source: main/trunk/greenstone3/bin/script/convert_coll_from_gs2.pl@ 26343

Last change on this file since 26343 was 26343, checked in by kjdon, 12 years ago

output the indexOPtions

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 29.0 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) {
[25731]157 # some metadata names need to be specially mapped to other names
158 # most of them however, can retain their original names
159 my $name = (defined $name_map{$entry}) ? $name_map{$entry} : $entry;
160 foreach my $lang (keys %{$collectionmeta->{$entry}}) {
161 my $value = $collectionmeta->{$entry}->{$lang};
162 if ($entry =~ /^icon/) {
163 $value = format_icon_value($value);
164 } else {
165 $value = tidy_up_display_item($value);
166 }
167 &output_display($collwriter, $name, $lang, $value);
168 }
[3979]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
[26343]262 # indexOptions
263 if (defined $collectcfg->{'indexoptions'}) {
264 foreach my $i (@{$collectcfg->{'indexoptions'}}) {
265 $collwriter->emptyTag('indexOption', 'name'=>$i);
266 }
267 }
268
[14754]269 #indexSubcollection
270 my $indexsubcollections = $collectcfg->{'indexsubcollections'};
271
272 if (defined $indexsubcollections) {
273 my $indexsubcollections_t= $collectcfg->{'indexsubcollections'};
274 foreach my $i ( @$indexsubcollections_t) {
275 $collwriter->startTag('indexSubcollection', 'name'=>$i);
276 &output_display($collwriter, 'name', $defaultlang, $i);
277 $collwriter->endTag('indexSubcollection');
278 }
279 }
280
281 #subcollection
282 my $subcollection = $collectcfg->{'subcollection'};
283 if (defined $subcollection){
284 foreach my $entry (keys %$subcollection){
285 my $value = $subcollection->{$entry};
286 $collwriter->emptyTag('subcollection','filter'=>$value,'name'=>$entry);
287 }
288 }
289
290 #indexlanguage
291 my $languages = $collectcfg->{'languages'};
292 if (defined $languages){
293 my $languages_t = $collectcfg->{'languages'};
294 foreach my $i (@$languages_t){
295 $collwriter->startTag('indexLanguage','name'=>$i);
296 &output_display($collwriter, 'name', $defaultlang, $i);
297 $collwriter->endTag('indexLanguage');
298 }
299 }
300
[13917]301 # level stuff for mgpp/lucene
302 if ($buildtype eq 'mgpp' || $buildtype eq 'lucene'){
303 foreach my $l (keys %$levelmap) {
304 $collwriter->startTag('level', 'name'=>$l);
305 #find the coll meta stuff
306 my $leveldisplay = ".$l";
307 foreach my $lang (keys %{$collectionmeta->{$leveldisplay}}) {
308 my $value = $collectionmeta->{$leveldisplay}->{$lang};
309 output_display($collwriter, 'name', $lang, $value);
310 }
311 $collwriter->endTag('level');
312 }
[14754]313 $collwriter->emptyTag('defaultLevel', 'name'=>$default_search_level);
[13917]314 }
[14754]315
316 # add in the search type
317 if (defined $format->{'SearchTypes'}){
318 $collwriter->startTag('format', 'name'=>"searchType");
319 $collwriter->charactersXML($format->{'SearchTypes'});
320 $collwriter->endTag('format');
321 }
322
[4184]323 # add in the format stuff
324 if (defined $format->{'SearchVList'}) {
[4969]325
[4184]326 $collwriter->startTag('format');
[10201]327 write_format($collwriter, $format->{'SearchVList'}, "document");
[4184]328 $collwriter->endTag('format');
329 }
[12979]330 elsif (defined $format->{'VList'}) {
331 $collwriter->startTag('format');
332 write_format($collwriter, $format->{'VList'}, "document");
333 $collwriter->endTag('format');
334 }
[3771]335
[4184]336 $collwriter->endTag('search');
[14754]337
338 # import plugins
[25669]339 # if ImagePlugin is added, then need to add in a replaceListRef element for gs2-image
340 my $contains_image_plugin = 0;
341
[14754]342 my $plugins = $collectcfg->{'plugin'};
