1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # buildConfigxml.pm --
|
---|
4 | # A component of the Greenstone digital library software
|
---|
5 | # from the New Zealand Digital Library Project at the
|
---|
6 | # University of Waikato, New Zealand.
|
---|
7 | #
|
---|
8 | # Copyright (C) 1999 New Zealand Digital Library Project
|
---|
9 | #
|
---|
10 | # This program is free software; you can redistribute it and/or modify
|
---|
11 | # it under the terms of the GNU General Public License as published by
|
---|
12 | # the Free Software Foundation; either version 2 of the License, or
|
---|
13 | # (at your option) any later version.
|
---|
14 | #
|
---|
15 | # This program is distributed in the hope that it will be useful,
|
---|
16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | # GNU General Public License for more details.
|
---|
19 | #
|
---|
20 | # You should have received a copy of the GNU General Public License
|
---|
21 | # along with this program; if not, write to the Free Software
|
---|
22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | #
|
---|
24 | ###########################################################################
|
---|
25 |
|
---|
26 | # 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.
|
---|
30 |
|
---|
31 | package buildConfigxml;
|
---|
32 |
|
---|
33 | use strict;
|
---|
34 | no strict 'refs';
|
---|
35 | no strict 'subs';
|
---|
36 |
|
---|
37 | use XMLParser;
|
---|
38 |
|
---|
39 |
|
---|
40 | # A mapping hash to resolve name discrepancy between gs2 and gs3.
|
---|
41 | my $nameMap = {"numDocs" => "numdocs",
|
---|
42 | "buildType" => "buildtype",
|
---|
43 | "orthogonalBuildTypes" => "orthogonalbuildtypes"
|
---|
44 | };
|
---|
45 |
|
---|
46 |
|
---|
47 | # A hash structure which is returned by sub read_cfg_file.
|
---|
48 | my $data = {};
|
---|
49 |
|
---|
50 | # use those unique attribute values to locate the text within the elements
|
---|
51 | my $currentLocation = "";
|
---|
52 | my $stringexp = q/^(buildType|numDocs)$/;
|
---|
53 | my $arrayexp = q/^(orthogonalBuildTypes)$/;
|
---|
54 |
|
---|
55 | my $indexmap_name = "";
|
---|
56 | my $haveindexfields = 0;
|
---|
57 |
|
---|
58 | # Reads in the model collection configuration file, collectionConfig.xml,
|
---|
59 | # into a structure which complies with the one used by gs2 (i.e. one read
|
---|
60 | # in by &cfgread::read_cfg_file).
|
---|
61 | sub read_cfg_file {
|
---|
62 | my ($filename) = @_;
|
---|
63 | $data = {};
|
---|
64 | if ($filename !~ /buildConfig\.xml$/ || !-f $filename) {
|
---|
65 | return undef;
|
---|
66 | }
|
---|
67 |
|
---|
68 | # Removed ProtocolEncoding (see MetadataXMLPlugin for details)
|
---|
69 |
|
---|
70 | # create XML::Parser object for parsing metadata.xml files
|
---|
71 | my $parser = new XML::Parser('Style' => 'Stream',
|
---|
72 | 'Pkg' => 'buildConfigxml',
|
---|
73 | 'Handlers' => {'Char' => \&Char,
|
---|
74 | 'Doctype' => \&Doctype
|
---|
75 | });
|
---|
76 |
|
---|
77 | if (!open (COLCFG, $filename)) {
|
---|
78 | print STDERR "buildConfigxml::read_cfg_file couldn't read the cfg file $filename\n";
|
---|
79 | } else {
|
---|
80 |
|
---|
81 | $parser->parsefile ($filename);# (COLCFG);
|
---|
82 | close (COLCFG);
|
---|
83 | }
|
---|
84 |
