source: main/trunk/greenstone2/perllib/buildConfigxml.pm@ 27818

Last change on this file since 27818 was 27818, checked in by kjdon, 11 years ago

adding in stuff for facets

  • Property svn:keywords set to Author Date Id Revision
File size: 24.0 KB
RevLine 
[15600]1###########################################################################
2#
[20096]3# buildConfigxml.pm --
[15600]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###########################################################################
[14741]25
[20105]26# reads in buildConfig.xml
27# Note, only implemented the bits that are currently used, eg by incremental
28# build code.
29# The resulting data is not a full representation on buildConfig.xml.
[15600]30
[20096]31package buildConfigxml;
[20105]32
[15600]33use strict;
34no strict 'refs';
35no strict 'subs';
36
[23895]37use XMLParser;
[15600]38
39
[17895]40# A mapping hash to resolve name discrepancy between gs2 and gs3.
[20105]41my $nameMap = {"numDocs" => "numdocs",
[24464]42 "buildType" => "buildtype",
43 "orthogonalBuildTypes" => "orthogonalbuildtypes"
[15600]44 };
[20105]45
46
[15600]47# A hash structure which is returned by sub read_cfg_file.
48my $data = {};
49
50# use those unique attribute values to locate the text within the elements
51my $currentLocation = "";
[20105]52my $stringexp = q/^(buildType|numDocs)$/;
[24464]53my $arrayexp = q/^(orthogonalBuildTypes)$/;
54
[20105]55my $indexmap_name = "";
56my $haveindexfields = 0;
[15600]57
[20105]58# Reads in the model collection configuration file, collectionConfig.xml,
59# into a structure which complies with the one used by gs2 (i.e. one read
60# in by &cfgread::read_cfg_file).
61sub read_cfg_file {
62 my ($filename) = @_;
63 $data = {};
64 if ($filename !~ /buildConfig\.xml$/ || !-f $filename) {
65 return undef;
66 }
[15600]67
[23895]68 # Removed ProtocolEncoding (see MetadataXMLPlugin for details)
69
[20105]70 # create XML::Parser object for parsing metadata.xml files
[23895]71 my $parser = new XML::Parser('Style' => 'Stream',
72 'Pkg' => 'buildConfigxml',
73 'Handlers' => {'Char' => \&Char,
[20105]74 'Doctype' => \&Doctype
75 });
[15600]76
[20105]77 if (!open (COLCFG, $filename)) {
78 print STDERR "buildConfigxml::read_cfg_file couldn't read the cfg file $filename\n";
79 } else {
[15600]80
[20105]81 $parser->parsefile ($filename);# (COLCFG);
82 close (COLCFG);
83 }
84
85 #&Display;
86 return $data;
87}
88
[15600]89sub StartTag {
90# Those marked with #@ will not be executed at the same time when this sub is being called
91# so that if/elsif is used to avoid unnecessary tests
92 my ($expat, $element) = @_;
[15619]93
[15600]94 my $name = $_{'name'};
[20105]95 my $shortname = $_{'shortname'};
[15600]96
[20105]97
[15600]98 #@ handling block metadataList
[24464]99 if (defined $name && (($name =~ m/$stringexp/) || ($name =~ m/$arrayexp/))) {
[15600]100 $currentLocation = $name;
[20105]101 # the value will be retrieved later in Text sub
[15600]102 }
103
[25758]104 #@ handle indexes - store indexmap (mg) or indexfields and indexfieldmap (mgpp/lucene/solr)
[20105]105 elsif ($element =~ /^indexList$/) {
106 # set up the data arrays
107 # this assumes that the build type has been read already, which is
108 # currently the order we save the file in.