[4184]343
[14754]344 if (defined $plugins){
345 $collwriter->startTag('import');
346 $collwriter->startTag('pluginList');
347 foreach my $pl (@$plugins) {
348 my ($pluginname) = @$pl[0];
[25669]349 if ($pluginname =~ m/^(ImagePlugin|ImagePlug|PagedImagePlugin)$/) {
350 $contains_image_plugin = 1;
351 }
[14754]352 $collwriter->startTag('plugin','name'=>$pluginname);
353
354 for (my $i=1; $i<scalar(@$pl); $i++) {
355 my $arg =@$pl[$i];
356 if ($arg =~ /^-/){
357 my $option_name=@$pl[$i];
358 my $option_value=@$pl[$i+1];
359 if (defined $option_value){
360 if ($option_value =~ /^-/){
361 $collwriter->startTag('option','name'=>$option_name);
362 $collwriter->endTag('option');
363 }else{
364 $collwriter->startTag('option','name'=>$option_name,'value'=>$option_value);
365 $collwriter->endTag('option');
366 }
367 }
368 }
369 }
370 $collwriter->endTag('plugin');
371 }
372 $collwriter->endTag('pluginList');
373 $collwriter->endTag('import');
374 }
375
[3979]376 $buildwriter->startTag('serviceRackList');
[3688]377
[4886]378 my @levels = ();
379 my $defaultlevel;
[13912]380 my $service_type = "MG";
381 if ($buildtype eq 'mgpp') {
382 $service_type = "MGPP";
383 } elsif ($buildtype eq "lucene") {
384 $service_type = "Lucene";
385 }
386
[14754]387 #indexSubcollectionList
388
389 my $subcollectionmap = $buildcfg->{'subcollectionmap'};
390 my $firstsubcollection = 1;
391 my $defaultsubcollection = "";
392 my @subcollist;
393 my $subcolmap = {};
394
395 if (defined $buildcfg->{'subcollectionmap'}) {
396 my $subcolmap_t = $buildcfg->{'subcollectionmap'};
397
398 foreach my $l (@$subcolmap_t) {
399 my @pair = split(/->/, $l);
400 $subcolmap->{$pair[0]} = $pair[1];
401 push @subcollist, $pair[0];
402 if ($firstsubcollection==1) {
403 $defaultsubcollection = $pair[1];
404 $firstsubcollection = 0;
405 }
406 }
407
408 }
409
410
[3947]411 #do the retrieve service
[13917]412 $buildwriter->startTag('serviceRack', 'name'=>"GS2".$service_type."Retrieve");
[13228]413 if ($buildtype eq 'mgpp' || $buildtype eq 'lucene') {
[13917]414 $buildwriter->emptyTag('defaultLevel', 'shortname'=>$default_retrieve_level);
[13115]415 } elsif ($buildtype eq "mg") {
[13912]416 $buildwriter->emptyTag('defaultIndex', 'shortname'=>$defaultindex);
[3688]417 }
[14754]418
419 if ((defined $defaultsubcollection) && ($defaultsubcollection ne "")) {
420 $buildwriter->emptyTag('defaultIndexSubcollection', 'shortname'=>$defaultsubcollection);
421 }
422
[10653]423 if (defined $indexstem) {
424 $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
425 }
[8967]426 # close off the Retrieve service
427 $buildwriter->endTag('serviceRack');
[3688]428
[3947]429 # add in the classifiers if needed
430
[3688]431 my $count = 1;
432 my $phind = 0;
433 my $started_classifiers = 0;
434 if (defined $collectcfg->{'classify'}) {
[4184]435 $collwriter->startTag('browse');
[10360]436 # add in default format if necessary
437 if (defined $format->{"VList"} || defined $format->{"HList"}) {
438 # global formats
439 $collwriter->startTag('format');
440 if (defined $format->{"VList"}) {
441 # VLIst applies to both classifier and doc nodes
442 write_format($collwriter, $format->{"VList"}, "document");
443 write_format($collwriter, $format->{"VList"}, "classifier");
444 }
445 if (defined $format->{"HList"}) {
446 # hlist is only for classifier nodes
447 write_format($collwriter, $format->{"HList"}, "horizontal");
448 }
449 $collwriter->endTag('format');
450 }
[3688]451 my $classifiers = $collectcfg->{'classify'};
[10360]452 foreach my $cl (@$classifiers) {
453 my $name = "CL$count";
[3688]454 $count++;
455 my ($classname) = @$cl[0];
[4184]456 if ($classname =~ /^phind$/i) {
