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 |
|
---|
35 | sub BEGIN {
|
---|
36 | @solrbuilder::ISA = ('lucenebuilder');
|
---|
37 | }
|
---|
38 |
|
---|
39 |
|
---|
40 | sub new {
|
---|
41 | my $class = shift(@_);
|
---|
42 | my $self = new lucenebuilder (@_);
|
---|
43 | $self = bless $self, $class;
|
---|
44 |
|
---|
45 | $self->{'buildtype'} = "solr";
|
---|
46 |
|
---|
47 | my $solr_passes_script = "solr_passes.pl";
|
---|
48 |
|
---|
49 | $self->{'solr_passes'} = "$solr_passes_script";
|
---|
50 | # Tack perl on the beginning to ensure execution
|
---|
51 | $self->{'solr_passes_exe'} = "\"".&util::get_perl_exec()."\" -S \"$solr_passes_script\"";
|
---|
52 | return $self;
|
---|
53 | }
|
---|
54 |
|
---|
55 |
|
---|
56 | sub default_buildproc {
|
---|
57 | my $self = shift (@_);
|
---|
58 |
|
---|
59 | return "solrbuildproc";
|
---|
60 | }
|
---|
61 |
|
---|
62 | # This writes a nice version of the text docs
|
---|
63 | #
|
---|
64 | # Essentially the same as the lucenebuilder.pm version, only using solr_passes
|
---|
65 | # => refactor and make better use of inheritence
|
---|
66 | #
|
---|
67 | sub compress_text
|
---|
68 | {
|
---|
69 | my $self = shift (@_);
|
---|
70 | # do nothing if we don't want compressed text
|
---|
71 | return if $self->{'no_text'};
|
---|
72 |
|
---|
73 | my ($textindex) = @_;
|
---|
74 |
|
---|
75 | # workaround to avoid hard-coding "solr" check into buildcol.pl
|
---|
76 | $textindex =~ s/^section://;
|
---|
77 |
|
---|
78 | my $outhandle = $self->{'outhandle'};
|
---|
79 |
|
---|
80 | # the text directory
|
---|
81 | my $text_dir = &FileUtils::filenameConcatenate($self->{'build_dir'}, "text");
|
---|
82 | my $build_dir = &FileUtils::filenameConcatenate($self->{'build_dir'},"");
|
---|
83 | &FileUtils::makeAllDirectories($text_dir);
|
---|
84 |
|
---|
85 | my $osextra = "";
|
---|
86 | if ($ENV{'GSDLOS'} =~ /^windows$/i)
|
---|
87 | {
|
---|
88 | $text_dir =~ s@/@\\@g;
|
---|
89 | }
|
---|
90 | else
|
---|
91 | {
|
---|
92 | if ($outhandle ne "STDERR")
|
---|
93 | {
|
---|
94 | # so solr_passes doesn't print to stderr if we redirect output
|
---|
95 | $osextra .= " 2>/dev/null";
|
---|
96 | }
|
---|
97 | }
|
---|
98 |
|
---|
99 | # Find the perl script to call to run solr
|
---|
100 | my $solr_passes = $self->{'solr_passes'};
|
---|
101 | my $solr_passes_exe = $self->{'solr_passes_exe'};
|
---|
102 |
|
---|
103 | my $solr_passes_sections = "Doc";
|
---|
104 |
|
---|
105 | my ($handle);
|
---|
106 |
|
---|
107 | if ($self->{'debug'})
|
---|
108 | {
|
---|
109 | $handle = *STDOUT;
|
---|
110 | }
|
---|
111 | else
|
---|
112 | {
|
---|
113 | my $site = $self->{'site'};
|
---|
114 | my $collect = $self->{'collection'};
|
---|
115 | my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
|
---|
116 | my $core = $core_prefix; # unused in this call to solr_passes
|
---|
117 |
|
---|
118 | $core = "building-".$core unless $self->{'incremental'}; # core points to building only for force_removeold
|
---|
119 |
|
---|
120 | print STDERR "Executable: $solr_passes_exe\n";
|
---|
121 | print STDERR "Sections: $solr_passes_sections\n";
|
---|
122 | print STDERR "Build Dir: $build_dir\n";
|
---|
123 | print STDERR "Cmd: $solr_passes_exe $core text \"$build_dir\" \"dummy\" $osextra\n";
|
---|
124 | if (!open($handle, "| $solr_passes_exe $core text \"$build_dir\" \"dummy\" $osextra"))
|
---|
125 | {
|
---|
126 | print STDERR "<FatalError name='NoRunSolrPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
