root/gsdl/trunk/perllib/mgppbuilder.pm @ 20418

Revision 20418, 27.6 KB (checked in by kjdon, 10 years ago)

language_metadata was sometimes with an underscore, and sometimes without. Have standardised to no underscore: languagemetadata

  • Property svn:keywords set to Author Date Id Revision
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 colcfg;
30use plugin;
31use strict; no strict 'refs';
32use util;
33
34
35sub BEGIN {
36    @mgppbuilder::ISA = ('basebuilder');
37}
38
39
40
41our %level_map = ('document'=>'Doc',
42          'section'=>'Sec',
43          'paragraph'=>'Para',
44          'Doc'=>'_textdocument_',
45          'Sec'=>'_textsection_',
46          'Para'=>'_textparagraph_');
47
48our %wanted_index_files = ('td'=>1,
49               't'=>1,
50               'tl'=>1,
51               'ti'=>1,
52               'idb'=>1,
53               'ib1'=>1,
54               'ib2'=>1,
55               'ib3'=>1,
56               'ib4'=>1,
57               'ib5'=>1,
58               'ib6'=>1,
59               'ib7'=>1,
60               'i'=>1,
61               'il'=>1,
62               'w'=>1,
63               'wa'=>1);
64
65
66my $maxdocsize = $basebuilder::maxdocsize;
67
68sub new {
69    my $class = shift(@_);
70
71    my $self = new basebuilder (@_);
72    $self = bless $self, $class;
73
74    #$self->{'indexfieldmap'} = \%static_indexfield_map;
75
76    # get the levels (Section, Paragraph) for indexing and compression
77    $self->{'levels'} = {};
78    $self->{'levelorder'} = ();
79    if (defined $self->{'collect_cfg'}->{'levels'}) {
80        foreach my $level ( @{$self->{'collect_cfg'}->{'levels'}} ){
81        $level =~ tr/A-Z/a-z/;
82            $self->{'levels'}->{$level} = 1;
83        push (@{$self->{'levelorder'}}, $level);
84        }
85    } else { # default to document
86    $self->{'levels'}->{'document'} = 1;
87    push (@{$self->{'levelorder'}}, 'document');
88    }
89   
90    $self->{'buildtype'} = "mgpp";
91
92    return $self;
93}
94
95sub generate_index_list {
96    my $self  = shift (@_);
97   
98    # sort out the indexes
99    #indexes are specified with spaces, but we put them into one index
100    my $indexes = $self->{'collect_cfg'}->{'indexes'};
101    if (defined $indexes) {
102    $self->{'collect_cfg'}->{'indexes'} = [];
103    push (@{$self->{'collect_cfg'}->{'indexes'}}, join(';', @$indexes).";");
104    }
105}
106
107sub generate_index_options {
108    my $self = shift (@_);
109
110    $self->SUPER::generate_index_options();
111
112    $self->{'casefold'} = 0;
113    $self->{'stem'} = 0;
114    $self->{'accentfold'} = 0;
115   
116    if (!defined($self->{'collect_cfg'}->{'indexoptions'})) {
117    # just use default options
118    $self->{'casefold'} = 1;
119    $self->{'stem'} = 1;
120    $self->{'accentfold'} = 1;
121    } else {
122    foreach my $option (@{$self->{'collect_cfg'}->{'indexoptions'}}) {
123        if ($option =~ /stem/) {
124        $self->{'stem'} = 1;
125        } elsif ($option =~ /casefold/) {
126        $self->{'casefold'} = 1;
127        } elsif ($option =~ /accentfold/) {
128        $self->{'accentfold'} = 1;
129        }
130    }
131    }
132   
133    # now we record this for the build cfg
134    $self->{'stemindexes'} = 0;
135    if ($self->{'casefold'}) {
136    $self->{'stemindexes'} += 1;
137    }
138    if ($self->{'stem'}) {
139    $self->{'stemindexes'} += 2;
140    }
141    if ($self->{'accentfold'}) {
142    $self->{'stemindexes'} += 4;
143    }
144   
145}
146
147sub default_buildproc {
148    my $self  = shift (@_);
149
150    return "mgppbuildproc";
151}
152
153sub compress_text {