[3688]457 $phind=1;
[4184]458 #should add it into coll config classifiers
[3688]459 next;
460 }
461
[9899]462 my $horizontalAtTop = &isHorizontalClassifier($database, $name);
[3688]463 if (not $started_classifiers) {
[8967]464 $buildwriter->startTag('serviceRack', 'name'=>'GS2Browse');
[10653]465 if (defined $indexstem) {
466 $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
467 }
468
[10360]469 $buildwriter->startTag('classifierList');
[3688]470 $started_classifiers = 1;
471 }
[4969]472 my $content = ''; #use buttonname first, then metadata
[4184]473 if ($classname eq "DateList") {
474 $content = "Date";
475 } else {
[10360]476 for (my $i=0; $i<scalar(@$cl); $i++) {
477 my $arg = @$cl[$i];
[4235]478 if ($arg eq "-buttonname"){
[4184]479 $content = @$cl[$i+1];
480 last;
[4886]481 } elsif ($arg eq "-metadata") {
[4235]482 $content = @$cl[$i+1];
[4184]483 }
[4235]484
[3688]485 }
486 }
[9899]487 if ($horizontalAtTop) {
488 $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content, 'horizontalAtTop'=>'true');
[14754]489
[9899]490 } else {
[4886]491 $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content);
[9899]492 }
[3688]493
[3947]494
[14754]495 # $collwriter->startTag('classifier', 'name'=>$name);
496 $collwriter->startTag('classifier', 'name'=>$classname);
497 for (my $i=1; $i<scalar(@$cl); $i++) {
498 my $arg =@$cl[$i];
499 if ($arg =~ /^-/){
500 my $option_name=@$cl[$i];
501 my $option_value=@$cl[$i+1];
502 if (defined $option_value){
503 if ($option_value=~ /^-/){
504 $collwriter->startTag('option','name'=>$option_name);
505 $collwriter->endTag('option');
506 }else{
507 $collwriter->startTag('option','name'=>$option_name,'value'=>$option_value);
508 $collwriter->endTag('option');
509 }
510 }
511 }
512 }
513
[10360]514 my $vlist = $name."VList";
515 my $hlist = $name."HList";
516 my $dlist = "";
[4184]517 if ($classname eq "DateList") {
518 $dlist = "DateList";
519 }
[10201]520 # need to work out how to split into classifier and document
[10360]521 if (defined $format->{$vlist} || defined $format->{$hlist} || defined $format->{$dlist}) {
[4184]522 $collwriter->startTag('format');
523 if (defined $format->{$vlist}) {
[10201]524 write_format($collwriter, $format->{$vlist}, "document");
[10360]525 write_format($collwriter, $format->{$vlist}, "classifier");
526 }
[4184]527 if (defined $format->{$hlist}) {
[10201]528 write_format($collwriter, $format->{$hlist}, "horizontal");
[10360]529 }
530
[4184]531 if (defined $format->{$dlist}) {
[10201]532 write_format($collwriter, $format->{$dlist}, "document");
[4184]533 }
534 $collwriter->endTag('format');
535 }
536 $collwriter->endTag('classifier');
[4969]537 } #foreach classifier
[3688]538 if ($started_classifiers) {
539 # end the classifiers
[3979]540 $buildwriter->endTag('classifierList');
[8967]541 # close off the Browse service
542 $buildwriter->endTag('serviceRack');
[3688]543 }
[3947]544
[4184]545 $collwriter->endTag('browse');
[3947]546 }
[4184]547
[8967]548
[3947]549 # the phind classifier is a separate service
550 if ($phind) {
551 # if phind classifier
[3979]552 $buildwriter->emptyTag('serviceRack', 'name'=>'PhindPhraseBrowse');
[3947]553 }
554
555 # do the search service
[13912]556 $buildwriter->startTag('serviceRack', 'name'=>'GS2'.$service_type.'Search');
[14754]557 #$buildwriter->emptyTag('defaultIndex', 'shortname'=>$defaultindex);
[13912]558 $buildwriter->startTag('indexList');
559 #for each index
560 foreach my $i (@indexlist) {
561 my $index = $indexmap->{$i};
562 $buildwriter->emptyTag('index', 'name'=>$i, 'shortname'=>$index);
563 }
564 $buildwriter->endTag('indexList');
565 if (defined $indexstem) {
566 $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
