root/main/trunk/greenstone2/perllib/basebuilder.pm @ 24495

Revision 24495, 25.4 KB (checked in by davidb, 8 years ago)

Introduction of $site variable passed into the XXbuilder.pm perl code for indexing (needed to support Solr in Greenstone3)

  • Property svn:keywords set to Author Date Id Revision
Line 
1###########################################################################
2#
3# basebuilder.pm -- base class for collection builders
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 basebuilder;
27
28use strict;
29no strict 'refs'; # allow filehandles to be variables and viceversa
30
31use classify;
32use cfgread;
33use colcfg;
34use dbutil;
35use plugin;
36use util;
37
38
39BEGIN {
40    # set autoflush on for STDERR and STDOUT so that mgpp
41    # doesn't get out of sync with plugins
42    STDOUT->autoflush(1);
43    STDERR->autoflush(1);
44}
45
46END {
47    STDOUT->autoflush(0);
48    STDERR->autoflush(0);
49}
50
51our $maxdocsize = 12000;
52
53# used to signify "gs2"(default) or "gs3"
54our $gs_mode = "gs2";
55
56sub new {
57    my ($class, $site, $collection, $source_dir, $build_dir, $verbosity,
58    $maxdocs, $debug, $keepold, $incremental, $incremental_mode,
59    $remove_empty_classifications,
60    $outhandle, $no_text, $failhandle, $gli) = @_;
61
62    $outhandle = *STDERR unless defined $outhandle;
63    $no_text = 0 unless defined $no_text;
64    $failhandle = *STDERR unless defined $failhandle;
65
66    # create a builder object
67    my $self = bless {'site'=>$site, # will be undef for Greenstone 2
68              'collection'=>$collection,
69              'source_dir'=>$source_dir,
70              'build_dir'=>$build_dir,
71              'verbosity'=>$verbosity,
72              'maxdocs'=>$maxdocs,
73              'debug'=>$debug,
74              'keepold'=>$keepold,
75              'incremental'=>$incremental,
76              'incremental_mode'=>$incremental_mode,
77              'remove_empty_classifications'=>$remove_empty_classifications,
78              'outhandle'=>$outhandle,
79              'no_text'=>$no_text,
80              'failhandle'=>$failhandle,
81              'notbuilt'=>{},    # indexes not built
82              'gli'=>$gli
83              }, $class;
84
85    $self->{'gli'} = 0 unless defined $self->{'gli'};
86   
87    # Read in the collection configuration file.
88    my ($colcfgname);
89    ($colcfgname, $gs_mode) = &colcfg::get_collect_cfg_name($outhandle);
90    $self->{'collect_cfg'} = &colcfg::read_collection_cfg ($colcfgname, $gs_mode);
91
92    if ($gs_mode eq "gs3") {
93    # read it in again to save the original form for later writing out
94    # of buildConfig.xml
95    # we use this preserve object because $self->{'collect_cfg'}->{'classify'} somewhat gets modified during the calling of &classify::load_classifiers.
96    $self->{'collect_cfg_preserve'} = &colcfg::read_collection_cfg ($colcfgname, $gs_mode);
97    }
98   
99    # get the database type for this collection from the collect.cfg file (may be undefined)
100    $self->{'infodbtype'} = $self->{'collect_cfg'}->{'infodbtype'} || &dbutil::get_default_infodb_type();
101
102
103    # load up any dontdb fields
104    $self->{'dontdb'} = {};
105    if (defined ($self->{'collect_cfg'}->{'dontgdbm'})) {
106    foreach my $dg (@{$self->{'collect_cfg'}->{'dontgdbm'}}) {
107        $self->{'dontdb'}->{$dg} = 1;
108    }
109    }
110
111    $self->{'maxnumeric'} = 4;
112    return $self;
113}
114
115# stuff has been moved here from new, so we can use subclass methods
116sub init {
117    my $self = shift(@_);
118   
119    my $outhandle = $self->{'outhandle'};
120    my $failhandle = $self->{'failhandle'};
121
122    $self->generate_index_list();
123    my $indexes = $self->{'collect_cfg'}->{'indexes'};
124    if (defined $indexes) {
125    # sort out subcollection indexes
126    if (defined $self->{'collect_cfg'}->{'indexsubcollections'}) {
127        $self->{'collect_cfg'}->{'indexes'} = [];
128        foreach my $subcollection (@{$self->{'collect_cfg'}->{'indexsubcollections'}}) {
129        foreach my $index (@$indexes) {
130            push (@{$self->{'collect_cfg'}->{'indexes'}}, "$index:$subcollection");