154
155    my $self = shift (@_);
156
157    # we don't do anything if we don't want compressed text
158    return if $self->{'no_text'};
159   
160    my ($textindex) = @_;
161
162    my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
163    my $exe = &util::get_os_exe ();
164    my $mgpp_passes_exe = &util::filename_cat($exedir, "mgpp_passes$exe");
165    my $mgpp_compression_dict_exe = &util::filename_cat($exedir, "mgpp_compression_dict$exe");
166    my $outhandle = $self->{'outhandle'};
167
168    my $maxnumeric = $self->{'maxnumeric'};
169   
170    &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, "text"));
171
172    my $collect_tail = &util::get_dirsep_tail($self->{'collection'});
173    my $basefilename = &util::filename_cat("text",$collect_tail);
174    my $fulltextprefix = &util::filename_cat ($self->{'build_dir'}, $basefilename);
175   
176    my $osextra = "";
177    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
178    $fulltextprefix =~ s@/@\\@g;
179    }
180    else {
181    $osextra = " -d /";
182    }
183
184
185    # define the section names and possibly the doc name for mgpasses
186    # the compressor doesn't need to know about paragraphs - never want to
187    # retrieve them
188   
189    # always use Doc and Sec levels
190    my $mgpp_passes_sections = "-J ". $level_map{"document"} ." -K " . $level_map{"section"} ." ";
191
192    print $outhandle "\n*** creating the compressed text\n" if ($self->{'verbosity'} >= 1);
193    print STDERR "<Stage name='CompressText'>\n" if $self->{'gli'};
194
195    # collect the statistics for the text
196    # -b $maxdocsize sets the maximum document size to be 12 meg
197    print $outhandle "\n    collecting text statistics (mgpp_passes -T1)\n"  if ($self->{'verbosity'} >= 1);
198    print STDERR "<Phase name='CollectTextStats'/>\n" if $self->{'gli'};
199
200    my ($handle);
201    if ($self->{'debug'}) {
202    $handle = *STDOUT;
203    }
204    else {
205    if (!-e "$mgpp_passes_exe" ||
206        !open($handle, "| mgpp_passes$exe  -M $maxnumeric $mgpp_passes_sections -f \"$fulltextprefix\" -T1 $osextra")) {
207        print STDERR "<FatalError name='NoRunMGPasses'>\n</Stage>\n" if $self->{'gli'};
208        die "mgppbuilder::compress_text - couldn't run $mgpp_passes_exe\n";
209    }
210    }
211   
212    my $db_level = "section";
213
214    $self->{'buildproc'}->set_output_handle ($handle);
215    $self->{'buildproc'}->set_mode ('text');
216    $self->{'buildproc'}->set_index ($textindex);
217    $self->{'buildproc'}->set_indexing_text (0);
218    #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
219    $self->{'buildproc'}->set_levels ($self->{'levels'});                     
220    $self->{'buildproc'}->set_db_level ($db_level);                       
221    $self->{'buildproc'}->reset();
222    &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
223           $self->{'buildproc'}, $self->{'maxdocs'});
224    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
225           "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
226    &plugin::end($self->{'pluginfo'});
227
228    close ($handle) unless $self->{'debug'};
229
230    $self->print_stats();
231
232    # create the compression dictionary
233    # the compression dictionary is built by assuming the stats are from a seed
234    # dictionary (-S), if a novel word is encountered it is spelled out (-H),
235    # and the resulting dictionary must be less than 5 meg with the most
236    # frequent words being put into the dictionary first (-2 -k 5120)
237    # note: these options are left over from mg version