127 | die "solrbuilder::build_index - couldn't run $solr_passes_exe\n$!\n";
|
---|
128 | }
|
---|
129 | }
|
---|
130 |
|
---|
131 | # stored text is always Doc and Sec levels
|
---|
132 | my $levels = { 'document' => 1, 'section' => 1 };
|
---|
133 | # always do database at section level
|
---|
134 | my $db_level = "section";
|
---|
135 |
|
---|
136 | # set up the document processr
|
---|
137 | $self->{'buildproc'}->set_output_handle ($handle);
|
---|
138 | $self->{'buildproc'}->set_mode ('text');
|
---|
139 | $self->{'buildproc'}->set_index ($textindex);
|
---|
140 | $self->{'buildproc'}->set_indexing_text (0);
|
---|
141 | #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
142 | $self->{'buildproc'}->set_levels ($levels);
|
---|
143 | $self->{'buildproc'}->set_db_level ($db_level);
|
---|
144 | $self->{'buildproc'}->reset();
|
---|
145 |
|
---|
146 | &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
147 | $self->{'buildproc'}, $self->{'maxdocs'});
|
---|
148 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
149 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
150 | &plugin::end($self->{'pluginfo'});
|
---|
151 |
|
---|
152 | close ($handle) unless $self->{'debug'};
|
---|
153 | $self->print_stats();
|
---|
154 |
|
---|
155 | print STDERR "</Stage>\n" if $self->{'gli'};
|
---|
156 | }
|
---|
157 |
|
---|
158 | #----
|
---|
159 |
|
---|
160 |
|
---|
161 |
|
---|
162 | sub filter_in_out_file
|
---|
163 | {
|
---|
164 | my ($in_filename,$out_filename,$replace_rules) = @_;
|
---|
165 |
|
---|
166 | if (open(SIN,"<$in_filename")) {
|
---|
167 |
|
---|
168 | if (open(SOUT,">$out_filename")) {
|
---|
169 |
|
---|
170 | my $line;
|
---|
171 | while (defined ($line=<SIN>)) {
|
---|
172 | chomp $line;
|
---|
173 |
|
---|
174 | my $done_insert = 0;
|
---|
175 | foreach my $rule (@$replace_rules) {
|
---|
176 | my $line_re = $rule->{'regexp'};
|
---|
177 | my $insert = $rule->{'insert'};
|
---|
178 |
|
---|
179 | if ($line =~ m/$line_re/) {
|
---|
180 | print SOUT $insert;
|
---|
181 | $done_insert = 1;
|
---|
182 | last;
|
---|
183 | }
|
---|
184 | }
|
---|
185 | if (!$done_insert) {
|
---|
186 | print SOUT "$line\n";;
|
---|
187 | }
|
---|
188 | }
|
---|
189 |
|
---|
190 | close(SOUT);
|
---|
191 | }
|
---|
192 | else {
|
---|
193 | print STDERR "Error: Failed to open $out_filename\n";
|
---|
194 | print STDERR " $!\n";
|
---|
195 | }
|
---|
196 |
|
---|
197 | close(SIN);
|
---|
198 | }
|
---|
199 | else {
|
---|
200 | print STDERR "Error: Failed to open $in_filename\n";
|
---|
201 | print STDERR " $!\n";
|
---|
202 | }
|
---|
203 |
|
---|
204 | }
|
---|
205 |
|
---|
206 | # We need to push the list of indexfield to shortname mappings through to the
|
---|
207 | # build_cfg as, unlike in MGPP, we need these mappings in advance to configure
|
---|
208 | # Lucene/Solr. Unfortunately the original function found in mgbuilder.pm makes
|
---|
209 | # a mess of this - it only output fields that have been processed (none have)
|
---|
210 | # and it has a hardcoded renaming for 'text' so it becomes 'TX' according to
|
---|
211 | # the schema but 'TE' according to XML sent to lucene_passes.pl/solr_passes.pl
|
---|
212 | # This version is dumber - just copy them all across verbatum - but works. We
|
---|
213 | # do still need to support the special case of 'allfields'
|
---|
214 | sub make_final_field_list
|
---|
215 | {
|
---|
216 | my $self = shift (@_);
|
---|
217 | $self->{'build_cfg'} = {};
|
---|
218 | my @indexfieldmap = ();
