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 |
|
---|
27 | package solrbuilder;
|
---|
28 |
|
---|
29 | use strict;
|
---|
30 | no strict 'refs';
|
---|
31 |
|
---|
32 | use lucenebuilder;
|
---|
33 | use solrserver;
|
---|
34 | use Config; # for getting the perlpath in the recommended way
|
---|
35 |
|
---|
36 | sub BEGIN {
|
---|
37 | @solrbuilder::ISA = ('lucenebuilder');
|
---|
38 | }
|
---|
39 |
|
---|
40 |
|
---|
41 | sub 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 |
|
---|
57 | sub 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 | #
|
---|
68 | sub 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 $text_dir = &FileUtils::filenameConcatenate($self->{'build_dir'}, "text");
|
---|
83 | my $build_dir = &FileUtils::filenameConcatenate($self->{'build_dir'},"");
|
---|
84 | &FileUtils::makeAllDirectories($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 | $core = "building-".$core unless $self->{'incremental'}; # core points to building only for force_removeold
|
---|
120 |
|
---|
121 | print STDERR "Executable: $solr_passes_exe\n";
|
---|
122 | print STDERR "Sections: $solr_passes_sections\n";
|
---|
123 | print STDERR "Build Dir: $build_dir\n";
|
---|
124 | print STDERR "Cmd: $solr_passes_exe $core text \"$build_dir\" \"dummy\" $osextra\n";
|
---|
125 | if (!open($handle, "| $solr_passes_exe $core text \"$build_dir\" \"dummy\" $osextra"))
|
---|
126 | {
|
---|
127 | print STDERR "<FatalError name='NoRunSolrPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
128 | die "solrbuilder::build_index - couldn't run $solr_passes_exe\n$!\n";
|
---|
129 | }
|
---|
130 | }
|
---|
131 |
|
---|
132 | # stored text is always Doc and Sec levels
|
---|
133 | my $levels = { 'document' => 1, 'section' => 1 };
|
---|
134 | # always do database at section level
|
---|
135 | my $db_level = "section";
|
---|
136 |
|
---|
137 | # set up the document processr
|
---|
138 | $self->{'buildproc'}->set_output_handle ($handle);
|
---|
139 | $self->{'buildproc'}->set_mode ('text');
|
---|
140 | $self->{'buildproc'}->set_index ($textindex);
|
---|
141 | $self->{'buildproc'}->set_indexing_text (0);
|
---|
142 | #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
143 | $self->{'buildproc'}->set_levels ($levels);
|
---|
144 | $self->{'buildproc'}->set_db_level ($db_level);
|
---|
145 | $self->{'buildproc'}->reset();
|
---|
146 |
|
---|
147 | &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
148 | $self->{'buildproc'}, $self->{'maxdocs'});
|
---|
149 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
150 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
151 | &plugin::end($self->{'pluginfo'});
|
---|
152 |
|
---|
153 | close ($handle) unless $self->{'debug'};
|
---|
154 | $self->print_stats();
|
---|
155 |
|
---|
156 | print STDERR "</Stage>\n" if $self->{'gli'};
|
---|
157 | }
|
---|
158 |
|
---|
159 | #----
|
---|
160 |
|
---|
161 |
|
---|
162 |
|
---|
163 | sub filter_in_out_file
|
---|
164 | {
|
---|
165 | my ($in_filename,$out_filename,$replace_rules) = @_;
|
---|
166 |
|
---|
167 | if (open(SIN,"<$in_filename")) {
|
---|
168 |
|
---|
169 | if (open(SOUT,">$out_filename")) {
|
---|
170 |
|
---|
171 | my $line;
|
---|
172 | while (defined ($line=<SIN>)) {
|
---|
173 | chomp $line;
|
---|
174 |
|
---|
175 | my $done_insert = 0;
|
---|
176 | foreach my $rule (@$replace_rules) {
|
---|
177 | my $line_re = $rule->{'regexp'};
|
---|
178 | my $insert = $rule->{'insert'};
|
---|
179 |
|
---|
180 | if ($line =~ m/$line_re/) {
|
---|
181 | print SOUT $insert;
|
---|
182 | $done_insert = 1;
|
---|
183 | last;
|
---|
184 | }
|
---|
185 | }
|
---|
186 | if (!$done_insert) {
|
---|
187 | print SOUT "$line\n";;
|
---|
188 | }
|
---|
189 | }
|
---|
190 |
|
---|
191 | close(SOUT);
|
---|
192 | }
|
---|