238    if (!$self->{'debug'}) {
239    print $outhandle "\n    creating the compression dictionary\n"  if ($self->{'verbosity'} >= 1);
240    print STDERR "<Phase name='CreatingCompress'/>\n" if $self->{'gli'};
241    if (!-e "$mgpp_compression_dict_exe") {
242        print STDERR "<FatalError name='NoRunMGCompress'/>\n</Stage>\n" if $self->{'gli'};
243        die "mgppbuilder::compress_text - couldn't run $mgpp_compression_dict_exe\n";
244    }
245    system ("mgpp_compression_dict$exe -f \"$fulltextprefix\" -S -H -2 -k 5120 $osextra");
246
247    if (!$self->{'debug'}) {
248        if (!-e "$mgpp_passes_exe" ||
249        !open ($handle, "| mgpp_passes$exe  -M $maxnumeric $mgpp_passes_sections -f \"$fulltextprefix\" -T2 $osextra")) {
250        print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
251        die "mgppbuilder::compress_text - couldn't run $mgpp_passes_exe\n";
252        }
253    }
254    }
255    else {
256    print STDERR "<Phase name='SkipCreatingComp'/>\n" if $self->{'gli'};
257    }
258
259    $self->{'buildproc'}->reset();
260    # compress the text
261    print $outhandle "\n    compressing the text (mgpp_passes -T2)\n"  if ($self->{'verbosity'} >= 1);
262    print STDERR "<Phase name='CompressingText'/>\n" if $self->{'gli'};
263
264    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
265           "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
266    close ($handle) unless $self->{'debug'};
267
268    $self->print_stats();
269    print STDERR "</Stage>\n" if $self->{'gli'};
270}
271
272
273sub build_indexes_extra {
274    my $self = shift(@_);
275    #define the final field lists
276    $self->make_final_field_list();
277}   
278
279# creates directory names for each of the index descriptions
280sub create_index_mapping {
281    my $self = shift (@_);
282    my ($indexes) = @_;
283
284    my %mapping = ();
285
286    return \%mapping if !(scalar @$indexes);
287
288    $mapping{'indexmaporder'} = [];
289    $mapping{'subcollectionmaporder'} = [];
290    $mapping{'languagemaporder'} = [];
291   
292    # dirnames is used to check for collisions. Start this off
293    # with the manditory directory names
294    my %dirnames = ('text'=>'text',
295            'extra'=>'extra');
296    my %pnames = ('index' => {}, 'subcollection' => {}, 'languages' => {});
297
298    foreach my $index (@$indexes) {
299    my ($fields, $subcollection, $languages) = split (":", $index);
300   
301    # we only ever have one index, and its called 'idx'
302    my $pindex = 'idx';
303   
304    # next comes a processed version of the subcollection if there is one.
305    my $psub = $self->process_field ($subcollection);
306    $psub = lc ($psub);
307
308    # next comes a processed version of the language if there is one.
309    my $plang = $self->process_field ($languages);
310    $plang = lc ($plang);
311
312    my $dirname = $pindex . $psub . $plang;
313
314    # check to be sure all index names are unique
315    while (defined ($dirnames{$dirname})) {
316        $dirname = $self->make_unique (\%pnames, $index, \$pindex, \$psub, \$plang);
317    }
318
319    $mapping{$index} = $dirname;
320
321    # store the mapping orders as well as the maps
322    # also put index, subcollection and language fields into the mapping thing -
323    # (the full index name (eg text:subcol:lang) is not used on
324    # the query page) -these are used for collectionmeta later on
325    if (!defined $mapping{'indexmap'}{"$fields"}) {
326        $mapping{'indexmap'}{"$fields"} = $pindex;
327        push (@{$mapping{'indexmaporder'}}, "$fields");
328        if (!defined $mapping{"$fields"}) {