|
---|
85 | #&Display;
|
---|
86 | return $data;
|
---|
87 | }
|
---|
88 |
|
---|
89 | sub StartTag {
|
---|
90 | # Those marked with #@ will not be executed at the same time when this sub is being called
|
---|
91 | # so that if/elsif is used to avoid unnecessary tests
|
---|
92 | my ($expat, $element) = @_;
|
---|
93 |
|
---|
94 | my $name = $_{'name'};
|
---|
95 | my $shortname = $_{'shortname'};
|
---|
96 |
|
---|
97 |
|
---|
98 | #@ handling block metadataList
|
---|
99 | if (defined $name && (($name =~ m/$stringexp/) || ($name =~ m/$arrayexp/))) {
|
---|
100 | $currentLocation = $name;
|
---|
101 | # the value will be retrieved later in Text sub
|
---|
102 | }
|
---|
103 |
|
---|
104 | #@ handle indexes - store indexmap (mg) or indexfields and indexfieldmap (mgpp/lucene/solr)
|
---|
105 | elsif ($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 | }
|
---|
114 | }
|
---|
115 | else {
|
---|
116 | # mgpp, lucene or solr
|
---|
117 | $indexmap_name = "indexfieldmap";
|
---|
118 | $haveindexfields = 1;
|
---|
119 | if (!defined $data->{"indexfieldmap"}) {
|
---|
120 | $data->{"indexfieldmap"} = [];
|
---|
121 | }
|
---|
122 | if (!defined $data->{"indexfields"}) {
|
---|
123 | $data->{"indexfields"} = [];
|
---|
124 | }
|
---|
125 |
|
---|
126 | }
|
---|
127 |
|
---|
128 | }
|
---|
129 |
|
---|
130 | elsif ($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'}}, $name;
|
---|
136 | }
|
---|
137 | }
|
---|
138 | }
|
---|
139 |
|
---|
140 |
|
---|
141 | }
|
---|
142 |
|
---|
143 | sub EndTag {
|
---|
144 | my ($expat, $element) = @_;
|
---|
145 | }
|
---|
146 |
|
---|
147 | sub Text {
|
---|
148 | if (defined $currentLocation) {
|
---|
149 | #@ Handling block metadataList(numDocs, buildType)
|
---|
150 | if ($currentLocation =~ /$stringexp/) {
|
---|
151 | #print $currentLocation;
|
---|
152 | my $key = $nameMap->{$currentLocation};
|
---|
153 | $data->{$key} = $_;
|
---|
154 | undef $currentLocation;
|
---|
155 | }
|
---|
156 | elsif ($currentLocation =~ /$arrayexp/) {
|
---|
157 | #print $currentLocation;
|
---|
158 | my $key = $nameMap->{$currentLocation};
|
---|
159 | push(@{$data->{$key}},$_);
|
---|
160 | undef $currentLocation;
|
---|
161 | }
|
---|
162 |
|
---|
163 | }
|
---|
164 | }
|
---|
165 |
|
---|
166 | # This sub is for debugging purposes
|
---|
167 | sub Display {
|
---|
168 |
|
---|
169 | print "NumDocs = ".$data->{'numdocs'}."\n" if (defined $data->{'numdocs'});
|
---|
170 | print "BuildType = ".$data->{'buildtype'}."\n" if (defined $data->{'buildtype'});
|
---|
171 | print "OrthogonalBuildTypes = ".join(",",@{$data->{'orthogonalbuildtypes'}})."\n" if (defined $data->{'orthogonalbuildtypes'});
|
---|
172 | print "IndexMap = ". join(" ",@{$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 $data->{'indexfields'});
|
---|
175 |
|
---|
176 | }
|
---|
177 |
|
---|
178 | # is this actually used??
|
---|
179 | sub Doctype {
|
---|
180 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
181 |
|
---|
182 | die if ($name !~ /^buildConfig$/);
|
---|
183 | }
|
---|
184 |
|
---|
185 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
---|
186 | # problem where $expat->{Text} is treated as the return value, slowing
|
---|
187 | # things down significantly in some cases.
