
214{
215 my $self = shift (@_);
216
217 # Replace the following marker:
218 #
219 # <!-- ##GREENSTONE-FIELDS## -->
220 #
221 # with lines of the form:
222 #
223 # <field name="<field>" type="string" ... />
224 #
225 # for each <field> in 'indexfieldmap'
226
227 my $schema_insert_xml = "";
228
229 foreach my $ifm (@{$self->{'build_cfg'}->{'indexfieldmap'}}) {
230
231 my ($field) = ($ifm =~ m/^.*->(.*)$/);
232
233 # Need special case for Long/Lat
234 # ... but for now treat everything as of type string
235
236 $schema_insert_xml .= " "; # indent
237 $schema_insert_xml .= "<field name=\"$field\" ";
238 $schema_insert_xml .= "type=\"string\" indexed=\"true\" ";
239 $schema_insert_xml .= "stored=\"false\" multiValued=\"true\" />\n";
240 }
241
242 # just the one rule to date
243 my $insert_rules
244 = [ { 'regexp' => "^\\s*<!--\\s*##GREENSTONE-FIELDS##\\s*-->\\s*\$",
245 'insert' => $schema_insert_xml } ];
246
247 my $solr_home = $ENV{'GEXT_SOLR'};
248## my $in_dirname = &util::filename_cat($solr_home,"etc","conf");
249 my $in_dirname = &util::filename_cat($solr_home,"conf");
250 my $schema_in_filename = &util::filename_cat($in_dirname,"schema.xml.in");
251
252
253 my $collect_home = $ENV{'GSDLCOLLECTDIR'};
254 my $out_dirname = &util::filename_cat($collect_home,"etc","conf");
255 my $schema_out_filename = &util::filename_cat($out_dirname,"schema.xml");
256
257 # make sure output conf directory exists
258 if (!-d $out_dirname) {
259 &util::mk_dir($out_dirname);
260 }
261
262 filter_in_out_file($schema_in_filename,$schema_out_filename,$insert_rules);
263
264 # now do the same for solrconfig.xml, stopwords, ...
265 # these are simpler, as they currently do not need any filtering
266
267 my @in_file_list = ( "solrconfig.xml", "stopwords.txt", "stopwords_en.txt",
268 "synonyms.txt", "protwords.txt" );
269
270 foreach my $file ( @in_file_list ) {
271 my $in_filename = &util::filename_cat($in_dirname,$file.".in");
272 my $out_filename = &util::filename_cat($out_dirname,$file);
273 filter_in_out_file($in_filename,$out_filename,[]);
274 }
275}
276
277
278sub pre_build_indexes
279{
280 my $self = shift (@_);
281 my ($indexname) = @_;
282 my $outhandle = $self->{'outhandle'};
283
284 # If the Solr/Jetty server is not already running, the following starts
285 # it up, and only returns when the server is "reading and listening"
286
287 my $solr_server = new solrserver($self->{'build_dir'});
288 $solr_server->start();
289 $self->{'solr_server'} = $solr_server;
290
291 my $indexes = [];
292 if (defined $indexname && $indexname =~ /\w/) {
293 push @$indexes, $indexname;
294 } else {
295 $indexes = $self->{'collect_cfg'}->{'indexes'};
296 }
297
298 # skip para-level check, as this is done in the main 'build_indexes'
299 # routine
300
301 my $all_metadata_specified = 0; # has the user added a 'metadata' index?
302 my $allfields_index = 0; # do we have an allfields index?
