source: main/trunk/greenstone3/bin/script/convert_coll_from_gs2.pl@ 25662

Last change on this file since 25662 was 25662, checked in by ak19, 12 years ago

Correcting minor typo in previous commit.

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 27.5 KB
RevLine 
[3688]1#!/usr/bin/perl -w
2
3BEGIN {
4 die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
[6307]5 die "GSDL3HOME not set\n" unless defined $ENV{'GSDL3HOME'};
[10326]6 die "GSDL3SRCHOME not set\n" unless defined $ENV{'GSDL3SRCHOME'};
[3688]7 die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
8 unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
[10326]9 unshift (@INC, "$ENV{'GSDL3SRCHOME'}/lib/perl/cpan");
[3688]10}
11
12use colcfg;
13use util;
14use parsargv;
15use FileHandle;
16use XML::Writer;
[10658]17#can't get this to work on windows
18#use GDBM_File;
[3688]19
[10360]20use strict;
21
[3688]22&main();
23sub print_usage() {
24 print STDOUT "Usage: convert_coll_from_gs2.pl [options] coll-name\n";
25 print STDOUT "options:\n";
26
[3872]27 print STDOUT " -collectdir Directory where collection lives.\n";
[4969]28 print STDOUT " -verbosity Controls the amount of output.\n";
29 print STDOUT " -defaultlang The language that is considered the default (for display text etc). defaults to 'en'\n\n";
[3688]30}
31
[13828]32
[3688]33sub main {
34
[4969]35 my ($defaultlang, $verbosity, $collectdir);
[3688]36 # note that no defaults are passed for most options as they're set
37 # later (after we check the collect.cfg file)
[4969]38 if (!&parsargv::parse(\@ARGV,
39 'verbosity/\d+/', \$verbosity,
40 'collectdir/.*/', \$collectdir,
41 'defaultlang/.*/', \$defaultlang)) {
[3688]42 &print_usage();
43 die "\n";
44 }
45
46 # get and check the collection name
[4969]47 my ($collection) = @ARGV;
48 if (!defined($collection) || $collection eq "") {
49 die "No collection specified\n";
[3688]50 }
[4969]51 if ($collection eq "gs2model") {
52 die "You cant convert the model collection\n";
53 }
[4184]54
[4969]55 if (!defined $collectdir || $collectdir eq "") {
56 $collectdir = &util::filename_cat ($ENV{'GSDLHOME'}, "collect");
57 }
58
59 if (!defined $defaultlang || $defaultlang eq "") {
60 $defaultlang = 'en';
61 }
62 # add on the coll name
63 $collectdir = &util::filename_cat ($collectdir, $collection);
64
[10360]65 my $collconfigfilename = &util::filename_cat ($collectdir, "etc", "collect.cfg");
[4969]66 print STDOUT "coll config=$collconfigfilename\n";
[3688]67 my $collectcfg;
68 if (-e $collconfigfilename) {
69 $collectcfg = &colcfg::read_collect_cfg ($collconfigfilename);
70
[3983]71 } else {
72 print STDERR "collect.cfg not found!!";
73 die "\n";
[3688]74 }
[4184]75
[3872]76
[10360]77 my $buildconfigfilename = &util::filename_cat ($collectdir, "index", "build.cfg");
[3688]78 my $buildcfg;
79 if (-e $buildconfigfilename) {
80 $buildcfg = &colcfg::read_build_cfg ($buildconfigfilename);
81
[3983]82 } else {
83 print STDERR "build.cfg not found!!";
84 die "\n";
[3688]85 }
[4184]86
[3688]87
[3979]88
[4969]89 my $colloutfile = &util::filename_cat ($collectdir, "etc", "collectionConfig.xml");
[3979]90 if (-e $colloutfile) {
91 print STDOUT "collectionConfig file already exists! overwriting it!\n";
92
93 }
94
[4969]95 my $buildoutfile = &util::filename_cat ($collectdir, "index", "buildConfig.xml");
[3979]96 if (-e $buildoutfile) {
[3688]97 print STDOUT "buildConfig file already exists! overwriting it!\n";
98
99 }
100
[10658]101# my $db_file = &util::filename_cat ($collectdir, "index", "text", "$collection.ldb");
[9899]102 my $database;
[10658]103# if (-e $db_file) {
104# $database = &open_database($db_file);
105# } else {
106# print STDERR "gdbm database file $db_file not found!!";
107# die "\n";
108# }
[9899]109
[3688]110 my $buildtype;
[4886]111 if (defined $buildcfg->{'buildtype'}) {
112 $buildtype = $buildcfg->{'buildtype'};