131        }
132        }
133    }
134   
135    # sort out language subindexes
136    if (defined $self->{'collect_cfg'}->{'languages'}) {
137        $indexes = $self->{'collect_cfg'}->{'indexes'};
138        $self->{'collect_cfg'}->{'indexes'} = [];
139        foreach my $language (@{$self->{'collect_cfg'}->{'languages'}}) {
140        foreach my $index (@$indexes) {
141            if (defined ($self->{'collect_cfg'}->{'indexsubcollections'})) {
142            push (@{$self->{'collect_cfg'}->{'indexes'}}, "$index:$language");
143            }
144            else { # add in an empty subcollection field
145            push (@{$self->{'collect_cfg'}->{'indexes'}}, "$index\:\:$language");
146            }
147        }
148        }
149    }
150    }
151   
152    if (defined($self->{'collect_cfg'}->{'indexes'})) {
153    # make sure that the same index isn't specified more than once
154    my %tmphash = ();
155    my @tmparray = @{$self->{'collect_cfg'}->{'indexes'}};
156    $self->{'collect_cfg'}->{'indexes'} = [];
157    foreach my $i (@tmparray) {
158        if (!defined ($tmphash{$i})) {
159        push (@{$self->{'collect_cfg'}->{'indexes'}}, $i);
160        $tmphash{$i} = 1;
161        }
162    }
163    } else {
164    $self->{'collect_cfg'}->{'indexes'} = [];
165    }
166
167    # check incremental against whether builder can cope or not.
168    if ($self->{'incremental'} && !$self->is_incremental_capable()) {
169    print $outhandle "WARNING: The indexer used is not capable of incremental building. Reverting to -removeold\n";
170    $self->{'keepold'} = 0;
171    $self->{'incremental'} = 0;
172    $self->{'incremental_mode'} = "none";
173   
174    }
175
176
177    # get the list of plugins for this collection
178    my $plugins = [];
179    if (defined $self->{'collect_cfg'}->{'plugin'}) {
180    $plugins = $self->{'collect_cfg'}->{'plugin'};
181    }
182   
183    # load all the plugins
184
185    #build up the extra global options for the plugins
186    my @global_opts = ();
187    if (defined $self->{'collect_cfg'}->{'separate_cjk'} && $self->{'collect_cfg'}->{'separate_cjk'} =~ /^true$/i) {
188    push @global_opts, "-separate_cjk";
189    }
190    $self->{'pluginfo'} = &plugin::load_plugins ($plugins, $self->{'verbosity'}, $outhandle, $failhandle, \@global_opts, $self->{'incremental_mode'});
191   
192    if (scalar(@{$self->{'pluginfo'}}) == 0) {
193    print $outhandle "No plugins were loaded.\n";
194    die "\n";
195    }
196
197    # get the list of classifiers for this collection
198    my $classifiers = [];
199    if (defined $self->{'collect_cfg'}->{'classify'}) {
200    $classifiers = $self->{'collect_cfg'}->{'classify'};
201    }
202
203    # load all the classifiers
204    $self->{'classifiers'} = &classify::load_classifiers ($classifiers, $self->{'build_dir'}, $outhandle);
205
206    # load up the document processor for building
207    # if a buildproc class has been created for this collection, use it
208    # otherwise, use the default buildproc for the builder we are initialising
209    my $buildprocdir = undef;
210    my $buildproctype;
211
212    my $collection = $self->{'collection'};
213    if (-e "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib/custombuildproc.pm") {
214    $buildprocdir = "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib";
215    $buildproctype = "custombuildproc";
216    } elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/custombuildproc.pm") {
217    $buildprocdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
218    $buildproctype = "custombuildproc";
219    } elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/${collection}buildproc.pm") {
220    $buildprocdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
221    $buildproctype = "${collection}buildproc";
222    } else {
223    $buildproctype = $self->default_buildproc();
224    }
225    if (defined $buildprocdir) {
226    require "$buildprocdir/$buildproctype.pm";
227    }
228    else {
229    require "$buildproctype.pm";
230    }
231
232    eval("\$self->{'buildproc'} = new $buildproctype(\$self->{'collection'}, " .