567 }
568
569 # index options
570 if ($buildtype eq 'mg' || $buildtype eq 'mgpp') {
571 $buildwriter->startTag('indexOptionList');
572 my $stemindexes = 3; # default is stem and casefold
573 if (defined $buildcfg->{'stemindexes'} && $buildcfg->{'stemindexes'} =~ /^\d+$/ ) {
574 $stemindexes = $buildcfg->{'stemindexes'};
575 }
576 $buildwriter->emptyTag('indexOption', 'name'=>'stemIndexes', 'value'=>$stemindexes);
577
578 my $maxnumeric = 4; # default
579 if (defined $buildcfg->{'maxnumeric'} && $buildcfg->{'maxnumeric'} =~ /^\d+$/) {
580 $maxnumeric = $buildcfg->{'maxnumeric'};
581 }
582 $buildwriter->emptyTag('indexOption', 'name'=>'maxnumeric', 'value'=>$maxnumeric);
583
584 $buildwriter->endTag('indexOptionList');
585 }
586
[13228]587 if ($buildtype eq 'mgpp' || $buildtype eq 'lucene') {
[3947]588
[13912]589 # level info
[14754]590 $buildwriter->emptyTag('defaultLevel', 'shortname'=>$default_search_level_shortname);
[13917]591 $buildwriter->emptyTag('defaultGDBMLevel', 'shortname'=>$default_retrieve_level);
[3979]592 $buildwriter->startTag('levelList');
[13917]593 foreach my $l (@levellist) {
594 my $level = $levelmap->{$l};
595 $buildwriter->emptyTag('level', 'name'=>$l, 'shortname'=>$level);
596 }
[3979]597 $buildwriter->endTag('levelList');
[3688]598
[5647]599 # do the search types if there
600 if (defined $collectcfg->{'searchtype'}) {
601 $buildwriter->startTag('searchTypeList');
[10360]602 foreach my $st (@{$collectcfg->{'searchtype'}}) {
[5647]603 $buildwriter->emptyTag('searchType', 'name'=>$st);
604 }
605 $buildwriter->endTag('searchTypeList');
[13228]606 } elsif (defined $format->{'SearchTypes'}) {
607 #check format statement
608 my $searchtype = $format->{'SearchTypes'};
609 $buildwriter->startTag('searchTypeList');
610 if ($searchtype =~ /form/) {
611 $buildwriter->emptyTag('searchType', 'name'=>'form');
612 }
613 if ($searchtype =~ /plain/) {
614 $buildwriter->emptyTag('searchType', 'name'=>'plain');
615 }
616 $buildwriter->endTag('searchTypeList');
[5647]617 }
[13912]618 }
[3947]619
[14754]620 #indexLanguageList
621 my $indexlanguages = $collectcfg->{'languages'};
622 my $firstindexlanguage = 1;
623 my $defaultindexlanguage_shortname;
624 if (defined $indexlanguages){
625 $buildwriter->startTag('indexLanguageList');
626 my $languages_t = $collectcfg->{'languages'};
627 foreach my $i (@$languages_t){
628 $buildwriter->startTag('indexLanguage','name'=>$i);
629 &output_display($buildwriter, 'name', $i, $i);
630 $buildwriter->endTag('indexLanguage');
631 if ($firstindexlanguage==1){
632 $defaultindexlanguage_shortname = $i;
633 $firstindexlanguage=0;
634 }
635 }
636 $buildwriter->endTag('indexLanguageList');
637 $buildwriter->startTag('defaultIndexLanguage', 'name'=>$defaultindexlanguage_shortname,'shortname'=>$defaultindexlanguage_shortname);
638 $buildwriter->endTag('defaultIndexLanguage');
639 }
[13115]640
[14754]641 # my $defaultsubcollection = "";
642 # my @subcollist;
643
644 if (scalar(@subcollist)>0){
645
646 $buildwriter->startTag('indexSubcollectionList');
647 foreach my $i (keys %$subcolmap){
648 my $short_name = $subcolmap->{$i};
649 $buildwriter->emptyTag('indexSubcollection', 'name'=>$i, 'shortname'=>$short_name);
650 }
651
652 $buildwriter->endTag('indexSubcollectionList');
653 $buildwriter->emptyTag('defaultIndexSubcollection', 'shortname'=>$defaultsubcollection);
654 }
655
656
[3979]657 $buildwriter->endTag('serviceRack');
[4969]658
[3979]659 $buildwriter->endTag('serviceRackList');
660 $buildwriter->endTag('buildConfig');
[11274]661
662 # we add in the default replace list just in case we have macros in the
663 # collection