[3688]113 } else {
114 $buildtype = 'mg';
115 }
[4184]116
[10653]117 my $indexstem = undef;
118 if (defined $buildcfg->{'indexstem'}) {
119 $indexstem = $buildcfg->{'indexstem'};
120 }
121
[3979]122 my $buildoutput = new IO::File(">$buildoutfile");
[25662]123 binmode($buildoutput,":utf8");
[4184]124 my $buildwriter = new XML::Writer(OUTPUT => $buildoutput, NEWLINES => 1);
[3979]125
[25646]126 $buildwriter->xmlDecl("UTF-8");
[9874]127 $buildwriter->startTag('buildConfig', 'xmlns:gsf'=>"http://www.greenstone.org/greenstone3/schema/ConfigFormat");
[4184]128
[3979]129 my $colloutput = new IO::File(">$colloutfile");
[25661]130 binmode($colloutput,":utf8");
[4184]131 my $collwriter = new XML::Writer(OUTPUT => $colloutput, NEWLINES => 1);
132
[25646]133 $collwriter->xmlDecl("UTF-8");
[14754]134 $collwriter->startTag('CollectionConfig', 'xmlns:gsf'=>"http://www.greenstone.org/greenstone3/schema/ConfigFormat", 'xmlns:xsl'=>'http://www.w3.org/1999/XSL/Transform');
[4184]135
[3979]136 #output the collection metadata to the collectionConfig file
137 $collwriter->startTag('metadataList');
138
[4969]139 my $creator = $collectcfg->{'creator'};
[14754]140 &output_metadata($collwriter,'default', 'creator', $creator);
[13836]141 my $public =$collectcfg->{'public'};
[14754]142 &output_metadata($collwriter,'default', 'public', $public);
143
[4969]144 $collwriter->endTag('metadataList');
145
146 #output the display collectionmeta to collectionConfig.xml
[3979]147
[4969]148 my $collectionmeta = $collectcfg->{'collectionmeta'};
[3979]149 if (defined $collectionmeta) {
[10360]150 my %name_map = ('collectionname', 'name',
[4969]151 'collectionextra', 'description',
152 'iconcollection', 'icon',
153 'iconcollectionsmall', 'smallicon');
154
155 $collwriter->startTag('displayItemList');
[10360]156 foreach my $entry ( keys %$collectionmeta) {
[3979]157 if (defined $name_map{$entry}) {
[10360]158 my $name= $name_map{$entry};
159 foreach my $lang (keys %{$collectionmeta->{$entry}}) {
160 my $value = $collectionmeta->{$entry}->{$lang};
[4184]161 if ($entry =~ /^icon/) {
162 $value = format_icon_value($value);
[12678]163 } else {
[13828]164 $value = tidy_up_display_item($value);
[4184]165 }
[4969]166 &output_display($collwriter, $name, $lang, $value);
[3979]167 }
168 }
169 }
[4969]170 $collwriter->endTag('displayItemList');
[3979]171 }
172
[4969]173 # output building metadata to build config file
[3872]174 my $numdocs = $buildcfg->{'numdocs'};
[3979]175 $buildwriter->startTag('metadataList');
[14754]176 &output_metadata($buildwriter,'', 'numDocs', $numdocs);
177 &output_metadata($buildwriter,'', 'buildType', $buildtype);
[3979]178 $buildwriter->endTag('metadataList');
[3872]179
180
[3771]181 #indexes
[13912]182 # maps index name to shortname
[3771]183 my $indexmap = {};
[13912]184 # keeps the order for indexes
185 my @indexlist = ();
[13917]186 my $defaultindex = "";
[4886]187 my $first = 1;
[13912]188 my $maptype = "indexfieldmap";
189 if ($buildtype eq "mg") {
190 $maptype = "indexmap";
191 }
192 if (defined $buildcfg->{$maptype}) {
193 my $indexmap_t = $buildcfg->{$maptype};
[10360]194 foreach my $i (@$indexmap_t) {
[3771]195 my ($k, $v) = $i =~ /^(.*)\-\>(.*)$/;
196 $indexmap->{$k} = $v;
[13912]197 push @indexlist, $k;
[14754]198 if ($first == 1) {
199 $defaultindex = $k;
[4886]200 $first = 0;
201 }
[3771]202 }
203 } else {
[25661]204 print STDERR "$maptype not defined\n";
[3771]205 }
[13912]206 # we use the shortname for default index
207 if (defined $collectcfg->{'defaultindex'}) {
208 $defaultindex = $collectcfg->{'defaultindex'};
[14754]209 #$defaultindex = $indexmap->{$defaultindex};
[3771]210 }
211
[13917]212 # levels
213 my $levelmap = {};
214 my @levellist = ();
215 my $default_search_level = "";
[14754]216 my $default_search_level_shortname = "";
[13917]217 my $default_retrieve_level = "Sec";
218 $first = 1;