233     "\$self->{'source_dir'}, \$self->{'build_dir'}, \$self->{'keepold'}, \$self->{'verbosity'}, \$self->{'outhandle'})");
234    die "$@" if $@;
235
236    # We call set_infodbtype() now so the buildproc knows the infodbtype for all phases of the build
237    $self->{'buildproc'}->set_infodbtype($self->{'infodbtype'});
238   
239   $self->generate_index_options();
240
241    if (!$self->{'debug'} && !$self->{'keepold'}) {
242    # remove any old builds
243    &util::rm_r($self->{'build_dir'});
244    &util::mk_all_dir($self->{'build_dir'});
245       
246    # make the text directory
247    my $textdir = "$self->{'build_dir'}/text";
248    &util::mk_all_dir($textdir);
249    }
250
251    if ($self->{'incremental'}) {
252    # some classes may need to do some additional initialisation
253    $self->init_for_incremental_build();
254    }
255   
256}
257
258sub is_incremental_capable
259{
260    # By default we return 'no' as the answer
261    # Safer to assume non-incremental to start with, and then override in
262    # inherited classes that are.
263
264    return 0;
265}
266
267# implement this in subclass if want to do additional initialisation for an
268# incremental build
269sub init_for_incremental_build {
270    my $self = shift (@_);
271}
272
273sub deinit {
274    my $self = shift (@_);
275   
276    &plugin::deinit($self->{'pluginfo'},$self->{'buildproc'});
277}
278
279sub generate_index_options {
280    my $self = shift (@_);
281
282    my $separate_cjk = 0;
283   
284    if (defined($self->{'collect_cfg'}->{'indexoptions'})) {
285    foreach my $option (@{$self->{'collect_cfg'}->{'indexoptions'}}) {
286        if ($option =~ /separate_cjk/) {
287        $separate_cjk = 1;
288        }
289    }
290    }
291    # set this for building
292    $self->{'buildproc'}->set_separate_cjk($separate_cjk);
293    # record it for build.cfg
294    $self->{'separate_cjk'} = $separate_cjk;
295}
296 
297sub set_sections_index_document_metadata {
298    my $self = shift (@_);
299    my ($index) = @_;
300 
301    $self->{'buildproc'}->set_sections_index_document_metadata($index);
302}
303
304sub set_maxnumeric {
305    my $self = shift (@_);
306    my ($maxnumeric) = @_;
307
308    $self->{'maxnumeric'} = $maxnumeric;
309}
310sub set_strip_html {
311    my $self = shift (@_);
312    my ($strip) = @_;
313   
314    $self->{'strip_html'} = $strip;
315    $self->{'buildproc'}->set_strip_html($strip);
316}
317
318sub compress_text {
319    my $self = shift (@_);
320    my ($textindex) = @_;
321
322    print STDERR "compress_text() should be implemented in subclass!!";
323    return;
324}
325
326
327sub build_indexes {
328    my $self = shift (@_);
329    my ($indexname) = @_;
330    my $outhandle = $self->{'outhandle'};
331
332    $self->pre_build_indexes();
333
334    my $indexes = [];
335    if (defined $indexname && $indexname =~ /\w/) {
336    push @$indexes, $indexname;
337    } else {
338    $indexes = $self->{'collect_cfg'}->{'indexes'};
339    }
340
341    # create the mapping between the index descriptions
342    # and their directory names (includes subcolls and langs)
343    $self->{'index_mapping'} = $self->create_index_mapping ($indexes);
344   
345    # build each of the indexes
346    foreach my $index (@$indexes) {
347    if ($self->want_built($index)) {
348        print $outhandle "\n*** building index $index in subdirectory " .