|
---|
219 | my @indexfields = ();
|
---|
220 |
|
---|
221 | # @todo support: $self->{'buildproc'}->{'extraindexfields'}
|
---|
222 | foreach my $fields (@{$self->{'collect_cfg'}->{'indexes'}})
|
---|
223 | {
|
---|
224 | # remove subcoll stuff
|
---|
225 | $fields =~ s/:.*$//;
|
---|
226 | foreach my $field (split(';', $fields))
|
---|
227 | {
|
---|
228 | my $shortname = 'ERROR';
|
---|
229 | if ($field eq 'allfields')
|
---|
230 | {
|
---|
231 | $shortname = 'ZZ';
|
---|
232 | }
|
---|
233 | elsif (defined $self->{'buildproc'}->{'indexfieldmap'}->{$field})
|
---|
234 | {
|
---|
235 | $shortname = $self->{'buildproc'}->{'indexfieldmap'}->{$field};
|
---|
236 | }
|
---|
237 | else
|
---|
238 | {
|
---|
239 | print STDERR 'Error! Couldn\'t find indexfieldmap for field: ' . $field . "\n";
|
---|
240 | }
|
---|
241 | push (@indexfieldmap, $field . '->' . $shortname);
|
---|
242 | push (@indexfields, $field);
|
---|
243 | }
|
---|
244 | }
|
---|
245 |
|
---|
246 | if (scalar @indexfieldmap)
|
---|
247 | {
|
---|
248 | $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
|
---|
249 | }
|
---|
250 |
|
---|
251 | if (scalar @indexfields)
|
---|
252 | {
|
---|
253 | $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
|
---|
254 | }
|
---|
255 | }
|
---|
256 |
|
---|
257 | # Generate solr schema.xml file based on indexmapfield and other associated
|
---|
258 | # config files
|
---|
259 | #
|
---|
260 | # Unlike make_auxiliary_files(), this needs to be done up-front (rather
|
---|
261 | # than at the end) so the data-types in schema.xml are correctly set up
|
---|
262 | # prior to document content being pumped through solr_passes.pl
|
---|
263 |
|
---|
264 |
|
---|
265 | sub premake_solr_auxiliary_files
|
---|
266 | {
|
---|
267 | my $self = shift (@_);
|
---|
268 |
|
---|
269 | # Replace the following marker:
|
---|
270 | #
|
---|
271 | # <!-- ##GREENSTONE-FIELDS## -->
|
---|
272 | #
|
---|
273 | # with lines of the form:
|
---|
274 | #
|
---|
275 | # <field name="<field>" type="string" ... />
|
---|
276 | #
|
---|
277 | # for each <field> in 'indexfieldmap'
|
---|
278 |
|
---|
279 | my $schema_insert_xml = "";
|
---|
280 |
|
---|
281 | foreach my $ifm (@{$self->{'build_cfg'}->{'indexfieldmap'}}) {
|
---|
282 |
|
---|
283 | my ($fullfieldname, $field) = ($ifm =~ m/^(.*)->(.*)$/);
|
---|
284 |
|
---|
285 | $schema_insert_xml .= " "; # indent
|
---|
286 | $schema_insert_xml .= "<field name=\"$field\" ";
|
---|
287 |
|
---|
288 | if($field eq "LA" || $field eq "LO")
|
---|
289 | {
|
---|
290 | $schema_insert_xml .= "type=\"location\" ";
|
---|
291 | }
|
---|
292 | # elsif ($field ne "ZZ" && $field ne "TX")
|
---|
293 | # {
|
---|
294 | # $schema_insert_xml .= "type=\"string\" ";
|
---|
295 | # }
|
---|
296 | else
|
---|
297 | {
|
---|
298 | #$schema_insert_xml .= "type=\"text_en_splitting\" ";
|
---|
299 |
|
---|
300 | # original default solr field type for all fields is text_en_splitting
|
---|
301 | my $solrfieldtype = "text_en_splitting";
|
---|
302 | if(defined $self->{'collect_cfg'}->{'indexfieldoptions'}->{$fullfieldname}->{'solrfieldtype'}) {
|
---|
303 | $solrfieldtype = $self->{'collect_cfg'}->{'indexfieldoptions'}->{$fullfieldname}->{'solrfieldtype'};
|
---|
304 | #print STDERR "@@@@#### found TYPE: $solrfieldtype\n";
|
---|
305 | }
|
---|
306 | $schema_insert_xml .= "type=\"$solrfieldtype\" ";
|
---|
307 |
|
---|
308 | }
|
---|
309 | # set termVectors=\"true\" when term vectors info is required,
|
---|