219 if ($buildtype eq "mgpp" || $buildtype eq "lucene") {
220 if (defined $buildcfg->{'levelmap'}) {
221 my $levelmap_t = $buildcfg->{'levelmap'};
222 foreach my $l (@$levelmap_t) {
223 my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
224 $levelmap->{$k} = $v;
225 push @levellist, $k;
226 if ($first) {
[14754]227 $default_search_level = $k;
228 $default_search_level_shortname = $v;
[13917]229 $first = 0;
230 }
231 }
232 }
233
234 if (defined $collectcfg->{'defaultlevel'}) {
235 $default_search_level = $collectcfg->{'defaultlevel'};
[14754]236 #$default_search_level = $levelmap->{$default_search_level};
237 $default_search_level_shortname = $levelmap->{$default_search_level};
[13917]238 }
239 if (defined $buildcfg->{'textlevel'}) {
240 $default_retrieve_level = $buildcfg->{'textlevel'};
241 }
242 }
[4184]243 # format stuff
244 my $format = $collectcfg->{'format'};
245
246 #output the search stuff to coll cfg
[14754]247 $collwriter->startTag('search','type'=>$buildtype);
[10360]248 foreach my $i (keys %$indexmap) {
[13912]249 $collwriter->startTag('index', 'name'=>$i);
[4184]250 #find the coll meta stuff
[10360]251 my $indexdisplay = ".$i";
252 foreach my $lang (keys %{$collectionmeta->{$indexdisplay}}) {
253 my $value = $collectionmeta->{$indexdisplay}->{$lang};
[14754]254 output_display($collwriter, 'name', $lang, $i);
[4184]255 }
256 $collwriter->endTag('index');
257 }
258
[14754]259 #output the defaultIndex to coll cfg
260 $collwriter->emptyTag('defaultIndex','name'=>$defaultindex);
261
262 #indexSubcollection
263 my $indexsubcollections = $collectcfg->{'indexsubcollections'};
264
265 if (defined $indexsubcollections) {
266 my $indexsubcollections_t= $collectcfg->{'indexsubcollections'};
267 foreach my $i ( @$indexsubcollections_t) {
268 $collwriter->startTag('indexSubcollection', 'name'=>$i);
269 &output_display($collwriter, 'name', $defaultlang, $i);
270 $collwriter->endTag('indexSubcollection');
271 }
272 }
273
274 #subcollection
275 my $subcollection = $collectcfg->{'subcollection'};
276 if (defined $subcollection){
277 foreach my $entry (keys %$subcollection){
278 my $value = $subcollection->{$entry};
279 $collwriter->emptyTag('subcollection','filter'=>$value,'name'=>$entry);
280 }
281 }
282
283 #indexlanguage
284 my $languages = $collectcfg->{'languages'};
285 if (defined $languages){
286 my $languages_t = $collectcfg->{'languages'};
287 foreach my $i (@$languages_t){
288 $collwriter->startTag('indexLanguage','name'=>$i);
289 &output_display($collwriter, 'name', $defaultlang, $i);
290 $collwriter->endTag('indexLanguage');
291 }
292 }
293
[13917]294 # level stuff for mgpp/lucene
295 if ($buildtype eq 'mgpp' || $buildtype eq 'lucene'){
296 foreach my $l (keys %$levelmap) {
297 $collwriter->startTag('level', 'name'=>$l);
298 #find the coll meta stuff
299 my $leveldisplay = ".$l";
300 foreach my $lang (keys %{$collectionmeta->{$leveldisplay}}) {
301 my $value = $collectionmeta->{$leveldisplay}->{$lang};
302 output_display($collwriter, 'name', $lang, $value);
303 }
304 $collwriter->endTag('level');
305 }
[14754]306 $collwriter->emptyTag('defaultLevel', 'name'=>$default_search_level);
[13917]307 }
[14754]308
309 # add in the search type
310 if (defined $format->{'SearchTypes'}){
311 $collwriter->startTag('format', 'name'=>"searchType");
312 $collwriter->charactersXML($format->{'SearchTypes'});
313 $collwriter->endTag('format');
314 }
315
[4184]316 # add in the format stuff
317 if (defined $format->{'SearchVList'}) {
[4969]318
[4184]319 $collwriter->startTag('format');
[10201]320 write_format($collwriter, $format->{'SearchVList'}, "document");
[4184]321 $collwriter->endTag('format');
322 }
[12979]323 elsif (defined $format->{'VList'}) {
324 $collwriter->startTag('format');
325 write_format($collwriter, $format->{'VList'}, "document");
326 $collwriter->endTag('format');
327 }
