1 | #!/usr/bin/perl -w
|
---|
2 |
|
---|
3 | BEGIN {
|
---|
4 | die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
|
---|
5 | die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
|
---|
6 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
|
---|
7 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
|
---|
8 | #unshift (@INC, "$ENV{'GSDLHOME'}/perllib/plugins");
|
---|
9 | # unshift (@INC, "$ENV{'GSDLHOME'}/perllib/classify");
|
---|
10 | }
|
---|
11 |
|
---|
12 | #use arcinfo;
|
---|
13 | use colcfg;
|
---|
14 | #use plugin;
|
---|
15 | use docprint;
|
---|
16 | use util;
|
---|
17 | use parsargv;
|
---|
18 | use FileHandle;
|
---|
19 | use XML::Writer;
|
---|
20 |
|
---|
21 | &main();
|
---|
22 | sub print_usage() {
|
---|
23 | print STDOUT "Usage: convert_coll_from_gs2.pl [options] coll-name\n";
|
---|
24 | print STDOUT "options:\n";
|
---|
25 |
|
---|
26 | print STDOUT " -collectdir Directory where collection lives.\n";
|
---|
27 |
|
---|
28 | }
|
---|
29 |
|
---|
30 | sub main {
|
---|
31 |
|
---|
32 | my $defaultlang = 'en';
|
---|
33 | my ($collectdir);
|
---|
34 | # note that no defaults are passed for most options as they're set
|
---|
35 | # later (after we check the collect.cfg file)
|
---|
36 | if (!parsargv::parse(\@ARGV,
|
---|
37 | 'verbosity/\d+/', \$verbosity,
|
---|
38 | 'collectdir/.*/', \$collectdir,
|
---|
39 | 'faillog/.*/', \$faillog)) {
|
---|
40 | &print_usage();
|
---|
41 | die "\n";
|
---|
42 | }
|
---|
43 |
|
---|
44 | # get and check the collection name
|
---|
45 | if (($collection = &util::use_collection(@ARGV, $collectdir)) eq "") {
|
---|
46 | &print_usage();
|
---|
47 | die "\n";
|
---|
48 | }
|
---|
49 |
|
---|
50 | $collconfigfilename = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "etc", "collect.cfg");
|
---|
51 | my $collectcfg;
|
---|
52 | if (-e $collconfigfilename) {
|
---|
53 | $collectcfg = &colcfg::read_collect_cfg ($collconfigfilename);
|
---|
54 |
|
---|
55 | } else {
|
---|
56 | print STDERR "collect.cfg not found!!";
|
---|
57 | die "\n";
|
---|
58 | }
|
---|
59 |
|
---|
60 |
|
---|
61 | $buildconfigfilename = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "index", "build.cfg");
|
---|
62 | my $buildcfg;
|
---|
63 | if (-e $buildconfigfilename) {
|
---|
64 | $buildcfg = &colcfg::read_build_cfg ($buildconfigfilename);
|
---|
65 |
|
---|
66 | } else {
|
---|
67 | print STDERR "build.cfg not found!!";
|
---|
68 | die "\n";
|
---|
69 | }
|
---|
70 |
|
---|
71 |
|
---|
72 |
|
---|
73 | my $colloutfile = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "etc", "collectionConfig.xml");
|
---|
74 | if (-e $colloutfile) {
|
---|
75 | print STDOUT "collectionConfig file already exists! overwriting it!\n";
|
---|
76 |
|
---|
77 | }
|
---|
78 |
|
---|
79 | my $buildoutfile = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "index", "buildConfig.xml");
|
---|
80 | if (-e $buildoutfile) {
|
---|
81 | print STDOUT "buildConfig file already exists! overwriting it!\n";
|
---|
82 |
|
---|
83 | }
|
---|
84 |
|
---|
85 | my $buildtype;
|
---|