310 | # see TermsResponse termResponse = solrResponse.getTermsResponse();
|
---|
311 | $schema_insert_xml .= "indexed=\"true\" stored=\"true\" multiValued=\"true\" termVectors=\"true\" termPositions=\"true\" termOffsets=\"true\" />\n";
|
---|
312 | }
|
---|
313 |
|
---|
314 | # just the one rule to date
|
---|
315 | my $insert_rules
|
---|
316 | = [ { 'regexp' => "^\\s*<!--\\s*##GREENSTONE-FIELDS##\\s*-->\\s*\$",
|
---|
317 | 'insert' => $schema_insert_xml } ];
|
---|
318 |
|
---|
319 | my $solr_home = $ENV{'GEXT_SOLR'};
|
---|
320 | ## my $in_dirname = &FileUtils::filenameConcatenate($solr_home,"etc","conf");
|
---|
321 | my $in_dirname = &FileUtils::filenameConcatenate($solr_home,"conf");
|
---|
322 | my $schema_in_filename = &FileUtils::filenameConcatenate($in_dirname,"schema.xml.in");
|
---|
323 |
|
---|
324 | my $collect_home = $ENV{'GSDLCOLLECTDIR'};
|
---|
325 | my $out_dirname = &FileUtils::filenameConcatenate($collect_home,"etc","conf");
|
---|
326 | my $schema_out_filename = &FileUtils::filenameConcatenate($out_dirname,"schema.xml");
|
---|
327 |
|
---|
328 | # make sure output conf directory exists
|
---|
329 | if (!&FileUtils::directoryExists($out_dirname)) {
|
---|
330 | &FileUtils::makeDirectory($out_dirname);
|
---|
331 | }
|
---|
332 |
|
---|
333 | filter_in_out_file($schema_in_filename,$schema_out_filename,$insert_rules);
|
---|
334 |
|
---|
335 | # now do the same for solrconfig.xml, stopwords, ...
|
---|
336 | # these are simpler, as they currently do not need any filtering
|
---|
337 |
|
---|
338 | my @in_file_list = ( "solrconfig.xml", "stopwords.txt", "stopwords_en.txt",
|
---|
339 | "synonyms.txt", "protwords.txt", "currency.xml", "elevate.xml" );
|
---|
340 |
|
---|
341 | foreach my $file ( @in_file_list ) {
|
---|
342 | my $in_filename = &FileUtils::filenameConcatenate($in_dirname,$file.".in");
|
---|
343 | my $out_filename = &FileUtils::filenameConcatenate($out_dirname,$file);
|
---|
344 |
|
---|
345 | if(&FileUtils::fileExists($in_filename)) {
|
---|
346 | filter_in_out_file($in_filename,$out_filename,[]);
|
---|
347 | }
|
---|
348 | }
|
---|
349 |
|
---|
350 | my @in_dir_list = ( "lang" );
|
---|
351 | foreach my $dir ( @in_dir_list ) {
|
---|
352 |
|
---|
353 | my $full_subdir_name = &FileUtils::filenameConcatenate($in_dirname,$dir);
|
---|
354 |
|
---|
355 | if(&FileUtils::directoryExists($full_subdir_name)) {
|
---|
356 | &FileUtils::copyFilesRecursiveNoSVN($full_subdir_name, $out_dirname);
|
---|
357 | }
|
---|
358 | }
|
---|
359 | }
|
---|
360 |
|
---|
361 |
|
---|
362 | sub pre_build_indexes
|
---|
363 | {
|
---|
364 | my $self = shift (@_);
|
---|
365 | my ($indexname) = @_;
|
---|
366 | my $outhandle = $self->{'outhandle'};
|
---|
367 |
|
---|
368 | # If the Solr/Jetty server is not already running, the following starts
|
---|
369 | # it up, and only returns when the server is "reading and listening"
|
---|
370 |
|
---|
371 | my $solr_server = new solrserver($self->{'build_dir'});
|
---|
372 | $solr_server->start();
|
---|
373 | $self->{'solr_server'} = $solr_server;
|
---|
374 |
|
---|
375 | my $indexes = [];
|
---|
376 | if (defined $indexname && $indexname =~ /\w/) {
|
---|
377 | push @$indexes, $indexname;
|
---|
378 | } else {
|
---|
379 | $indexes = $self->{'collect_cfg'}->{'indexes'};
|
---|
380 | }
|
---|
381 |
|
---|
382 | # skip para-level check, as this is done in the main 'build_indexes'
|
---|
383 | # routine
|
---|
384 |
|
---|
385 | my $all_metadata_specified = 0; # has the user added a 'metadata' index?
