source: gsdl/trunk/perllib/mgppbuilder.pm@ 14666

Last change on this file since 14666 was 14666, checked in by anna, 15 years ago

Undoing commit of r14657.

  • Property svn:keywords set to Author Date Id Revision
File size: 28.4 KB
Line 
1###########################################################################
2#
3# mgppbuilder.pm -- MGBuilder object
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
26package mgppbuilder;
27
28use basebuilder;
29use classify;
30use cfgread;
31use colcfg;
32use plugin;
33use util;
34use FileHandle;
35
36sub BEGIN {
37 @mgppbuilder::ISA = ('basebuilder');
38}
39
40
41
42our %level_map = ('document'=>'Doc',
43 'section'=>'Sec',
44 'paragraph'=>'Para',
45 'Doc'=>'_textdocument_',
46 'Sec'=>'_textsection_',
47 'Para'=>'_textparagraph_');
48
49our %wanted_index_files = ('td'=>1,
50 't'=>1,
51 'tl'=>1,
52 'ti'=>1,
53 'idb'=>1,
54 'ib1'=>1,
55 'ib2'=>1,
56 'ib3'=>1,
57 'ib4'=>1,
58 'ib5'=>1,
59 'ib6'=>1,
60 'ib7'=>1,
61 'i'=>1,
62 'il'=>1,
63 'w'=>1,
64 'wa'=>1);
65
66# change this so a user can add their own ones in via a file or cfg
67#add AND, OR, NOT NEAR to this list - these cannot be used as field names
68#also add the level names (Doc, Sec, Para)
69our %static_indexfield_map = ('Title'=>'TI',
70 'TI'=>1,
71 'Subject'=>'SU',
72 'SU'=>1,
73 'Creator'=>'CR',
74 'CR'=>1,
75 'Organization'=>'ORG',
76 'ORG'=>1,
77 'Source'=>'SO',
78 'SO'=>1,
79 'Howto'=>'HT',
80 'HT'=>1,
81 'ItemTitle'=>'IT',
82 'IT'=>1,
83 'ProgNumber'=>'PN',
84 'PN'=>1,
85 'People'=>'PE',
86 'PE'=>1,
87 'Coverage'=>'CO',
88 'CO'=>1,
89 'allfields'=>'ZZ',
90 'ZZ'=>1,
91 'text'=>'TX',
92 'TX'=>1,
93 'AND'=>1,
94 'OR'=>1,
95 'NOT'=>1,
96 'NEAR'=>1,
97 'Doc'=>1,
98 'Sec'=>1,
99 'Para'=>1);
100
101my $maxdocsize = $basebuilder::maxdocsize;
102
103sub new {
104 my $class = shift(@_);
105
106 my $self = new basebuilder (@_);
107 $self = bless $self, $class;
108
109 $self->{'indexfieldmap'} = \%static_indexfield_map;
110
111 # get the levels (Section, Paragraph) for indexing and compression
112 $self->{'levels'} = {};
113 $self->{'levelorder'} = ();
114 if (defined $self->{'collect_cfg'}->{'levels'}) {
115 foreach my $level ( @{$self->{'collect_cfg'}->{'levels'}} ){
116 $level =~ tr/A-Z/a-z/;
117 $self->{'levels'}->{$level} = 1;
118 push (@{$self->{'levelorder'}}, $level);
119 }
120 } else { # default to document
121 $self->{'levels'}->{'document'} = 1;
122 push (@{$self->{'levelorder'}}, 'document');
123 }
124
125 $self->{'buildtype'} = "mgpp";
126
127 return $self;
128}
129
130sub generate_index_list {
131 my $self = shift (@_);
132
133 # sort out the indexes
134 #indexes are specified with spaces, but we put them into one index
135 my $indexes = $self->{'collect_cfg'}->{'indexes'};
136 $self->{'collect_cfg'}->{'indexes'} = [];
137 push (@{$self->{'collect_cfg'}->{'indexes'}}, join(';', @$indexes).";");
138}
139
140sub generate_index_options {
141 my $self = shift (@_);
142
143 $self->{'casefold'} = 0;
144 $self->{'stem'} = 0;
145 $self->{'accentfold'} = 0;
146
147 if (!defined($self->{'collect_cfg'}->{'indexoptions'})) {
148 # just use default options
149 $self->{'casefold'} = 1;
150 $self->{'stem'} = 1;
151 $self->{'accentfold'} = 1;
152 } else {