303
304 # Using a hashmap here would duplications, but while more space
305 # efficient, it's not entirely clear it would be more computationally
306 # efficient
307 my @all_fields = ();
308
309 foreach my $index (@$indexes) {
310 if ($self->want_built($index)) {
311
312 # get the parameters for the output
313 # split on : just in case there is subcoll and lang stuff
314 my ($fields) = split (/:/, $index);
315
316 foreach my $field (split (/;/, $fields)) {
317 if ($field eq "metadata") {
318 $all_metadata_specified = 1;
319 }
320 else {
321 push(@all_fields,$field);
322 }
323 }
324 }
325 }
326
327 if ($all_metadata_specified) {
328
329 # (Unforunately) we need to process all the documents in the collection
330 # to figure out what the metadata_field_mapping is
331
332 # set up the document processr
333 $self->{'buildproc'}->set_output_handle (undef);
334 $self->{'buildproc'}->set_mode ('index_field_mapping');
335 $self->{'buildproc'}->reset();
336
337 &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
338 $self->{'buildproc'}, $self->{'maxdocs'});
339 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
340 "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
341 &plugin::end($self->{'pluginfo'});
342
343 }
344
345 else {
346 # Field mapping solely dependent of entries in 'indexes'
347
348 # No need to explicitly handle "allfields" as create_shortname()
349 # will get a fix on it through it's static_indexfield_map
350
351 my $buildproc = $self->{'buildproc'};
352
353 foreach my $field (@all_fields) {
354 if (!defined $buildproc->{'indexfieldmap'}->{$field}) {
355 my $shortname = $buildproc->create_shortname($field);
356 $buildproc->{'indexfieldmap'}->{$field} = $shortname;
357 $buildproc->{'indexfieldmap'}->{$shortname} = 1;
358 }
359 }
360 }
361
362 # Write out solr 'schema.xml' (and related) file
363 #
364 $self->make_final_field_list();
365 $self->premake_solr_auxiliary_files();
366
367 # Now update the solr-core information in solr.xml
368 # => at most two cores <colname>-Doc and <colname>-Sec
369
370 my $site = $self->{'site'};
371 my $collect = $self->{'collection'};
372 my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
373
374 # my $idx = $self->{'index_mapping'}->{$index};
375 my $idx = "idx";
376
377 my $build_dir = $self->{'build_dir'};
378
379 foreach my $level (keys %{$self->{'levels'}}) {
380
381 my ($pindex) = $level =~ /^(.)/;
382
383 my $index_dir = $pindex.$idx;
384 my $core = "$core_prefix-$index_dir";
385
386 my $force_removeold = ($self->{'incremental'}) ? 0 : 1;
387 if ($force_removeold) {
388 print $outhandle "\n-removeold set (new index will be created)\n";
389
390 my $full_index_dir = &util::filename_cat($build_dir,$index_dir);
391 &util::rm_r($full_index_dir);
392 &util::mk_dir($full_index_dir);
393 }
394
395 # if collect==core already in solr.xml (check with STATUS)
396 # => use RELOAD call to refresh fields now expressed in schema.xml
397 #
398 # else
399 # => use CREATE API to add to solr.xml
400
401 my $check_core_exists = $solr_server->admin_ping_core($core);
402
403 if ($check_core_exists) {
404 print $outhandle "Reloading Solr core: $core\n";
405 $solr_server->admin_reload_core($core);
406 }
407 else {
408 print $outhandle "Creating Solr core: $core\n";
409 $solr_server->admin_create_core($core);
410 }
411 }
412
413}
414
415# Essentially the same as the lucenebuilder.pm version, only using solr_passes
416# => refactor and make better use of inheritence
417
418sub build_index {
419 my $self = shift (@_);
420 my ($index,$llevel) = @_;
421 my $outhandle = $self->{'outhandle'};
422 my $build_dir = $self->{'build_dir'};
423
424 # get the full index directory path and make sure it exists
425 my $indexdir = $self->{'index_mapping'}->{$index};
426 &util::mk_all_dir (&util::filename_cat($build_dir, $indexdir));
427
428 # Find the perl script to call to run solr
429 my $solr_passes = $self->{'solr_passes'};
430 my $solr_passes_exe = $self->{'solr_passes_exe'};
431
432 # define the section names for solrpasses
433 # define the section names and possibly the doc name for solrpasses
434 my $solr_passes_sections = $llevel;
435
436 my $osextra = "";
437 if ($ENV{'GSDLOS'} =~ /^windows$/i) {
438 $build_dir =~ s@/@\\@g;
439 } else {
440 if ($outhandle ne "STDERR") {
441 # so solr_passes doesn't print to stderr if we redirect output
442 $osextra .= " 2>/dev/null";
443 }
444 }
445
446 # get the index expression if this index belongs
447 # to a subcollection
448 my $indexexparr = [];
449 my $langarr = [];
450
451 # there may be subcollection info, and language info.