|
---|
386 | my $allfields_index = 0; # do we have an allfields index?
|
---|
387 |
|
---|
388 | # Using a hashmap here would avoid duplications, but while more space
|
---|
389 | # efficient, it's not entirely clear it would be more computationally
|
---|
390 | # efficient
|
---|
391 | my @all_fields = ();
|
---|
392 |
|
---|
393 | foreach my $index (@$indexes) {
|
---|
394 | if ($self->want_built($index)) {
|
---|
395 |
|
---|
396 | # get the parameters for the output
|
---|
397 | # split on : just in case there is subcoll and lang stuff
|
---|
398 | my ($fields) = split (/:/, $index);
|
---|
399 |
|
---|
400 | foreach my $field (split (/;/, $fields)) {
|
---|
401 | if ($field eq "metadata") {
|
---|
402 | $all_metadata_specified = 1;
|
---|
403 | }
|
---|
404 | else {
|
---|
405 | push(@all_fields,$field);
|
---|
406 | }
|
---|
407 | }
|
---|
408 | }
|
---|
409 | }
|
---|
410 |
|
---|
411 | if ($all_metadata_specified) {
|
---|
412 |
|
---|
413 | # (Unforunately) we need to process all the documents in the collection
|
---|
414 | # to figure out what the metadata_field_mapping is
|
---|
415 |
|
---|
416 | # set up the document processr
|
---|
417 | $self->{'buildproc'}->set_output_handle (undef);
|
---|
418 | $self->{'buildproc'}->set_mode ('index_field_mapping');
|
---|
419 | $self->{'buildproc'}->reset();
|
---|
420 |
|
---|
421 | &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
422 | $self->{'buildproc'}, $self->{'maxdocs'});
|
---|
423 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
424 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
425 | &plugin::end($self->{'pluginfo'});
|
---|
426 |
|
---|
427 | }
|
---|
428 |
|
---|
429 | else {
|
---|
430 | # Field mapping solely dependent of entries in 'indexes'
|
---|
431 |
|
---|
432 | # No need to explicitly handle "allfields" as create_shortname()
|
---|
433 | # will get a fix on it through it's static_indexfield_map
|
---|
434 |
|
---|
435 | my $buildproc = $self->{'buildproc'};
|
---|
436 |
|
---|
437 | foreach my $field (@all_fields)
|
---|
438 | {
|
---|
439 | if (!defined $buildproc->{'indexfieldmap'}->{$field})
|
---|
440 | {
|
---|
441 | my $shortname = '';
|
---|
442 | if (defined $buildproc->{'fieldnamemap'}->{$field})
|
---|
443 | {
|
---|
444 | $shortname = $buildproc->{'fieldnamemap'}->{$field};
|
---|
445 | }
|
---|
446 | else
|
---|
447 | {
|
---|
448 | $shortname = $buildproc->create_shortname($field);
|
---|
449 | }
|
---|
450 | $buildproc->{'indexfieldmap'}->{$field} = $shortname;
|
---|
451 | $buildproc->{'indexfieldmap'}->{$shortname} = 1;
|
---|
452 | }
|
---|
453 | }
|
---|
454 | }
|
---|
455 |
|
---|
456 | # Write out solr 'schema.xml' (and related) file
|
---|
457 | #
|
---|
458 | $self->make_final_field_list();
|
---|
459 | $self->premake_solr_auxiliary_files();
|
---|
460 |
|
---|
461 | # Now update the solr-core information in solr.xml
|
---|
462 | # => at most two cores <colname>-Doc and <colname>-Sec
|
---|
463 |
|
---|
464 | my $site = $self->{'site'};
|
---|
465 | my $collect = $self->{'collection'};
|
---|
466 | my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
|
---|
467 |
|
---|
468 | # my $idx = $self->{'index_mapping'}->{$index};
|
---|
469 | my $idx = "idx";
|
---|
470 |
|
---|
471 | my $build_dir = $self->{'build_dir'};
|
---|
472 |
|
---|
473 | foreach my $level (keys %{$self->{'levels'}}) {
|
---|
474 |
|
---|
475 | my ($pindex) = $level =~ /^(.)/;
|
---|
476 |
|
---|
477 | my $index_dir = $pindex.$idx;
|
---|
478 | my $core = "$core_prefix-$index_dir";
|
---|
479 |
|
---|
480 | # force_removeold == opposite of being run in 'incremental' mode