109 if ($data->{'buildtype'} eq "mg") {
110 $indexmap_name = "indexmap";
111 if (!defined $data->{"indexmap"}) {
112 $data->{"indexmap"} = [];
113 }
[15619]114 }
[20105]115 else {
[25758]116 # mgpp, lucene or solr
[20105]117 $indexmap_name = "indexfieldmap";
118 $haveindexfields = 1;
119 if (!defined $data->{"indexfieldmap"}) {
120 $data->{"indexfieldmap"} = [];
121 }
122 if (!defined $data->{"indexfields"}) {
123 $data->{"indexfields"} = [];
124 }
[17895]125
[20105]126 }
127
[17895]128 }
[15619]129
[20105]130 elsif ($element =~ /index/) {
131 # store each index in the map
132 if (defined $name && defined $shortname) {
133 push @{$data->{$indexmap_name}}, "$name->$shortname";
134 if ($haveindexfields) {
135 push @{$data->{'indexfields'}}, $name;
136 }
[15600]137 }
138 }
139
140
141}
142
143sub EndTag {
144 my ($expat, $element) = @_;
145}
146
147sub Text {
[15619]148 if (defined $currentLocation) {
[20105]149 #@ Handling block metadataList(numDocs, buildType)
[24464]150 if ($currentLocation =~ /$stringexp/) {
[15619]151 #print $currentLocation;
152 my $key = $nameMap->{$currentLocation};
153 $data->{$key} = $_;
154 undef $currentLocation;
[20105]155 }
[24464]156 elsif ($currentLocation =~ /$arrayexp/) {
157 #print $currentLocation;
158 my $key = $nameMap->{$currentLocation};
159 push(@{$data->{$key}},$_);
160 undef $currentLocation;
161 }
162
[15619]163 }
[15600]164}
[15619]165
[15600]166# This sub is for debugging purposes
167sub Display {
[20105]168
169 print "NumDocs = ".$data->{'numdocs'}."\n" if (defined $data->{'numdocs'});
170 print "BuildType = ".$data->{'buildtype'}."\n" if (defined $data->{'buildtype'});
[24464]171 print "OrthogonalBuildTypes = ".join(",",@{$data->{'orthogonalbuildtypes'}})."\n" if (defined $data->{'orthogonalbuildtypes'});
[20105]172 print "IndexMap = ". join(" ",@{$data->{'indexmap'}})."\n" if (defined $data->{'indexmap'});
173 print "IndexFieldMap = ". join(" ",@{$data->{'indexfieldmap'}})."\n" if (defined $data->{'indexfieldmap'});
174 print "IndexFields = ". join(" ",@{$data->{'indexfields'}})."\n" if (defined $data->{'indexfields'});
175
[15600]176}
[20105]177
178# is this actually used??
[15600]179sub Doctype {
180 my ($expat, $name, $sysid, $pubid, $internal) = @_;
181
[20105]182 die if ($name !~ /^buildConfig$/);
[15600]183}
184
185# This Char function overrides the one in XML::Parser::Stream to overcome a
186# problem where $expat->{Text} is treated as the return value, slowing
187# things down significantly in some cases.