153 foreach my $option (@{$self->{'collect_cfg'}->{'indexoptions'}}) {
154 if ($option =~ /stem/) {
155 $self->{'stem'} = 1;
156 } elsif ($option =~ /casefold/) {
157 $self->{'casefold'} = 1;
158 } elsif ($option =~ /accentfold/) {
159 $self->{'accentfold'} = 1;
160 }
161 }
162 }
163
164 # now we record this for the build cfg
165 $self->{'stemindexes'} = 0;
166 if ($self->{'casefold'}) {
167 $self->{'stemindexes'} += 1;
168 }
169 if ($self->{'stem'}) {
170 $self->{'stemindexes'} += 2;
171 }
172 if ($self->{'accentfold'}) {
173 $self->{'stemindexes'} += 4;
174 }
175
176}
177
178sub default_buildproc {
179 my $self = shift (@_);
180
181 return "mgppbuildproc";
182}
183
184sub compress_text {
185
186 my $self = shift (@_);
187
188 # we don't do anything if we don't want compressed text
189 return if $self->{'no_text'};
190
191 my ($textindex) = @_;
192
193 my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
194 my $exe = &util::get_os_exe ();
195 my $mgpp_passes_exe = &util::filename_cat($exedir, "mgpp_passes$exe");
196 my $mgpp_compression_dict_exe = &util::filename_cat($exedir, "mgpp_compression_dict$exe");
197 my $outhandle = $self->{'outhandle'};
198
199 my $maxnumeric = $self->{'maxnumeric'};
200
201 &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, "text"));
202
203 my $basefilename = "text/$self->{'collection'}";
204 my $fulltextprefix = &util::filename_cat ($self->{'build_dir'}, $basefilename);
205
206 my $osextra = "";
207 if ($ENV{'GSDLOS'} =~ /^windows$/i) {
208 $fulltextprefix =~ s@/@\\@g;
209 }
210 else {
211 $osextra = " -d /";
212 }
213
214
215 # define the section names and possibly the doc name for mgpasses
216 # the compressor doesn't need to know about paragraphs - never want to
217 # retrieve them
218
219 # always use Doc and Sec levels
220 my $mgpp_passes_sections = "-J ". $level_map{"document"} ." -K " . $level_map{"section"} ." ";
221
222 print $outhandle "\n*** creating the compressed text\n" if ($self->{'verbosity'} >= 1);
223 print STDERR "<Stage name='CompressText'>\n" if $self->{'gli'};
224
225 # collect the statistics for the text
226 # -b $maxdocsize sets the maximum document size to be 12 meg
227 print $outhandle "\n collecting text statistics (mgpp_passes -T1)\n" if ($self->{'verbosity'} >= 1);
228 print STDERR "<Phase name='CollectTextStats'/>\n" if $self->{'gli'};
229
230 my ($handle);
231 if ($self->{'debug'}) {
232 $handle = STDOUT;
233 } else {
234 if (!-e "$mgpp_passes_exe" ||
235 !open (PIPEOUT, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fulltextprefix\" -T1 $osextra")) {
236 print STDERR "<FatalError name='NoRunMGPasses'>\n</Stage>\n" if $self->{'gli'};
237 die "mgppbuilder::compress_text - couldn't run $mgpp_passes_exe\n";
238 }
239 $handle = mgppbuilder::PIPEOUT;
240 }
241
242 my $gdbm_level = "section";
243
244 $self->{'buildproc'}->set_output_handle ($handle);
245 $self->{'buildproc'}->set_mode ('text');
246 $self->{'buildproc'}->set_index ($textindex);
247 $self->{'buildproc'}->set_indexing_text (0);
248 $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
249 $self->{'buildproc'}->set_levels ($self->{'levels'});
250 $self->{'buildproc'}->set_gdbm_level ($gdbm_level);
251 $self->{'buildproc'}->reset();
252 &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
