| 1 | #!/usr/bin/perl -w
|
|---|
| 2 |
|
|---|
| 3 | BEGIN {
|
|---|
| 4 | die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
|
|---|
| 5 | die "GSDL3HOME not set\n" unless defined $ENV{'GSDL3HOME'};
|
|---|
| 6 | die "GSDL3SRCHOME not set\n" unless defined $ENV{'GSDL3SRCHOME'};
|
|---|
| 7 | die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
|
|---|
| 8 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
|
|---|
| 9 | unshift (@INC, "$ENV{'GSDL3SRCHOME'}/lib/perl/cpan");
|
|---|
| 10 | }
|
|---|
| 11 |
|
|---|
| 12 | use colcfg;
|
|---|
| 13 | use util;
|
|---|
| 14 | use parsargv;
|
|---|
| 15 | use FileHandle;
|
|---|
| 16 | use XML::Writer;
|
|---|
| 17 | #can't get this to work on windows
|
|---|
| 18 | #use GDBM_File;
|
|---|
| 19 |
|
|---|
| 20 | use strict;
|
|---|
| 21 |
|
|---|
| 22 | &main();
|
|---|
| 23 | sub print_usage() {
|
|---|
| 24 | print STDOUT "Usage: convert_coll_from_gs2.pl [options] coll-name\n";
|
|---|
| 25 | print STDOUT "options:\n";
|
|---|
| 26 |
|
|---|
| 27 | print STDOUT " -collectdir Directory where collection lives.\n";
|
|---|
| 28 | print STDOUT " -verbosity Controls the amount of output.\n";
|
|---|
| 29 | print STDOUT " -defaultlang The language that is considered the default (for display text etc). defaults to 'en'\n\n";
|
|---|
| 30 | }
|
|---|
| 31 |
|
|---|
| 32 |
|
|---|
| 33 | sub main {
|
|---|
| 34 |
|
|---|
| 35 | my ($defaultlang, $verbosity, $collectdir);
|
|---|
| 36 | # note that no defaults are passed for most options as they're set
|
|---|
| 37 | # later (after we check the collect.cfg file)
|
|---|
| 38 | if (!&parsargv::parse(\@ARGV,
|
|---|
| 39 | 'verbosity/\d+/', \$verbosity,
|
|---|
| 40 | 'collectdir/.*/', \$collectdir,
|
|---|
| 41 | 'defaultlang/.*/', \$defaultlang)) {
|
|---|
| 42 | &print_usage();
|
|---|
| 43 | die "\n";
|
|---|
| 44 | }
|
|---|
| 45 |
|
|---|
| 46 | # get and check the collection name
|
|---|
| 47 | my ($collection) = @ARGV;
|
|---|
| 48 | if (!defined($collection) || $collection eq "") {
|
|---|
| 49 | die "No collection specified\n";
|
|---|
| 50 | }
|
|---|
| 51 | if ($collection eq "gs2model") {
|
|---|
| 52 | die "You cant convert the model collection\n";
|
|---|
| 53 | }
|
|---|
| 54 |
|
|---|
| 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 |
|
|---|
| 65 | my $collconfigfilename = &util::filename_cat ($collectdir, "etc", "collect.cfg");
|
|---|
| 66 | print STDOUT "coll config=$collconfigfilename\n";
|
|---|
| 67 | my $collectcfg;
|
|---|
| 68 | if (-e $collconfigfilename) {
|
|---|
| 69 | $collectcfg = &colcfg::read_collect_cfg ($collconfigfilename);
|
|---|
| 70 |
|
|---|
| 71 | } else {
|
|---|
| 72 | print STDERR "collect.cfg not found!!";
|
|---|
| 73 | die "\n";
|
|---|
| 74 | }
|
|---|
| 75 |
|
|---|
| 76 |
|
|---|
| 77 | my $buildconfigfilename = &util::filename_cat ($collectdir, "index", "build.cfg");
|
|---|
| 78 | my $buildcfg;
|
|---|
| 79 | if (-e $buildconfigfilename) {
|
|---|
| 80 | $buildcfg = &colcfg::read_build_cfg ($buildconfigfilename);
|
|---|
| 81 |
|
|---|
| 82 | } else {
|
|---|
| 83 | print STDERR "build.cfg not found!!";
|
|---|
| 84 | die "\n";
|
|---|
| 85 | }
|
|---|
| 86 |
|
|---|
| 87 |
|
|---|