329        $mapping{"$fields"} = $pindex;
330        }   
331    }
332    if ($psub =~ /\w/ && !defined ($mapping{'subcollectionmap'}{$subcollection})) {
333        $mapping{'subcollectionmap'}{$subcollection} = $psub;
334        push (@{$mapping{'subcollectionmaporder'}}, $subcollection);
335        $mapping{$subcollection} = $psub;
336    }
337    if ($plang =~ /\w/ && !defined ($mapping{'languagemap'}{$languages})) {
338        $mapping{'languagemap'}{$languages} = $plang;
339        push (@{$mapping{'languagemaporder'}}, $languages);
340        $mapping{$languages} = $plang;
341    }
342    $dirnames{$dirname} = $index;
343    $pnames{'index'}->{$pindex} = "$fields";
344    $pnames{'subcollection'}->{$psub} = $subcollection;
345    $pnames{'languages'}->{$plang} = $languages;
346    }
347
348    return \%mapping;
349}
350
351sub make_unique {
352    my $self = shift (@_);
353    my ($namehash, $index, $indexref, $subref, $langref) = @_;
354    my ($fields, $subcollection, $languages) = split (":", $index);
355
356    if ($namehash->{'index'}->{$$indexref} ne "$fields") {
357    $self->get_next_version ($indexref);
358    } elsif ($namehash->{'subcollection'}->{$$subref} ne $subcollection) {
359    $self->get_next_version ($subref);
360    } elsif ($namehash->{'languages'}->{$$langref} ne $languages) {
361    $self->get_next_version ($langref);
362    }
363    return "$$indexref$$subref$$langref";
364}   
365
366
367sub build_index {
368    my $self = shift (@_);
369    my ($index) = @_;
370    my $outhandle = $self->{'outhandle'};
371
372    # get the full index directory path and make sure it exists
373    my $indexdir = $self->{'index_mapping'}->{$index};
374    &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, $indexdir));
375
376    my $collect_tail = &util::get_dirsep_tail($self->{'collection'});
377    my $fullindexprefix = &util::filename_cat ($self->{'build_dir'},
378                           $indexdir,
379                           $collect_tail);
380    my $fulltextprefix = &util::filename_cat ($self->{'build_dir'}, "text",
381                           $collect_tail);
382
383    # get any os specific stuff
384    my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
385
386    my $exe = &util::get_os_exe ();
387    my $mgpp_passes_exe = &util::filename_cat($exedir, "mgpp_passes$exe");
388
389    # define the section names for mgpasses
390    my $mgpp_passes_sections = "-J ". $level_map{"document"} ." -K " . $level_map{"section"} ." ";
391    if ($self->{'levels'}->{'paragraph'}) {
392    $mgpp_passes_sections .= "-K " . $level_map{'paragraph'}. " ";
393    }
394
395    my $mgpp_perf_hash_build_exe =
396    &util::filename_cat($exedir, "mgpp_perf_hash_build$exe");
397    my $mgpp_weights_build_exe =
398    &util::filename_cat ($exedir, "mgpp_weights_build$exe");
399    my $mgpp_invf_dict_exe =
400    &util::filename_cat ($exedir, "mgpp_invf_dict$exe");
401    my $mgpp_stem_idx_exe =
402    &util::filename_cat ($exedir, "mgpp_stem_idx$exe");
403
404    my $maxnumeric = $self->{'maxnumeric'};
405
406    my $osextra = "";
407    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
408    $fullindexprefix =~ s@/@\\@g;
409    } else {
410    $osextra = " -d /";
411    if ($outhandle ne "STDERR") {
412        # so mgpp_passes doesn't print to stderr if we redirect output
413        $osextra .= " 2>/dev/null";
414    }
415    }
416 
417    # get the index expression if this index belongs
418    # to a subcollection
419    my $indexexparr = [];
420    my $langarr = [];
421    # there may be subcollection info, and language info.