664 $collwriter->emptyTag('replaceListRef', 'id'=>'gs2-standard');
[25669]665 $collwriter->emptyTag('replaceListRef', 'id'=>'gs2-image') if $contains_image_plugin;
[14754]666 $collwriter->endTag('CollectionConfig');
[4969]667 $collwriter->end();
[3979]668 $buildwriter->end();
669 $buildoutput->close();
670 $colloutput->close();
[9899]671 &close_database($database);
[3979]672}
[3688]673
[3979]674
[4969]675sub output_metadata {
[14754]676 my ($writer, $lang, $metaname, $metavalue) = @_;
677 $lang = 'en' if $lang eq 'default';
678 if ($lang ne ""){
679 $writer->startTag('metadata', 'lang'=>$lang, 'name'=>$metaname);
680 }else{
681 $writer->startTag('metadata', 'name'=>$metaname);
682 }
[4184]683 $writer->characters($metavalue);
684 $writer->endTag('metadata');
685}
686
687sub output_display {
[4969]688 my ($writer, $name, $lang, $value) = @_;
[4184]689 $lang = 'en' if $lang eq 'default';
690 if ($lang =~ /^\[/) {
691 ($lang) = $lang =~ /\[l=(.*)\]/;
[3979]692 }
[14754]693
[4969]694 $writer->startTag('displayItem', 'name'=>$name, 'lang'=>$lang);
[4184]695 $writer->characters($value);
[4969]696 $writer->endTag('displayItem');
[3688]697}
[4184]698sub format_icon_value {
699 my ($value) = @_;
700 if ($value =~ /^_/) {
[13828]701 my ($newvalue) = $value =~ /images\/(.*)$/;
[4184]702 if ($newvalue) {
703 return $newvalue;
704 }
705 }
706 return $value;
707}
[3688]708
[13828]709sub tidy_up_display_item {
[12678]710 my ($value) = @_;
711 # remove \n
712 $value =~ s/\\n//g;
713 # replace \' with '
714 $value =~ s/\\\'/\'/g;
715 # replace \" with "
716 $value =~ s/\\\"/\"/g;
[13828]717 # replace _httpprefix_ with _httpsite_
[13843]718 $value =~ s/_httpprefix_/_httpsite_/g;
719 $value =~ s/_gwcgi_//g;
720 $value =~ s/[a-z][a-z]?=_cgiarg[a-z][a-z]?_&?//g;
721 $value =~ s/&p=/&sa=/g;
[12678]722 return $value;
723}
724
[26124]725sub format_if_or {
726 my ($format, $node_type) = @_;
727
728 # while we find nested if/or statements, recurse to find more nested if/or statements,
729 # and try to expand (process) these nested statements starting from innermost going to outermost
730
731 while($format =~ m/^.*\{(?:If|Or)\}\{[^\}\{]*\{/) { # contains nested if/or statement, expand it
732
733 my ($prefix, $nested_to_process, $suffix) = $format =~ m/^(.*\{(?:If|Or)\}\{[^\}\{]*)(\{[^\}]*\}\s*\{[^\}]*\})(.*)$/g; # recursion step
734
735 #print STDERR "prefix: |$prefix|\n\nnested: |$nested_to_process|\n\nsuffix: |$suffix|\n\n";
736 $format = $prefix . &format_if_or($nested_to_process, $node_type) . $suffix;
737 }
738
739 if($format =~ m/\{(If|Or)\}\{[^\}\{]*\}/g) { # base step: contains if/or statement(s), but none nested
740 # expand them
741 $format =~ s/\{If\}\{([^\}]*)\}/&format_if($1, $node_type)/eg;
742 $format =~ s/\{Or\}\{([^\}]*)\}/&format_or($1)/eg;
743 }
744 return $format;
745}
746
[4184]747sub write_format {
[10201]748 my ($writer, $old_format, $node_type) = @_;
[7471]749 # replace \' with '
750 $old_format =~ s/\\\'/\'/g;
[12236]751 # replace \" with "
752 $old_format =~ s/\\\"/\"/g;
[4184]753 #convert [] to <gsf:...>
[26124]754 # now handles nested {If} and {Or}
755 $old_format = &format_if_or($old_format, $node_type);
[4184]756 $old_format =~ s/\[Text\]/\<gsf:text\/\>/g;
757 $old_format =~ s/\[num\]/\<gsf:num\/\>/g;
[10782]758 $old_format =~ s/\[link\]/\<gsf:link type=\'$node_type\'\>/g;
[4184]759 $old_format =~ s/\[\/link\]/\<\/gsf:link\>/g;
760 $old_format =~ s/\[srclink\]/\<gsf:link type=\'source\'\>/g;
761 $old_format =~ s/\[\/srclink\]/\<\/gsf:link\>/g;
[10782]762 $old_format =~ s/\[icon\]/\<gsf:icon type=\'$node_type\'\/\>/g;
[4184]763 $old_format =~ s/\[srcicon\]/\<gsf:icon type=\'source\'\/\>/g;
[10360]764
765 # what to do with hightlight??