193 | else {
|
---|
194 | print STDERR "Error: Failed to open $out_filename\n";
|
---|
195 | print STDERR " $!\n";
|
---|
196 | }
|
---|
197 |
|
---|
198 | close(SIN);
|
---|
199 | }
|
---|
200 | else {
|
---|
201 | print STDERR "Error: Failed to open $in_filename\n";
|
---|
202 | print STDERR " $!\n";
|
---|
203 | }
|
---|
204 |
|
---|
205 | }
|
---|
206 |
|
---|
207 | # We need to push the list of indexfield to shortname mappings through to the
|
---|
208 | # build_cfg as, unlike in MGPP, we need these mappings in advance to configure
|
---|
209 | # Lucene/Solr. Unfortunately the original function found in mgbuilder.pm makes
|
---|
210 | # a mess of this - it only output fields that have been processed (none have)
|
---|
211 | # and it has a hardcoded renaming for 'text' so it becomes 'TX' according to
|
---|
212 | # the schema but 'TE' according to XML sent to lucene_passes.pl/solr_passes.pl
|
---|
213 | # This version is dumber - just copy them all across verbatum - but works. We
|
---|
214 | # do still need to support the special case of 'allfields'
|
---|
215 | sub make_final_field_list
|
---|
216 | {
|
---|
217 | my $self = shift (@_);
|
---|
218 | $self->{'build_cfg'} = {};
|
---|
219 | my @indexfieldmap = ();
|
---|
220 | my @indexfields = ();
|
---|
221 |
|
---|
222 | # @todo support: $self->{'buildproc'}->{'extraindexfields'}
|
---|
223 | foreach my $fields (@{$self->{'collect_cfg'}->{'indexes'}})
|
---|
224 | {
|
---|
225 | # remove subcoll stuff
|
---|
226 | $fields =~ s/:.*$//;
|
---|
227 | foreach my $field (split(';', $fields))
|
---|
228 | {
|
---|
229 | my $shortname = 'ERROR';
|
---|
230 | if ($field eq 'allfields')
|
---|
231 | {
|
---|
232 | $shortname = 'ZZ';
|
---|
233 | }
|
---|
234 | elsif (defined $self->{'buildproc'}->{'indexfieldmap'}->{$field})
|
---|
235 | {
|
---|
236 | $shortname = $self->{'buildproc'}->{'indexfieldmap'}->{$field};
|
---|
237 | }
|
---|
238 | else
|
---|
239 | {
|
---|
240 | print STDERR 'Error! Couldn\'t find indexfieldmap for field: ' . $field . "\n";
|
---|
241 | }
|
---|
242 | push (@indexfieldmap, $field . '->' . $shortname);
|
---|
243 | push (@indexfields, $field);
|
---|
244 | }
|
---|
245 | }
|
---|
246 |
|
---|
247 | if (scalar @indexfieldmap)
|
---|
248 | {
|
---|
249 | $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
|
---|
250 | }
|
---|
251 |
|
---|
252 | if (scalar @indexfields)
|
---|
253 | {
|
---|
254 | $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
|
---|
255 | }
|
---|
256 | }
|
---|
257 |
|
---|
258 | # Generate solr schema.xml file based on indexmapfield and other associated
|
---|
259 | # config files
|
---|
260 | #
|
---|
261 | # Unlike make_auxiliary_files(), this needs to be done up-front (rather
|
---|
262 | # than at the end) so the data-types in schema.xml are correctly set up
|
---|
263 | # prior to document content being pumped through solr_passes.pl
|
---|
264 |
|
---|
265 |
|
---|
266 | sub premake_solr_auxiliary_files
|
---|
267 | {
|
---|
268 | my $self = shift (@_);
|
---|
269 |
|
---|
270 | # Replace the following marker:
|
---|
271 | #
|
---|
272 | # <!-- ##GREENSTONE-FIELDS## -->
|
---|
273 | #
|
---|
274 | # with lines of the form:
|
---|
275 | #
|
---|
276 | # <field name="<field>" type="string" ... />
|
---|
277 | #
|
---|
278 | # for each <field> in 'indexfieldmap'
|
---|
279 |
|
---|
280 | my $schema_insert_xml = "";
|
---|
281 |
|
---|
282 | foreach my $ifm (@{$self->{'build_cfg'}->{'indexfieldmap'}}) {
|
---|
283 |
|
---|
284 | my ($field) = ($ifm =~ m/^.*->(.*)$/);
|
---|
285 |
|
---|
286 | $schema_insert_xml .= " "; # indent
|
---|
287 | $schema_insert_xml .= "<field name=\"$field\" ";
|
---|
288 |
|
---|
289 | if($field eq "LA" || $field eq "LO")
|
---|
290 | {
|
---|
291 | $schema_insert_xml .= "type=\"location\" ";
