source: main/trunk/greenstone2/perllib/collConfigxml.pm@ 28034

Last change on this file since 28034 was 28034, checked in by kjdon, 11 years ago

handling importOption and buildOption - they are not options in a importOptions list, just xxOptions at the top level
  • Property svn:keywords set to Author Date Id Revision
File size: 14.9 KB
Line 
1###########################################################################
2#
3# collConfigxml.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 configuration files of xml form
27
28package collConfigxml;
29use strict;
30no strict 'refs';
31no strict 'subs';
32
33use XMLParser;
34
35# A mapping hash to resolve name discrepancy between gs2 and gs3.
36# the first item is the gs3 element name from collectionConfig, the second one
37# is the internal name for the option
38my $nameMap = {"key" => "value",
39 "creator" => "creator",
40 "maintainer" => "maintainer",
41 "public" => "public",
42 "infodb" => "infodbtype",
43 "defaultIndex" => "defaultindex",
44 "defaultLevel" => "defaultlevel",
45 "name" => "collectionname",
46 "description" => "collectionextra",
47 "smallicon" => "iconcollectionsmall",
48 "icon" => "iconcollection",
49 "level" => "levels",
50 "classifier" => "classify",
51 "indexSubcollection" => "indexsubcollections",
52 "indexLanguage" => "languages",
53 "defaultIndexLanguage" => "defaultlanguage",
54 "index" => "indexes",
55 "sort" => "sortfields",
56 "facet" => "facetfields",
57 "plugin" => "plugin",
58 "plugout" => "plugout",
59 "indexOption" => "indexoptions",
60 "searchType" => "searchtype",
61 "languageMetadata" => "languagemetadata",
62 "buildType" => "buildtype",
63 "orthogonalBuildTypes" => "orthogonalbuildtypes",
64 };
65# A hash structure which is returned by sub read_cfg_file.
66my $data = {};
67
68my $repeatedBlock = q/^(browse|pluginList)$/;
69
70# use those unique attribute values to locate the text within the elements
71# creator, public, maintainer and within a displayItem.
72my $currentLocation = "";
73my $stringexp = q/^(creator|maintainer|public|buildType)$/;
74my $displayItemNames = q/^(name|description)$/;
75
76# these options get set at top level
77my $topleveloptionexp = q/^(importOption|buildOption)$/;
78
79# For storing the attributes during the StartTag subroutine, so that
80# we can use it later in Text (or EndTag) subroutines
81my $currentAttrRef = undef;
82
83my $currentLevel = "";
84
85# Count the elements with same name within the same block
86# ("plugin", "option")
87my $currentIndex = 0;
88my $arrayexp = q/^(index|sort|facet|level|indexOption|indexSubcollection|indexLanguage|orthogonalBuildTypes)$/;
89my $arrayarrayexp = q/^(plugin|classifier)$/; #|buildOption)$/;
90my $hashexp = q/^(subcollection)$/; # add other element names that should be represented by hash expressions here
91my $hashhashexp = q/^(displayItem)$/; # add other (collectionmeta) element names that should be represented by hashes of hashes here.
92
93my $defaults = q/^(defaultIndex|defaultLevel|defaultIndexLanguage|languageMetadata)$/;
94
95# Reads in the model collection configuration file, collectionConfig.xml,
96# into a structure which complies with the one used by gs2 (i.e. one read
97# in by &cfgread::read_cfg_file).
98sub read_cfg_file {
99 my ($filename) = @_;
100 $data = {};
101 if ($filename !~ /collectionConfig\.xml$/ || !-f $filename) {
102 return undef;
103 }
104
105 # Removed ProtocolEncoding (see MetadataXMLPlugin for details)
106
107 # create XML::Parser object for parsing metadata.xml files
108 my $parser = new XML::Parser('Style' => 'Stream',
109 'Pkg' => 'collConfigxml',
110 'Handlers' => {'Char' => \&Char,
111 'Doctype' => \&Doctype
112 });
113 if (!open (COLCFG, $filename)) {
114 print STDERR "cfgread::read_cfg_file couldn't read the cfg file $filename\n";
115 } else {
116
117 $parser->parsefile ($filename);# (COLCFG);
118 close (COLCFG);
119 }
120
121 #&Display;
122 return $data;
123}
124
125sub StartTag {
126# Those marked with #@ will not be executed at the same time when this sub is being called
127# so that if/elsif is used to avoid unnecessary tests
128 my ($expat, $element) = @_;
129
130 # See http://search.cpan.org/~msergeant/XML-Parser-2.36/Parser.pm#Stream
131 # %_ is a hash of all the attributes of this element, we want to store them so we can use the attributes
132 # when the textnode contents of the element are parsed in the subroutine Text (that's the handler for Text).