422    my ($fields, $subcollection, $language) = split (":", $index);
423    my @subcollections = ();
424    @subcollections = split /,/, $subcollection if (defined $subcollection);
425
426    foreach $subcollection (@subcollections) {
427    if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
428        push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
429    }
430    }
431   
432    # add expressions for languages if this index belongs to
433    # a language subcollection - only put languages expressions for the
434    # ones we want in the index
435   
436    my @languages = ();
437    my $languagemetadata = "Language";
438    if (defined ($self->{'collect_cfg'}->{'languagemetadata'})) {
439    $languagemetadata = $self->{'collect_cfg'}->{'languagemetadata'};
440    }
441    @languages = split /,/, $language if (defined $language);
442    foreach my $language (@languages) {
443    my $not=0;
444    if ($language =~ s/^\!//) {
445        $not = 1;
446    }
447    if($not) {
448        push (@$langarr, "!$language");
449    } else {
450        push (@$langarr, "$language");
451    }
452    }
453
454    # Build index dictionary. Uses verbatim stem method
455    print $outhandle "\n    creating index dictionary (mgpp_passes -I1)\n"  if ($self->{'verbosity'} >= 1);
456    print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
457    my ($handle);
458    if ($self->{'debug'}) {
459    $handle = *STDOUT;
460    }
461    else {
462    if (!-e "$mgpp_passes_exe" ||
463        !open($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fullindexprefix\" -I1 $osextra")) {
464        print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
465        die "mgppbuilder::build_index - couldn't run $mgpp_passes_exe\n";
466    }
467    }
468       
469    # db_level is always section
470    my $db_level = "section";
471
472    # set up the document processr
473    $self->{'buildproc'}->set_output_handle ($handle);
474    $self->{'buildproc'}->set_mode ('text');
475    $self->{'buildproc'}->set_index ($index, $indexexparr);
476    $self->{'buildproc'}->set_index_languages ($languagemetadata, $langarr) if (defined $language);
477    $self->{'buildproc'}->set_indexing_text (1);
478    #$self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
479    $self->{'buildproc'}->set_levels ($self->{'levels'});
480    $self->{'buildproc'}->set_db_level ($db_level);   
481   
482    $self->{'buildproc'}->reset();
483    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
484           "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
485    close ($handle) unless $self->{'debug'};
486
487    $self->print_stats();
488
489    # 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.
490    # we check on the .id file - index dictionary
491    my $dict_file = "$fullindexprefix.id";
492    if (!-e $dict_file) {
493    print $outhandle "mgppbuilder::build_index - Couldn't create index $index\n";
494    print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
495    $self->{'notbuilt'}->{$index}=1;
496    return;
497    }
498
499    if (!$self->{'debug'}) {
500    # create the perfect hash function
501    if (!-e "$mgpp_perf_hash_build_exe") {
502        print STDERR "<FatalError name='NoRunMGHash'/>\n</Stage>\n" if $self->{'gli'};
503        die "mgppbuilder::build_index - couldn't run $mgpp_perf_hash_build_exe\n";
504    }
505    system ("mgpp_perf_hash_build$exe -f \"$fullindexprefix\" $osextra");
506
507    if (!-e "$mgpp_passes_exe" ||
508        !open ($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fullindexprefix\" -I2 $osextra")) {
509        print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