|
---|
292 | }
|
---|
293 | # elsif ($field ne "ZZ" && $field ne "TX")
|
---|
294 | # {
|
---|
295 | # $schema_insert_xml .= "type=\"string\" ";
|
---|
296 | # }
|
---|
297 | else
|
---|
298 | {
|
---|
299 | $schema_insert_xml .= "type=\"text_en_splitting\" ";
|
---|
300 | }
|
---|
301 | $schema_insert_xml .= "indexed=\"true\" stored=\"false\" multiValued=\"true\" />\n";
|
---|
302 | #$schema_insert_xml .= "indexed=\"true\" stored=\"true\" multiValued=\"true\" />\n";
|
---|
303 | }
|
---|
304 |
|
---|
305 | # just the one rule to date
|
---|
306 | my $insert_rules
|
---|
307 | = [ { 'regexp' => "^\\s*<!--\\s*##GREENSTONE-FIELDS##\\s*-->\\s*\$",
|
---|
308 | 'insert' => $schema_insert_xml } ];
|
---|
309 |
|
---|
310 | my $solr_home = $ENV{'GEXT_SOLR'};
|
---|
311 | ## my $in_dirname = &FileUtils::filenameConcatenate($solr_home,"etc","conf");
|
---|
312 | my $in_dirname = &FileUtils::filenameConcatenate($solr_home,"conf");
|
---|
313 | my $schema_in_filename = &FileUtils::filenameConcatenate($in_dirname,"schema.xml.in");
|
---|
314 |
|
---|
315 | my $collect_home = $ENV{'GSDLCOLLECTDIR'};
|
---|
316 | my $out_dirname = &FileUtils::filenameConcatenate($collect_home,"etc","conf");
|
---|
317 | my $schema_out_filename = &FileUtils::filenameConcatenate($out_dirname,"schema.xml");
|
---|
318 |
|
---|
319 | # make sure output conf directory exists
|
---|
320 | if (!&FileUtils::directoryExists($out_dirname)) {
|
---|
321 | &FileUtils::makeDirectory($out_dirname);
|
---|
322 | }
|
---|
323 |
|
---|
324 | filter_in_out_file($schema_in_filename,$schema_out_filename,$insert_rules);
|
---|
325 |
|
---|
326 | # now do the same for solrconfig.xml, stopwords, ...
|
---|
327 | # these are simpler, as they currently do not need any filtering
|
---|
328 |
|
---|
329 | my @in_file_list = ( "solrconfig.xml", "stopwords.txt", "stopwords_en.txt",
|
---|
330 | "synonyms.txt", "protwords.txt", "currency.xml", "elevate.xml" );
|
---|
331 |
|
---|
332 | foreach my $file ( @in_file_list ) {
|
---|
333 | my $in_filename = &FileUtils::filenameConcatenate($in_dirname,$file.".in");
|
---|
334 | my $out_filename = &FileUtils::filenameConcatenate($out_dirname,$file);
|
---|
335 |
|
---|
336 | if(&FileUtils::fileExists($in_filename)) {
|
---|
337 | filter_in_out_file($in_filename,$out_filename,[]);
|
---|
338 | }
|
---|
339 | }
|
---|
340 |
|
---|
341 | my @in_dir_list = ( "lang" );
|
---|
342 | foreach my $dir ( @in_dir_list ) {
|
---|
343 |
|
---|
344 | my $full_subdir_name = &FileUtils::filenameConcatenate($in_dirname,$dir);
|
---|
345 |
|
---|
346 | if(&FileUtils::directoryExists($full_subdir_name)) {
|
---|
347 | &FileUtils::copyFilesRecursiveNoSVN($full_subdir_name, $out_dirname);
|
---|
348 | }
|
---|
349 | }
|
---|
350 | }
|
---|
351 |
|
---|
352 |
|
---|
353 | sub pre_build_indexes
|
---|
354 | {
|
---|
355 | my $self = shift (@_);
|
---|
356 | my ($indexname) = @_;
|
---|
357 | my $outhandle = $self->{'outhandle'};
|
---|
358 |
|
---|
359 | # If the Solr/Jetty server is not already running, the following starts
|
---|
360 | # it up, and only returns when the server is "reading and listening"
|
---|
361 |
|
---|
362 | my $solr_server = new solrserver($self->{'build_dir'});
|
---|
363 | $solr_server->start();
|
---|
364 | $self->{'solr_server'} = $solr_server;
|
---|
365 |
|
---|
366 | my $indexes = [];
|
---|
367 | if (defined $indexname && $indexname =~ /\w/) {
|
---|
368 | push @$indexes, $indexname;
|
---|
369 | } else {
|
---|
370 | $indexes = $self->{'collect_cfg'}->{'indexes'};
|
---|
371 | }
|
---|
372 |
|
---|
373 | # skip para-level check, as this is done in the main 'build_indexes'
|
---|
374 | # routine
|
---|
375 |
|
---|
376 | my $all_metadata_specified = 0; # has the user added a 'metadata' index?