133 $currentAttrRef = \%_;
134
135 my $name = $_{'name'};
136 my $value = $_{'value'};
137 my $type = $_{'type'};
138 my $orthogonal = $_{'orthogonal'};
139
140 # for subcollections
141 my $filter = $_{'filter'};
142
143 # was this just a flax thing??
144 my $assigned = $_{'assigned'};
145
146 #@ Marking repeated block
147 if ($element =~ /$repeatedBlock/) {
148 $currentIndex = 0;
149 }
150
151 #@ handling block metadataList
152 elsif (defined $name and $name =~ /$stringexp/){
153 $currentLocation = $name;
154 }
155 #@ handling default search index/level/indexLanguage and languageMetadata
156 elsif ($element =~ /$defaults/) {
157 if (defined $name and $name =~ /\w/) {
158 $data->{$nameMap->{$element}} = $name;
159 }
160 }
161
162 #@ handling the displayItems name and description (known as collectionname and collectionextra in GS2)
163 elsif($element eq "displayItemList") {
164 $currentLevel = "displayItemList"; # storing the parent if it is displayItemList
165 }
166 elsif($element =~ /$hashhashexp/) { # can expand on this to check for other collectionmeta elements
167 if((!defined $assigned) || (defined $assigned and $assigned =~ /\w/ and $assigned eq "true")) {
168 # either when there is no "assigned" attribute, or when assigned=true (for displayItems):
169 $currentLocation = $name;
170 }
171 }
172
173 #@ Handling database type: gdbm or gdbm-txtgz, later jdbm.
174 elsif ($element eq "infodb") {
175 $data->{'infodbtype'} = $type;
176 }
177
178 #@ Handling indexer: mgpp/mg/lucene; stringexp
179 #@ Handling orthogonal indexers: audioDB; arrayexp
180 elsif ($element eq "search") {
181 if ((defined $orthogonal) && ($orthogonal =~ m/^(true|on|1)$/i)) {
182 push(@{$data->{'orthogonalbuildtypes'}},$type);
183 }
184 else {
185 $data->{'buildtype'} = $type;
186 }
187 }
188
189 elsif ($element eq "store_metadata_coverage")
190 {
191## print STDERR "*&*&*&*&*& HERE &*&*&*&*&*&*";
192 $data->{'store_metadata_coverage'} = $value;
193 }
194
195 #@ Handling searchtype: plain,form; arrayexp
196 #elsif ($element eq "format" and defined $name and $name =~ /searchType/) {
197 #@ Handling searchtype: plain, form
198 #$currentLocation = $name;
199 #}
200
201 #@ Handle index|sort|facet|level|indexOption|indexSubcollection|indexLanguage
202 elsif ($element =~ /$arrayexp/) {
203 my $key = $nameMap->{$element}; #
204 if (!defined $data->{$key}) {
205 $data->{$key} = [];
206 }
207
208 if (defined $name) {
209 push (@{$data->{$key}},$name);
210 }
211 }
212 # importOption and buildOption, just stored at top level, name=value,
213 # as per gs2 version
214 elsif ($element =~ /$topleveloptionexp/) {
215 if (defined $name) {
216 if (!defined $value) {
217 # flag option, set to true
218 $value = "true";
219 }
220 $data->{$name} = $value;
221 }
222 }
223
224 #@ plugout options
225 elsif ($element eq "plugout") {
226 $currentLevel = "plugout";
227 my $key = $nameMap->{$currentLevel};
228 if (!defined $data->{$key}) {
229 $data->{$key} = [];
230 }