253 $self->{'buildproc'}, $self->{'maxdocs'});
254 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
255 "", {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
256 &plugin::end($self->{'pluginfo'});
257 close (PIPEOUT);
258
259 close ($handle) unless $self->{'debug'};
260
261 $self->print_stats();
262
263 # create the compression dictionary
264 # the compression dictionary is built by assuming the stats are from a seed
265 # dictionary (-S), if a novel word is encountered it is spelled out (-H),
266 # and the resulting dictionary must be less than 5 meg with the most
267 # frequent words being put into the dictionary first (-2 -k 5120)
268 # note: these options are left over from mg version
269 if (!$self->{'debug'}) {
270 print $outhandle "\n creating the compression dictionary\n" if ($self->{'verbosity'} >= 1);
271 print STDERR "<Phase name='CreatingCompress'/>\n" if $self->{'gli'};
272 if (!-e "$mgpp_compression_dict_exe") {
273 print STDERR "<FatalError name='NoRunMGCompress'/>\n</Stage>\n" if $self->{'gli'};
274 die "mgppbuilder::compress_text - couldn't run $mgpp_compression_dict_exe\n";
275 }
276 system ("mgpp_compression_dict$exe -f \"$fulltextprefix\" -S -H -2 -k 5120 $osextra");
277
278 if (!$self->{'debug'}) {
279 if (!-e "$mgpp_passes_exe" ||
280 !open ($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fulltextprefix\" -T2 $osextra")) {
281 print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
282 die "mgppbuilder::compress_text - couldn't run $mgpp_passes_exe\n";
283 }
284 }
285 }
286 else {
287 print STDERR "<Phase name='SkipCreatingComp'/>\n" if $self->{'gli'};
288 }
289
290 $self->{'buildproc'}->reset();
291 # compress the text
292 print $outhandle "\n compressing the text (mgpp_passes -T2)\n" if ($self->{'verbosity'} >= 1);
293 print STDERR "<Phase name='CompressingText'/>\n" if $self->{'gli'};
294
295 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
296 "", {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
297 close ($handle) unless $self->{'debug'};
298
299 $self->print_stats();
300 print STDERR "</Stage>\n" if $self->{'gli'};
301}
302
303
304sub build_indexes_extra {
305 my $self = shift(@_);
306 #define the final field lists
307 $self->make_final_field_list();
308}
309
310# creates directory names for each of the index descriptions
311sub create_index_mapping {
312 my $self = shift (@_);
313 my ($indexes) = @_;
314
315 my %mapping = ();
316
317 $mapping{'indexmaporder'} = [];
318 $mapping{'subcollectionmaporder'} = [];
319 $mapping{'languagemaporder'} = [];
320
321 # dirnames is used to check for collisions. Start this off
322 # with the manditory directory names
323 my %dirnames = ('text'=>'text',
324 'extra'=>'extra');
325 my %pnames = ('index' => {}, 'subcollection' => {}, 'languages' => {});
326
327 foreach my $index (@$indexes) {
328 my ($fields, $subcollection, $languages) = split (":", $index);
329
330 # we only ever have one index, and its called 'idx'
331 my $pindex = 'idx';
332
333 # next comes a processed version of the subcollection if there is one.
334 my $psub = $self->process_field ($subcollection);
335 $psub = lc ($psub);
336
337 # next comes a processed version of the language if there is one.