510        die "mgppbuilder::build_index - couldn't run $mgpp_passes_exe\n";
511    }
512    }
513   
514    # invert the text
515    print $outhandle "\n    inverting the text (mgpp_passes -I2)\n"  if ($self->{'verbosity'} >= 1);
516    print STDERR "<Phase name='InvertingText'/>\n" if $self->{'gli'};
517    $self->{'buildproc'}->reset();
518    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
519           "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
520
521    $self->print_stats ();
522   
523    if (!$self->{'debug'}) {
524
525    close ($handle);
526   
527    # create the weights file
528    print $outhandle "\n    create the weights file\n"  if ($self->{'verbosity'} >= 1);
529    print STDERR "<Phase name='CreateTheWeights'/>\n" if $self->{'gli'};
530    if (!-e "$mgpp_weights_build_exe") {
531        print STDERR "<FatalError name='NoRunMGWeights'/>\n</Stage>\n" if $self->{'gli'};
532        die "mgppbuilder::build_index - couldn't run $mgpp_weights_build_exe\n";
533    }
534    system ("mgpp_weights_build$exe -f \"$fullindexprefix\" $osextra");
535
536    # create 'on-disk' stemmed dictionary
537    print $outhandle "\n    creating 'on-disk' stemmed dictionary\n"  if ($self->{'verbosity'} >= 1);
538    if (!-e "$mgpp_invf_dict_exe") {
539        print STDERR "<FatalError name='NoRunMGInvf'/>\n</Stage>\n" if $self->{'gli'};
540        die "mgppbuilder::build_index - couldn't run $mgpp_invf_dict_exe\n";
541    }
542    system ("mgpp_invf_dict$exe -f \"$fullindexprefix\" $osextra" );
543
544
545    # creates stem index files for the various stemming methods
546    print $outhandle "\n    creating stem indexes\n"  if ($self->{'verbosity'} >= 1);
547    print STDERR "<Phase name='CreatingStemIndx'/>\n" if $self->{'gli'};
548    if (!-e "$mgpp_stem_idx_exe") {
549        print STDERR "<FatalError name='NoRunMGStem'/>\n</Stage>\n" if $self->{'gli'};
550        die "mgppbuilder::build_index - couldn't run $mgpp_stem_idx_exe\n";
551    }
552    my $accent_folding_enabled = 1;
553    if ($self->{'accentfold'}) {
554        # the first time we do this, we test for accent folding enabled
555        if (system ("mgpp_stem_idx$exe -b 4096 -s4 -f \"$fullindexprefix\" $osextra") == 2) {
556        # accent folding has not been enabled in mgpp
557        $accent_folding_enabled = 0;
558        $self->{'stemindexes'} -= 4;
559        }
560    }
561    if ($self->{'casefold'}) {
562        system ("mgpp_stem_idx$exe -b 4096 -s1 -f \"$fullindexprefix\" $osextra");
563        if ($accent_folding_enabled && $self->{'accentfold'}) {
564        system ("mgpp_stem_idx$exe -b 4096 -s5 -f \"$fullindexprefix\" $osextra");
565        }
566    }
567    if ($self->{'stem'}) {
568        system ("mgpp_stem_idx$exe -b 4096 -s2 -f \"$fullindexprefix\" $osextra");
569        if ($accent_folding_enabled && $self->{'accentfold'}) {
570        system ("mgpp_stem_idx$exe -b 4096 -s6 -f \"$fullindexprefix\" $osextra");
571        }
572    }
573    if ($self->{'casefold'} && $self->{'stem'}) {
574        system ("mgpp_stem_idx$exe -b 4096 -s3 -f \"$fullindexprefix\" $osextra");
575        if ($accent_folding_enabled && $self->{'accentfold'}) {
576        system ("mgpp_stem_idx$exe -b 4096 -s7 -f \"$fullindexprefix\" $osextra");
577        }
578    }
579
580    # remove unwanted files
581    my $tmpdir = &util::filename_cat ($self->{'build_dir'}, $indexdir);
582    opendir (DIR, $tmpdir) || die
583        "mgppbuilder::build_index - couldn't read directory $tmpdir\n";
584    foreach my $file (readdir(DIR)) {
585        next if $file =~ /^\./;
586        my ($suffix) = $file =~ /\.([^\.]+)$/;
587        if (defined $suffix && !defined $wanted_index_files{$suffix}) {
588        # delete it!