231 if(defined $name and $name ne ""){
232 push (@{$data->{$key}},$name);
233 }
234 else{
235 push (@{$data->{$key}},"GreenstoneXMLPlugout");
236 }
237 }
238 if ($currentLevel eq "plugout" and $element eq "option") {
239 my $key = $nameMap->{$currentLevel};
240 if (defined $name and $name ne ""){
241 push (@{$data->{$key}},$name);
242 }
243 if (defined $value and $value ne ""){
244 push (@{$data->{$key}},$value);
245 }
246 }
247
248 #@ use hash of hash of strings: hashexp
249 elsif ($element =~ /$hashexp/) {
250 if (!defined $data->{$element}) {
251 $data->{$element} = {};
252 }
253 if (defined $name and $name =~ /\w/) {
254 if (defined $filter and $filter =~ /\w/) {
255 $data->{$element}->{$name} = $filter;
256
257 }
258 }
259 }
260
261 #@ Handling each classifier/plugin element
262 elsif ($element =~ /$arrayarrayexp/) {
263 # find the gs2 mapping name
264 $currentLevel = $element;
265 my $key = $nameMap->{$element};
266
267 # define an array of array of strings foreach $k (@{$data->{$key}}) {
268 if (!defined $data->{$key}) {
269 $data->{$key} = [];
270 }
271
272 # Push classifier/plugin name (e.g. AZList) into $data as the first string
273 push (@{$data->{$key}->[$currentIndex]},$name);
274 if (defined $value and $value =~ /\w/) {
275 push (@{$data->{$key}->[$currentIndex]}, $value);
276 print "$value\n";
277 }
278 #print $currentIndex."indexup\n";
279 }
280
281 #@ Handling the option elements in each classifier/plugin element (as the following strings)
282 elsif ($currentLevel =~ /$arrayarrayexp/ and $element eq "option") {
283 # find the gs2 mapping name for classifier and plugin
284 my $key = $nameMap->{$currentLevel};
285
286 if (defined $name and $name =~ /\w/) {
287 push (@{$data->{$key}->[$currentIndex]}, $name);
288 }
289 if (defined $value and $value =~ /\w/) {
290 push (@{$data->{$key}->[$currentIndex]}, $value);
291 }
292
293 }
294
295
296}
297
298sub EndTag {
299 my ($expat, $element) = @_;
300 my $endTags = q/^(browse|pluginList|displayItemList|indexOption)$/; #|buildOptionList)$/;
301 if ($element =~ /$endTags/) {
302 $currentIndex = 0;
303 $currentLevel = "";
304 }
305
306 # $arrayarrayexp contains classifier|plugin
307 elsif($element =~ /$arrayarrayexp/ ){
308 $currentIndex = $currentIndex + 1;
309 }
310}
311
312sub Text {
313 if (defined $currentLocation) {
314 #@ Handling block metadataList(creator, maintainer, public)
315 if($currentLocation =~ /$stringexp/){
316 #print $currentLocation;
317 my $key = $nameMap->{$currentLocation};
318 $data->{$key} = $_;
319 undef $currentLocation;
320 }
321
322 #@ Handling displayItem metadata that are children of displayItemList
323 # that means we will be getting the collection's name and possibly description ('collectionextra' in GS2).
324 elsif($currentLevel eq "displayItemList" && $currentLocation =~ /$displayItemNames/) {
325 my $lang = $currentAttrRef->{'lang'};
326 my $name = $currentAttrRef->{'name'};
327
328 # this is how data->collectionmeta's language is set in Greenstone 2.