|
---|
188 | sub Char {
|
---|
189 | if ($]<5.008) {
|
---|
190 | use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+ and Perl 5.6
|
---|
191 | }
|
---|
192 | $_[0]->{'Text'} .= $_[1];
|
---|
193 | return undef;
|
---|
194 | }
|
---|
195 |
|
---|
196 |
|
---|
197 |
|
---|
198 | sub write_line {
|
---|
199 | my ($filehandle, $line) = @_;
|
---|
200 | print $filehandle join ("", @$line), "\n";
|
---|
201 | }
|
---|
202 |
|
---|
203 | sub search_and_retrieve_settings
|
---|
204 | {
|
---|
205 | my ($buildcfg,$collectcfg) = @_;
|
---|
206 |
|
---|
207 | my $settings = {};
|
---|
208 |
|
---|
209 | if (defined $buildcfg->{"buildtype"}) {
|
---|
210 | $settings->{'buildtype'} = $buildcfg->{"buildtype"};
|
---|
211 | } else {
|
---|
212 | $settings->{'buildtype'} = "mgpp";
|
---|
213 | }
|
---|
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 |
|
---|
223 | if (defined $buildcfg->{"numdocs"}) {
|
---|
224 | $settings->{'numdocs'} = $buildcfg->{"numdocs"};
|
---|
225 | }
|
---|
226 | else {
|
---|
227 | $settings->{'numdocs'} = 0;
|
---|
228 | }
|
---|
229 |
|
---|
230 | my $service_type = "MGPP";
|
---|
231 | if ($buildtype eq "mg") {
|
---|
232 | $service_type = "MG";
|
---|
233 | } elsif ($buildtype eq "lucene") {
|
---|
234 | $service_type = "Lucene";
|
---|
235 | } elsif ($buildtype eq "solr") {
|
---|
236 | $service_type = "Solr";
|
---|
237 | }
|
---|
238 | $settings->{'service_type'} = $service_type;
|
---|
239 |
|
---|
240 |
|
---|
241 | if (defined $buildcfg->{"infodbtype"}) {
|
---|
242 | $settings->{'infodbtype'} = $buildcfg->{'infodbtype'};
|
---|
243 | }
|
---|
244 | else {
|
---|
245 | $settings->{'infodbtype'} = "gdbm";
|
---|
246 | }
|
---|
247 |
|
---|
248 |
|
---|
249 | #--
|
---|
250 | # indexes
|
---|
251 | #--
|
---|
252 |
|
---|
253 | my $indexmap = {}; # maps index name to shortname
|
---|
254 | my $indexlist = []; # keeps the order for indexes
|
---|
255 | my $defaultindex = "";
|
---|
256 | my $maptype = ($buildtype eq "mg") ? "indexmap" : "indexfieldmap";
|
---|
257 |
|
---|
258 | if (defined $buildcfg->{$maptype}) {
|
---|
259 | my $first = 1;
|
---|
260 | my $indexmap_t = $buildcfg->{$maptype};
|
---|
261 | foreach my $i (@$indexmap_t) {
|
---|
262 | my ($k, $v) = $i =~ /^(.*)\-\>(.*)$/;
|
---|
263 | $indexmap->{$k} = $v;
|
---|
264 | push @$indexlist, $k;
|
---|
265 | if ($first) {
|
---|
266 | $defaultindex = $v;
|
---|
267 | $first = 0;
|
---|
268 | }
|
---|
269 | }
|
---|
270 | # now if the user has assigned a default index, we use it
|
---|
271 | if (defined $collectcfg->{"defaultindex"}) {
|
---|
272 | $defaultindex = $indexmap->{$collectcfg->{"defaultindex"}};
|
---|
273 | }
|
---|
274 | } else {
|
---|
275 | print STDERR "$maptype not defined\n";
|
---|
276 | }
|
---|
277 |
|
---|
278 | $settings->{'num_indexes'} = $buildcfg->{'num_indexes'};
|
---|
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 | }
|
---|
303 | }
|
---|
304 |
|
---|
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) = $l =~ m/^(.*)\-\>(.*)$/;
|
---|
312 |
|
---|
313 | $settings->{'default_subcol'} = $v;
|
---|
314 | }
|
---|
315 | }
|
---|
316 |
|
---|
317 |
|
---|
318 | #--
|
---|
319 | # indexstem
|
---|
320 | #--
|
---|
321 | if (defined $buildcfg->{'indexstem'}) {
|
---|
322 | $settings->{'indexstem'} = $buildcfg->{'indexstem'};
|
---|
323 | }
|
---|
324 |
|
---|
325 | #--
|
---|
326 | # levelList
|
---|
327 | #--
|
---|
328 |
|
---|
329 | my $levelmap = {};
|
---|
330 | my $levellist = [];
|
---|
331 | my $default_search_level = "Doc";
|
---|
332 | my $default_retrieve_level = "Doc";
|
---|
333 | my $default_db_level = "Doc";
|
---|
334 |
|
---|
335 | if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
|
---|
336 | if (defined $buildcfg->{'levelmap'}) {
|
---|
337 | my $first = 1;
|
---|
338 |
|
---|
339 | my $levelmap_t = $buildcfg->{'levelmap'};
|
---|
340 | foreach my $l (@$levelmap_t) {
|
---|
341 | my ($key, $val) = $l =~ /^(.*)\-\>(.*)$/;
|
---|
342 | $levelmap->{$key} = $val;
|
---|
343 | push @$levellist, $key;
|
---|
344 | if ($first) {
|
---|
345 | # let default search level follow the first level in the level list
|
---|
346 | $default_search_level = $val;
|
---|
347 | # retrieve/database levels may get modified later if text level is defined
|
---|
348 | $default_retrieve_level = $val;
|
---|
349 | $default_db_level = $val;
|
---|
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.