| 88 |
|
|---|
| 89 | my $colloutfile = &util::filename_cat ($collectdir, "etc", "collectionConfig.xml");
|
|---|
| 90 | if (-e $colloutfile) {
|
|---|
| 91 | print STDOUT "collectionConfig file already exists! overwriting it!\n";
|
|---|
| 92 |
|
|---|
| 93 | }
|
|---|
| 94 |
|
|---|
| 95 | my $buildoutfile = &util::filename_cat ($collectdir, "index", "buildConfig.xml");
|
|---|
| 96 | if (-e $buildoutfile) {
|
|---|
| 97 | print STDOUT "buildConfig file already exists! overwriting it!\n";
|
|---|
| 98 |
|
|---|
| 99 | }
|
|---|
| 100 |
|
|---|
| 101 | # my $db_file = &util::filename_cat ($collectdir, "index", "text", "$collection.ldb");
|
|---|
| 102 | my $database;
|
|---|
| 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 | # }
|
|---|
| 109 |
|
|---|
| 110 | my $buildtype;
|
|---|
| 111 | if (defined $buildcfg->{'buildtype'}) {
|
|---|
| 112 | $buildtype = $buildcfg->{'buildtype'};
|
|---|
| 113 | } else {
|
|---|
| 114 | $buildtype = 'mg';
|
|---|
| 115 | }
|
|---|
| 116 |
|
|---|
| 117 | my $indexstem = undef;
|
|---|
| 118 | if (defined $buildcfg->{'indexstem'}) {
|
|---|
| 119 | $indexstem = $buildcfg->{'indexstem'};
|
|---|
| 120 | }
|
|---|
| 121 |
|
|---|
| 122 | my $buildoutput = new IO::File(">$buildoutfile");
|
|---|
| 123 | my $buildwriter = new XML::Writer(OUTPUT => $buildoutput, NEWLINES => 1);
|
|---|
| 124 |
|
|---|
| 125 | $buildwriter->startTag('buildConfig', 'xmlns:gsf'=>"http://www.greenstone.org/greenstone3/schema/ConfigFormat");
|
|---|
| 126 |
|
|---|
| 127 | my $colloutput = new IO::File(">$colloutfile");
|
|---|
| 128 | my $collwriter = new XML::Writer(OUTPUT => $colloutput, NEWLINES => 1);
|
|---|
| 129 |
|
|---|
| 130 | $collwriter->startTag('collectionConfig', 'xmlns:gsf'=>"http://www.greenstone.org/greenstone3/schema/ConfigFormat", 'xmlns:xsl'=>'http://www.w3.org/1999/XSL/Transform');
|
|---|
| 131 |
|
|---|
| 132 | #output the collection metadata to the collectionConfig file
|
|---|
| 133 | $collwriter->startTag('metadataList');
|
|---|
| 134 |
|
|---|
| 135 | my $creator = $collectcfg->{'creator'};
|
|---|
| 136 | &output_metadata($collwriter, 'creator', $creator);
|
|---|
| 137 | my $public =$collectcfg->{'public'};
|
|---|
| 138 | &output_metadata($collwriter, 'public', $public);
|
|---|
| 139 | $collwriter->endTag('metadataList');
|
|---|
| 140 |
|
|---|
| 141 | #output the display collectionmeta to collectionConfig.xml
|
|---|
| 142 |
|
|---|
| 143 | my $collectionmeta = $collectcfg->{'collectionmeta'};
|
|---|
| 144 | if (defined $collectionmeta) {
|
|---|
| 145 | my %name_map = ('collectionname', 'name',
|
|---|
| 146 | 'collectionextra', 'description',
|
|---|
| 147 | 'iconcollection', 'icon',
|
|---|
| 148 | 'iconcollectionsmall', 'smallicon');
|
|---|
| 149 |
|
|---|
| 150 | $collwriter->startTag('displayItemList');
|
|---|
| 151 | foreach my $entry ( keys %$collectionmeta) {
|
|---|
| 152 | if (defined $name_map{$entry}) {
|
|---|
| 153 | my $name= $name_map{$entry};
|
|---|
| 154 | foreach my $lang (keys %{$collectionmeta->{$entry}}) {
|
|---|
| 155 | my $value = $collectionmeta->{$entry}->{$lang};
|
|---|
| 156 | if ($entry =~ /^icon/) {
|
|---|
| 157 | $value = format_icon_value($value);