329 # Need to be consistent, since export.pl accesses these values all in the same way
330 if(!defined $lang) {
331 $lang = 'default';
332 } else {
333 $lang = "[l=$lang]";
334 }
335
336 if(defined $name and $name =~ /$displayItemNames/) { # attribute name = 'name' || 'description'
337 # using $nameMap->$name resolves to 'collectionname' if $name='name' and 'collectionextra' if $name='description'
338 $data->{'collectionmeta'}->{$nameMap->{$name}}->{$lang} = $_; # the value is the Text parsed
339 #print STDERR "***Found: $nameMap->{$name} collectionmeta, lang is $lang. Value: $data->{'collectionmeta'}->{$nameMap->{$name}}->{$lang}\n";
340 }
341 undef $currentLocation;
342 }
343
344 #@ Handling searchtype: plain,form; arrayexp
345 elsif (defined $currentLocation and $currentLocation =~ /searchType/) {
346 # map 'searchType' into 'searchtype'
347 my $key = $nameMap->{$currentLocation};
348 # split it by ','
349 my ($plain, $form) = split (",", $_);
350
351 if (!defined $data->{$key}) {
352 $data->{$key} = [];
353 }
354 if (defined $plain and $plain =~ /\w/) {
355 push @{ $data->{$key} }, $plain;
356 }
357 if (defined $form and $form =~ /\w/) {
358 push @{ $data->{$key} }, $form;
359 }
360 }
361 }
362}
363
364# This sub is for debugging purposes
365sub Display {
366 # metadataList
367 foreach my $k (keys %{$data}) {
368 print STDERR "*** metadatalist key $k\n";
369 }
370
371 print STDERR "*** creator: ".$data->{'creator'}."\n" if (defined $data->{'creator'});
372 print STDERR "*** maintainer: ".$data->{"maintainer"}."\n" if (defined $data->{"maintainer"});
373 print STDERR "*** public: ".$data->{"public"}."\n" if (defined $data->{"public"});
374 print STDERR "*** default index: ".$data->{"defaultindex"}."\n" if (defined $data->{"defaultindex"});
375 print STDERR "*** default level: ".$data->{"defaultlevel"}."\n" if (defined $data->{"defaultlevel"});
376 print STDERR "*** build type: ".$data->{"buildtype"}."\n" if (defined $data->{"buildtype"});
377 print STDERR "*** orthogonal build types: ".join(",",$data->{"orthogonalbuildtypes"})."\n" if (defined $data->{"orthogonalbuildtypes"});
378 print STDERR "*** search types: \n";
379 print STDERR join(",",@{$data->{"searchtype"}})."\n" if (defined $data->{"searchtype"});
380 print STDERR "*** levels: \n";
381 print STDERR join(",",@{$data->{'levels'}})."\n" if (defined $data->{'levels'});
382 print STDERR "*** index subcollections: \n";
383 print STDERR join(",",@{$data->{'indexsubcollections'}})."\n" if (defined $data->{'indexsubcollections'});
384 print STDERR "*** indexes: \n";
385 print STDERR join(",",@{$data->{'indexes'}})."\n" if (defined $data->{'indexes'});
386 print STDERR "*** index options: \n";
387 print STDERR join(",",@{$data->{'indexoptions'}})."\n" if (defined $data->{'indexoptions'});
388 print STDERR "*** languages: \n";
389 print STDERR join(",",@{$data->{'languages'}})."\n" if (defined $data->{'languages'});
390 print STDERR "*** language metadata: \n";
391 print STDERR join(",",@{$data->{'languagemetadata'}})."\n" if (defined $data->{'languagemetadata'});
392
393 print STDERR "*** Plugins: \n";
394 if (defined $data->{'plugin'}) {
395 foreach $a (@{$data->{'plugin'}}) {
396 print join(",",@$a);
397 print "\n";
398 }
399 }
400
401 #print STDERR "*** Build options: \n";
402 #if (defined $data->{'store_metadata_coverage'}) {
403 #foreach $a (@{$data->{'store_metadata_coverage'}}) {
404 # print join(",",@$a,@$_);
405 # print "\n";
406 #}
407 #}
408
409 if (defined $data->{'classify'}) {
410 print STDERR "*** Classifiers: \n";
411 map { print join(",",@$_)."\n"; } @{$data->{'classify'}};
412 }
413
414 if (defined $data->{'subcollection'}) {
415 foreach my $key (keys %{$data->{'subcollection'}}) {
416 print "subcollection ".$key." ".$data->{'subcollection'}->{$key}."\n";
417 }
418 }
419}
420# is this actually used??
421sub Doctype {
422 my ($expat, $name, $sysid, $pubid, $internal) = @_;
423
424 die if ($name !~ /^CollectionConfig$/);
425}
426
427# This Char function overrides the one in XML::Parser::Stream to overcome a
428# problem where $expat->{Text} is treated as the return value, slowing
429# things down significantly in some cases.
430sub Char {
431 if ($]<5.008) {
432 use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+ and Perl 5.6
433 }
434 $_[0]->{'Text'} .= $_[1];
435 return undef;
436}
437
438
439
440
441#########################################################
442
4431;
Note: See TracBrowser for help on using the repository browser.