188sub Char {
189 if ($]<5.008) {
190 use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+ and Perl 5.6
191 }
192 $_[0]->{'Text'} .= $_[1];
193 return undef;
194}
[15619]195
[15600]196
197
198sub write_line {
199 my ($filehandle, $line) = @_;
200 print $filehandle join ("", @$line), "\n";
201}
202
[24464]203sub search_and_retrieve_settings
204{
205 my ($buildcfg,$collectcfg) = @_;
[15600]206
[24464]207 my $settings = {};
[15600]208
209 if (defined $buildcfg->{"buildtype"}) {
[24464]210 $settings->{'buildtype'} = $buildcfg->{"buildtype"};
[15600]211 } else {
[24464]212 $settings->{'buildtype'} = "mgpp";
[15600]213 }
[24464]214 my $buildtype = $settings->{'buildtype'};
215
216 if (defined $collectcfg->{"orthogonalbuildtypes"}) {
217 # Note the use of collectcfg, not bulidcfg
218 $settings->{'orthogonalbuildtypes'} = $collectcfg->{"orthogonalbuildtypes"};
219 } else {
220 $settings->{'orthogonalbuildtypes '}= [];
221 }
222
[15600]223 if (defined $buildcfg->{"numdocs"}) {
[24464]224 $settings->{'numdocs'} = $buildcfg->{"numdocs"};
[15600]225 }
[24464]226 else {
227 $settings->{'numdocs'} = 0;
[21783]228 }
[23939]229
[15600]230 my $service_type = "MGPP";
231 if ($buildtype eq "mg") {
232 $service_type = "MG";
233 } elsif ($buildtype eq "lucene") {
234 $service_type = "Lucene";
[25758]235 } elsif ($buildtype eq "solr") {
236 $service_type = "Solr";
[15600]237 }
[24464]238 $settings->{'service_type'} = $service_type;
[15600]239
240
[24464]241 if (defined $buildcfg->{"infodbtype"}) {
242 $settings->{'infodbtype'} = $buildcfg->{'infodbtype'};
[15600]243 }
[24464]244 else {
245 $settings->{'infodbtype'} = "gdbm";
[21439]246 }
[15600]247
[24464]248
249 #--
250 # indexes
251 #--
[15600]252
[24464]253 my $indexmap = {}; # maps index name to shortname
254 my $indexlist = []; # keeps the order for indexes
[15600]255 my $defaultindex = "";
[24464]256 my $maptype = ($buildtype eq "mg") ? "indexmap" : "indexfieldmap";
[15600]257
258 if (defined $buildcfg->{$maptype}) {
[24464]259 my $first = 1;
[15600]260 my $indexmap_t = $buildcfg->{$maptype};
[24464]261 foreach my $i (@$indexmap_t) {
262 my ($k, $v) = $i =~ /^(.*)\-\>(.*)$/;
[15600]263 $indexmap->{$k} = $v;
[24464]264 push @$indexlist, $k;
[15600]265 if ($first) {
266 $defaultindex = $v;
267 $first = 0;
[24464]268 }
[15600]269 }
[24464]270 # now if the user has assigned a default index, we use it
[15600]271 if (defined $collectcfg->{"defaultindex"}) {
[24464]272 $defaultindex = $indexmap->{$collectcfg->{"defaultindex"}};
[15600]273 }
274 } else {
[26094]275 print STDERR "$maptype not defined\n";
[24464]276 }
[15600]277
[26094]278 $settings->{'num_indexes'} = $buildcfg->{'num_indexes'};
[24464]279 $settings->{'defaultindex'} = $defaultindex;
280 $settings->{'indexmap'} = $indexmap;
281 $settings->{'indexlist'} = $indexlist;
282
283 #--
284 # default lang
285 #--
286 $settings->{'default_lang'} = "";
287 $settings->{'default_lang_short'} = "";
288
289 if (defined $buildcfg->{"languagemap"}) {
290 my $langmap_t = $buildcfg->{"languagemap"};
291 if ((defined $langmap_t) && (scalar(@$langmap_t)>=1)) {
292 my $l = $langmap_t->[0];
293 my ($k, $v) = $l =~ m/^(.*)\-\>(.*)$/;
294 $settings->{'default_lang'} = $k; #name
295 $settings->{'default_lang_short'} = $v; #short name
296 }
297
298 # now if the user has assigned a default language (as "en", "ru" etc.)