338 my $plang = $self->process_field ($languages);
339 $plang = lc ($plang);
340
341 my $dirname = $pindex . $psub . $plang;
342
343 # check to be sure all index names are unique
344 while (defined ($dirnames{$dirname})) {
345 $dirname = $self->make_unique (\%pnames, $index, \$pindex, \$psub, \$plang);
346 }
347
348 $mapping{$index} = $dirname;
349
350 # store the mapping orders as well as the maps
351 # also put index, subcollection and language fields into the mapping thing -
352 # (the full index name (eg text:subcol:lang) is not used on
353 # the query page) -these are used for collectionmeta later on
354 if (!defined $mapping{'indexmap'}{"$fields"}) {
355 $mapping{'indexmap'}{"$fields"} = $pindex;
356 push (@{$mapping{'indexmaporder'}}, "$fields");
357 if (!defined $mapping{"$fields"}) {
358 $mapping{"$fields"} = $pindex;
359 }
360 }
361 if ($psub =~ /\w/ && !defined ($mapping{'subcollectionmap'}{$subcollection})) {
362 $mapping{'subcollectionmap'}{$subcollection} = $psub;
363 push (@{$mapping{'subcollectionmaporder'}}, $subcollection);
364 $mapping{$subcollection} = $psub;
365 }
366 if ($plang =~ /\w/ && !defined ($mapping{'languagemap'}{$languages})) {
367 $mapping{'languagemap'}{$languages} = $plang;
368 push (@{$mapping{'languagemaporder'}}, $languages);
369 $mapping{$languages} = $plang;
370 }
371 $dirnames{$dirname} = $index;
372 $pnames{'index'}->{$pindex} = "$fields";
373 $pnames{'subcollection'}->{$psub} = $subcollection;
374 $pnames{'languages'}->{$plang} = $languages;
375 }
376
377 return \%mapping;
378}
379
380sub make_unique {
381 my $self = shift (@_);
382 my ($namehash, $index, $indexref, $subref, $langref) = @_;
383 my ($fields, $subcollection, $languages) = split (":", $index);
384
385 if ($namehash->{'index'}->{$$indexref} ne "$fields") {
386 $self->get_next_version ($indexref);
387 } elsif ($namehash->{'subcollection'}->{$$subref} ne $subcollection) {
388 $self->get_next_version ($subref);
389 } elsif ($namehash->{'languages'}->{$$langref} ne $languages) {
390 $self->get_next_version ($langref);
391 }
392 return "$$indexref$$subref$$langref";
393}
394
395
396sub build_index {
397 my $self = shift (@_);
398 my ($index) = @_;
399 my $outhandle = $self->{'outhandle'};
400
401 # get the full index directory path and make sure it exists
402 my $indexdir = $self->{'index_mapping'}->{$index};
403 &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, $indexdir));
404 my $fullindexprefix = &util::filename_cat ($self->{'build_dir'},
405 $indexdir,
406 $self->{'collection'});
407 my $fulltextprefix = &util::filename_cat ($self->{'build_dir'}, "text",
408 $self->{'collection'});
409
410 # get any os specific stuff
411 my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
412
413 my $exe = &util::get_os_exe ();
414 my $mgpp_passes_exe = &util::filename_cat($exedir, "mgpp_passes$exe");
415
416 # define the section names for mgpasses
417 my $mgpp_passes_sections = "-J ". $level_map{"document"} ." -K " . $level_map{"section"} ." ";
418 if ($self->{'levels'}->{'paragraph'}) {
419 $mgpp_passes_sections .= "-K " . $level_map{'paragraph'}. " ";
420 }
421
422 my $mgpp_perf_hash_build_exe =
423 &util::filename_cat($exedir, "mgpp_perf_hash_build$exe");
424 my $mgpp_weights_build_exe =
425 &util::filename_cat ($exedir, "mgpp_weights_build$exe");
426 my $mgpp_invf_dict_exe =
427 &util::filename_cat ($exedir, "mgpp_invf_dict$exe");
428 my $mgpp_stem_idx_exe =
429 &util::filename_cat ($exedir, "mgpp_stem_idx$exe");
430
431 my $maxnumeric = $self->{'maxnumeric'};
432
433 my $osextra = "";
434 if ($ENV{'GSDLOS'} =~ /^windows$/i) {
435 $fullindexprefix =~ s@/@\\@g;
436 } else {
437 $osextra = " -d /";
438 if ($outhandle ne "STDERR") {
439 # so mgpp_passes doesn't print to stderr if we redirect output
440 $osextra .= " 2>/dev/null";
441 }
442 }
443
444 # get the index expression if this index belongs
445 # to a subcollection
446 my $indexexparr = [];
447 my $langarr = [];
448 # there may be subcollection info, and language info.