589        print $outhandle "deleting $file\n" if $self->{'verbosity'} > 2;
590        #&util::rm (&util::filename_cat ($tmpdir, $file));
591        }
592    }
593    closedir (DIR);
594    }
595    print STDERR "</Stage>\n" if $self->{'gli'};
596}   
597
598
599sub get_collection_meta_indexes
600{
601    my $self = shift(@_);
602    my $collection_infodb = shift(@_);
603
604    # define the indexed field mapping if not already done so (ie if infodb called separately from build_index)
605    if (!defined $self->{'build_cfg'}) {
606    $self->read_final_field_list();
607    }
608
609    # first do the collection meta stuff - everything without a dot
610    my $collmetadefined = 0;
611    my $metadata_entry;
612    if (defined $self->{'collect_cfg'}->{'collectionmeta'}) {
613    $collmetadefined = 1;
614    }
615
616    #add the index field macros to [collection]
617    # eg <TI>Title
618    #    <SU>Subject
619    # these now come from collection meta. if that is not defined, uses the metadata name
620    my $collmeta = "";
621    if (defined $self->{'build_cfg'}->{'indexfields'}) {
622    foreach my $longfield (@{$self->{'build_cfg'}->{'indexfields'}}){
623        my $shortfield = $self->{'buildproc'}->{'indexfieldmap'}->{$longfield};
624        next if $shortfield eq 1;
625       
626        # we need to check if some coll meta has been defined - don't output
627        # any that have
628        $collmeta = ".$longfield";
629        if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{$collmeta}) {
630        if ($longfield eq "allfields") {
631            $collection_infodb->{$shortfield} = [ "_query:textallfields_" ];
632        } elsif ($longfield eq "text") {
633            $collection_infodb->{$shortfield} = [ "_query:texttextonly_" ];
634        } else {
635            $collection_infodb->{$shortfield} = [ $longfield ];
636        }
637        }
638    }
639    }
640
641    # now add the level names
642    my $level_entry = "";
643    foreach my $level (@{$self->{'collect_cfg'}->{'levels'}}) {
644    $collmeta = ".$level"; # based on the original specification
645    $level =~ tr/A-Z/a-z/; # make it lower case
646    my $levelid = $level_map{$level}; # find the actual value we used in the index
647    if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{$collmeta}) {
648        # use the default macro
649        $collection_infodb->{$levelid} = [ $level_map{$levelid} ];
650    }
651    }
652   
653    # now add subcoll meta
654    my $subcoll_entry = "";
655    my $shortname = "";
656    my $one_entry = "";
657    foreach my $subcoll (@{$self->{'index_mapping'}->{'subcollectionmaporder'}}) {
658    $shortname = $self->{'index_mapping'}->{$subcoll};
659    if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{".$subcoll"}) {
660        $collection_infodb->{$shortname} = [ $subcoll ];
661    }
662    }
663
664    # now add language meta
665    my $lang_entry = "";
666    foreach my $lang (@{$self->{'index_mapping'}->{'languagemaporder'}}) {
667    $shortname = $self->{'index_mapping'}->{$lang};
668    if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{".$lang"}) {
669        $collection_infodb->{$shortname} = [ $lang ];
670    }
671    }
672}
673
674
675# default is to output the metadata sets (prefixes) used in collection
676sub output_collection_meta
677{
678    my $self = shift(@_);
679    my $infodb_handle = shift(@_);
680
681    my %collection_infodb = ();
682    $self->get_collection_meta_sets(\%collection_infodb);
683    $self->get_collection_meta_indexes(\%collection_infodb);
684    &dbutil::write_infodb_entry($self->{'infodbtype'}, $infodb_handle, "collection", \%collection_infodb);
685}
686
687
688# at the end of building, we have an indexfieldmap with all the mappings,
689# plus some extras, and indexmap with any indexes in it that weren't
690# specified in the index definition.  we want to make an ordered list of
691# fields that are indexed, and a list of mappings that are used. this will
692# be used for the build.cfg file, and for collection meta definition we
693# store these in a build.cfg bit
694sub make_final_field_list {
695    my $self = shift (@_);
696   
697    $self->{'build_cfg'} = {};
698   
699    # store the indexfieldmap information
700    my @indexfieldmap = ();
701    my @indexfields = ();
702    my $specifiedfields = {};
703    my @specifiedfieldorder = ();
704
705    # go through the index definition and add each thing to a map, so we
706    # can easily check if it is already specified - when doing the
707    # metadata, we print out all the individual fields, but some may
708    # already be specified in the index definition, so we dont want to add
709    # those again.