299 if (defined $collectcfg->{"defaultlanguage"}) {
300 $settings->{'default_lang'} = $collectcfg->{"defaultlanguage"};
301 # what about default_lang_short ?? ####
302 }
[22485]303 }
[15600]304
[24464]305 # default subcol
306 $settings->{'default_subcol'} = "";
307 if (defined $buildcfg->{'subcollectionmap'}) {
308 my $subcolmap_t = $buildcfg->{'subcollectionmap'};
309 if ((defined $subcolmap_t) && (scalar(@$subcolmap_t)>=1)) {
310 my $l = $subcolmap_t->[0];
311 my ($k, $v) = $l =~ m/^(.*)\-\>(.*)$/;
[20065]312
[24464]313 $settings->{'default_subcol'} = $v;
[15600]314 }
315 }
316
[24464]317
318 #--
319 # indexstem
320 #--
321 if (defined $buildcfg->{'indexstem'}) {
322 $settings->{'indexstem'} = $buildcfg->{'indexstem'};
323 }
324
325 #--
326 # levelList
327 #--
328
[15600]329 my $levelmap = {};
[24464]330 my $levellist = [];
[15600]331 my $default_search_level = "Doc";
332 my $default_retrieve_level = "Doc";
[15685]333 my $default_db_level = "Doc";
[24464]334
[25758]335 if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
[15600]336 if (defined $buildcfg->{'levelmap'}) {
[24464]337 my $first = 1;
338
[15600]339 my $levelmap_t = $buildcfg->{'levelmap'};
340 foreach my $l (@$levelmap_t) {
341 my ($key, $val) = $l =~ /^(.*)\-\>(.*)$/;
342 $levelmap->{$key} = $val;
[24464]343 push @$levellist, $key;
[15600]344 if ($first) {
345 # let default search level follow the first level in the level list
346 $default_search_level = $val;
[15685]347 # retrieve/database levels may get modified later if text level is defined
[15600]348 $default_retrieve_level = $val;
[15685]349 $default_db_level = $val;
[15600]350 $first = 0;
351 }
352 }
353 }
354 # the default level assigned by the user is no longer ignored [Shaoqun], but the retrievel level stays the same.
[26094]355 if (defined $collectcfg->{"defaultlevel"}) {
356 $default_search_level = $levelmap->{$collectcfg->{"defaultlevel"}};
[15600]357 # $default_retrieve_level = $default_search_level;
[26094]358 }
[15600]359
360 if (defined $buildcfg->{'textlevel'}) {
[15685]361 # let the retrieve/database levels always follow the textlevel
[15600]362 $default_retrieve_level = $buildcfg->{'textlevel'};
[15685]363 $default_db_level = $buildcfg->{'textlevel'};
[15600]364
365 }
366 }
[24464]367 $settings->{'levelmap'} = $levelmap;
368 $settings->{'levellist'} = $levellist;
[26094]369 $settings->{'default_search_level'} = $default_search_level if $default_search_level;
[24464]370 $settings->{'default_retrieve_level'} = $default_retrieve_level;
371 $settings->{'default_db_level'} = $default_db_level;
372
[27803]373 # sort field list
374 ######
375
376 my $sortmap = {}; # maps index name to shortname
377 my $sortlist = []; # keeps the order for indexes
378
379 if (defined ($buildcfg->{"indexsortfieldmap"})) {
380 my $sortmap_t = $buildcfg->{"indexsortfieldmap"};
381 foreach my $s (@$sortmap_t) {
382 my ($k, $v) = $s =~ /^(.*)\-\>(.*)$/;
383 $sortmap->{$v} = $k;
384 }
385 }
386 if (defined ($buildcfg->{"indexsortfields"})) {
387 $sortlist = $buildcfg->{"indexsortfields"};
388 }
389
390 $settings->{'sortlist'} = $sortlist;
391 $settings->{'sortmap'} = $sortmap;
392
[27818]393 # facet field list
394 ######
[27803]395
[27818]396 my $facetmap = {}; # maps index name to shortname