[3771]328
[4184]329 $collwriter->endTag('search');
[14754]330
331 # import plugins
332
333 my $plugins = $collectcfg->{'plugin'};
[4184]334
[14754]335 if (defined $plugins){
336 $collwriter->startTag('import');
337 $collwriter->startTag('pluginList');
338 foreach my $pl (@$plugins) {
339 my ($pluginname) = @$pl[0];
340 $collwriter->startTag('plugin','name'=>$pluginname);
341
342 for (my $i=1; $i<scalar(@$pl); $i++) {
343 my $arg =@$pl[$i];
344 if ($arg =~ /^-/){
345 my $option_name=@$pl[$i];
346 my $option_value=@$pl[$i+1];
347 if (defined $option_value){
348 if ($option_value =~ /^-/){
349 $collwriter->startTag('option','name'=>$option_name);
350 $collwriter->endTag('option');
351 }else{
352 $collwriter->startTag('option','name'=>$option_name,'value'=>$option_value);
353 $collwriter->endTag('option');
354 }
355 }
356 }
357 }
358 $collwriter->endTag('plugin');
359 }
360 $collwriter->endTag('pluginList');
361 $collwriter->endTag('import');
362 }
363
[3979]364 $buildwriter->startTag('serviceRackList');
[3688]365
[4886]366 my @levels = ();
367 my $defaultlevel;
[13912]368 my $service_type = "MG";
369 if ($buildtype eq 'mgpp') {
370 $service_type = "MGPP";
371 } elsif ($buildtype eq "lucene") {
372 $service_type = "Lucene";
373 }
374
[14754]375 #indexSubcollectionList
376
377 my $subcollectionmap = $buildcfg->{'subcollectionmap'};
378 my $firstsubcollection = 1;
379 my $defaultsubcollection = "";
380 my @subcollist;
381 my $subcolmap = {};
382
383 if (defined $buildcfg->{'subcollectionmap'}) {
384 my $subcolmap_t = $buildcfg->{'subcollectionmap'};
385
386 foreach my $l (@$subcolmap_t) {
387 my @pair = split(/->/, $l);
388 $subcolmap->{$pair[0]} = $pair[1];
389 push @subcollist, $pair[0];
390 if ($firstsubcollection==1) {
391 $defaultsubcollection = $pair[1];
392 $firstsubcollection = 0;
393 }
394 }
395
396 }
397
398
[3947]399 #do the retrieve service
[13917]400 $buildwriter->startTag('serviceRack', 'name'=>"GS2".$service_type."Retrieve");
[13228]401 if ($buildtype eq 'mgpp' || $buildtype eq 'lucene') {
[13917]402 $buildwriter->emptyTag('defaultLevel', 'shortname'=>$default_retrieve_level);
[13115]403 } elsif ($buildtype eq "mg") {
[13912]404 $buildwriter->emptyTag('defaultIndex', 'shortname'=>$defaultindex);
[3688]405 }
[14754]406
407 if ((defined $defaultsubcollection) && ($defaultsubcollection ne "")) {
408 $buildwriter->emptyTag('defaultIndexSubcollection', 'shortname'=>$defaultsubcollection);
409 }
410
[10653]411 if (defined $indexstem) {
412 $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
413 }
[8967]414 # close off the Retrieve service
415 $buildwriter->endTag('serviceRack');
[3688]416
[3947]417 # add in the classifiers if needed
418
[3688]419 my $count = 1;
420 my $phind = 0;
421 my $started_classifiers = 0;
422 if (defined $collectcfg->{'classify'}) {
[4184]423 $collwriter->startTag('browse');
[10360]424 # add in default format if necessary
425 if (defined $format->{"VList"} || defined $format->{"HList"}) {
426 # global formats
427 $collwriter->startTag('format');
428 if (defined $format->{"VList"}) {
429 # VLIst applies to both classifier and doc nodes
430 write_format($collwriter, $format->{"VList"}, "document");
431 write_format($collwriter, $format->{"VList"}, "classifier");
432 }
433 if (defined $format->{"HList"}) {
434 # hlist is only for classifier nodes
435 write_format($collwriter, $format->{"HList"}, "horizontal");
436 }
437 $collwriter->endTag('format');
438 }
[3688]439 my $classifiers = $collectcfg->{'classify'};
[10360]440 foreach my $cl (@$classifiers) {
441 my $name = "CL$count";
[3688]442 $count++;
443 my ($classname) = @$cl[0];
[4184]444 if ($classname =~ /^phind$/i) {
[3688]445 $phind=1;
[4184]446 #should add it into coll config classifiers
[3688]447 next;
448 }
449
[9899]450 my $horizontalAtTop = &isHorizontalClassifier($database, $name);