349        "$self->{'index_mapping'}->{$index}\n" if ($self->{'verbosity'} >= 1);
350        print STDERR "<Stage name='Index' source='$index'>\n" if $self->{'gli'};
351        $self->build_index($index);
352    } else {
353        print $outhandle "\n*** ignoring index $index\n" if ($self->{'verbosity'} >= 1);
354    }
355    }
356
357    $self->post_build_indexes();
358
359}
360
361# implement this in subclass if want to do extra stuff at before building
362# all the indexes
363sub pre_build_indexes {
364    my $self = shift(@_);
365    my ($indexname) = @_; # optional parameter
366}
367
368# implement this in subclass if want to do extra stuff at the end of building
369# all the indexes
370sub post_build_indexes {
371    my $self = shift(@_);   
372}
373
374sub build_index {
375    my $self = shift (@_);
376    my ($index) = @_;
377   
378    print STDERR "build_index should be implemented in subclass\n";
379    return;
380}
381
382
383
384sub make_infodatabase {
385    my $self = shift (@_);
386    my $outhandle = $self->{'outhandle'};
387
388    print STDERR "BuildDir: $self->{'build_dir'}\n";
389
390    my $textdir = &util::filename_cat($self->{'build_dir'}, "text");
391    my $assocdir = &util::filename_cat($self->{'build_dir'}, "assoc");
392    &util::mk_all_dir ($textdir);
393    &util::mk_all_dir ($assocdir);
394
395    # Get info database file path
396    my $infodb_type = $self->{'infodbtype'};
397    my $infodb_file_path = &dbutil::get_infodb_file_path($infodb_type, $self->{'collection'}, $textdir);
398
399    print $outhandle "\n*** creating the info database and processing associated files\n"
400    if ($self->{'verbosity'} >= 1);
401    print STDERR "<Stage name='CreateInfoData'>\n" if $self->{'gli'};
402
403    # init all the classifiers
404    &classify::init_classifiers ($self->{'classifiers'});
405
406    my $reconstructed_docs = undef;
407    my $database_recs = undef;
408
409    if ($self->{'incremental'}) {
410    $database_recs = {};
411
412    &dbutil::read_infodb_file($infodb_type, $infodb_file_path, $database_recs);
413    }
414
415   
416    # Important (for memory usage reasons) that we obtain the filehandle
417    # here for writing out to the database, rather than after
418    # $reconstructed_docs has been set up (assuming -incremental is on)
419    #
420    # This is because when we open a pipe to txt2db [using open()]
421    # this triggers a fork() followed by exec().  $reconstructed_docs
422    # can get very large, and so if we did the open() after this, it means
423    # the fork creates a clone of the *large* process image which (admittedly)
424    # is then quickly replaced in the execve() with the much smaller image for
425    # 'txt2db'.  The trouble is, in that seismic second caused by
426    # the fork(), the system really does need to have all that memory available
427    # even though it isn't ultimately used.  The result is an out of memory
428    # error.
429
430    my ($infodb_handle);
431    if ($self->{'debug'}) {
432    $infodb_handle = *STDOUT;
433    }
434    else {
435    $infodb_handle = &dbutil::open_infodb_write_handle($infodb_type, $infodb_file_path);
436    if (!defined($infodb_handle))
437    {
438        print STDERR "<FatalError name='NoRunText2DB'/>\n</Stage>\n" if $self->{'gli'};
439        die "builder::make_infodatabase - couldn't open infodb write handle\n";
440    }
441    }
442
443    if ($self->{'incremental'}) {
444    # reconstruct doc_obj metadata from database for all docs
445    $reconstructed_docs
446        = &classify::reconstruct_doc_objs_metadata($infodb_type,
447                               $infodb_file_path,
448                               $database_recs);
449    }
450
451    # set up the document processor
452
453    $self->{'buildproc'}->set_output_handle ($infodb_handle);
454    $self->{'buildproc'}->set_mode ('infodb');
455    $self->{'buildproc'}->set_assocdir ($assocdir);
456    $self->{'buildproc'}->set_dontdb ($self->{'dontdb'});
457    $self->{'buildproc'}->set_classifiers ($self->{'classifiers'});
458    $self->{'buildproc'}->set_indexing_text (0);
459    $self->{'buildproc'}->set_store_text(1);
460    $self->{'buildproc'}->set_store_metadata_coverage ($self->{'collect_cfg'}->{'store_metadata_coverage'});
461
462    # make_infodatabase needs full reset even for incremental build
463    # as incremental works by reconstructing all docs from the database and
464    # then adding in the new ones