|
---|
377 | my $allfields_index = 0; # do we have an allfields index?
|
---|
378 |
|
---|
379 | # Using a hashmap here would avoid duplications, but while more space
|
---|
380 | # efficient, it's not entirely clear it would be more computationally
|
---|
381 | # efficient
|
---|
382 | my @all_fields = ();
|
---|
383 |
|
---|
384 | foreach my $index (@$indexes) {
|
---|
385 | if ($self->want_built($index)) {
|
---|
386 |
|
---|
387 | # get the parameters for the output
|
---|
388 | # split on : just in case there is subcoll and lang stuff
|
---|
389 | my ($fields) = split (/:/, $index);
|
---|
390 |
|
---|
391 | foreach my $field (split (/;/, $fields)) {
|
---|
392 | if ($field eq "metadata") {
|
---|
393 | $all_metadata_specified = 1;
|
---|
394 | }
|
---|
395 | else {
|
---|
396 | push(@all_fields,$field);
|
---|
397 | }
|
---|
398 | }
|
---|
399 | }
|
---|
400 | }
|
---|
401 |
|
---|
402 | if ($all_metadata_specified) {
|
---|
403 |
|
---|
404 | # (Unforunately) we need to process all the documents in the collection
|
---|
405 | # to figure out what the metadata_field_mapping is
|
---|
406 |
|
---|
407 | # set up the document processr
|
---|
408 | $self->{'buildproc'}->set_output_handle (undef);
|
---|
409 | $self->{'buildproc'}->set_mode ('index_field_mapping');
|
---|
410 | $self->{'buildproc'}->reset();
|
---|
411 |
|
---|
412 | &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
413 | $self->{'buildproc'}, $self->{'maxdocs'});
|
---|
414 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
415 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
416 | &plugin::end($self->{'pluginfo'});
|
---|
417 |
|
---|
418 | }
|
---|
419 |
|
---|
420 | else {
|
---|
421 | # Field mapping solely dependent of entries in 'indexes'
|
---|
422 |
|
---|
423 | # No need to explicitly handle "allfields" as create_shortname()
|
---|
424 | # will get a fix on it through it's static_indexfield_map
|
---|
425 |
|
---|
426 | my $buildproc = $self->{'buildproc'};
|
---|
427 |
|
---|
428 | foreach my $field (@all_fields)
|
---|
429 | {
|
---|
430 | if (!defined $buildproc->{'indexfieldmap'}->{$field})
|
---|
431 | {
|
---|
432 | my $shortname = '';
|
---|
433 | if (defined $buildproc->{'fieldnamemap'}->{$field})
|
---|
434 | {
|
---|
435 | $shortname = $buildproc->{'fieldnamemap'}->{$field};
|
---|
436 | }
|
---|
437 | else
|
---|
438 | {
|
---|
439 | $shortname = $buildproc->create_shortname($field);
|
---|
440 | }
|
---|
441 | $buildproc->{'indexfieldmap'}->{$field} = $shortname;
|
---|
442 | $buildproc->{'indexfieldmap'}->{$shortname} = 1;
|
---|
443 | }
|
---|
444 | }
|
---|
445 | }
|
---|
446 |
|
---|
447 | # Write out solr 'schema.xml' (and related) file
|
---|
448 | #
|
---|
449 | $self->make_final_field_list();
|
---|
450 | $self->premake_solr_auxiliary_files();
|
---|
451 |
|
---|
452 | # Now update the solr-core information in solr.xml
|
---|
453 | # => at most two cores <colname>-Doc and <colname>-Sec
|
---|
454 |
|
---|
455 | my $site = $self->{'site'};
|
---|
456 | my $collect = $self->{'collection'};
|
---|
457 | my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
|
---|
458 |
|
---|
459 | # my $idx = $self->{'index_mapping'}->{$index};
|
---|
460 | my $idx = "idx";
|
---|
461 |
|
---|
462 | my $build_dir = $self->{'build_dir'};
|
---|
463 |
|
---|
464 | foreach my $level (keys %{$self->{'levels'}}) {
|
---|
465 |
|
---|
466 | my ($pindex) = $level =~ /^(.)/;
|
---|
467 |
|
---|
468 | my $index_dir = $pindex.$idx;
|
---|
469 | my $core = "$core_prefix-$index_dir";
|
---|
470 |
|
---|
471 | # force_removeold == opposite of being run in 'incremental' mode