449 my ($fields, $subcollection, $language) = split (":", $index);
450 my @subcollections = ();
451 @subcollections = split /,/, $subcollection if (defined $subcollection);
452
453 foreach $subcollection (@subcollections) {
454 if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
455 push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
456 }
457 }
458
459 # add expressions for languages if this index belongs to
460 # a language subcollection - only put languages expressions for the
461 # ones we want in the index
462
463 my @languages = ();
464 my $language_metadata = "Language";
465 if (defined ($self->{'collect_cfg'}->{'language_metadata'})) {
466 $language_metadata = $self->{'collect_cfg'}->{'language_metadata'};
467 }
468 @languages = split /,/, $language if (defined $language);
469 foreach my $language (@languages) {
470 my $not=0;
471 if ($language =~ s/^\!//) {
472 $not = 1;
473 }
474 if($not) {
475 push (@$langarr, "!$language");
476 } else {
477 push (@$langarr, "$language");
478 }
479 }
480
481 # Build index dictionary. Uses verbatim stem method
482 print $outhandle "\n creating index dictionary (mgpp_passes -I1)\n" if ($self->{'verbosity'} >= 1);
483 print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
484 my ($handle);
485 if ($self->{'debug'}) {
486 $handle = STDOUT;
487 } else {
488 if (!-e "$mgpp_passes_exe" ||
489 !open (PIPEOUT, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fullindexprefix\" -I1 $osextra")) {
490 print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
491 die "mgppbuilder::build_index - couldn't run $mgpp_passes_exe\n";
492 }
493 $handle = mgppbuilder::PIPEOUT;
494 }
495
496 # gdbm_level is always section
497 my $gdbm_level = "section";
498
499 # set up the document processr
500 $self->{'buildproc'}->set_output_handle ($handle);
501 $self->{'buildproc'}->set_mode ('text');
502 $self->{'buildproc'}->set_index ($index, $indexexparr);
503 $self->{'buildproc'}->set_index_languages ($language_metadata, $langarr) if (defined $language);
504 $self->{'buildproc'}->set_indexing_text (1);
505 $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
506 $self->{'buildproc'}->set_levels ($self->{'levels'});
507 $self->{'buildproc'}->set_gdbm_level ($gdbm_level);
508
509 $self->{'buildproc'}->reset();
510 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
511 "", {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
512 close ($handle) unless $self->{'debug'};
513
514 $self->print_stats();
515
516 # now we check to see if the required files have been produced - if not we quit building this index so the whole process doesn't crap out.
517 # we check on the .id file - index dictionary
518 my $dict_file = "$fullindexprefix.id";
519 if (!-e $dict_file) {
520 print $outhandle "mgppbuilder::build_index - Couldn't create index $index\n";
521 print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
522 $self->{'notbuilt'}->{$index}=1;
523 return;
524 }
525
526 if (!$self->{'debug'}) {
527 # create the perfect hash function
528 if (!-e "$mgpp_perf_hash_build_exe") {
529 print STDERR "<FatalError name='NoRunMGHash'/>\n</Stage>\n" if $self->{'gli'};
530 die "mgppbuilder::build_index - couldn't run $mgpp_perf_hash_build_exe\n";
531 }
532 system ("mgpp_perf_hash_build$exe -f \"$fullindexprefix\" $osextra");
533
534 if (!-e "$mgpp_passes_exe" ||
535 !open ($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fullindexprefix\" -I2 $osextra")) {
536 print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
537 die "mgppbuilder::build_index - couldn't run $mgpp_passes_exe\n";
538 }
539 }
540
541 # invert the text