|
---|
355 | if (defined $collectcfg->{"defaultlevel"}) {
|
---|
356 | $default_search_level = $levelmap->{$collectcfg->{"defaultlevel"}};
|
---|
357 | # $default_retrieve_level = $default_search_level;
|
---|
358 | }
|
---|
359 |
|
---|
360 | if (defined $buildcfg->{'textlevel'}) {
|
---|
361 | # let the retrieve/database levels always follow the textlevel
|
---|
362 | $default_retrieve_level = $buildcfg->{'textlevel'};
|
---|
363 | $default_db_level = $buildcfg->{'textlevel'};
|
---|
364 |
|
---|
365 | }
|
---|
366 | }
|
---|
367 | $settings->{'levelmap'} = $levelmap;
|
---|
368 | $settings->{'levellist'} = $levellist;
|
---|
369 | $settings->{'default_search_level'} = $default_search_level if $default_search_level;
|
---|
370 | $settings->{'default_retrieve_level'} = $default_retrieve_level;
|
---|
371 | $settings->{'default_db_level'} = $default_db_level;
|
---|
372 |
|
---|
373 |
|
---|
374 | return $settings;
|
---|
375 | }
|
---|
376 |
|
---|
377 |
|
---|
378 | sub write_search_servicerack
|
---|
379 | {
|
---|
380 | my ($buildcfg,$settings) = @_;
|
---|
381 |
|
---|
382 | my $buildtype = $settings->{'buildtype'};
|
---|
383 | my $infodbtype = $settings->{'infodbtype'};
|
---|
384 | my $service_type = $settings->{'service_type'};
|
---|
385 |
|
---|
386 | # there's no searching and therefore no search services if there are no indexes
|
---|
387 | return if($settings->{'num_indexes'} <= 0);
|
---|
388 |
|
---|
389 | # do the search service
|
---|
390 | &write_line('COLCFG', ["<serviceRack name=\"GS2", $service_type, "Search\">"]);
|
---|
391 | if (defined $buildcfg->{'indexstem'}) {
|
---|
392 | my $indexstem = $buildcfg->{'indexstem'};
|
---|
393 | &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
|
---|
394 | }
|
---|
395 | if (defined $buildcfg->{'infodbtype'}) {
|
---|
396 | &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
|
---|
397 | }
|
---|
398 |
|
---|
399 | #indexes
|
---|
400 | my $indexmap = $settings->{'indexmap'};
|
---|
401 | my $indexlist = $settings->{'indexlist'};
|
---|
402 | my $defaultindex = $settings->{'defaultindex'};
|
---|
403 |
|
---|
404 | #for each index in indexList, write them out
|
---|
405 | &write_line('COLCFG', ["<indexList>"]);
|
---|
406 | foreach my $i (@$indexlist) {
|
---|
407 | my $index = $indexmap->{$i};
|
---|
408 | &write_line('COLCFG', ["<index name=\"", $i, "\" ", "shortname=\"", $index, "\" />"]);
|
---|
409 | }
|
---|
410 | &write_line('COLCFG', ["</indexList>"]);
|
---|
411 |
|
---|
412 |
|
---|
413 | #$defaultindex = "ZZ" if (!$defaultindex); # index allfields by default
|
---|
414 | if ($defaultindex) {
|
---|
415 | &write_line('COLCFG', ["<defaultIndex shortname=\"", $defaultindex, "\" />"]);
|
---|
416 | }
|
---|
417 |
|
---|
418 |
|
---|
419 | # do indexOptionList
|
---|
420 | if ($buildtype eq "mg" || $buildtype eq "mgpp") {
|
---|
421 | &write_line('COLCFG', ["<indexOptionList>"]);
|
---|
422 | my $stemindexes = 3; # default is stem and casefold
|
---|
423 | if (defined $buildcfg->{'stemindexes'} && $buildcfg->{'stemindexes'} =~ /^\d+$/ ) {
|
---|
424 | $stemindexes = $buildcfg->{'stemindexes'};
|
---|
425 | }
|
---|