|
---|
481 | my $force_removeold = ($self->{'incremental'}) ? 0 : 1;
|
---|
482 |
|
---|
483 | if ($force_removeold) {
|
---|
484 | print $outhandle "\n-removeold set (new index will be created)\n";
|
---|
485 |
|
---|
486 | # create cores under temporary core names, corresponding to building directory
|
---|
487 | $core = "building-".$core;
|
---|
488 |
|
---|
489 | my $full_index_dir = &FileUtils::filenameConcatenate($build_dir,$index_dir);
|
---|
490 | &FileUtils::removeFilesRecursive($full_index_dir);
|
---|
491 | &FileUtils::makeDirectory($full_index_dir);
|
---|
492 |
|
---|
493 | my $full_tlog_dir = &FileUtils::filenameConcatenate($full_index_dir, "tlog");
|
---|
494 | &FileUtils::makeDirectory($full_tlog_dir);
|
---|
495 |
|
---|
496 | # Solr then wants an "index" folder within this general index area!
|
---|
497 | # my $full_index_index_dir = &FileUtils::filenameConcatenate($full_index_dir,"index");
|
---|
498 | # &FileUtils::makeDirectory($full_index_index_dir);
|
---|
499 |
|
---|
500 |
|
---|
501 | # now go on and create new index
|
---|
502 | print $outhandle "Creating Solr core: $core\n";
|
---|
503 | $solr_server->admin_create_core($core);
|
---|
504 |
|
---|
505 | }
|
---|
506 | else {
|
---|
507 | # if collect==core already in solr.xml (check with STATUS)
|
---|
508 | # => use RELOAD call to refresh fields now expressed in schema.xml
|
---|
509 | #
|
---|
510 | # else
|
---|
511 | # => use CREATE API to add to solr.xml
|
---|
512 |
|
---|
513 | my $check_core_exists = $solr_server->admin_ping_core($core);
|
---|
514 |
|
---|
515 | if ($check_core_exists) {
|
---|
516 | print $outhandle "Reloading Solr core: $core\n";
|
---|
517 | $solr_server->admin_reload_core($core);
|
---|
518 | }
|
---|
519 | else {
|
---|
520 | print $outhandle "Creating Solr core: $core\n";
|
---|
521 | $solr_server->admin_create_core($core);
|
---|
522 | }
|
---|
523 | }
|
---|
524 | }
|
---|
525 |
|
---|
526 | }
|
---|
527 |
|
---|
528 | # Essentially the same as the lucenebuilder.pm version, only using solr_passes
|
---|
529 | # => refactor and make better use of inheritence
|
---|
530 |
|
---|
531 | sub build_index {
|
---|
532 | my $self = shift (@_);
|
---|
533 | my ($index,$llevel) = @_;
|
---|
534 | my $outhandle = $self->{'outhandle'};
|
---|
535 | my $build_dir = $self->{'build_dir'};
|
---|
536 |
|
---|
537 | # get the full index directory path and make sure it exists
|
---|
538 | my $indexdir = $self->{'index_mapping'}->{$index};
|
---|
539 | &FileUtils::makeAllDirectories(&FileUtils::filenameConcatenate($build_dir, $indexdir));
|
---|
540 |
|
---|
541 | # Find the perl script to call to run solr
|
---|
542 | my $solr_passes = $self->{'solr_passes'};
|
---|
543 | my $solr_passes_exe = $self->{'solr_passes_exe'};
|
---|
544 |
|
---|
545 | # define the section names for solrpasses
|
---|
546 | # define the section names and possibly the doc name for solrpasses
|
---|
547 | my $solr_passes_sections = $llevel;
|
---|
548 |
|
---|
549 | my $osextra = "";
|
---|
550 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
---|
551 | $build_dir =~ s@/@\\@g;
|
---|
552 | } else {
|
---|
553 | if ($outhandle ne "STDERR") {
|
---|
554 | # so solr_passes doesn't print to stderr if we redirect output
|
---|
555 | $osextra .= " 2>/dev/null";
|
---|
556 | }
|
---|
557 | }
|
---|
558 |
|
---|
559 | # get the index expression if this index belongs
|
---|
560 | # to a subcollection
|
---|
561 | my $indexexparr = [];
|
---|
562 | my $langarr = [];
|
---|
563 |
|
---|
564 | # there may be subcollection info, and language info.