542 print $outhandle "\n inverting the text (mgpp_passes -I2)\n" if ($self->{'verbosity'} >= 1);
543 print STDERR "<Phase name='InvertingText'/>\n" if $self->{'gli'};
544 $self->{'buildproc'}->reset();
545 &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
546 "", {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
547
548 $self->print_stats ();
549
550 if (!$self->{'debug'}) {
551
552 close ($handle);
553
554 # create the weights file
555 print $outhandle "\n create the weights file\n" if ($self->{'verbosity'} >= 1);
556 print STDERR "<Phase name='CreateTheWeights'/>\n" if $self->{'gli'};
557 if (!-e "$mgpp_weights_build_exe") {
558 print STDERR "<FatalError name='NoRunMGWeights'/>\n</Stage>\n" if $self->{'gli'};
559 die "mgppbuilder::build_index - couldn't run $mgpp_weights_build_exe\n";
560 }
561 system ("mgpp_weights_build$exe -f \"$fullindexprefix\" $osextra");
562
563 # create 'on-disk' stemmed dictionary
564 print $outhandle "\n creating 'on-disk' stemmed dictionary\n" if ($self->{'verbosity'} >= 1);
565 if (!-e "$mgpp_invf_dict_exe") {
566 print STDERR "<FatalError name='NoRunMGInvf'/>\n</Stage>\n" if $self->{'gli'};
567 die "mgppbuilder::build_index - couldn't run $mgpp_invf_dict_exe\n";
568 }
569 system ("mgpp_invf_dict$exe -f \"$fullindexprefix\" $osextra" );
570
571
572 # creates stem index files for the various stemming methods
573 print $outhandle "\n creating stem indexes\n" if ($self->{'verbosity'} >= 1);
574 print STDERR "<Phase name='CreatingStemIndx'/>\n" if $self->{'gli'};
575 if (!-e "$mgpp_stem_idx_exe") {
576 print STDERR "<FatalError name='NoRunMGStem'/>\n</Stage>\n" if $self->{'gli'};
577 die "mgppbuilder::build_index - couldn't run $mgpp_stem_idx_exe\n";
578 }
579 my $accent_folding_enabled = 1;
580 if ($self->{'accentfold'}) {
581 # the first time we do this, we test for accent folding enabled
582 if (system ("mgpp_stem_idx$exe -b 4096 -s4 -f \"$fullindexprefix\" $osextra") == 2) {
583 # accent folding has not been enabled in mgpp
584 $accent_folding_enabled = 0;
585 $self->{'stemindexes'} -= 4;
586 }
587 }
588 if ($self->{'casefold'}) {
589 system ("mgpp_stem_idx$exe -b 4096 -s1 -f \"$fullindexprefix\" $osextra");
590 if ($accent_folding_enabled && $self->{'accentfold'}) {
591 system ("mgpp_stem_idx$exe -b 4096 -s5 -f \"$fullindexprefix\" $osextra");
592 }
593 }
594 if ($self->{'stem'}) {
595 system ("mgpp_stem_idx$exe -b 4096 -s2 -f \"$fullindexprefix\" $osextra");
596 if ($accent_folding_enabled && $self->{'accentfold'}) {
597 system ("mgpp_stem_idx$exe -b 4096 -s6 -f \"$fullindexprefix\" $osextra");
598 }
599 }
600 if ($self->{'casefold'} && $self->{'stem'}) {
601 system ("mgpp_stem_idx$exe -b 4096 -s3 -f \"$fullindexprefix\" $osextra");
602 if ($accent_folding_enabled && $self->{'accentfold'}) {
603 system ("mgpp_stem_idx$exe -b 4096 -s7 -f \"$fullindexprefix\" $osextra");
604 }
605 }
606
607 # remove unwanted files
608 my $tmpdir = &util::filename_cat ($self->{'build_dir'}, $indexdir);
609 opendir (DIR, $tmpdir) || die
610 "mgppbuilder::build_index - couldn't read directory $tmpdir\n";
611 foreach my $file (readdir(DIR)) {
612 next if $file =~ /^\./;
613 my ($suffix) = $file =~ /\.([^\.]+)$/;
614 if (defined $suffix && !defined $wanted_index_files{$suffix}) {
615 # delete it!
616 print $outhandle "deleting $file\n" if $self->{'verbosity'} > 2;
617 #&util::rm (&util::filename_cat ($tmpdir, $file));
618 }
619 }
620 closedir (DIR);
621 }
622 print STDERR "</Stage>\n" if $self->{'gli'};
623}
624
625# now only outputs stuff if you can't generate it from collectionmeta - e.g. if someone has specified 'metadata' as an index.