710
711    my $field;
712    foreach $field (@{$self->{'collect_cfg'}->{'indexes'}}) {
713    # remove subcoll stuff
714    my $parts = $field;
715    $parts =~ s/:.*$//;
716    # *************
717    my @fs = split(';', $parts);
718    foreach my $f(@fs) {
719        if (!defined $specifiedfields->{$f}) {
720        $specifiedfields->{$f}=1;
721        push (@specifiedfieldorder, "$f");
722        }
723    }
724    }
725   
726    #add all fields bit
727    my $ifm = $self->{'buildproc'}->{'indexfieldmap'};
728   
729    foreach $field (@specifiedfieldorder) {
730    if ($field eq "metadata") {
731        foreach my $newfield (keys %{$self->{'buildproc'}->{'indexfields'}}) {
732        if (!defined $specifiedfields->{$newfield}) {
733            push (@indexfieldmap, "$newfield\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$newfield}");
734            push (@indexfields, "$newfield");
735        }
736        }
737
738    } elsif ($field eq 'text') {
739        push (@indexfieldmap, "text\-\>TX");
740        push (@indexfields, "text");
741    } elsif ($field eq 'allfields') {
742        push (@indexfieldmap, "allfields\-\>ZZ");
743        push (@indexfields, "allfields");
744    } else {
745        # we only add in the ones that have been processed
746        if (defined $ifm->{$field}) {
747        push (@indexfieldmap, "$field\-\>$ifm->{$field}");
748        push (@indexfields, "$field");
749        }
750
751       
752    }
753    }
754
755    if (scalar @indexfieldmap) {
756    $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
757    }
758    if (scalar @indexfields) {
759    $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
760    }
761}
762
763
764# recreate the field list from the build.cfg file, look first in building,
765# then in index to find it. if there is no build.cfg, we can't do the field
766# list (there is unlikely to be any index anyway.)
767sub read_final_field_list {
768    my $self = shift (@_);
769    $self->{'build_cfg'} = {};
770    my @indexfieldmap = ();
771    my @indexfields = ();
772    my @indexmap = ();
773
774    # we read the stuff in from the build.cfg file - if its there
775    my $buildcfg = $self->read_build_cfg();
776    return unless defined $buildcfg;
777
778    my $field;
779    if (defined $buildcfg->{'indexfields'}) {
780    foreach $field (@{$buildcfg->{'indexfields'}}) {
781        push (@indexfields, "$field");
782    }
783    }
784
785    if (defined $buildcfg->{'indexfieldmap'}) {
786    foreach $field (@{$buildcfg->{'indexfieldmap'}}) {
787        push (@indexfieldmap, "$field");
788        my ($f, $v) = $field =~ /^(.*)\-\>(.*)$/;
789        $self->{'buildproc'}->{'indexfieldmap'}->{$f} = $v;
790    }
791    }       
792
793    if (defined $buildcfg->{'indexmap'}) {
794    foreach $field (@{$buildcfg->{'indexmap'}}) {
795        push (@indexmap, "$field");
796    }
797    }       
798
799    $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
800    $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
801    $self->{'build_cfg'}->{'indexmap'} = \@indexmap;
802}
803
804
805sub build_cfg_extra {
806    my $self = shift (@_);
807    my ($build_cfg) = @_;
808
809    $build_cfg->{'numsections'} = $self->{'buildproc'}->get_num_sections();
810   
811    # store the level info
812    my @indexlevels = ();
813    my @levelmap = ();
814    foreach my $l (@{$self->{'levelorder'}}) {
815    push (@indexlevels, $level_map{$l});
816    push (@levelmap, "$l\-\>$level_map{$l}");
817    }
818    $build_cfg->{'indexlevels'} = \@indexlevels;
819    $build_cfg->{'levelmap'} = \@levelmap;
820
821    # text level (and database level) is always section
822    $build_cfg->{'textlevel'} = $level_map{'section'};
823   
824}
825
8261;
827
828
Note: See TracBrowser for help on using the browser.