465    $self->{'buildproc'}->zero_reset();
466
467    $self->{'buildproc'}->{'mdprefix_fields'} = {};
468   
469    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
470           "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'},0, $self->{'gli'});
471
472    if ($self->{'incremental'}) {
473    # create flat classify structure, ready for new docs to be added
474    foreach my $doc_obj ( @$reconstructed_docs ) {
475        if (! defined $self->{'buildproc'}->{'dont_process_reconstructed'}->{$doc_obj->get_OID()}) {
476        print $outhandle "  Adding reconstructed ", $doc_obj->get_OID(), " into classify structures\n";
477        $self->{'buildproc'}->process($doc_obj,undef);
478        }
479    }
480    }
481    # this has changed to only output collection meta if its
482    # not in the config file
483    $self->output_collection_meta($infodb_handle);
484   
485    # output classification information
486    &classify::output_classify_info ($self->{'classifiers'}, $infodb_type, $infodb_handle,
487                     $self->{'remove_empty_classifications'},
488                     $self->{'gli'});
489
490    # Output classifier reverse lookup, used in incremental deletion
491    ####&classify::print_reverse_lookup($infodb_handle);
492
493    # output doclist
494    my @doc_list = $self->{'buildproc'}->get_doc_list();
495    my $browselist_infodb = { 'hastxt' => [ "0" ],
496                  'childtype' => [ "VList" ],
497                  'numleafdocs' => [ scalar(@doc_list) ],
498                  'thistype' => [ "Invisible" ],
499                  'contains' => [ join(";", @doc_list) ] };
500    &dbutil::write_infodb_entry($infodb_type, $infodb_handle, "browselist", $browselist_infodb);
501
502    &dbutil::close_infodb_write_handle($infodb_type, $infodb_handle) if !$self->{'debug'};
503   
504    if ($infodb_type eq "gdbm-txtgz") {
505    my $gdb_infodb_file_path = &dbutil::get_infodb_file_path("gdbm", $self->{'collection'}, $textdir);
506    if (-e $gdb_infodb_file_path) {
507        &util::rm($gdb_infodb_file_path);
508    }
509    }
510    print STDERR "</Stage>\n" if $self->{'gli'};
511}
512
513sub make_auxiliary_files {
514    my $self = shift (@_);
515    my ($index);
516    my $build_cfg = {};
517    # subclasses may have already defined stuff in here
518    if (defined $self->{'build_cfg'}) {
519    $build_cfg = $self->{'build_cfg'};
520    }
521
522    my $outhandle = $self->{'outhandle'};
523
524    print $outhandle "\n*** creating auxiliary files \n" if ($self->{'verbosity'} >= 1);
525    print STDERR "<Stage name='CreatingAuxilary'>\n" if $self->{'gli'};
526
527    # get the text directory
528    &util::mk_all_dir ($self->{'build_dir'});
529
530    # store the build date
531    $build_cfg->{'builddate'} = time;
532    $build_cfg->{'buildtype'} = $self->{'buildtype'};
533    $build_cfg->{'indexstem'} = &util::get_dirsep_tail($self->{'collection'});
534    $build_cfg->{'stemindexes'} = $self->{'stemindexes'};
535    if ($self->{'separate_cjk'}) {
536    $build_cfg->{'separate_cjk'} = "true";
537    }
538   
539    # store the number of documents and number of bytes
540    $build_cfg->{'numdocs'} = $self->{'buildproc'}->get_num_docs();
541    $build_cfg->{'numsections'} = $self->{'buildproc'}->get_num_sections();
542    $build_cfg->{'numbytes'} = $self->{'buildproc'}->get_num_bytes();
543
544    # store the mapping between the index names and the directory names
545    # the index map is used to determine what indexes there are, so any that are not built should not be put into the map.
546    my @indexmap = ();
547    foreach my $index (@{$self->{'index_mapping'}->{'indexmaporder'}}) {
548    if (not defined ($self->{'notbuilt'}->{$index})) {
549        push (@indexmap, "$index\-\>$self->{'index_mapping'}->{'indexmap'}->{$index}");
550    }
551    }
552    $build_cfg->{'indexmap'} = \@indexmap if scalar (@indexmap);
553
554    my @subcollectionmap = ();
555    foreach my $subcollection (@{$self->{'index_mapping'}->{'subcollectionmaporder'}}) {
556    push (@subcollectionmap, "$subcollection\-\>" .
557          $self->{'index_mapping'}->{'subcollectionmap'}->{$subcollection});
558    }
559    $build_cfg->{'subcollectionmap'} = \@subcollectionmap if scalar (@subcollectionmap);
560
561    my @languagemap = ();
562    foreach my $language (@{$self->{'index_mapping'}->{'languagemaporder'}}) {
563    push (@languagemap, "$language\-\>" .