|
|---|
| 158 | } else {
|
|---|
| 159 | $value = tidy_up_display_item($value);
|
|---|
| 160 | }
|
|---|
| 161 | &output_display($collwriter, $name, $lang, $value);
|
|---|
| 162 | }
|
|---|
| 163 | }
|
|---|
| 164 | }
|
|---|
| 165 | $collwriter->endTag('displayItemList');
|
|---|
| 166 | }
|
|---|
| 167 |
|
|---|
| 168 | # output building metadata to build config file
|
|---|
| 169 | my $numdocs = $buildcfg->{'numdocs'};
|
|---|
| 170 | $buildwriter->startTag('metadataList');
|
|---|
| 171 | &output_metadata($buildwriter, 'numDocs', $numdocs);
|
|---|
| 172 | &output_metadata($buildwriter, 'buildType', $buildtype);
|
|---|
| 173 | $buildwriter->endTag('metadataList');
|
|---|
| 174 |
|
|---|
| 175 |
|
|---|
| 176 | #indexes
|
|---|
| 177 | my $indexmap = {};
|
|---|
| 178 | my $firstindex = "";
|
|---|
| 179 | my $first = 1;
|
|---|
| 180 | if (defined $buildcfg->{'indexmap'}) {
|
|---|
| 181 | my $indexmap_t = $buildcfg->{'indexmap'};
|
|---|
| 182 | foreach my $i (@$indexmap_t) {
|
|---|
| 183 | my ($k, $v) = $i =~ /^(.*)\-\>(.*)$/;
|
|---|
| 184 | $indexmap->{$k} = $v;
|
|---|
| 185 | if ($first) {
|
|---|
| 186 | $firstindex = $v;
|
|---|
| 187 | $first = 0;
|
|---|
| 188 | }
|
|---|
| 189 | }
|
|---|
| 190 | } else {
|
|---|
| 191 | print STDERR "indexmap not defined";
|
|---|
| 192 | }
|
|---|
| 193 | my $defaultindex;
|
|---|
| 194 | if ($buildtype eq "mg") {
|
|---|
| 195 | if (defined $collectcfg->{'defaultindex'}) {
|
|---|
| 196 | $defaultindex = $collectcfg->{'defaultindex'};
|
|---|
| 197 | $defaultindex = $indexmap->{$defaultindex};
|
|---|
| 198 | } else {
|
|---|
| 199 | # use the first one
|
|---|
| 200 | $defaultindex = $firstindex;
|
|---|
| 201 | }
|
|---|
| 202 | }
|
|---|
| 203 | elsif ($buildtype eq "mgpp") {
|
|---|
| 204 | $defaultindex = "idx";
|
|---|
| 205 | }
|
|---|
| 206 | elsif ($buildtype eq "lucene") {
|
|---|
| 207 | # lucene not implemented yet
|
|---|
| 208 | $defaultindex = "idx";
|
|---|
| 209 | }
|
|---|
| 210 |
|
|---|
| 211 | # format stuff
|
|---|
| 212 | my $format = $collectcfg->{'format'};
|
|---|
| 213 |
|
|---|
| 214 | #output the search stuff to coll cfg
|
|---|
| 215 | $collwriter->startTag('search');
|
|---|
| 216 | foreach my $i (keys %$indexmap) {
|
|---|
| 217 | my $shortname = $indexmap->{$i};
|
|---|
| 218 | $collwriter->startTag('index', 'name'=>$shortname);
|
|---|
| 219 | #find the coll meta stuff
|
|---|
| 220 | my $indexdisplay = ".$i";
|
|---|
| 221 | foreach my $lang (keys %{$collectionmeta->{$indexdisplay}}) {
|
|---|
| 222 | my $value = $collectionmeta->{$indexdisplay}->{$lang};
|
|---|
| 223 | output_display($collwriter, 'name', $lang, $value);
|
|---|
| 224 | }
|
|---|
| 225 | $collwriter->endTag('index');
|
|---|
| 226 |
|
|---|
| 227 | }
|
|---|
| 228 |
|
|---|
| 229 | # add in the format stuff
|
|---|
| 230 | if (defined $format->{'SearchVList'}) {
|
|---|
| 231 |
|
|---|
| 232 | $collwriter->startTag('format');
|
|---|
| 233 | write_format($collwriter, $format->{'SearchVList'}, "document");
|
|---|
| 234 | $collwriter->endTag('format');
|
|---|
| 235 | }
|
|---|
| 236 | elsif (defined $format->{'VList'}) {
|
|---|
| 237 | $collwriter->startTag('format');
|
|---|