397 my $facetlist = []; # keeps the order for indexes
[27803]398
[27818]399 if (defined ($buildcfg->{"indexfacetfieldmap"})) {
400 my $facetmap_t = $buildcfg->{"indexfacetfieldmap"};
401 foreach my $s (@$facetmap_t) {
402 my ($k, $v) = $s =~ /^(.*)\-\>(.*)$/;
403 $facetmap->{$v} = $k;
404 }
405 }
406 if (defined ($buildcfg->{"indexfacetfields"})) {
407 $facetlist = $buildcfg->{"indexfacetfields"};
408 }
409
410 $settings->{'facetlist'} = $facetlist;
411 $settings->{'facetmap'} = $facetmap;
412
413
[24464]414 return $settings;
415}
416
417
418sub write_search_servicerack
419{
420 my ($buildcfg,$settings) = @_;
421
422 my $buildtype = $settings->{'buildtype'};
423 my $infodbtype = $settings->{'infodbtype'};
424 my $service_type = $settings->{'service_type'};
425
[26094]426 # there's no searching and therefore no search services if there are no indexes
427 return if($settings->{'num_indexes'} <= 0);
428
[24464]429 # do the search service
430 &write_line('COLCFG', ["<serviceRack name=\"GS2", $service_type, "Search\">"]);
431 if (defined $buildcfg->{'indexstem'}) {
432 my $indexstem = $buildcfg->{'indexstem'};
433 &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
434 }
435 if (defined $buildcfg->{'infodbtype'}) {
436 &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
437 }
438
439 #indexes
440 my $indexmap = $settings->{'indexmap'};
441 my $indexlist = $settings->{'indexlist'};
442 my $defaultindex = $settings->{'defaultindex'};
443
444 #for each index in indexList, write them out
445 &write_line('COLCFG', ["<indexList>"]);
446 foreach my $i (@$indexlist) {
447 my $index = $indexmap->{$i};
448 &write_line('COLCFG', ["<index name=\"", $i, "\" ", "shortname=\"", $index, "\" />"]);
449 }
450 &write_line('COLCFG', ["</indexList>"]);
451
452
453 #$defaultindex = "ZZ" if (!$defaultindex); # index allfields by default
454 if ($defaultindex) {
455 &write_line('COLCFG', ["<defaultIndex shortname=\"", $defaultindex, "\" />"]);
456 }
457
458
459 # do indexOptionList
460 if ($buildtype eq "mg" || $buildtype eq "mgpp") {
461 &write_line('COLCFG', ["<indexOptionList>"]);
462 my $stemindexes = 3; # default is stem and casefold
463 if (defined $buildcfg->{'stemindexes'} && $buildcfg->{'stemindexes'} =~ /^\d+$/ ) {
464 $stemindexes = $buildcfg->{'stemindexes'};
465 }
466 &write_line('COLCFG', ["<indexOption name=\"stemIndexes\" value=\"", $stemindexes, "\" />"]);
467
468 my $maxnumeric = 4; # default
469 if (defined $buildcfg->{'maxnumeric'} && $buildcfg->{'maxnumeric'} =~ /^\d+$/) {
470 $maxnumeric = $buildcfg->{'maxnumeric'};
471 }
472 &write_line('COLCFG', ["<indexOption name=\"maxnumeric\" value=\"", $maxnumeric, "\" />"]);
473 &write_line('COLCFG', ["</indexOptionList>"]);
474 }
475
476 #--
477 # levelList
478 #--
479 my $levelmap = $settings->{'levelmap'};
480 my $levellist = $settings->{'levellist'};
481 my $default_search_level = $settings->{'default_search_level'};
482 my $default_retrieve_level = $settings->{'default_retrieve_level'};
483 my $default_db_level = $settings->{'default_db_level'};
484
[15600]485 #for each level in levelList, write them out
[21783]486 if ($buildtype ne "mg") {
[24464]487 &write_line('COLCFG', ["<levelList>"]);
488 foreach my $lv (@$levellist) {