|
---|
565 | my ($fields, $subcollection, $language) = split (":", $index);
|
---|
566 | my @subcollections = ();
|
---|
567 | @subcollections = split /,/, $subcollection if (defined $subcollection);
|
---|
568 |
|
---|
569 | foreach $subcollection (@subcollections) {
|
---|
570 | if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
|
---|
571 | push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
|
---|
572 | }
|
---|
573 | }
|
---|
574 |
|
---|
575 | # add expressions for languages if this index belongs to
|
---|
576 | # a language subcollection - only put languages expressions for the
|
---|
577 | # ones we want in the index
|
---|
578 | my @languages = ();
|
---|
579 | my $languagemetadata = "Language";
|
---|
580 | if (defined ($self->{'collect_cfg'}->{'languagemetadata'})) {
|
---|
581 | $languagemetadata = $self->{'collect_cfg'}->{'languagemetadata'};
|
---|
582 | }
|
---|
583 | @languages = split /,/, $language if (defined $language);
|
---|
584 | foreach my $language (@languages) {
|
---|
585 | my $not=0;
|
---|
586 | if ($language =~ s/^\!//) {
|
---|
587 | $not = 1;
|
---|
588 | }
|
---|
589 | if($not) {
|
---|
590 | push (@$langarr, "!$language");
|
---|
591 | } else {
|
---|
592 | push (@$langarr, "$language");
|
---|
593 | }
|
---|
594 | }
|
---|
595 |
|
---|
596 | # Build index dictionary. Uses verbatim stem method
|
---|
597 | print $outhandle "\n creating index dictionary (solr_passes -I1)\n" if ($self->{'verbosity'} >= 1);
|
---|
598 | print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
|
---|
599 | my ($handle);
|
---|
600 |
|
---|
601 | if ($self->{'debug'}) {
|
---|
602 | $handle = *STDOUT;
|
---|
603 | } else {
|
---|
604 | my $site = $self->{'site'};
|
---|
605 | my $collect = $self->{'collection'};
|
---|
606 | my $core_prefix = (defined $site) ? "$site-$collect" : $collect;
|
---|
607 | my $ds_idx = $self->{'index_mapping'}->{$index};
|
---|
608 | my $core = "$core_prefix-$ds_idx";
|
---|
609 |
|
---|
610 | $core = "building-".$core unless $self->{'incremental'}; # core points to building only for force_removeold
|
---|
611 |
|
---|
612 | print STDERR "Cmd: $solr_passes_exe $core index \"$build_dir\" \"$indexdir\" $osextra\n";
|
---|
613 | if (!open($handle, "| $solr_passes_exe $core index \"$build_dir\" \"$indexdir\" $osextra")) {
|
---|
614 | print STDERR "<FatalError name='NoRunSolrPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
615 | die "solrbuilder::build_index - couldn't run $solr_passes_exe\n!$\n";
|
---|
616 | }
|
---|
617 | }
|
---|
618 |
|
---|
619 | my $store_levels = $self->{'levels'};
|
---|
620 | my $db_level = "section"; #always
|
---|
621 | my $dom_level = "";
|
---|
622 | foreach my $key (keys %$store_levels) {
|
---|
623 | if ($mgppbuilder::level_map{$key} eq $llevel) {
|
---|
624 | $dom_level = $key;
|
---|
625 | }
|
---|
626 | }
|
---|
627 | if ($dom_level eq "") {
|
---|
628 | print STDERR "Warning: unrecognized tag level $llevel\n";
|
---|
629 | $dom_level = "document";
|
---|
630 | }
|
---|
631 |
|
---|
632 | my $local_levels = { $dom_level => 1 }; # work on one level at a time
|
---|
633 |
|
---|
634 | # set up the document processr
|
---|