564          $self->{'index_mapping'}->{'languagemap'}->{$language});
565    }
566    $build_cfg->{'languagemap'} = \@languagemap if scalar (@languagemap);
567
568    my @notbuilt = ();
569    foreach my $nb (keys %{$self->{'notbuilt'}}) {
570    push (@notbuilt, $nb);
571    }
572    $build_cfg->{'notbuilt'} = \@notbuilt if scalar (@notbuilt);
573
574    $build_cfg->{'maxnumeric'} = $self->{'maxnumeric'};
575
576    $build_cfg->{'infodbtype'} = $self->{'infodbtype'};
577   
578    # write out the earliestDatestamp information needed for OAI
579    my $archivedir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "archives");
580    if(!-d $archivedir) {
581    $archivedir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "export");
582    }
583    my $earliestDatestampFile = &util::filename_cat ($archivedir, "earliestDatestamp");
584    my $earliestDatestamp = 0;
585    if (open(FIN,"<$earliestDatestampFile")) {
586    {
587        # slurp in file as a single line
588        local $/ = undef;
589        $earliestDatestamp = <FIN>;
590        #&unicode::ensure_utf8(\$earliestDatestamp); # turn any high bytes that aren't valid utf-8 into utf-8.
591    }
592    close(FIN);
593    }
594    else {
595    print $outhandle "Warning: unable to read collection's earliestDatestamp from $earliestDatestampFile.\n";
596    print $outhandle "Setting value to 0.\n";
597    }
598    $build_cfg->{'earliestdatestamp'} = $earliestDatestamp;
599   
600    $self->build_cfg_extra($build_cfg);
601
602    if ($gs_mode eq "gs2") {
603      &colcfg::write_build_cfg(&util::filename_cat($self->{'build_dir'},"build.cfg"), $build_cfg);
604    }
605    if ($gs_mode eq "gs3") {
606
607      &colcfg::write_build_cfg_xml(&util::filename_cat($self->{'build_dir'}, "buildConfig.xml"), $build_cfg, $self->{'collect_cfg_preserve'});
608    }   
609
610    print STDERR "</Stage>\n" if $self->{'gli'};
611}
612
613# implement this in subclass if want to add extra stuff to build.cfg
614sub build_cfg_extra {
615   my $self = shift(@_);
616   my ($build_cfg) = @_;
617   
618}
619
620
621sub collect_specific {
622    my $self = shift (@_);
623}
624
625sub want_built {
626    my $self = shift (@_);
627    my ($index) = @_;
628
629    if (defined ($self->{'collect_cfg'}->{'dontbuild'})) {
630    foreach my $checkstr (@{$self->{'collect_cfg'}->{'dontbuild'}}) {
631        if ($index =~ /^$checkstr$/) {
632        $self->{'notbuilt'}->{$index} = 1;
633        return 0;
634        }
635    }
636    }
637
638    return 1;
639}
640
641sub create_index_mapping {
642    my $self = shift (@_);
643    my ($indexes) = @_;
644
645    print STDERR "create_index_mapping should be implemented in subclass\n";
646    my %mapping = ();
647    return \%mapping;
648}
649
650# returns a processed version of a field.
651# if the field has only one component the processed
652# version will contain the first character and next consonant
653# of that componant - otherwise it will contain the first
654# character of the first two components
655# only uses letdig (\w) characters now
656sub process_field {
657    my $self = shift (@_);
658    my ($field) = @_;
659
660    return "" unless (defined ($field) && $field =~ /\S/);
661   
662    my ($a, $b);
663    my @components = split /,/, $field;
664    if (scalar @components >= 2) {
665    # pick the first letdig from the first two field names
666    ($a) = $components[0] =~ /^[^\w]*(\w)/;
667    ($b) = $components[1] =~ /^[^\w]*(\w)/;
668    } else {
669    # pick the first two letdig chars
670    ($a, $b) = $field =~ /^[^\w]*(\w)[^\w]*?(\w)/i;
671    }
672    # there may not have been any letdigs...
673    $a = 'a' unless defined $a;
674    $b = '0' unless defined $b;
675   
676    my $newfield = "$a$b";
677    if ($newfield =~ /^\d\d$/) {
678    # digits only - Greenstone runtime doesn't like this.