489 my $level = $levelmap->{$lv};
490 &write_line('COLCFG', ["<level name=\"", $lv, "\" shortname=\"", $level, "\" />"]);
491 }
492 &write_line('COLCFG', ["</levelList>"]);
[15600]493 }
494 # add in defaultLevel as the same level as indexLevelList, making the reading job easier
[25758]495 if ($buildtype eq "lucene" || $buildtype eq "mgpp" || $buildtype eq "solr") {
[15600]496 &write_line('COLCFG', ["<defaultLevel shortname=\"", $default_search_level, "\" />"]);
497 }
[25758]498 if ($buildtype eq "lucene" || $buildtype eq "mgpp" || $buildtype eq "solr") {
[15685]499 &write_line('COLCFG', ["<defaultDBLevel shortname=\"", $default_db_level, "\" />"]);
[15600]500 }
[27803]501
502 # do sort list
503 if ($buildtype eq "lucene" || $buildtype eq "solr") {
504 my $sortlist = $settings->{'sortlist'};
505 my $sortmap = $settings->{'sortmap'};
506 &write_line('COLCFG', ["<sortList>"]);
507 foreach my $sf (@$sortlist) {
508 my $sortf = $sortmap->{$sf};
509 &write_line('COLCFG', ["<sort name=\"", $sortf, "\" shortname=\"", $sf, "\" />"]);
510 }
511 &write_line('COLCFG', ["</sortList>"]);
512 }
513
514 # do facet list
515 if ($buildtype eq "solr") {
516 &write_line('COLCFG', ["<facetList>"]);
517 my $facetlist = $settings->{'facetlist'};
518 my $facetmap = $settings->{'facetmap'};
519 foreach my $ff (@$facetlist) {
520 my $facetf = $facetmap->{$ff};
[27818]521 &write_line('COLCFG', ["<facet name=\"", $facetf, "\" shortname=\"", $ff, "\" />"]);
[27803]522 }
523 &write_line('COLCFG', ["</facetList>"]);
524 }
[15600]525 # do searchTypeList
[25758]526 if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
[15600]527 &write_line('COLCFG', ["<searchTypeList>"]);
528
529 if (defined $buildcfg->{"searchtype"}) {
530 my $searchtype_t = $buildcfg->{"searchtype"};
531 foreach my $s (@$searchtype_t) {
532 &write_line('COLCFG', ["<searchType name=\"", $s, "\" />"]);
533 }
534 } else {
535 &write_line('COLCFG', ["<searchType name=\"plain\" />"]);
536 &write_line('COLCFG', ["<searchType name=\"form\" />"]);
537 }
538 &write_line('COLCFG', ["</searchTypeList>"]);
539 }
540
541 # do indexLanguageList [in collect.cfg: languages; in build.cfg: languagemap]
[24464]542 my $default_lang = $settings->{'default_lang'};
543 my $default_lang_short = $settings->{'default_lang_short'};
[15600]544 if (defined $buildcfg->{"languagemap"}) {
545 &write_line('COLCFG', ["<indexLanguageList>"]);
546
547 my $langmap_t = $buildcfg->{"languagemap"};
548 foreach my $l (@$langmap_t) {
549 my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
550
551 &write_line('COLCFG', ["<indexLanguage name=\"", $k, "\" shortname=\"", $v, "\" />"]);
552 }
553
554 &write_line('COLCFG', ["</indexLanguageList>"]);
[24464]555
[15600]556 &write_line('COLCFG', ["<defaultIndexLanguage name=\"", $default_lang, "\" shortname=\"", $default_lang_short, "\" />"]);
557 }
558
[24464]559 # do indexSubcollectionList
560 my $default_subcol = $settings->{'default_subcol'};
[15600]561
562 if (defined $buildcfg->{'subcollectionmap'}) {
563 &write_line('COLCFG', ["<indexSubcollectionList>"]);
564 my $subcolmap = {};
565 my @subcollist = ();
[24464]566
[15600]567 my $subcolmap_t = $buildcfg->{'subcollectionmap'};
568 foreach my $l (@$subcolmap_t) {
569 my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