[3688]451 if (not $started_classifiers) {
[8967]452 $buildwriter->startTag('serviceRack', 'name'=>'GS2Browse');
[10653]453 if (defined $indexstem) {
454 $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
455 }
456
[10360]457 $buildwriter->startTag('classifierList');
[3688]458 $started_classifiers = 1;
459 }
[4969]460 my $content = ''; #use buttonname first, then metadata
[4184]461 if ($classname eq "DateList") {
462 $content = "Date";
463 } else {
[10360]464 for (my $i=0; $i<scalar(@$cl); $i++) {
465 my $arg = @$cl[$i];
[4235]466 if ($arg eq "-buttonname"){
[4184]467 $content = @$cl[$i+1];
468 last;
[4886]469 } elsif ($arg eq "-metadata") {
[4235]470 $content = @$cl[$i+1];
[4184]471 }
[4235]472
[3688]473 }
474 }
[9899]475 if ($horizontalAtTop) {
476 $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content, 'horizontalAtTop'=>'true');
[14754]477
[9899]478 } else {
[4886]479 $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content);
[9899]480 }
[3688]481
[3947]482
[14754]483 # $collwriter->startTag('classifier', 'name'=>$name);
484 $collwriter->startTag('classifier', 'name'=>$classname);
485 for (my $i=1; $i<scalar(@$cl); $i++) {
486 my $arg =@$cl[$i];
487 if ($arg =~ /^-/){
488 my $option_name=@$cl[$i];
489 my $option_value=@$cl[$i+1];
490 if (defined $option_value){
491 if ($option_value=~ /^-/){
492 $collwriter->startTag('option','name'=>$option_name);
493 $collwriter->endTag('option');
494 }else{
495 $collwriter->startTag('option','name'=>$option_name,'value'=>$option_value);
496 $collwriter->endTag('option');
497 }
498 }
499 }
500 }
501
[10360]502 my $vlist = $name."VList";
503 my $hlist = $name."HList";
504 my $dlist = "";
[4184]505 if ($classname eq "DateList") {
506 $dlist = "DateList";
507 }
[10201]508 # need to work out how to split into classifier and document
[10360]509 if (defined $format->{$vlist} || defined $format->{$hlist} || defined $format->{$dlist}) {
[4184]510 $collwriter->startTag('format');
511 if (defined $format->{$vlist}) {
[10201]512 write_format($collwriter, $format->{$vlist}, "document");
[10360]513 write_format($collwriter, $format->{$vlist}, "classifier");
514 }
[4184]515 if (defined $format->{$hlist}) {
[10201]516 write_format($collwriter, $format->{$hlist}, "horizontal");
[10360]517 }
518
[4184]519 if (defined $format->{$dlist}) {
[10201]520 write_format($collwriter, $format->{$dlist}, "document");
[4184]521 }
522 $collwriter->endTag('format');
523 }
524 $collwriter->endTag('classifier');
[4969]525 } #foreach classifier
[3688]526 if ($started_classifiers) {
527 # end the classifiers
[3979]528 $buildwriter->endTag('classifierList');
[8967]529 # close off the Browse service
530 $buildwriter->endTag('serviceRack');
[3688]531 }
[3947]532
[4184]533 $collwriter->endTag('browse');
[3947]534 }
[4184]535
[8967]536
[3947]537 # the phind classifier is a separate service
538 if ($phind) {
539 # if phind classifier
[3979]540 $buildwriter->emptyTag('serviceRack', 'name'=>'PhindPhraseBrowse');
[3947]541 }
542
543 # do the search service
[13912]544 $buildwriter->startTag('serviceRack', 'name'=>'GS2'.$service_type.'Search');
[14754]545 #$buildwriter->emptyTag('defaultIndex', 'shortname'=>$defaultindex);
[13912]546 $buildwriter->startTag('indexList');
547 #for each index
548 foreach my $i (@indexlist) {
549 my $index = $indexmap->{$i};
550 $buildwriter->emptyTag('index', 'name'=>$i, 'shortname'=>$index);
551 }
552 $buildwriter->endTag('indexList');
553 if (defined $indexstem) {
554 $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
555 }
556
557 # index options
558 if ($buildtype eq 'mg' || $buildtype eq 'mgpp') {
559 $buildwriter->startTag('indexOptionList');
560 my $stemindexes = 3; # default is stem and casefold
561 if (defined $buildcfg->{'stemindexes'} && $buildcfg->{'stemindexes'} =~ /^\d+$/ ) {
562 $stemindexes = $buildcfg->{'stemindexes'};
563 }