626sub output_collection_meta {
627 my $self = shift(@_);
628 my ($handle) = @_;
629
630 # define the indexed field mapping if not already done so (ie if infodb called separately from build_index)
631 if (!defined $self->{'build_cfg'}) {
632 $self->read_final_field_list();
633 }
634
635 # do the collection info
636 print $handle "[collection]\n";
637
638 # first do the collection meta stuff - everything without a dot
639 my $collmetadefined = 0;
640 my $metadata_entry;
641 if (defined $self->{'collect_cfg'}->{'collectionmeta'}) {
642 $collmetadefined = 1;
643 }
644
645 #add the index field macros to [collection]
646 # eg <TI>Title
647 # <SU>Subject
648 # these now come from collection meta. if that is not defined, usses the metadata name
649 my $field_entry="";
650 my $collmeta = "";
651 foreach my $longfield (@{$self->{'build_cfg'}->{'indexfields'}}){
652 my $shortfield = $self->{'buildproc'}->{'indexfieldmap'}->{$longfield};
653 next if $shortfield eq 1;
654
655 # we need to check if some coll meta has been defined - don't output
656 # any that have
657 $collmeta = ".$longfield";
658 if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{$collmeta}) {
659 if ($longfield eq "allfields") {
660 $field_entry .= "<$shortfield>_query:textallfields_\n";
661 } elsif ($longfield eq "text") {
662 $field_entry .= "<$shortfield>_query:texttextonly_\n";
663 } else {
664 $field_entry .= "<$shortfield>$longfield\n";
665 }
666 }
667 }
668 print $handle $field_entry;
669
670 # now add the level names
671 my $level_entry = "";
672 foreach my $level (@{$self->{'collect_cfg'}->{'levels'}}) {
673 $collmeta = ".$level"; # based on the original specification
674 $level =~ tr/A-Z/a-z/; # make it lower case
675 my $levelid = $level_map{$level}; # find the actual value we used in the index
676 if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{$collmeta}) {
677 # use the default macro
678 $level_entry .= "<$levelid>" . $level_map{$levelid} . "\n";
679 }
680 }
681 print $handle $level_entry;
682
683 # now add subcoll meta
684 my $subcoll_entry = "";
685 my $shortname = "";
686 my $one_entry = "";
687 foreach my $subcoll (@{$self->{'index_mapping'}->{'subcollectionmaporder'}}) {
688 $shortname = $self->{'index_mapping'}->{$subcoll};
689 if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{".$subcoll"}) {
690 $subcoll_entry .= "<$shortname>$subcoll\n";
691 }
692 }
693 print $handle $subcoll_entry;
694
695 # now add language meta
696 my $lang_entry = "";
697 foreach my $lang (@{$self->{'index_mapping'}->{'languagemaporder'}}) {
698 $shortname = $self->{'index_mapping'}->{$lang};
699 if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{".$lang"}) {
700 $lang_entry .= "<$shortname>$lang\n";
701 }
702 }
703 print $handle $lang_entry;
704 # end the collection entry
705 print $handle "\n" . ('-' x 70) . "\n";
706
707
708}
709
710# at the end of building, we have an indexfieldmap with all the mappings,
711# plus some extras, and indexmap with any indexes in it that weren't
712# specified in the index definition. we want to make an ordered list of
713# fields that are indexed, and a list of mappings that are used. this will
714# be used for the build.cfg file, and for collection meta definition we
715# store these in a build.cfg bit
716sub make_final_field_list {
717 my $self = shift (@_);
718
719 $self->{'build_cfg'} = {};
720
721 # store the indexfieldmap information
722 my @indexfieldmap = ();
723 my @indexfields = ();
724 my $specifiedfields = {};
725 my @specifiedfieldorder = ();
726
727 # go through the index definition and add each thing to a map, so we
728 # can easily check if it is already specified - when doing the
729 # metadata, we print out all the individual fields, but some may
730 # already be specified in the index definition, so we dont want to add
731 # those again.