| 238 | write_format($collwriter, $format->{'VList'}, "document");
|
|---|
| 239 | $collwriter->endTag('format');
|
|---|
| 240 | }
|
|---|
| 241 |
|
|---|
| 242 | $collwriter->endTag('search');
|
|---|
| 243 |
|
|---|
| 244 | $buildwriter->startTag('serviceRackList');
|
|---|
| 245 |
|
|---|
| 246 | my @levels = ();
|
|---|
| 247 | my $defaultlevel;
|
|---|
| 248 | my $service_name = "";
|
|---|
| 249 | #do the retrieve service
|
|---|
| 250 | # assume mgpp or mg
|
|---|
| 251 | if ($buildtype eq 'mgpp' || $buildtype eq 'lucene') {
|
|---|
| 252 | #for each level
|
|---|
| 253 | if (defined $buildcfg->{'indexlevels'}) {
|
|---|
| 254 | push @levels, @{$buildcfg->{'indexlevels'}};
|
|---|
| 255 |
|
|---|
| 256 | if (defined $buildcfg->{'textlevel'}) {
|
|---|
| 257 | $defaultlevel = $buildcfg->{'textlevel'};
|
|---|
| 258 | } else {
|
|---|
| 259 | $defaultlevel = $levels[0];
|
|---|
| 260 | }
|
|---|
| 261 | } else { #use levels from collect.cfg - must be an old collection
|
|---|
| 262 | @levels = ('Document');
|
|---|
| 263 | $defaultlevel = 'Document';
|
|---|
| 264 | if (defined $collectcfg->{'levels'}) {
|
|---|
| 265 | foreach my $l (@{$collectcfg->{'levels'}}) {
|
|---|
| 266 | if ($l eq "Section") {
|
|---|
| 267 | $defaultlevel = 'Section';
|
|---|
| 268 | }
|
|---|
| 269 | push @levels, $l;
|
|---|
| 270 | }
|
|---|
| 271 | }
|
|---|
| 272 | }
|
|---|
| 273 | if ($buildtype eq 'mgpp') {
|
|---|
| 274 | $service_name = "MGPP";
|
|---|
| 275 | } else {
|
|---|
| 276 | $service_name = "Lucene";
|
|---|
| 277 | }
|
|---|
| 278 | $buildwriter->startTag('serviceRack', 'name'=>"GS2".$service_name."Retrieve");
|
|---|
| 279 | $buildwriter->emptyTag('defaultLevel', 'name'=>$defaultlevel);
|
|---|
| 280 |
|
|---|
| 281 |
|
|---|
| 282 | } elsif ($buildtype eq "mg") {
|
|---|
| 283 | $buildwriter->startTag('serviceRack', 'name'=>'GS2MGRetrieve');
|
|---|
| 284 | $buildwriter->emptyTag('defaultIndex', 'name'=>$defaultindex);
|
|---|
| 285 |
|
|---|
| 286 | }
|
|---|
| 287 | if (defined $indexstem) {
|
|---|
| 288 | $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
|
|---|
| 289 | }
|
|---|
| 290 | # close off the Retrieve service
|
|---|
| 291 | $buildwriter->endTag('serviceRack');
|
|---|
| 292 |
|
|---|
| 293 | # add in the classifiers if needed
|
|---|
| 294 |
|
|---|
| 295 | my $count = 1;
|
|---|
| 296 | my $phind = 0;
|
|---|
| 297 | my $started_classifiers = 0;
|
|---|
| 298 | if (defined $collectcfg->{'classify'}) {
|
|---|
| 299 | $collwriter->startTag('browse');
|
|---|
| 300 | # add in default format if necessary
|
|---|
| 301 | if (defined $format->{"VList"} || defined $format->{"HList"}) {
|
|---|
| 302 | # global formats
|
|---|
| 303 | $collwriter->startTag('format');
|
|---|
| 304 | if (defined $format->{"VList"}) {
|
|---|
| 305 | # VLIst applies to both classifier and doc nodes
|
|---|
| 306 | write_format($collwriter, $format->{"VList"}, "document");
|
|---|
| 307 | write_format($collwriter, $format->{"VList"}, "classifier");
|
|---|
| 308 | }
|
|---|
| 309 | if (defined $format->{"HList"}) {
|
|---|
| 310 | # hlist is only for classifier nodes
|
|---|
| 311 | write_format($collwriter, $format->{"HList"}, "horizontal");