|
---|
472 | my $force_removeold = ($self->{'incremental'}) ? 0 : 1;
|
---|
473 |
|
---|
474 | if ($force_removeold) {
|
---|
475 | print $outhandle "\n-removeold set (new index will be created)\n";
|
---|
476 |
|
---|
477 | # create cores under temporary core names, corresponding to building directory
|
---|
478 | $core = "building-".$core;
|
---|
479 |
|
---|
480 | my $full_index_dir = &FileUtils::filenameConcatenate($build_dir,$index_dir);
|
---|
481 | &FileUtils::removeFilesRecursive($full_index_dir);
|
---|
482 | &FileUtils::makeDirectory($full_index_dir);
|
---|
483 |
|
---|
484 | my $full_tlog_dir = &FileUtils::filenameConcatenate($full_index_dir, "tlog");
|
---|
485 | &FileUtils::makeDirectory($full_tlog_dir);
|
---|
486 |
|
---|
487 | # Solr then wants an "index" folder within this general index area!
|
---|
488 | # my $full_index_index_dir = &FileUtils::filenameConcatenate($full_index_dir,"index");
|
---|
489 | # &FileUtils::makeDirectory($full_index_index_dir);
|
---|
490 |
|
---|
491 |
|
---|
492 | # now go on and create new index
|
---|
493 | print $outhandle "Creating Solr core: $core\n";
|
---|
494 | $solr_server->admin_create_core($core);
|
---|
495 |
|
---|
496 | }
|
---|
497 | else {
|
---|
498 | # if collect==core already in solr.xml (check with STATUS)
|
---|
499 | # => use RELOAD call to refresh fields now expressed in schema.xml
|
---|
500 | #
|
---|
501 | # else
|
---|
502 | # => use CREATE API to add to solr.xml
|
---|
503 |
|
---|
504 | my $check_core_exists = $solr_server->admin_ping_core($core);
|
---|
505 |
|
---|
506 | if ($check_core_exists) {
|
---|
507 | print $outhandle "Reloading Solr core: $core\n";
|
---|
508 | $solr_server->admin_reload_core($core);
|
---|
509 | }
|
---|
510 | else {
|
---|
511 | print $outhandle "Creating Solr core: $core\n";
|
---|
512 | $solr_server->admin_create_core($core);
|
---|
513 | }
|
---|
514 | }
|
---|
515 | }
|
---|
516 |
|
---|
517 | }
|
---|
518 |
|
---|
519 | # Essentially the same as the lucenebuilder.pm version, only using solr_passes
|
---|
520 | # => refactor and make better use of inheritence
|
---|
521 |
|
---|
522 | sub build_index {
|
---|
523 | my $self = shift (@_);
|
---|
524 | my ($index,$llevel) = @_;
|
---|
525 | my $outhandle = $self->{'outhandle'};
|
---|
526 | my $build_dir = $self->{'build_dir'};
|
---|
527 |
|
---|
528 | # get the full index directory path and make sure it exists
|
---|
529 | my $indexdir = $self->{'index_mapping'}->{$index};
|
---|
530 | &FileUtils::makeAllDirectories(&FileUtils::filenameConcatenate($build_dir, $indexdir));
|
---|
531 |
|
---|
532 | # Find the perl script to call to run solr
|
---|
533 | my $solr_passes = $self->{'solr_passes'};
|
---|
534 | my $solr_passes_exe = $self->{'solr_passes_exe'};
|
---|
535 |
|
---|
536 | # define the section names for solrpasses
|
---|
537 | # define the section names and possibly the doc name for solrpasses
|
---|
538 | my $solr_passes_sections = $llevel;
|
---|
539 |
|
---|
540 | my $osextra = "";
|
---|
541 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
---|
542 | $build_dir =~ s@/@\\@g;
|
---|
543 | } else {
|
---|
544 | if ($outhandle ne "STDERR") {
|
---|
545 | # so solr_passes doesn't print to stderr if we redirect output
|
---|
546 | $osextra .= " 2>/dev/null";
|
---|
547 | }
|
---|
548 | }
|
---|
549 |
|
---|
550 | # get the index expression if this index belongs
|
---|
551 | # to a subcollection
|
---|
552 | my $indexexparr = [];
|
---|
553 | my $langarr = [];
|
---|
554 |
|
---|
555 | # there may be subcollection info, and language info.