564 $buildwriter->emptyTag('indexOption', 'name'=>'stemIndexes', 'value'=>$stemindexes);
565
566 my $maxnumeric = 4; # default
567 if (defined $buildcfg->{'maxnumeric'} && $buildcfg->{'maxnumeric'} =~ /^\d+$/) {
568 $maxnumeric = $buildcfg->{'maxnumeric'};
569 }
570 $buildwriter->emptyTag('indexOption', 'name'=>'maxnumeric', 'value'=>$maxnumeric);
571
572 $buildwriter->endTag('indexOptionList');
573 }
574
[13228]575 if ($buildtype eq 'mgpp' || $buildtype eq 'lucene') {
[3947]576
[13912]577 # level info
[14754]578 $buildwriter->emptyTag('defaultLevel', 'shortname'=>$default_search_level_shortname);
[13917]579 $buildwriter->emptyTag('defaultGDBMLevel', 'shortname'=>$default_retrieve_level);
[3979]580 $buildwriter->startTag('levelList');
[13917]581 foreach my $l (@levellist) {
582 my $level = $levelmap->{$l};
583 $buildwriter->emptyTag('level', 'name'=>$l, 'shortname'=>$level);
584 }
[3979]585 $buildwriter->endTag('levelList');
[3688]586
[5647]587 # do the search types if there
588 if (defined $collectcfg->{'searchtype'}) {
589 $buildwriter->startTag('searchTypeList');
[10360]590 foreach my $st (@{$collectcfg->{'searchtype'}}) {
[5647]591 $buildwriter->emptyTag('searchType', 'name'=>$st);
592 }
593 $buildwriter->endTag('searchTypeList');
[13228]594 } elsif (defined $format->{'SearchTypes'}) {
595 #check format statement
596 my $searchtype = $format->{'SearchTypes'};
597 $buildwriter->startTag('searchTypeList');
598 if ($searchtype =~ /form/) {
599 $buildwriter->emptyTag('searchType', 'name'=>'form');
600 }
601 if ($searchtype =~ /plain/) {
602 $buildwriter->emptyTag('searchType', 'name'=>'plain');
603 }
604 $buildwriter->endTag('searchTypeList');
[5647]605 }
[13912]606 }
[3947]607
[14754]608 #indexLanguageList
609 my $indexlanguages = $collectcfg->{'languages'};
610 my $firstindexlanguage = 1;
611 my $defaultindexlanguage_shortname;
612 if (defined $indexlanguages){
613 $buildwriter->startTag('indexLanguageList');
614 my $languages_t = $collectcfg->{'languages'};
615 foreach my $i (@$languages_t){
616 $buildwriter->startTag('indexLanguage','name'=>$i);
617 &output_display($buildwriter, 'name', $i, $i);
618 $buildwriter->endTag('indexLanguage');
619 if ($firstindexlanguage==1){
620 $defaultindexlanguage_shortname = $i;
621 $firstindexlanguage=0;
622 }
623 }
624 $buildwriter->endTag('indexLanguageList');
625 $buildwriter->startTag('defaultIndexLanguage', 'name'=>$defaultindexlanguage_shortname,'shortname'=>$defaultindexlanguage_shortname);
626 $buildwriter->endTag('defaultIndexLanguage');
627 }
[13115]628
[14754]629 # my $defaultsubcollection = "";
630 # my @subcollist;
631
632 if (scalar(@subcollist)>0){
633
634 $buildwriter->startTag('indexSubcollectionList');
635 foreach my $i (keys %$subcolmap){
636 my $short_name = $subcolmap->{$i};
637 $buildwriter->emptyTag('indexSubcollection', 'name'=>$i, 'shortname'=>$short_name);
638 }
639
640 $buildwriter->endTag('indexSubcollectionList');
641 $buildwriter->emptyTag('defaultIndexSubcollection', 'shortname'=>$defaultsubcollection);
642 }
643
644
[3979]645 $buildwriter->endTag('serviceRack');
[4969]646
[3979]647 $buildwriter->endTag('serviceRackList');
648 $buildwriter->endTag('buildConfig');
[11274]649
650 # we add in the default replace list just in case we have macros in the
651 # collection
652 $collwriter->emptyTag('replaceListRef', 'id'=>'gs2-standard');
[14754]653 $collwriter->endTag('CollectionConfig');
[4969]654 $collwriter->end();
[3979]655 $buildwriter->end();
656 $buildoutput->close();
657 $colloutput->close();
[9899]658 &close_database($database);
[3979]659}
[3688]660
[3979]661
[4969]662sub output_metadata {
[14754]663 my ($writer, $lang, $metaname, $metavalue) = @_;
664 $lang = 'en' if $lang eq 'default';
665 if ($lang ne ""){
666 $writer->startTag('metadata', 'lang'=>$lang, 'name'=>$metaname);
667 }else{
668 $writer->startTag('metadata', 'name'=>$metaname);