|
|---|
| 312 | }
|
|---|
| 313 | $collwriter->endTag('format');
|
|---|
| 314 | }
|
|---|
| 315 | my $classifiers = $collectcfg->{'classify'};
|
|---|
| 316 | foreach my $cl (@$classifiers) {
|
|---|
| 317 | my $name = "CL$count";
|
|---|
| 318 | $count++;
|
|---|
| 319 | my ($classname) = @$cl[0];
|
|---|
| 320 | if ($classname =~ /^phind$/i) {
|
|---|
| 321 | $phind=1;
|
|---|
| 322 | #should add it into coll config classifiers
|
|---|
| 323 | next;
|
|---|
| 324 | }
|
|---|
| 325 |
|
|---|
| 326 | my $horizontalAtTop = &isHorizontalClassifier($database, $name);
|
|---|
| 327 | if (not $started_classifiers) {
|
|---|
| 328 | $buildwriter->startTag('serviceRack', 'name'=>'GS2Browse');
|
|---|
| 329 | if (defined $indexstem) {
|
|---|
| 330 | $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
|
|---|
| 331 | }
|
|---|
| 332 |
|
|---|
| 333 | $buildwriter->startTag('classifierList');
|
|---|
| 334 | $started_classifiers = 1;
|
|---|
| 335 | }
|
|---|
| 336 | my $content = ''; #use buttonname first, then metadata
|
|---|
| 337 | if ($classname eq "DateList") {
|
|---|
| 338 | $content = "Date";
|
|---|
| 339 | } else {
|
|---|
| 340 | for (my $i=0; $i<scalar(@$cl); $i++) {
|
|---|
| 341 | my $arg = @$cl[$i];
|
|---|
| 342 | if ($arg eq "-buttonname"){
|
|---|
| 343 | $content = @$cl[$i+1];
|
|---|
| 344 | last;
|
|---|
| 345 | } elsif ($arg eq "-metadata") {
|
|---|
| 346 | $content = @$cl[$i+1];
|
|---|
| 347 | }
|
|---|
| 348 |
|
|---|
| 349 | }
|
|---|
| 350 | }
|
|---|
| 351 | if ($horizontalAtTop) {
|
|---|
| 352 | $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content, 'horizontalAtTop'=>'true');
|
|---|
| 353 | } else {
|
|---|
| 354 | $buildwriter->emptyTag('classifier', 'name'=>$name, 'content'=>$content);
|
|---|
| 355 | }
|
|---|
| 356 |
|
|---|
| 357 |
|
|---|
| 358 | $collwriter->startTag('classifier', 'name'=>$name);
|
|---|
| 359 | my $vlist = $name."VList";
|
|---|
| 360 | my $hlist = $name."HList";
|
|---|
| 361 | my $dlist = "";
|
|---|
| 362 | if ($classname eq "DateList") {
|
|---|
| 363 | $dlist = "DateList";
|
|---|
| 364 | }
|
|---|
| 365 | # need to work out how to split into classifier and document
|
|---|
| 366 | if (defined $format->{$vlist} || defined $format->{$hlist} || defined $format->{$dlist}) {
|
|---|
| 367 | $collwriter->startTag('format');
|
|---|
| 368 | if (defined $format->{$vlist}) {
|
|---|
| 369 | write_format($collwriter, $format->{$vlist}, "document");
|
|---|
| 370 | write_format($collwriter, $format->{$vlist}, "classifier");
|
|---|
| 371 | }
|
|---|
| 372 | if (defined $format->{$hlist}) {
|
|---|
| 373 | write_format($collwriter, $format->{$hlist}, "horizontal");
|
|---|
| 374 | }
|
|---|
| 375 |
|
|---|
| 376 | if (defined $format->{$dlist}) {
|
|---|
| 377 | write_format($collwriter, $format->{$dlist}, "document");
|
|---|
| 378 | }
|
|---|
| 379 | $collwriter->endTag('format');
|
|---|
| 380 | }
|
|---|
| 381 | $collwriter->endTag('classifier');
|
|---|
| 382 | } #foreach classifier
|
|---|
| 383 | if ($started_classifiers) {
|
|---|
| 384 | # end the classifiers
|
|---|
| 385 | $buildwriter->endTag('classifierList');
|
|---|
| 386 | # close off the Browse service
|
|---|