426 | &write_line('COLCFG', ["<indexOption name=\"stemIndexes\" value=\"", $stemindexes, "\" />"]);
|
---|
427 |
|
---|
428 | my $maxnumeric = 4; # default
|
---|
429 | if (defined $buildcfg->{'maxnumeric'} && $buildcfg->{'maxnumeric'} =~ /^\d+$/) {
|
---|
430 | $maxnumeric = $buildcfg->{'maxnumeric'};
|
---|
431 | }
|
---|
432 | &write_line('COLCFG', ["<indexOption name=\"maxnumeric\" value=\"", $maxnumeric, "\" />"]);
|
---|
433 | &write_line('COLCFG', ["</indexOptionList>"]);
|
---|
434 | }
|
---|
435 |
|
---|
436 | #--
|
---|
437 | # levelList
|
---|
438 | #--
|
---|
439 | my $levelmap = $settings->{'levelmap'};
|
---|
440 | my $levellist = $settings->{'levellist'};
|
---|
441 | my $default_search_level = $settings->{'default_search_level'};
|
---|
442 | my $default_retrieve_level = $settings->{'default_retrieve_level'};
|
---|
443 | my $default_db_level = $settings->{'default_db_level'};
|
---|
444 |
|
---|
445 | #for each level in levelList, write them out
|
---|
446 | if ($buildtype ne "mg") {
|
---|
447 | &write_line('COLCFG', ["<levelList>"]);
|
---|
448 | foreach my $lv (@$levellist) {
|
---|
449 | my $level = $levelmap->{$lv};
|
---|
450 | &write_line('COLCFG', ["<level name=\"", $lv, "\" shortname=\"", $level, "\" />"]);
|
---|
451 | }
|
---|
452 | &write_line('COLCFG', ["</levelList>"]);
|
---|
453 | }
|
---|
454 | # add in defaultLevel as the same level as indexLevelList, making the reading job easier
|
---|
455 | if ($buildtype eq "lucene" || $buildtype eq "mgpp" || $buildtype eq "solr") {
|
---|
456 | &write_line('COLCFG', ["<defaultLevel shortname=\"", $default_search_level, "\" />"]);
|
---|
457 | }
|
---|
458 | if ($buildtype eq "lucene" || $buildtype eq "mgpp" || $buildtype eq "solr") {
|
---|
459 | &write_line('COLCFG', ["<defaultDBLevel shortname=\"", $default_db_level, "\" />"]);
|
---|
460 | }
|
---|
461 | # do searchTypeList
|
---|
462 | if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
|
---|
463 | &write_line('COLCFG', ["<searchTypeList>"]);
|
---|
464 |
|
---|
465 | if (defined $buildcfg->{"searchtype"}) {
|
---|
466 | my $searchtype_t = $buildcfg->{"searchtype"};
|
---|
467 | foreach my $s (@$searchtype_t) {
|
---|
468 | &write_line('COLCFG', ["<searchType name=\"", $s, "\" />"]);
|
---|
469 | }
|
---|
470 | } else {
|
---|
471 | &write_line('COLCFG', ["<searchType name=\"plain\" />"]);
|
---|
472 | &write_line('COLCFG', ["<searchType name=\"form\" />"]);
|
---|
473 | }
|
---|
474 | &write_line('COLCFG', ["</searchTypeList>"]);
|
---|
475 | }
|
---|
476 |
|
---|
477 | # do indexLanguageList [in collect.cfg: languages; in build.cfg: languagemap]
|
---|
478 | my $default_lang = $settings->{'default_lang'};
|
---|
479 | my $default_lang_short = $settings->{'default_lang_short'};
|
---|
480 | if (defined $buildcfg->{"languagemap"}) {
|
---|
481 | &write_line('COLCFG', ["<indexLanguageList>"]);
|
---|
482 |
|
---|
483 | my $langmap_t = $buildcfg->{"languagemap"};
|
---|
484 | foreach my $l (@$langmap_t) {
|
---|
485 | my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
|
---|
486 |
|
---|
487 | &write_line('COLCFG', ["<indexLanguage name=\"", $k, "\" shortname=\"", $v, "\" />"]);
|
---|
488 | }
|