452 my ($fields, $subcollection, $language) = split (":", $index);
453 my @subcollections = ();
454 @subcollections = split /,/, $subcollection if (defined $subcollection);
455
456 foreach $subcollection (@subcollections) {
457 if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
458 push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
459 }
460 }
461
462 # add expressions for languages if this index belongs to
463 # a language subcollection - only put languages expressions for the
464 # ones we want in the index
465 my @languages = ();
466 my $languagemetadata = "Language";
467 if (defined ($self->{'collect_cfg'}->{'languagemetadata'})) {
468 $languagemetadata = $self->{'collect_cfg'}->{'languagemetadata'};
469 }
470 @languages = split /,/, $language if (defined $language);
471 foreach my $language (@languages) {
472 my $not=0;
473 if ($language =~ s/^\!//) {
474 $not = 1;
475 }
476 if($not) {
477 push (@$langarr, "!$language");
478 } else {
479 push (@$langarr, "$language");
480 }
481 }
482
483 # Build index dictionary. Uses verbatim stem method
484 print $outhandle "\n creating index dictionary (solr_passes -I1)\n" if ($self->{'verbosity'} >= 1);
485 print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
486 my ($handle);
487
488 if ($self->{'debug'}) {
489 $handle = *STDOUT;
490 } else {
491 my $site = $self->{'site'};
492 my $collect = $self->{'collection'};
493 my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
494 my $ds_idx = $self->{'index_mapping'}->{$index};
495 my $core = "$core_prefix-$ds_idx";
496
497 print STDERR "Cmd: $solr_passes_exe $core index \"$build_dir\" \"$indexdir\" $osextra\n";
498 if (!open($handle, "| $solr_passes_exe $core index \"$build_dir\" \"$indexdir\" $osextra")) {
499 print STDERR "<FatalError name='NoRunSolrPasses'/>\n</Stage>\n" if $self->{'gli'};
500 die "solrbuilder::build_index - couldn't run $solr_passes_exe\n!$\n";
501 }
502 }
503
504 my $store_levels = $self->{'levels'};
505 my $db_level = "section"; #always
506 my $dom_level = "";
507 foreach my $key (keys %$store_levels) {
508 if ($mgppbuilder::level_map{$key} eq $llevel) {
509 $dom_level = $key;
510 }
511 }
512 if ($dom_level eq "") {
513 print STDERR "Warning: unrecognized tag level $llevel\n";
514 $dom_level = "document";
515 }
516
517 my $local_levels = { $dom_level => 1 }; # work on one level at a time
518
519 # set up the document processr
520 $self->{'buildproc'}->set_output_handle ($handle);
521 $self->{'buildproc'}->set_mode ('text');
522 $self->{'buildproc'}->set_index ($index, $indexexparr);
523 $self->{'buildproc'}->set_index_languages ($languagemetadata, $langarr) if (defined $language);
524 $self->{'buildproc'}->set_indexing_text (1);
525 #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
526 $self->{'buildproc'}->set_levels ($local_levels);
527 $self->{'buildproc'}->set_db_level($db_level);
528 $self->{'buildproc'}->reset();
529
530 print $handle "<update>\n";
531
532 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
533 "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
534
535
536 print $handle "</update>\n";
537
538 close ($handle) unless $self->{'debug'};
539
540 $self->print_stats();
541
542 $self->{'buildproc'}->set_levels ($store_levels);
543 print STDERR "</Stage>\n" if $self->{'gli'};
544
545}
546
547
548sub post_build_indexes {
549 my $self = shift(@_);
550
551 # deliberately override to prevent the mgpp post_build_index() calling
552 # $self->make_final_field_list()
553 # as this has been done in our pre_build_indexes() phase for solr
554
555
556 # Also need to stop the Solr/jetty server if it was explicitly started
557 # in pre_build_indexes()
558
559 my $solr_server = $self->{'solr_server'};
560
561 if ($solr_server->explicitly_started()) {
562 $solr_server->stop();
563 }
564
565 $self->{'solr_server'} = undef;
566
567}
568
569
5701;
571
572
Note: See TracBrowser for help on using the repository browser.