766 $old_format =~ s/\[\/?highlight\]//g;
[4184]767
[4969]768 #now do the rest of the [] which are assumed to be metadata
[10360]769 $old_format =~ s/\[([^\]]*)\]/&format_metadata($1)/eg;
770
771 # some html tidy
[4184]772 #turn <br> into <br />
773 $old_format =~ s/\<br\>/\<br \/\>/g;
774 #turn <p> into <p />
775 $old_format =~ s/\<p\>/\<p \/\>/g;
776
777 #put quotes around any atts
778 $old_format =~ s/=([a-z]+)([> ])/=\'$1\'$2/g;
779
[10201]780 if ($node_type eq "document") {
781 $writer->startTag('gsf:template', 'match'=>'documentNode');
782 $writer->charactersXML($old_format);
783 $writer->endTag('gsf:template');
784 } elsif ($node_type eq "classifier") {
785 $writer->startTag('gsf:template', 'match'=>'classifierNode');
786 $writer->charactersXML($old_format);
787 $writer->endTag('gsf:template');
788 } elsif ($node_type eq "horizontal") {
789 $writer->startTag('gsf:template', 'match'=>'classifierNode', 'mode'=>'horizontal');
790 $writer->charactersXML($old_format);
791 $writer->endTag('gsf:template');
[4184]792
[10201]793 }
[4184]794}
795
[10360]796sub format_metadata {
797 my ($metadata_string) = @_;
798
799 #print STDERR "original meta = $metadata_string\n";
800
801 # what shall we do with cgisafe??
802 my $cgisafe = $metadata_string =~ s/^cgisafe://;
803
804 my ($select) = $metadata_string =~ /^(parent|sibling)/;
805 $metadata_string =~ s/^(parent|sibling)//;
806 my ($scope, $delim);
807
808 if ($select) {
809 ($scope) = $metadata_string =~ /^\((Top|All)/;
810 $metadata_string =~ s/^\((Top|All)\)?//;
811 if ($scope) {
[10782]812 ($delim) = $metadata_string =~ /^\'([^\']*)\'\)/;
813 $metadata_string =~ s/^\'([^\']*)\'\)//;
[10360]814 }
815 }
816 $metadata_string =~ s/^://;
817 # remove ex.