---|
489 |
|
---|
490 | &write_line('COLCFG', ["</indexLanguageList>"]);
|
---|
491 |
|
---|
492 | &write_line('COLCFG', ["<defaultIndexLanguage name=\"", $default_lang, "\" shortname=\"", $default_lang_short, "\" />"]);
|
---|
493 | }
|
---|
494 |
|
---|
495 | # do indexSubcollectionList
|
---|
496 | my $default_subcol = $settings->{'default_subcol'};
|
---|
497 |
|
---|
498 | if (defined $buildcfg->{'subcollectionmap'}) {
|
---|
499 | &write_line('COLCFG', ["<indexSubcollectionList>"]);
|
---|
500 | my $subcolmap = {};
|
---|
501 | my @subcollist = ();
|
---|
502 |
|
---|
503 | my $subcolmap_t = $buildcfg->{'subcollectionmap'};
|
---|
504 | foreach my $l (@$subcolmap_t) {
|
---|
505 | my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
|
---|
506 | $subcolmap->{$k} = $v;
|
---|
507 | push @subcollist, $k;
|
---|
508 | }
|
---|
509 |
|
---|
510 | foreach my $sl (@subcollist) {
|
---|
511 | my $subcol = $subcolmap->{$sl};
|
---|
512 | &write_line('COLCFG', ["<indexSubcollection name=\"", $sl, "\" shortname=\"", $subcol, "\" />"]);
|
---|
513 | }
|
---|
514 |
|
---|
515 | &write_line('COLCFG', ["</indexSubcollectionList>"]);
|
---|
516 | &write_line('COLCFG', ["<defaultIndexSubcollection shortname=\"", $default_subcol, "\" />"]);
|
---|
517 | }
|
---|
518 |
|
---|
519 | # close off search service
|
---|
520 | &write_line('COLCFG', ["</serviceRack>"]);
|
---|
521 |
|
---|
522 | }
|
---|
523 |
|
---|
524 |
|
---|
525 | sub write_orthogonalsearch_serviceracks
|
---|
526 | {
|
---|
527 | my ($buildcfg,$settings) = @_;
|
---|
528 |
|
---|
529 | #return if($settings->{'num_indexes'} <= 0); # no search if no indexes
|
---|
530 |
|
---|
531 | my $infodbtype = $settings->{'infodbtype'};
|
---|
532 |
|
---|
533 | my $orthogonalbuildtypes = $settings->{'orthogonalbuildtypes'};
|
---|
534 |
|
---|
535 | foreach my $obt (@$orthogonalbuildtypes) {
|
---|
536 | $obt =~ s/^(.)/\u$1/; # capitialize initial letter
|
---|
537 | $obt =~ s/-(.)/\u$1/g; # change any hyphenated words to cap next letter
|
---|
538 |
|
---|
539 | &write_line('COLCFG', ["<serviceRack name=\"GS2", $obt, "Search\">"]);
|
---|
540 |
|
---|
541 | &write_line('COLCFG',["<databaseType name=\"",$infodbtype,"\" />"]);
|
---|
542 | &write_line('COLCFG', ["</serviceRack>"]);
|
---|
543 | }
|
---|
544 | }
|
---|
545 |
|
---|
546 |
|
---|
547 |
|
---|
548 | sub write_retrieve_servicerack
|
---|
549 | {
|
---|
550 | my ($buildcfg,$settings) = @_;
|
---|
551 |
|
---|
552 | my $buildtype = $settings->{'buildtype'};
|
---|
553 | my $infodbtype = $settings->{'infodbtype'};
|
---|
554 |
|
---|
555 | my $service_type = $settings->{'service_type'};
|
---|
556 |
|
---|
557 | # do the retrieve service
|
---|
558 | &write_line('COLCFG', ["<serviceRack name=\"GS2", $service_type, "Retrieve\">"]);
|
---|
559 |
|
---|
560 | # do default index
|
---|
561 | if (defined $buildcfg->{"languagemap"}) {
|
---|
562 | my $default_lang = $settings->{'default_lang'};
|
---|
563 | &write_line('COLCFG', ["<defaultIndexLanguage shortname=\"", $default_lang, "\" />"]);
|
---|
564 | }
|
---|
565 | if (defined $buildcfg->{'subcollectionmap'}) {
|
---|
566 | my $default_subcol = $settings->{'default_subcol'};