| 387 | $buildwriter->endTag('serviceRack');
|
|---|
| 388 | }
|
|---|
| 389 |
|
|---|
| 390 | $collwriter->endTag('browse');
|
|---|
| 391 | }
|
|---|
| 392 |
|
|---|
| 393 |
|
|---|
| 394 | # the phind classifier is a separate service
|
|---|
| 395 | if ($phind) {
|
|---|
| 396 | # if phind classifier
|
|---|
| 397 | $buildwriter->emptyTag('serviceRack', 'name'=>'PhindPhraseBrowse');
|
|---|
| 398 | }
|
|---|
| 399 |
|
|---|
| 400 | # do the search service
|
|---|
| 401 | if ($buildtype eq 'mgpp' || $buildtype eq 'lucene') {
|
|---|
| 402 |
|
|---|
| 403 | $buildwriter->startTag('serviceRack', 'name'=>'GS2'.$service_name.'Search');
|
|---|
| 404 | $buildwriter->emptyTag('defaultLevel', 'name'=>$defaultlevel);
|
|---|
| 405 | $buildwriter->startTag('levelList');
|
|---|
| 406 | foreach my $level (@levels) {
|
|---|
| 407 | $buildwriter->emptyTag('level', 'name'=>$level);
|
|---|
| 408 | }
|
|---|
| 409 | $buildwriter->endTag('levelList');
|
|---|
| 410 |
|
|---|
| 411 | #fieldlist
|
|---|
| 412 | my $fieldmap = {};
|
|---|
| 413 | my $fieldlist = ();
|
|---|
| 414 | if (defined $buildcfg->{'indexfields'}) {
|
|---|
| 415 | push @$fieldlist, @{$buildcfg->{'indexfields'}};
|
|---|
| 416 | if (defined $buildcfg->{'indexfieldmap'}) {
|
|---|
| 417 | my $fieldmap_t = $buildcfg->{'indexfieldmap'};
|
|---|
| 418 | foreach my $f (@$fieldmap_t) {
|
|---|
| 419 | my ($k, $v) = $f =~ /^(.*)\-\>(.*)$/;
|
|---|
| 420 | $fieldmap->{$k} = $v;
|
|---|
| 421 | }
|
|---|
| 422 | }
|
|---|
| 423 | $buildwriter->startTag('fieldList');
|
|---|
| 424 | foreach my $f (@$fieldlist) {
|
|---|
| 425 | my $field = $fieldmap->{$f};
|
|---|
| 426 | $buildwriter->emptyTag('field', 'shortname'=>$field, 'name'=>$f);
|
|---|
| 427 | }
|
|---|
| 428 | $buildwriter->endTag('fieldList');
|
|---|
| 429 | } else {
|
|---|
| 430 | print STDERR "indexfieldmap not defined";
|
|---|
| 431 | }
|
|---|
| 432 |
|
|---|
| 433 | # do the search types if there
|
|---|
| 434 | if (defined $collectcfg->{'searchtype'}) {
|
|---|
| 435 | $buildwriter->startTag('searchTypeList');
|
|---|
| 436 | foreach my $st (@{$collectcfg->{'searchtype'}}) {
|
|---|
| 437 | $buildwriter->emptyTag('searchType', 'name'=>$st);
|
|---|
| 438 | }
|
|---|
| 439 | $buildwriter->endTag('searchTypeList');
|
|---|
| 440 | } elsif (defined $format->{'SearchTypes'}) {
|
|---|
| 441 | #check format statement
|
|---|
| 442 | my $searchtype = $format->{'SearchTypes'};
|
|---|
| 443 | $buildwriter->startTag('searchTypeList');
|
|---|
| 444 | if ($searchtype =~ /form/) {
|
|---|
| 445 | $buildwriter->emptyTag('searchType', 'name'=>'form');
|
|---|
| 446 | }
|
|---|
| 447 | if ($searchtype =~ /plain/) {
|
|---|
| 448 | $buildwriter->emptyTag('searchType', 'name'=>'plain');
|
|---|
| 449 | }
|
|---|
| 450 | $buildwriter->endTag('searchTypeList');
|
|---|
| 451 | }
|
|---|
| 452 | } elsif ($buildtype eq 'mg') {
|
|---|
| 453 | $buildwriter->startTag('serviceRack', 'name'=>'GS2MGSearch');
|
|---|
| 454 | }
|
|---|
| 455 |
|
|---|
| 456 | if (defined $indexstem) {
|
|---|
| 457 | $buildwriter->emptyTag('indexStem', 'name'=>$indexstem);
|
|---|
| 458 | }
|
|---|
| 459 |
|
|---|
| 460 | # index options
|
|---|