818 $metadata_string =~ s/^ex\.//;
819
820 #print STDERR "select=$select, scope=$scope, delim=|$delim|, meta = $metadata_string\n";
821
822 my $new_format = "<gsf:metadata name='$metadata_string' ";
823 if (defined $select) {
824 if ($select eq "sibling") {
825 $new_format .= "multiple='true' ";
826 if (defined $delim) {
827 $new_format .= "separator='$delim' ";
828 }
829 } elsif ($select eq "parent"){
830 if (defined $scope) {
831 if ($scope eq "Top") {
832 $new_format .= "select='root' ";
833 } elsif ($scope eq "All") {
834 $new_format .= "select='ancestors' ";
835 if (defined $delim) {
836 $new_format .= "separator='$delim' ";
837 }
838 }
839 } else {
840 $new_format .= "select='parent' ";
841 }
842 }
843 }
844 $new_format .= "/>";
845 #print STDERR "$new_format\n";
846 return $new_format;
847
848}
849
850sub format_if {
851
852 my ($if_string, $node_type) = @_;
[10782]853 #print STDERR "if string = $if_string\n";
[10360]854
[10782]855 my @parts = split /,/, $if_string;
856 my $test = $parts[0];
857 my $true_option = $parts[1];
858 my $false_option;
859 if (scalar (@parts) == 3) {
860 $false_option = $parts[2];
861 }
862 $test =~ s/^\s*//;
863 $test =~ s/\s*$//;
864 my ($test_meta, $test_type, $test_value);
865 if ($test =~ /^(\[.+\])$/) {
866 $test_meta = $1;
867 $test_type = 'exists';
868 } else {
869 my ($lhs, $exp, $rhs) = $test =~ /^(.+)\s+(eq|ne|lt|gt|le|ge|sw|ew)\s+(.+)$/;
870 #print STDERR "lhs, exp, rhs = $lhs, $exp, $rhs\n";
871 if ($exp eq "eq") {
872 $test_type = "equals";
873 } elsif ($exp eq "sw") {
874 $test_type = "startsWith";
875 } elsif ($exp eq "ew") {
876 $test_type = "endsWith";
877 } elsif ($exp eq "ne") {
878 $test_type = "notEquals";
879 } elsif ($exp eq "lt") {
880 $test_type = "lessThan";
881 }elsif ($exp eq "gt") {
882 $test_type = "greaterThan";
883 }elsif ($exp eq "le") {
884 $test_type = "lessThanOrEquals";
885 }elsif ($exp eq "ge") {
886 $test_type = "greaterThanOrEquals";
887 }
888 if ($lhs =~ /^\[.+\]$/) {
889 $test_meta = $lhs;
890 $test_value = $rhs;
891 } else {
892 # assume rhs has meta
893 $test_meta = $rhs;
894 $test_value = $lhs;
895 }
896
897 #remove beginning and end quotes
898 $test_value =~ s/^[\'\"]//;
899 $test_value =~ s/[\'\"]$//;
900 }
901 my $test_atts = "test='$test_type' ";
902 if (defined $test_value) {
903 $test_atts .= "test-value='$test_value' ";
904 }
905 #print STDERR "test, true, false = $test, $true_option, $false_option\n";
906 my $new_format = "<gsf:switch>$test_meta";
907 $new_format .= "<gsf:when $test_atts>$true_option</gsf:when>";
908 if (defined $false_option) {
909 $new_format .="<gsf:otherwise>$false_option</gsf:otherwise>";
910 }
911 $new_format .= "</gsf:switch>";
912
913 #print STDERR "new format = $new_format\n";
914 return $new_format;
[10360]915}
916
917sub format_or {
918 my ($or_string) = @_;
919 my @meta_list = split (',', $or_string);
920 return "" unless scalar (@meta_list);
921 my $new_format = "<gsf:choose-metadata>";
922 foreach my $m (@meta_list) {
923 if ($m =~ /^\[(.*)\]$/) {
924 $new_format .= &format_metadata($1);
925 } else {
926 # a default value
927 $new_format .= "<gsf:default>$m</gsf:default>";
928 last;
929 }
930 }
931 $new_format .= "</gsf:choose-metadata>";
932 return $new_format;
933}
934
[9899]935sub open_database {
936 my ($db_file) = @_;
937
938 my $database = ();
[10658]939# tie (%$database, 'GDBM_File', $db_file, GDBM_READER, 0400) ||
940# die "Couldn't open database $db_file\n";
[9899]941
942 return $database;
943}
944
945sub close_database {
946 my ($database) = @_;
[10360]947 untie %$database;
[9899]948}
949sub isHorizontalClassifier {
950 my ($database, $name) = @_;
951
[10658]952 return 0; # can't get this to work for windows
[10360]953 my $record = $database->{$name};
[9899]954 my ($childtype) = $record =~ /<childtype>(\w*)/;
955 if ($childtype eq "HList") {
956 return 1;
957 }
958 return 0;
959}
[3688]960#$writer->startTag('');
961#$writer->endTag('');
962#$writer->characters();
963#$writer->emptyTag('');
964
9651;
Note: See TracBrowser for help on using the repository browser.