|
---|
567 | &write_line('COLCFG', ["<defaultIndexSubcollection shortname=\"", $default_subcol, "\" />"]);
|
---|
568 | }
|
---|
569 | if ($buildtype eq "mg") {
|
---|
570 | my $defaultindex = $settings->{'defaultindex'};
|
---|
571 | &write_line('COLCFG', ["<defaultIndex shortname=\"", $defaultindex, "\" />"]);
|
---|
572 | }
|
---|
573 |
|
---|
574 | if (defined $buildcfg->{'indexstem'}) {
|
---|
575 | my $indexstem = $buildcfg->{'indexstem'};
|
---|
576 | &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
|
---|
577 | }
|
---|
578 | if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
|
---|
579 | my $default_retrieve_level = $settings->{'default_retrieve_level'};
|
---|
580 | &write_line('COLCFG', ["<defaultLevel shortname=\"", $default_retrieve_level, "\" />"]);
|
---|
581 | }
|
---|
582 | if (defined $buildcfg->{'infodbtype'}) {
|
---|
583 | &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
|
---|
584 | }
|
---|
585 |
|
---|
586 | &write_line('COLCFG', ["</serviceRack>"]);
|
---|
587 |
|
---|
588 | }
|
---|
589 |
|
---|
590 |
|
---|
591 | # Create the buildConfig.xml file for a specific collection
|
---|
592 | sub write_cfg_file {
|
---|
593 | # this sub is called in make_auxiliary_files() in basebuilder.pm
|
---|
594 | # the received args: $buildoutfile - destination file: buildConfig.xml
|
---|
595 | # $buildcfg - all build options,
|
---|
596 | # $collectcfg - contents of collectionConfig.xml read in by read_cfg_file sub in buildConfigxml.pm.
|
---|
597 | my ($buildoutfile, $buildcfg, $collectcfg) = @_;
|
---|
598 | my $line = [];
|
---|
599 |
|
---|
600 | if (!open (COLCFG, ">$buildoutfile")) {
|
---|
601 | print STDERR "buildConfigxml::write_cfg_file couldn't write the build config file $buildoutfile\n";
|
---|
602 | die;
|
---|
603 | }
|
---|
604 |
|
---|
605 | my $settings = search_and_retrieve_settings($buildcfg,$collectcfg);
|
---|
606 |
|
---|
607 | my $buildtype = $settings->{'buildtype'};
|
---|
608 | my $orthogonalbuildtypes = $settings->{'orthogonalbuildtypes'};
|
---|
609 | my $numdocs = $settings->{'numdocs'};
|
---|
610 |
|
---|
611 | &write_line('COLCFG', ["<buildConfig xmlns:gsf=\"http://www.greenstone.org/greenstone3/schema/ConfigFormat\">"]);
|
---|
612 |
|
---|
613 | # output building metadata to build config file
|
---|
614 | &write_line('COLCFG', ["<metadataList>"]);
|
---|
615 | &write_line('COLCFG', ["<metadata name=\"numDocs\">", $numdocs, "</metadata>"]);
|
---|
616 | &write_line('COLCFG', ["<metadata name=\"buildType\">", $buildtype, "</metadata>"]);
|
---|
617 | foreach my $obt (@$orthogonalbuildtypes) {
|
---|
618 | &write_line('COLCFG', ["<metadata name=\"orthogonalBuildTypes\">", $obt, "</metadata>"]);
|
---|
619 | }
|
---|
620 |
|
---|
621 | if (defined $buildcfg->{'indexstem'}) {
|
---|
622 | &write_line('COLCFG', ["<metadata name=\"indexStem\">", $buildcfg->{"indexstem"}, "</metadata>"]);
|
---|
623 | }
|
---|
624 | if (defined $buildcfg->{'infodbtype'}) {
|
---|
625 | &write_line('COLCFG', ["<metadata name=\"infodbType\">", $buildcfg->{"infodbtype"}, "</metadata>"]);
|
---|
626 | }
|
---|
627 | if (defined $buildcfg->{'earliestdatestamp'}) {