679    $newfield = "a$a";
680    }
681    return $newfield;
682   
683}
684
685sub get_next_version {
686    my $self = shift (@_);
687    my ($nameref) = @_;
688    my $num=0;
689    if ($$nameref =~ /(\d\d)$/) {
690    $num = $1; $num ++;
691    $$nameref =~ s/\d\d$/$num/;
692    } elsif ($$nameref =~ /(\d)$/) {
693    $num = $1;
694    if ($num == 9) {$$nameref =~ s/\d$/10/;}
695    else {$num ++; $$nameref =~ s/\d$/$num/;}
696    } else {
697    $$nameref =~ s/.$/0/;
698    }
699}
700
701
702
703sub get_collection_meta_sets
704{
705    my $self = shift(@_);
706    my $collection_infodb = shift(@_);
707
708    my $mdprefix_fields = $self->{'buildproc'}->{'mdprefix_fields'};
709    foreach my $prefix (keys %$mdprefix_fields)
710    {
711    push(@{$collection_infodb->{"metadataset"}}, $prefix);
712
713    foreach my $field (keys %{$mdprefix_fields->{$prefix}})
714    {
715        push(@{$collection_infodb->{"metadatalist-$prefix"}}, $field);
716
717        my $val = $mdprefix_fields->{$prefix}->{$field};
718        push(@{$collection_infodb->{"metadatafreq-$prefix-$field"}}, $val);
719    }
720    }
721}
722
723
724# default is to output the metadata sets (prefixes) used in collection
725sub output_collection_meta
726{
727    my $self = shift(@_);
728    my $infodb_handle = shift(@_);
729
730    my %collection_infodb = ();
731    $self->get_collection_meta_sets(\%collection_infodb);
732    &dbutil::write_infodb_entry($self->{'infodbtype'}, $infodb_handle, "collection", \%collection_infodb);
733}
734
735# sometimes we need to read in an existing build.cfg - for example,
736# if doing each stage of building separately, or when doing incremental
737# building
738sub read_build_cfg {
739    my $self = shift(@_);
740
741    my $buildconfigfilename;
742   
743    if ($gs_mode eq "gs2") {
744    $buildconfigfilename = "build.cfg";
745    } else {
746    $buildconfigfilename = "buildConfig.xml";
747    }
748   
749    my $buildconfigfile = &util::filename_cat($self->{'build_dir'}, $buildconfigfilename);
750   
751    if (!-e $buildconfigfile) {
752    # try the index dir - but do we know where it is?? try here
753    $buildconfigfile  = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "index", $buildconfigfilename);
754    if (!-e $buildconfigfile) {
755        #we cant find a config file - just ignore the field list
756        return undef;
757    }
758    }
759    return &colcfg::read_building_cfg( $buildconfigfile, $gs_mode);
760   
761}
762
763sub print_stats {
764    my $self = shift (@_);
765
766    my $outhandle = $self->{'outhandle'};
767    my $indexing_text = $self->{'buildproc'}->get_indexing_text();
768    my $index = $self->{'buildproc'}->get_index();
769    my $num_bytes = $self->{'buildproc'}->get_num_bytes();
770    my $num_processed_bytes = $self->{'buildproc'}->get_num_processed_bytes();
771
772    if ($indexing_text) {
773    print $outhandle "Stats (Creating index $index)\n";
774    } else {
775    print $outhandle "Stats (Compressing text from $index)\n";
776    }
777    print $outhandle "Total bytes in collection: $num_bytes\n";
778    print $outhandle "Total bytes in $index: $num_processed_bytes\n";
779
780    if ($num_processed_bytes < 50 && ($indexing_text || !$self->{'no_text'})) {
781   
782    if ($self->{'incremental'}) {
783        if ($num_processed_bytes == 0) {
784        if ($indexing_text) {
785            print $outhandle "No additional text was added to $index\n";
786        } elsif (!$self->{'no_text'}) {
787            print $outhandle "No additional text was compressed\n";
788        }   
789        }   
790    }
791    else {
792        print $outhandle "***************\n";
793        if ($indexing_text) {
794        print $outhandle "WARNING: There is very little or no text to process for $index\n";
795        } elsif (!$self->{'no_text'}) {
796        print $outhandle "WARNING: There is very little or no text to compress\n";
797        }     
798        print $outhandle "         Was this your intention?\n";
799        print $outhandle "***************\n";
800    }
801
802    }
803
804}
805
806 
8071;
808
Note: See TracBrowser for help on using the browser.