| 461 | if ($buildtype eq 'mg' || $buildtype eq 'mgpp') {
|
|---|
| 462 | $buildwriter->startTag('indexOptionList');
|
|---|
| 463 | my $stemindexes = 3; # default is stem and casefold
|
|---|
| 464 | if (defined $buildcfg->{'stemindexes'} && $buildcfg->{'stemindexes'} =~ /^\d+$/ ) {
|
|---|
| 465 | $stemindexes = $buildcfg->{'stemindexes'};
|
|---|
| 466 | }
|
|---|
| 467 | $buildwriter->emptyTag('indexOption', 'name'=>'stemIndexes', 'value'=>$stemindexes);
|
|---|
| 468 |
|
|---|
| 469 | my $maxnumeric = 4; # default
|
|---|
| 470 | if (defined $buildcfg->{'maxnumeric'} && $buildcfg->{'maxnumeric'} =~ /^\d+$/) {
|
|---|
| 471 | $maxnumeric = $buildcfg->{'maxnumeric'};
|
|---|
| 472 | }
|
|---|
| 473 | $buildwriter->emptyTag('indexOption', 'name'=>'maxnumeric', 'value'=>$maxnumeric);
|
|---|
| 474 |
|
|---|
| 475 | $buildwriter->endTag('indexOptionList');
|
|---|
| 476 | }
|
|---|
| 477 | $buildwriter->emptyTag('defaultIndex', 'name'=>$defaultindex);
|
|---|
| 478 | $buildwriter->startTag('indexList');
|
|---|
| 479 | #for each index
|
|---|
| 480 | foreach my $i (keys %$indexmap) {
|
|---|
| 481 | my $index = $indexmap->{$i};
|
|---|
| 482 | $buildwriter->emptyTag('index', 'name'=>$index);
|
|---|
| 483 | }
|
|---|
| 484 | $buildwriter->endTag('indexList');
|
|---|
| 485 |
|
|---|
| 486 | $buildwriter->endTag('serviceRack');
|
|---|
| 487 |
|
|---|
| 488 | $buildwriter->endTag('serviceRackList');
|
|---|
| 489 | $buildwriter->endTag('buildConfig');
|
|---|
| 490 |
|
|---|
| 491 | # we add in the default replace list just in case we have macros in the
|
|---|
| 492 | # collection
|
|---|
| 493 | $collwriter->emptyTag('replaceListRef', 'id'=>'gs2-standard');
|
|---|
| 494 | $collwriter->endTag('collectionConfig');
|
|---|
| 495 | $collwriter->end();
|
|---|
| 496 | $buildwriter->end();
|
|---|
| 497 | $buildoutput->close();
|
|---|
| 498 | $colloutput->close();
|
|---|
| 499 | &close_database($database);
|
|---|
| 500 | }
|
|---|
| 501 |
|
|---|
| 502 |
|
|---|
| 503 | sub output_metadata {
|
|---|
| 504 | my ($writer, $metaname, $metavalue) = @_;
|
|---|
| 505 | $writer->startTag('metadata', 'name'=>$metaname);
|
|---|
| 506 | $writer->characters($metavalue);
|
|---|
| 507 | $writer->endTag('metadata');
|
|---|
| 508 | }
|
|---|
| 509 |
|
|---|
| 510 | sub output_display {
|
|---|
| 511 | my ($writer, $name, $lang, $value) = @_;
|
|---|
| 512 | $lang = 'en' if $lang eq 'default';
|
|---|
| 513 | if ($lang =~ /^\[/) {
|
|---|
| 514 | ($lang) = $lang =~ /\[l=(.*)\]/;
|
|---|
| 515 | }
|
|---|
| 516 | $writer->startTag('displayItem', 'name'=>$name, 'lang'=>$lang);
|
|---|
| 517 | $writer->characters($value);
|
|---|
| 518 | $writer->endTag('displayItem');
|
|---|
| 519 | }
|
|---|
| 520 | sub format_icon_value {
|
|---|
| 521 | my ($value) = @_;
|
|---|
| 522 | if ($value =~ /^_/) {
|
|---|
| 523 | my ($newvalue) = $value =~ /images\/(.*)$/;
|
|---|
| 524 | if ($newvalue) {
|
|---|
| 525 | return $newvalue;
|
|---|
| 526 | }
|
|---|
| 527 | }
|
|---|
| 528 | return $value;
|
|---|
| 529 | }
|
|---|
| 530 |
|
|---|
| 531 | sub tidy_up_display_item {
|
|---|
| 532 | my ($value) = @_;
|
|---|
| 533 | # remove \n
|
|---|
| 534 | $value =~ s/\\n//g;
|
|---|