635 | $self->{'buildproc'}->set_output_handle ($handle);
|
---|
636 | $self->{'buildproc'}->set_mode ('text');
|
---|
637 | $self->{'buildproc'}->set_index ($index, $indexexparr);
|
---|
638 | $self->{'buildproc'}->set_index_languages ($languagemetadata, $langarr) if (defined $language);
|
---|
639 | $self->{'buildproc'}->set_indexing_text (1);
|
---|
640 | #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
641 | $self->{'buildproc'}->set_levels ($local_levels);
|
---|
642 | if (defined $self->{'collect_cfg'}->{'sortfields'}) {
|
---|
643 | $self->{'buildproc'}->set_sortfields ($self->{'collect_cfg'}->{'sortfields'});
|
---|
644 | }
|
---|
645 | if (defined $self->{'collect_cfg'}->{'facetfields'}) {
|
---|
646 | $self->{'buildproc'}->set_facetfields ($self->{'collect_cfg'}->{'facetfields'});
|
---|
647 | }
|
---|
648 | $self->{'buildproc'}->set_db_level($db_level);
|
---|
649 | $self->{'buildproc'}->reset();
|
---|
650 |
|
---|
651 | print $handle "<update>\n";
|
---|
652 |
|
---|
653 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
654 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
655 |
|
---|
656 | print $handle "</update>\n";
|
---|
657 |
|
---|
658 | close ($handle) unless $self->{'debug'};
|
---|
659 |
|
---|
660 | $self->print_stats();
|
---|
661 |
|
---|
662 | $self->{'buildproc'}->set_levels ($store_levels);
|
---|
663 | print STDERR "</Stage>\n" if $self->{'gli'};
|
---|
664 |
|
---|
665 | }
|
---|
666 |
|
---|
667 |
|
---|
668 | sub post_build_indexes {
|
---|
669 | my $self = shift(@_);
|
---|
670 |
|
---|
671 | # deliberately override to prevent the mgpp post_build_index() calling
|
---|
672 | # $self->make_final_field_list()
|
---|
673 | # as this has been done in our pre_build_indexes() phase for solr
|
---|
674 |
|
---|
675 |
|
---|
676 | # Also need to stop the Solr server (be it tomcat or jetty) if it was explicitly started
|
---|
677 | # in pre_build_indexes()
|
---|
678 |
|
---|
679 | my $solr_server = $self->{'solr_server'};
|
---|
680 |
|
---|
681 | if ($solr_server->explicitly_started()) {
|
---|
682 | $solr_server->stop();
|
---|
683 | }
|
---|
684 |
|
---|
685 | $self->{'solr_server'} = undef;
|
---|
686 |
|
---|
687 | }
|
---|
688 |
|
---|
689 | sub build_cfg_extra {
|
---|
690 | my $self = shift (@_);
|
---|
691 | my ($build_cfg) = @_;
|
---|
692 |
|
---|
693 | $self->lucenebuilder::build_cfg_extra($build_cfg);
|
---|
694 |
|
---|
695 | # need to add in facet stuff
|
---|
696 | my @facetfields = ();
|
---|
697 | my @facetfieldmap = ();
|
---|
698 |
|
---|
699 | foreach my $sf (@{$self->{'buildproc'}->{'facetfields'}}) {
|
---|
700 | if ($sf eq "rank") {
|
---|
701 | push(@facetfields, $sf);
|
---|
702 | } elsif ($self->{'buildproc'}->{'actualsortfields'}->{$sf}) {
|
---|
703 | my $shortname = $self->{'buildproc'}->{'sortfieldnamemap'}->{$sf};
|
---|
704 | push(@facetfields, $shortname);
|
---|
705 | push (@facetfieldmap, "$sf\-\>$shortname");
|
---|
706 | }
|
---|
707 |
|
---|
708 | }
|
---|
709 | $build_cfg->{'indexfacetfields'} = \@facetfields;
|
---|
710 | $build_cfg->{'indexfacetfieldmap'} = \@facetfieldmap;
|
---|
711 | }
|
---|
712 | 1;
|
---|
713 |
|
---|
714 |
|
---|