669 }
[4184]670 $writer->characters($metavalue);
671 $writer->endTag('metadata');
672}
673
674sub output_display {
[4969]675 my ($writer, $name, $lang, $value) = @_;
[4184]676 $lang = 'en' if $lang eq 'default';
677 if ($lang =~ /^\[/) {
678 ($lang) = $lang =~ /\[l=(.*)\]/;
[3979]679 }
[14754]680
[4969]681 $writer->startTag('displayItem', 'name'=>$name, 'lang'=>$lang);
[4184]682 $writer->characters($value);
[4969]683 $writer->endTag('displayItem');
[3688]684}
[4184]685sub format_icon_value {
686 my ($value) = @_;
687 if ($value =~ /^_/) {
[13828]688 my ($newvalue) = $value =~ /images\/(.*)$/;
[4184]689 if ($newvalue) {
690 return $newvalue;
691 }
692 }
693 return $value;
694}
[3688]695
[13828]696sub tidy_up_display_item {
[12678]697 my ($value) = @_;
698 # remove \n
699 $value =~ s/\\n//g;
700 # replace \' with '
701 $value =~ s/\\\'/\'/g;
702 # replace \" with "
703 $value =~ s/\\\"/\"/g;
[13828]704 # replace _httpprefix_ with _httpsite_
[13843]705 $value =~ s/_httpprefix_/_httpsite_/g;
706 $value =~ s/_gwcgi_//g;
707 $value =~ s/[a-z][a-z]?=_cgiarg[a-z][a-z]?_&?//g;
708 $value =~ s/&p=/&sa=/g;
[12678]709 return $value;
710}
711
[4184]712sub write_format {
[10201]713 my ($writer, $old_format, $node_type) = @_;
[7471]714 # replace \' with '
715 $old_format =~ s/\\\'/\'/g;
[12236]716 # replace \" with "
717 $old_format =~ s/\\\"/\"/g;
[4184]718 #convert [] to <gsf:...>
[10360]719 # assume no nesting {If} or {Or} for now
720 $old_format =~ s/\{If\}\{([^\}]*)\}/&format_if($1, $node_type)/eg;
721 $old_format =~ s/\{Or\}\{([^\}]*)\}/&format_or($1)/eg;
[4184]722 $old_format =~ s/\[Text\]/\<gsf:text\/\>/g;
723 $old_format =~ s/\[num\]/\<gsf:num\/\>/g;
[10782]724 $old_format =~ s/\[link\]/\<gsf:link type=\'$node_type\'\>/g;
[4184]725 $old_format =~ s/\[\/link\]/\<\/gsf:link\>/g;
726 $old_format =~ s/\[srclink\]/\<gsf:link type=\'source\'\>/g;
727 $old_format =~ s/\[\/srclink\]/\<\/gsf:link\>/g;
[10782]728 $old_format =~ s/\[icon\]/\<gsf:icon type=\'$node_type\'\/\>/g;
[4184]729 $old_format =~ s/\[srcicon\]/\<gsf:icon type=\'source\'\/\>/g;
[10360]730
731 # what to do with hightlight??
732 $old_format =~ s/\[\/?highlight\]//g;
[4184]733
[4969]734 #now do the rest of the [] which are assumed to be metadata
[10360]735 $old_format =~ s/\[([^\]]*)\]/&format_metadata($1)/eg;
736
737 # some html tidy
[4184]738 #turn <br> into <br />
739 $old_format =~ s/\<br\>/\<br \/\>/g;
740 #turn <p> into <p />
741 $old_format =~ s/\<p\>/\<p \/\>/g;
742
743 #put quotes around any atts
744 $old_format =~ s/=([a-z]+)([> ])/=\'$1\'$2/g;
745
[10201]746 if ($node_type eq "document") {
747 $writer->startTag('gsf:template', 'match'=>'documentNode');
748 $writer->charactersXML($old_format);
749 $writer->endTag('gsf:template');
750 } elsif ($node_type eq "classifier") {
751 $writer->startTag('gsf:template', 'match'=>'classifierNode');
752 $writer->charactersXML($old_format);
753 $writer->endTag('gsf:template');
754 } elsif ($node_type eq "horizontal") {
755 $writer->startTag('gsf:template', 'match'=>'classifierNode', 'mode'=>'horizontal');
756 $writer->charactersXML($old_format);
757 $writer->endTag('gsf:template');
[4184]758
[10201]759 }
[4184]760}
761
[10360]762sub format_metadata {
763 my ($metadata_string) = @_;
764
765 #print STDERR "original meta = $metadata_string\n";
766
767 # what shall we do with cgisafe??
768 my $cgisafe = $metadata_string =~ s/^cgisafe://;
769
770 my ($select) = $metadata_string =~ /^(parent|sibling)/;
771 $metadata_string =~ s/^(parent|sibling)//;
772 my ($scope, $delim);
773
774 if ($select) {
775 ($scope) = $metadata_string =~ /^\((Top|All)/;
776 $metadata_string =~ s/^\((Top|All)\)?//;
777 if ($scope) {
[10782]778 ($delim) = $metadata_string =~ /^\'([^\']*)\'\)/;
779 $metadata_string =~ s/^\'([^\']*)\'\)//;
[10360]780 }
781 }
782 $metadata_string =~ s/^://;
783 # remove ex.