|
---|
556 | my ($fields, $subcollection, $language) = split (":", $index);
|
---|
557 | my @subcollections = ();
|
---|
558 | @subcollections = split /,/, $subcollection if (defined $subcollection);
|
---|
559 |
|
---|
560 | foreach $subcollection (@subcollections) {
|
---|
561 | if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
|
---|
562 | push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
|
---|
563 | }
|
---|
564 | }
|
---|
565 |
|
---|
566 | # add expressions for languages if this index belongs to
|
---|
567 | # a language subcollection - only put languages expressions for the
|
---|
568 | # ones we want in the index
|
---|
569 | my @languages = ();
|
---|
570 | my $languagemetadata = "Language";
|
---|
571 | if (defined ($self->{'collect_cfg'}->{'languagemetadata'})) {
|
---|
572 | $languagemetadata = $self->{'collect_cfg'}->{'languagemetadata'};
|
---|
573 | }
|
---|
574 | @languages = split /,/, $language if (defined $language);
|
---|
575 | foreach my $language (@languages) {
|
---|
576 | my $not=0;
|
---|
577 | if ($language =~ s/^\!//) {
|
---|
578 | $not = 1;
|
---|
579 | }
|
---|
580 | if($not) {
|
---|
581 | push (@$langarr, "!$language");
|
---|
582 | } else {
|
---|
583 | push (@$langarr, "$language");
|
---|
584 | }
|
---|
585 | }
|
---|
586 |
|
---|
587 | # Build index dictionary. Uses verbatim stem method
|
---|
588 | print $outhandle "\n creating index dictionary (solr_passes -I1)\n" if ($self->{'verbosity'} >= 1);
|
---|
589 | print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
|
---|
590 | my ($handle);
|
---|
591 |
|
---|
592 | if ($self->{'debug'}) {
|
---|
593 | $handle = *STDOUT;
|
---|
594 | } else {
|
---|
595 | my $site = $self->{'site'};
|
---|
596 | my $collect = $self->{'collection'};
|
---|
597 | my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
|
---|
598 | my $ds_idx = $self->{'index_mapping'}->{$index};
|
---|
599 | my $core = "$core_prefix-$ds_idx";
|
---|
600 |
|
---|
601 | $core = "building-".$core unless $self->{'incremental'}; # core points to building only for force_removeold
|
---|
602 |
|
---|
603 | print STDERR "Cmd: $solr_passes_exe $core index \"$build_dir\" \"$indexdir\" $osextra\n";
|
---|
604 | if (!open($handle, "| $solr_passes_exe $core index \"$build_dir\" \"$indexdir\" $osextra")) {
|
---|
605 | print STDERR "<FatalError name='NoRunSolrPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
606 | die "solrbuilder::build_index - couldn't run $solr_passes_exe\n!$\n";
|
---|
607 | }
|
---|
608 | }
|
---|
609 |
|
---|
610 | my $store_levels = $self->{'levels'};
|
---|
611 | my $db_level = "section"; #always
|
---|
612 | my $dom_level = "";
|
---|
613 | foreach my $key (keys %$store_levels) {
|
---|
614 | if ($mgppbuilder::level_map{$key} eq $llevel) {
|
---|
615 | $dom_level = $key;
|
---|
616 | }
|
---|
617 | }
|
---|
618 | if ($dom_level eq "") {
|
---|
619 | print STDERR "Warning: unrecognized tag level $llevel\n";
|
---|
620 | $dom_level = "document";
|
---|
621 | }
|
---|
622 |
|
---|
623 | my $local_levels = { $dom_level => 1 }; # work on one level at a time
|
---|
624 |
|
---|
625 | # set up the document processr
|
---|