570 $subcolmap->{$k} = $v;
571 push @subcollist, $k;
572 }
[24464]573
[15600]574 foreach my $sl (@subcollist) {
575 my $subcol = $subcolmap->{$sl};
576 &write_line('COLCFG', ["<indexSubcollection name=\"", $sl, "\" shortname=\"", $subcol, "\" />"]);
577 }
578
579 &write_line('COLCFG', ["</indexSubcollectionList>"]);
580 &write_line('COLCFG', ["<defaultIndexSubcollection shortname=\"", $default_subcol, "\" />"]);
581 }
582
583 # close off search service
584 &write_line('COLCFG', ["</serviceRack>"]);
585
[24464]586}
587
588
589sub write_orthogonalsearch_serviceracks
590{
591 my ($buildcfg,$settings) = @_;
592
[26094]593 #return if($settings->{'num_indexes'} <= 0); # no search if no indexes
594
[24464]595 my $infodbtype = $settings->{'infodbtype'};
596
597 my $orthogonalbuildtypes = $settings->{'orthogonalbuildtypes'};
598
599 foreach my $obt (@$orthogonalbuildtypes) {
600 $obt =~ s/^(.)/\u$1/; # capitialize initial letter
601 $obt =~ s/-(.)/\u$1/g; # change any hyphenated words to cap next letter
602
603 &write_line('COLCFG', ["<serviceRack name=\"GS2", $obt, "Search\">"]);
604
605 &write_line('COLCFG',["<databaseType name=\"",$infodbtype,"\" />"]);
606 &write_line('COLCFG', ["</serviceRack>"]);
607 }
608}
609
610
611
612sub write_retrieve_servicerack
613{
614 my ($buildcfg,$settings) = @_;
615
616 my $buildtype = $settings->{'buildtype'};
617 my $infodbtype = $settings->{'infodbtype'};
618
619 my $service_type = $settings->{'service_type'};
620
[15600]621 # do the retrieve service
622 &write_line('COLCFG', ["<serviceRack name=\"GS2", $service_type, "Retrieve\">"]);
623
624 # do default index
625 if (defined $buildcfg->{"languagemap"}) {
[24464]626 my $default_lang = $settings->{'default_lang'};
[15600]627 &write_line('COLCFG', ["<defaultIndexLanguage shortname=\"", $default_lang, "\" />"]);
628 }
629 if (defined $buildcfg->{'subcollectionmap'}) {
[24464]630 my $default_subcol = $settings->{'default_subcol'};
[15600]631 &write_line('COLCFG', ["<defaultIndexSubcollection shortname=\"", $default_subcol, "\" />"]);
632 }
633 if ($buildtype eq "mg") {
[24464]634 my $defaultindex = $settings->{'defaultindex'};
[15600]635 &write_line('COLCFG', ["<defaultIndex shortname=\"", $defaultindex, "\" />"]);
636 }
637
638 if (defined $buildcfg->{'indexstem'}) {
639 my $indexstem = $buildcfg->{'indexstem'};
640 &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
641 }
[25758]642 if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
[24464]643 my $default_retrieve_level = $settings->{'default_retrieve_level'};
[15600]644 &write_line('COLCFG', ["<defaultLevel shortname=\"", $default_retrieve_level, "\" />"]);
645 }
[21439]646 if (defined $buildcfg->{'infodbtype'}) {
647 &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
648 }
649
[15600]650 &write_line('COLCFG', ["</serviceRack>"]);
651
[24464]652}
653
654
655# Create the buildConfig.xml file for a specific collection
656sub write_cfg_file {
657 # this sub is called in make_auxiliary_files() in basebuilder.pm
658 # the received args: $buildoutfile - destination file: buildConfig.xml
659 # $buildcfg - all build options,
660 # $collectcfg - contents of collectionConfig.xml read in by read_cfg_file sub in buildConfigxml.pm.