784 $metadata_string =~ s/^ex\.//;
785
786 #print STDERR "select=$select, scope=$scope, delim=|$delim|, meta = $metadata_string\n";
787
788 my $new_format = "<gsf:metadata name='$metadata_string' ";
789 if (defined $select) {
790 if ($select eq "sibling") {
791 $new_format .= "multiple='true' ";
792 if (defined $delim) {
793 $new_format .= "separator='$delim' ";
794 }
795 } elsif ($select eq "parent"){
796 if (defined $scope) {
797 if ($scope eq "Top") {
798 $new_format .= "select='root' ";
799 } elsif ($scope eq "All") {
800 $new_format .= "select='ancestors' ";
801 if (defined $delim) {
802 $new_format .= "separator='$delim' ";
803 }
804 }
805 } else {
806 $new_format .= "select='parent' ";
807 }
808 }
809 }
810 $new_format .= "/>";
811 #print STDERR "$new_format\n";
812 return $new_format;
813
814}
815
816sub format_if {
817
818 my ($if_string, $node_type) = @_;
[10782]819 #print STDERR "if string = $if_string\n";
[10360]820
[10782]821 my @parts = split /,/, $if_string;
822 my $test = $parts[0];
823 my $true_option = $parts[1];
824 my $false_option;
825 if (scalar (@parts) == 3) {
826 $false_option = $parts[2];
827 }
828 $test =~ s/^\s*//;
829 $test =~ s/\s*$//;
830 my ($test_meta, $test_type, $test_value);
831 if ($test =~ /^(\[.+\])$/) {
832 $test_meta = $1;
833 $test_type = 'exists';
834 } else {
835 my ($lhs, $exp, $rhs) = $test =~ /^(.+)\s+(eq|ne|lt|gt|le|ge|sw|ew)\s+(.+)$/;
836 #print STDERR "lhs, exp, rhs = $lhs, $exp, $rhs\n";
837 if ($exp eq "eq") {
838 $test_type = "equals";
839 } elsif ($exp eq "sw") {
840 $test_type = "startsWith";
841 } elsif ($exp eq "ew") {
842 $test_type = "endsWith";
843 } elsif ($exp eq "ne") {
844 $test_type = "notEquals";
845 } elsif ($exp eq "lt") {
846 $test_type = "lessThan";
847 }elsif ($exp eq "gt") {
848 $test_type = "greaterThan";
849 }elsif ($exp eq "le") {
850 $test_type = "lessThanOrEquals";
851 }elsif ($exp eq "ge") {
852 $test_type = "greaterThanOrEquals";
853 }
854 if ($lhs =~ /^\[.+\]$/) {
855 $test_meta = $lhs;
856 $test_value = $rhs;
857 } else {
858 # assume rhs has meta
859 $test_meta = $rhs;
860 $test_value = $lhs;
861 }
862
863 #remove beginning and end quotes
864 $test_value =~ s/^[\'\"]//;
865 $test_value =~ s/[\'\"]$//;
866 }
867 my $test_atts = "test='$test_type' ";
868 if (defined $test_value) {
869 $test_atts .= "test-value='$test_value' ";
870 }
871 #print STDERR "test, true, false = $test, $true_option, $false_option\n";
872 my $new_format = "<gsf:switch>$test_meta";
873 $new_format .= "<gsf:when $test_atts>$true_option</gsf:when>";
874 if (defined $false_option) {
875 $new_format .="<gsf:otherwise>$false_option</gsf:otherwise>";
876 }
877 $new_format .= "</gsf:switch>";
878
879 #print STDERR "new format = $new_format\n";
880 return $new_format;
[10360]881}
882
883sub format_or {
884 my ($or_string) = @_;
885 my @meta_list = split (',', $or_string);
886 return "" unless scalar (@meta_list);
887 my $new_format = "<gsf:choose-metadata>";
888 foreach my $m (@meta_list) {
889 if ($m =~ /^\[(.*)\]$/) {
890 $new_format .= &format_metadata($1);
891 } else {
892 # a default value
893 $new_format .= "<gsf:default>$m</gsf:default>";
894 last;
895 }
896 }
897 $new_format .= "</gsf:choose-metadata>";
898 return $new_format;
899}
900
[9899]901sub open_database {
902 my ($db_file) = @_;
903
904 my $database = ();
[10658]905# tie (%$database, 'GDBM_File', $db_file, GDBM_READER, 0400) ||
906# die "Couldn't open database $db_file\n";
[9899]907
908 return $database;
909}
910
911sub close_database {
912 my ($database) = @_;
[10360]913 untie %$database;
[9899]914}
915sub isHorizontalClassifier {
916 my ($database, $name) = @_;
917
[10658]918 return 0; # can't get this to work for windows
[10360]919 my $record = $database->{$name};
[9899]920 my ($childtype) = $record =~ /<childtype>(\w*)/;
921 if ($childtype eq "HList") {
922 return 1;
923 }
924 return 0;
925}
[3688]926#$writer->startTag('');
927#$writer->endTag('');
928#$writer->characters();
929#$writer->emptyTag('');
930
9311;
Note: See TracBrowser for help on using the repository browser.