86 | if (defined $collectcfg->{'buildtype'}) {
|
---|
87 | $buildtype = $collectcfg->{'buildtype'};
|
---|
88 | } else {
|
---|
89 | $buildtype = 'mg';
|
---|
90 | }
|
---|
91 |
|
---|
92 | my $buildoutput = new IO::File(">$buildoutfile");
|
---|
93 | my $buildwriter = new XML::Writer(OUTPUT => $buildoutput);
|
---|
94 |
|
---|
95 | $buildwriter->startTag('buildConfig');
|
---|
96 |
|
---|
97 | my $colloutput = new IO::File(">$colloutfile");
|
---|
98 | my $collwriter = new XML::Writer(OUTPUT => $colloutput);
|
---|
99 |
|
---|
100 | $collwriter->startTag('collectionConfig');
|
---|
101 |
|
---|
102 | #output the collection metadata to the collectionConfig file
|
---|
103 | my $collectionmeta = $collectcfg->{'collectionmeta'};
|
---|
104 | $collwriter->startTag('metadataList');
|
---|
105 |
|
---|
106 | %name_map = (
|
---|
107 | 'collectionname', 'colName',
|
---|
108 | 'collectionextra', 'colDescription',
|
---|
109 | 'iconcollection', 'colIcon',
|
---|
110 | 'iconcollectionsmall', 'colIconSmall');
|
---|
111 |
|
---|
112 | if (defined $collectionmeta) {
|
---|
113 | print STDOUT "coll meta defined \n";
|
---|
114 | foreach $entry ( keys %$collectionmeta) {
|
---|
115 | print STDOUT "entry=$entry\n";
|
---|
116 | if (defined $name_map{$entry}) {
|
---|
117 | $name= $name_map{$entry};
|
---|
118 | print STDOUT "name=$name\n";
|
---|
119 | foreach $lang (keys %{$collectionmeta->{$entry}}) {
|
---|
120 | print STDOUT "lang=$lang\n";
|
---|
121 | outputmetadata($collwriter, $name, $lang, $collectionmeta->{$entry}->{$lang});
|
---|
122 | }
|
---|
123 | }
|
---|
124 | }
|
---|
125 | }
|
---|
126 | $collwriter->endTag('metadataList');
|
---|
127 |
|
---|
128 | #num docs
|
---|
129 | my $numdocs = $buildcfg->{'numdocs'};
|
---|
130 | $buildwriter->startTag('metadataList');
|
---|
131 | $buildwriter->startTag('metadata', 'name'=>'numDocs');
|
---|
132 | $buildwriter->characters($numdocs);
|
---|
133 | $buildwriter->endTag('metadata');
|
---|
134 | $buildwriter->endTag('metadataList');
|
---|
135 |
|
---|
136 |
|
---|
137 | #indexes
|
---|
138 | my $indexmap = {};
|
---|
139 | if (defined $buildcfg->{'indexmap'}) {
|
---|
140 | $indexmap_t = $buildcfg->{'indexmap'};
|
---|
141 | foreach $i (@$indexmap_t) {
|
---|
142 | my ($k, $v) = $i =~ /^(.*)\-\>(.*)$/;
|
---|
143 | $indexmap->{$k} = $v;
|
---|
144 | }
|
---|
145 | } else {
|
---|
146 | print STDERR "indexmap not defined";
|
---|
147 | }
|
---|
148 | my $defaultindex;
|
---|
149 | if (defined $collectcfg->{'defaultindex'}) {
|
---|
150 | $defaultindex = $collectcfg->{'defaultindex'};
|
---|
151 | } else {
|
---|
152 | print STDERR "defaultindex not defined";
|
---|
153 | }
|
---|
154 | $defaultindex = $indexmap->{$defaultindex};
|
---|
155 |
|
---|
156 |
|
---|
157 | $buildwriter->startTag('serviceRackList');
|
---|
158 |
|
---|
159 | my @levels = ('Document');
|
---|
160 | my $defaultlevel = 'Document';
|
---|
161 |
|
---|
162 | #do the retrieve service
|
---|
163 | # assume mgpp or mg
|
---|
164 | if ($buildtype eq 'mgpp') {
|
---|
165 | #for each level
|
---|
166 | if (defined $collectcfg->{'levels'}) {