|
---|
628 | &write_line('COLCFG', ["<metadata name=\"earliestdatestamp\">", $buildcfg->{"earliestdatestamp"}, "</metadata>"]);
|
---|
629 | }
|
---|
630 |
|
---|
631 | &write_line('COLCFG', ["</metadataList>"]);
|
---|
632 |
|
---|
633 | # output serviceRackList
|
---|
634 | &write_line('COLCFG', ["<serviceRackList>"]);
|
---|
635 |
|
---|
636 | write_search_servicerack($buildcfg,$settings);
|
---|
637 |
|
---|
638 | # add in orthogonalbuildtypes
|
---|
639 | write_orthogonalsearch_serviceracks($buildcfg,$settings);
|
---|
640 |
|
---|
641 | write_retrieve_servicerack($buildcfg,$settings);
|
---|
642 |
|
---|
643 | # do the browse service
|
---|
644 | my $count = 1;
|
---|
645 | my $phind = 0;
|
---|
646 | my $started_classifiers = 0;
|
---|
647 |
|
---|
648 | my $classifiers = $collectcfg->{"classify"};
|
---|
649 | foreach my $cl (@$classifiers) {
|
---|
650 | my $name = "CL$count";
|
---|
651 | $count++;
|
---|
652 | my ($classname) = @$cl[0];
|
---|
653 | if ($classname =~ /^phind$/i) {
|
---|
654 | $phind=1;
|
---|
655 | #should add it into coll config classifiers
|
---|
656 | next;
|
---|
657 | }
|
---|
658 |
|
---|
659 | if (not $started_classifiers) {
|
---|
660 | &write_line('COLCFG', ["<serviceRack name=\"GS2Browse\">"]);
|
---|
661 | if (defined $buildcfg->{'indexstem'}) {
|
---|
662 | my $indexstem = $buildcfg->{'indexstem'};
|
---|
663 | &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
|
---|
664 | }
|
---|
665 | if (defined $buildcfg->{'infodbtype'}) {
|
---|
666 | my $infodbtype = $buildcfg->{'infodbtype'};
|
---|
667 | &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
|
---|
668 | }
|
---|
669 | &write_line('COLCFG', ["<classifierList>"]);
|
---|
670 | $started_classifiers = 1;
|
---|
671 | }
|
---|
672 | my $content = ''; #use buttonname first, then metadata
|
---|
673 | if ($classname eq "DateList") {
|
---|
674 | $content = "Date";
|
---|
675 | } else {
|
---|
676 | for (my $j=0; $j<scalar(@$cl); $j++) {
|
---|
677 | my $arg = @$cl[$j];
|
---|
678 | if ($arg eq "-buttonname"){
|
---|
679 | $content = @$cl[$j+1];
|
---|
680 | last;
|
---|
681 | } elsif ($arg eq "-metadata") {
|
---|
682 | $content = @$cl[$j+1];
|
---|
683 | }
|
---|
684 |
|
---|
685 | }
|
---|
686 | }
|
---|
687 | &write_line('COLCFG', ["<classifier name=\"", $name, "\" content=\"", $content, "\" />"]);
|
---|
688 | }
|
---|
689 | if ($started_classifiers) {
|
---|
690 | # end the classifiers
|
---|
691 | &write_line('COLCFG', ["</classifierList>"]);
|
---|
692 | # close off the Browse service
|
---|
693 | &write_line('COLCFG', ["</serviceRack>"]);
|
---|
694 | }
|
---|
695 |
|
---|
696 | # the phind classifier is a separate service
|
---|
697 | if ($phind) {
|
---|
698 | # if phind classifier
|
---|
699 | &write_line('COLCFG', ["<serviceRack name=\"PhindPhraseBrowse\" />"]);
|
---|
700 | }
|
---|
701 |
|
---|
702 |
|
---|
703 | &write_line('COLCFG', ["</serviceRackList>"]);
|
---|
704 | &write_line('COLCFG', ["</buildConfig>"]);
|
---|
705 |
|
---|
706 | close (COLCFG);
|
---|
707 | }
|
---|
708 |
|
---|
709 |
|
---|
710 | #########################################################
|
---|
711 |
|
---|
712 | 1;
|
---|