732
733 my $field;
734 foreach $field (@{$self->{'collect_cfg'}->{'indexes'}}) {
735 # remove subcoll stuff
736 my $parts = $field;
737 $parts =~ s/:.*$//;
738 # *************
739 my @fs = split(';', $parts);
740 foreach my $f(@fs) {
741 if (!defined $specifiedfields->{$f}) {
742 $specifiedfields->{$f}=1;
743 push (@specifiedfieldorder, "$f");
744 }
745 }
746 }
747
748 #add all fields bit
749 foreach $field (@specifiedfieldorder) {
750 if ($field eq "metadata") {
751 foreach my $newfield (keys %{$self->{'buildproc'}->{'indexfields'}}) {
752 if (!defined $specifiedfields->{$newfield}) {
753 push (@indexfieldmap, "$newfield\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$newfield}");
754 push (@indexfields, "$newfield");
755 }
756 }
757
758 } elsif ($field eq 'text') {
759 push (@indexfieldmap, "text\-\>TX");
760 push (@indexfields, "text");
761 } elsif ($field eq 'allfields') {
762 push (@indexfieldmap, "allfields\-\>ZZ");
763 push (@indexfields, "allfields");
764 } else {
765
766 my $ifm = $self->{'buildproc'}->{'indexfieldmap'};
767
768 if (defined $ifm->{$field}) {
769 push (@indexfieldmap, "$field\-\>$ifm->{$field}");
770 push (@indexfields, "$field");
771 }
772
773
774 }
775 }
776
777 $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
778 $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
779
780}
781
782
783# recreate the field list from the build.cfg file, look first in building,
784# then in index to find it. if there is no build.cfg, we can't do the field
785# list (there is unlikely to be any index anyway.)
786sub read_final_field_list {
787 my $self = shift (@_);
788 $self->{'build_cfg'} = {};
789 my @indexfieldmap = ();
790 my @indexfields = ();
791 my @indexmap = ();
792
793 if (scalar(keys %{$self->{'buildproc'}->{'indexfieldmap'}}) == 0) {
794 # set the default mapping
795 $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
796 }
797 # we read the stuff in from the build.cfg file - if its there
798 my $buildconfigfile = &util::filename_cat($self->{'build_dir'}, "build.cfg");
799
800 if (!-e $buildconfigfile) {
801 # try the index dir - but do we know where it is?? try here
802 $buildconfigfile = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "index", "build.cfg");
803 if (!-e $buildconfigfile) {
804 #we cant find a config file - just ignore the field list
805 return;
806 }
807 }
808
809 my $buildcfg = &colcfg::read_build_cfg( $buildconfigfile);
810 my $field;
811 if (defined $buildcfg->{'indexfields'}) {
812 foreach $field (@{$buildcfg->{'indexfields'}}) {
813 push (@indexfields, "$field");
814 }
815 }
816
817 if (defined $buildcfg->{'indexfieldmap'}) {
818 foreach $field (@{$buildcfg->{'indexfieldmap'}}) {
819 push (@indexfieldmap, "$field");
820 my ($f, $v) = $field =~ /^(.*)\-\>(.*)$/;
821 $self->{'buildproc'}->{'indexfieldmap'}->{$f} = $v;
822 }
823 }
824
825 if (defined $buildcfg->{'indexmap'}) {
826 foreach $field (@{$buildcfg->{'indexmap'}}) {
827 push (@indexmap, "$field");
828 }
829 }
830
831 $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
832 $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
833 $self->{'build_cfg'}->{'indexmap'} = \@indexmap;
834}
835
836
837sub build_cfg_extra {
838 my $self = shift (@_);
839 my ($build_cfg) = @_;
840
841 $build_cfg->{'numsections'} = $self->{'buildproc'}->get_num_sections();
842
843 # store the level info
844 my @indexlevels = ();
845 my @levelmap = ();
846 foreach my $l (@{$self->{'levelorder'}}) {
847 push (@indexlevels, $level_map{$l});
848 push (@levelmap, "$l\-\>$level_map{$l}");
849 }
850 $build_cfg->{'indexlevels'} = \@indexlevels;
851 $build_cfg->{'levelmap'} = \@levelmap;
852
853 # text level (and gdbm level) is always section
854 $build_cfg->{'textlevel'} = $level_map{'section'};
855
856}
857
8581;
859
860
Note: See TracBrowser for help on using the repository browser.