|
---|
167 | push @levels, @{$collectcfg->{'levels'}};
|
---|
168 |
|
---|
169 | foreach $l(@levels){
|
---|
170 | $defaultlevel = 'Section' if $l eq 'Section';
|
---|
171 | }
|
---|
172 | }
|
---|
173 |
|
---|
174 | $buildwriter->startTag('serviceRack', 'name'=>'GS2MGPPRetrieve');
|
---|
175 | $buildwriter->emptyTag('defaultLevel', 'name'=>$defaultlevel);
|
---|
176 | $buildwriter->startTag('levelList');
|
---|
177 | foreach $level (@levels) {
|
---|
178 | $buildwriter->emptyTag('level', 'name'=>$level);
|
---|
179 | }
|
---|
180 | $buildwriter->endTag('levelList');
|
---|
181 |
|
---|
182 |
|
---|
183 | } else {
|
---|
184 | $buildwriter->startTag('serviceRack', 'name'=>'GS2MGRetrieve');
|
---|
185 | $buildwriter->emptyTag('defaultIndex', 'name'=>$defaultindex);
|
---|
186 |
|
---|
187 | }
|
---|
188 |
|
---|
189 | # add in the classifiers if needed
|
---|
190 |
|
---|
191 | my $count = 1;
|
---|
192 | my $phind = 0;
|
---|
193 | my $started_classifiers = 0;
|
---|
194 | if (defined $collectcfg->{'classify'}) {
|
---|
195 | my $classifiers = $collectcfg->{'classify'};
|
---|
196 | foreach $cl (@$classifiers) {
|
---|
197 | print STDERR "cl=$cl\n";
|
---|
198 | $name = "CL$count";
|
---|
199 | $count++;
|
---|
200 | my ($classname) = @$cl[0];
|
---|
201 | if ($classname eq "Phind") {
|
---|
202 | $phind=1;
|
---|
203 | next;
|
---|
204 | }
|
---|
205 |
|
---|
206 | my $document_interleave = "true";
|
---|
207 | my $orientation = "vertical";
|
---|
208 | if ($classname eq "AZList") { #there may be others
|
---|
209 | $document_interleave = "false";
|
---|
210 | $orientation = "horizontal";
|
---|
211 | }
|
---|
212 | if (not $started_classifiers) {
|
---|
213 | $buildwriter->startTag('classifierList');
|
---|
214 | $started_classifiers = 1;
|
---|
215 | }
|
---|
216 | my $content = ''; #use metadata
|
---|
217 |
|
---|
218 | for ($i=0; $i<scalar(@$cl); $i++) {
|
---|
219 | $arg = @$cl[$i];
|
---|
220 | if ($arg eq "-metadata") {
|
---|
221 | $content = @$cl[$i+1];
|
---|
222 | last;
|
---|
223 | }
|
---|
224 | }
|
---|
225 |
|
---|
226 | $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content, 'orientation'=>$orientation, 'documentInterleave'=>$document_interleave);
|
---|
227 |
|
---|
228 | } #foreach classifier
|
---|
229 | if ($started_classifiers) {
|
---|
230 | # end the classifiers
|
---|
231 | $buildwriter->endTag('classifierList');
|
---|
232 | }
|
---|
233 |
|
---|
234 | }
|
---|
235 | # close off the Retrieve service
|
---|
236 | $buildwriter->endTag('serviceRack');
|
---|
237 |
|
---|
238 | # the phind classifier is a separate service
|
---|
239 | if ($phind) {
|
---|
240 | # if phind classifier
|
---|
241 | $buildwriter->emptyTag('serviceRack', 'name'=>'PhindPhraseBrowse');
|
---|
242 | }
|
---|
243 |
|
---|
244 | # do the search service
|
---|
245 | if ($buildtype eq 'mgpp') {
|
---|
246 |
|
---|
247 | $buildwriter->startTag('serviceRack', 'name'=>'GS2MGPPSearch');
|
---|
248 | $buildwriter->emptyTag('defaultLevel', 'name'=>$defaultlevel);
|
---|
249 | $buildwriter->startTag('levelList');