626 | $self->{'buildproc'}->set_output_handle ($handle);
|
---|
627 | $self->{'buildproc'}->set_mode ('text');
|
---|
628 | $self->{'buildproc'}->set_index ($index, $indexexparr);
|
---|
629 | $self->{'buildproc'}->set_index_languages ($languagemetadata, $langarr) if (defined $language);
|
---|
630 | $self->{'buildproc'}->set_indexing_text (1);
|
---|
631 | #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
632 | $self->{'buildproc'}->set_levels ($local_levels);
|
---|
633 | if (defined $self->{'collect_cfg'}->{'sortfields'}) {
|
---|
634 | $self->{'buildproc'}->set_sortfields ($self->{'collect_cfg'}->{'sortfields'});
|
---|
635 | }
|
---|
636 | if (defined $self->{'collect_cfg'}->{'facetfields'}) {
|
---|
637 | $self->{'buildproc'}->set_facetfields ($self->{'collect_cfg'}->{'facetfields'});
|
---|
638 | }
|
---|
639 | $self->{'buildproc'}->set_db_level($db_level);
|
---|
640 | $self->{'buildproc'}->reset();
|
---|
641 |
|
---|
642 | print $handle "<update>\n";
|
---|
643 |
|
---|
644 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
645 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
646 |
|
---|
647 | print $handle "</update>\n";
|
---|
648 |
|
---|
649 | close ($handle) unless $self->{'debug'};
|
---|
650 |
|
---|
651 | $self->print_stats();
|
---|
652 |
|
---|
653 | $self->{'buildproc'}->set_levels ($store_levels);
|
---|
654 | print STDERR "</Stage>\n" if $self->{'gli'};
|
---|
655 |
|
---|
656 | }
|
---|
657 |
|
---|
658 |
|
---|
659 | sub post_build_indexes {
|
---|
660 | my $self = shift(@_);
|
---|
661 |
|
---|
662 | # deliberately override to prevent the mgpp post_build_index() calling
|
---|
663 | # $self->make_final_field_list()
|
---|
664 | # as this has been done in our pre_build_indexes() phase for solr
|
---|
665 |
|
---|
666 |
|
---|
667 | # Also need to stop the Solr/jetty server if it was explicitly started
|
---|
668 | # in pre_build_indexes()
|
---|
669 |
|
---|
670 | my $solr_server = $self->{'solr_server'};
|
---|
671 |
|
---|
672 | if ($solr_server->explicitly_started()) {
|
---|
673 | $solr_server->stop();
|
---|
674 | }
|
---|
675 |
|
---|
676 | $self->{'solr_server'} = undef;
|
---|
677 |
|
---|
678 | }
|
---|
679 |
|
---|
680 | sub build_cfg_extra {
|
---|
681 | my $self = shift (@_);
|
---|
682 | my ($build_cfg) = @_;
|
---|
683 |
|
---|
684 | $self->lucenebuilder::build_cfg_extra($build_cfg);
|
---|
685 |
|
---|
686 | # need to add in facet stuff
|
---|
687 | my @facetfields = ();
|
---|
688 | my @facetfieldmap = ();
|
---|
689 |
|
---|
690 | foreach my $sf (@{$self->{'buildproc'}->{'facetfields'}}) {
|
---|
691 | if ($sf eq "rank") {
|
---|
692 | push(@facetfields, $sf);
|
---|
693 | } elsif ($self->{'buildproc'}->{'actualsortfields'}->{$sf}) {
|
---|
694 | my $shortname = $self->{'buildproc'}->{'sortfieldnamemap'}->{$sf};
|
---|
695 | push(@facetfields, $shortname);
|
---|
696 | push (@facetfieldmap, "$sf\-\>$shortname");
|
---|
697 | }
|
---|
698 |
|
---|
699 | }
|
---|
700 | $build_cfg->{'indexfacetfields'} = \@facetfields;
|
---|
701 | $build_cfg->{'indexfacetfieldmap'} = \@facetfieldmap;
|
---|
702 | }
|
---|
703 | 1;
|
---|
704 |
|
---|
705 |
|
---|