661 my ($buildoutfile, $buildcfg, $collectcfg) = @_;
662 my $line = [];
663
664 if (!open (COLCFG, ">$buildoutfile")) {
665 print STDERR "buildConfigxml::write_cfg_file couldn't write the build config file $buildoutfile\n";
666 die;
667 }
668
669 my $settings = search_and_retrieve_settings($buildcfg,$collectcfg);
670
671 my $buildtype = $settings->{'buildtype'};
672 my $orthogonalbuildtypes = $settings->{'orthogonalbuildtypes'};
673 my $numdocs = $settings->{'numdocs'};
674
675 &write_line('COLCFG', ["<buildConfig xmlns:gsf=\"http://www.greenstone.org/greenstone3/schema/ConfigFormat\">"]);
676
677 # output building metadata to build config file
678 &write_line('COLCFG', ["<metadataList>"]);
679 &write_line('COLCFG', ["<metadata name=\"numDocs\">", $numdocs, "</metadata>"]);
680 &write_line('COLCFG', ["<metadata name=\"buildType\">", $buildtype, "</metadata>"]);
681 foreach my $obt (@$orthogonalbuildtypes) {
682 &write_line('COLCFG', ["<metadata name=\"orthogonalBuildTypes\">", $obt, "</metadata>"]);
683 }
684
685 if (defined $buildcfg->{'indexstem'}) {
686 &write_line('COLCFG', ["<metadata name=\"indexStem\">", $buildcfg->{"indexstem"}, "</metadata>"]);
687 }
688 if (defined $buildcfg->{'infodbtype'}) {
689 &write_line('COLCFG', ["<metadata name=\"infodbType\">", $buildcfg->{"infodbtype"}, "</metadata>"]);
690 }
691 if (defined $buildcfg->{'earliestdatestamp'}) {
692 &write_line('COLCFG', ["<metadata name=\"earliestdatestamp\">", $buildcfg->{"earliestdatestamp"}, "</metadata>"]);
693 }
694
695 &write_line('COLCFG', ["</metadataList>"]);
696
697 # output serviceRackList
698 &write_line('COLCFG', ["<serviceRackList>"]);
699
700 write_search_servicerack($buildcfg,$settings);
701
702 # add in orthogonalbuildtypes
703 write_orthogonalsearch_serviceracks($buildcfg,$settings);
704
705 write_retrieve_servicerack($buildcfg,$settings);
706
[15600]707 # do the browse service
708 my $count = 1;
709 my $phind = 0;
710 my $started_classifiers = 0;
711
712 my $classifiers = $collectcfg->{"classify"};
713 foreach my $cl (@$classifiers) {
714 my $name = "CL$count";
715 $count++;
716 my ($classname) = @$cl[0];
717 if ($classname =~ /^phind$/i) {
718 $phind=1;
719 #should add it into coll config classifiers
720 next;
721 }
722
723 if (not $started_classifiers) {
724 &write_line('COLCFG', ["<serviceRack name=\"GS2Browse\">"]);
725 if (defined $buildcfg->{'indexstem'}) {
726 my $indexstem = $buildcfg->{'indexstem'};
727 &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
728 }
[21439]729 if (defined $buildcfg->{'infodbtype'}) {
730 my $infodbtype = $buildcfg->{'infodbtype'};
731 &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
732 }
[15600]733 &write_line('COLCFG', ["<classifierList>"]);
734 $started_classifiers = 1;
735 }
736 my $content = ''; #use buttonname first, then metadata
737 if ($classname eq "DateList") {
738 $content = "Date";
739 } else {
740 for (my $j=0; $j<scalar(@$cl); $j++) {
741 my $arg = @$cl[$j];
742 if ($arg eq "-buttonname"){
743 $content = @$cl[$j+1];
744 last;
745 } elsif ($arg eq "-metadata") {
746 $content = @$cl[$j+1];
747 }
748
749 }
750 }
751 &write_line('COLCFG', ["<classifier name=\"", $name, "\" content=\"", $content, "\" />"]);
752 }
753 if ($started_classifiers) {
754 # end the classifiers
755 &write_line('COLCFG', ["</classifierList>"]);
756 # close off the Browse service
757 &write_line('COLCFG', ["</serviceRack>"]);
758 }
759
760 # the phind classifier is a separate service
761 if ($phind) {
762 # if phind classifier
763 &write_line('COLCFG', ["<serviceRack name=\"PhindPhraseBrowse\" />"]);
764 }
765
766
767 &write_line('COLCFG', ["</serviceRackList>"]);
768 &write_line('COLCFG', ["</buildConfig>"]);
769
770 close (COLCFG);
771 }
772
773
774#########################################################
775
7761;
Note: See TracBrowser for help on using the repository browser.