|
---|
250 | foreach $level (@levels) {
|
---|
251 | $buildwriter->emptyTag('level', 'name'=>$level);
|
---|
252 | }
|
---|
253 | $buildwriter->endTag('levelList');
|
---|
254 |
|
---|
255 | #fieldlist
|
---|
256 | print STDOUT "trying fields\n";
|
---|
257 | my $fieldmap = {};
|
---|
258 | if (defined $buildcfg->{'indexfieldmap'}) {
|
---|
259 | print STDOUT "doing fields\n";
|
---|
260 | $fieldmap_t = $buildcfg->{'indexfieldmap'};
|
---|
261 | foreach $f (@$fieldmap_t) {
|
---|
262 | my ($k, $v) = $f =~ /^(.*)\-\>(.*)$/;
|
---|
263 | $fieldmap->{$k} = $v;
|
---|
264 | }
|
---|
265 |
|
---|
266 | $buildwriter->startTag('fieldList');
|
---|
267 | foreach $f (keys %$fieldmap) {
|
---|
268 | $field = $fieldmap->{$f};
|
---|
269 | $buildwriter->emptyTag('field', 'shortname'=>$field, 'name'=>$f);
|
---|
270 | }
|
---|
271 | $buildwriter->endTag('fieldList');
|
---|
272 | } else {
|
---|
273 | print STDERR "indexfieldmap not defined";
|
---|
274 | }
|
---|
275 |
|
---|
276 | } elsif ($buildtype eq 'mg') {
|
---|
277 | $buildwriter->startTag('serviceRack', 'name'=>'GS2MGSearch');
|
---|
278 | }
|
---|
279 |
|
---|
280 |
|
---|
281 | $buildwriter->emptyTag('defaultIndex', 'name'=>$defaultindex);
|
---|
282 | $buildwriter->startTag('indexList');
|
---|
283 | #for each index
|
---|
284 | foreach $i (keys %$indexmap) {
|
---|
285 | $index = $indexmap->{$i};
|
---|
286 | $buildwriter->emptyTag('index', 'name'=>$index);
|
---|
287 | }
|
---|
288 | $buildwriter->endTag('indexList');
|
---|
289 |
|
---|
290 | $buildwriter->endTag('serviceRack');
|
---|
291 |
|
---|
292 | $buildwriter->endTag('serviceRackList');
|
---|
293 | $buildwriter->endTag('buildConfig');
|
---|
294 | $collwriter->endTag('collectionConfig');
|
---|
295 | $buildwriter->end();
|
---|
296 | $buildoutput->close();
|
---|
297 | $colloutput->close();
|
---|
298 | }
|
---|
299 |
|
---|
300 |
|
---|
301 | sub outputmetadata {
|
---|
302 | my ($collwriter, $metaname, $lang, $metavalue) = @_;
|
---|
303 | $lang = 'en' if $lang eq 'default';
|
---|
304 | if ($lang =~ /^\[/) {
|
---|
305 | ($lang) = $lang =~ /\[l=(..)\]/;
|
---|
306 | print STDOUT "new lang = $lang\n";
|
---|
307 | }
|
---|
308 | my $newvalue;
|
---|
309 | if ($name =~ /colIcon/) {
|
---|
310 | #may need to translate the value
|
---|
311 | ($newvalue) = $metavalue =~ /^_httpprefix_\/collect\/.*\/images\/(.*)$/;
|
---|
312 | ($newvalue) = $metavalue =~ /^_httpcollection_\/images\/(.*)$/ if not defined $newvalue;
|
---|
313 | $newvalue = $metavalue if not defined $newvalue;
|
---|
314 | print STDOUT "old value=$metavalue. new value = $newvalue\n";
|
---|
315 | } else {
|
---|
316 | $newvalue = $metavalue;
|
---|
317 | }
|
---|
318 | $collwriter->startTag('metadata', 'name'=>$metaname, 'lang'=>$lang);
|
---|
319 | $collwriter->characters($newvalue);
|
---|
320 | $collwriter->endTag('metadata');
|
---|
321 | }
|
---|
322 |
|
---|
323 | #$writer->startTag('');
|
---|
324 | #$writer->endTag('');
|
---|
325 | #$writer->characters();
|
---|
326 | #$writer->emptyTag('');
|
---|
327 |
|
---|
328 | 1;
|
---|