| 535 | # replace \' with '
|
|---|
| 536 | $value =~ s/\\\'/\'/g;
|
|---|
| 537 | # replace \" with "
|
|---|
| 538 | $value =~ s/\\\"/\"/g;
|
|---|
| 539 | # replace _httpprefix_ with _httpsite_
|
|---|
| 540 | $value =~ s/_httpprefix_/_httpsite_/;
|
|---|
| 541 | $value =~ s/_gwcgi_//;
|
|---|
| 542 | $value =~ s/[a-z][a-z]?=_cgiarg[a-z][a-z]?_&?//;
|
|---|
| 543 | return $value;
|
|---|
| 544 | }
|
|---|
| 545 |
|
|---|
| 546 | sub write_format {
|
|---|
| 547 | my ($writer, $old_format, $node_type) = @_;
|
|---|
| 548 | # replace \' with '
|
|---|
| 549 | $old_format =~ s/\\\'/\'/g;
|
|---|
| 550 | # replace \" with "
|
|---|
| 551 | $old_format =~ s/\\\"/\"/g;
|
|---|
| 552 | #convert [] to <gsf:...>
|
|---|
| 553 | # assume no nesting {If} or {Or} for now
|
|---|
| 554 | $old_format =~ s/\{If\}\{([^\}]*)\}/&format_if($1, $node_type)/eg;
|
|---|
| 555 | $old_format =~ s/\{Or\}\{([^\}]*)\}/&format_or($1)/eg;
|
|---|
| 556 | $old_format =~ s/\[Text\]/\<gsf:text\/\>/g;
|
|---|
| 557 | $old_format =~ s/\[num\]/\<gsf:num\/\>/g;
|
|---|
| 558 | $old_format =~ s/\[link\]/\<gsf:link type=\'$node_type\'\>/g;
|
|---|
| 559 | $old_format =~ s/\[\/link\]/\<\/gsf:link\>/g;
|
|---|
| 560 | $old_format =~ s/\[srclink\]/\<gsf:link type=\'source\'\>/g;
|
|---|
| 561 | $old_format =~ s/\[\/srclink\]/\<\/gsf:link\>/g;
|
|---|
| 562 | $old_format =~ s/\[icon\]/\<gsf:icon type=\'$node_type\'\/\>/g;
|
|---|
| 563 | $old_format =~ s/\[srcicon\]/\<gsf:icon type=\'source\'\/\>/g;
|
|---|
| 564 |
|
|---|
| 565 | # what to do with hightlight??
|
|---|
| 566 | $old_format =~ s/\[\/?highlight\]//g;
|
|---|
| 567 |
|
|---|
| 568 | #now do the rest of the [] which are assumed to be metadata
|
|---|
| 569 | $old_format =~ s/\[([^\]]*)\]/&format_metadata($1)/eg;
|
|---|
| 570 |
|
|---|
| 571 | # some html tidy
|
|---|
| 572 | #turn <br> into <br />
|
|---|
| 573 | $old_format =~ s/\<br\>/\<br \/\>/g;
|
|---|
| 574 | #turn <p> into <p />
|
|---|
| 575 | $old_format =~ s/\<p\>/\<p \/\>/g;
|
|---|
| 576 |
|
|---|
| 577 | #put quotes around any atts
|
|---|
| 578 | $old_format =~ s/=([a-z]+)([> ])/=\'$1\'$2/g;
|
|---|
| 579 |
|
|---|
| 580 | if ($node_type eq "document") {
|
|---|
| 581 | $writer->startTag('gsf:template', 'match'=>'documentNode');
|
|---|
| 582 | $writer->charactersXML($old_format);
|
|---|
| 583 | $writer->endTag('gsf:template');
|
|---|
| 584 | } elsif ($node_type eq "classifier") {
|
|---|
| 585 | $writer->startTag('gsf:template', 'match'=>'classifierNode');
|
|---|
| 586 | $writer->charactersXML($old_format);
|
|---|
| 587 | $writer->endTag('gsf:template');
|
|---|
| 588 | } elsif ($node_type eq "horizontal") {
|
|---|
| 589 | $writer->startTag('gsf:template', 'match'=>'classifierNode', 'mode'=>'horizontal');
|
|---|
| 590 | $writer->charactersXML($old_format);
|
|---|
| 591 | $writer->endTag('gsf:template');
|
|---|
| 592 |
|
|---|
| 593 | }
|
|---|
| 594 | }
|
|---|
| 595 |
|
|---|
| 596 | sub format_metadata {
|
|---|
| 597 | my ($metadata_string) = @_;
|
|---|
| 598 |
|
|---|
| 599 | #print STDERR "original meta = $metadata_string\n";
|
|---|
| 600 |
|
|---|
| 601 | # what shall we do with cgisafe??
|
|---|
| 602 | my $cgisafe = $metadata_string =~ s/^cgisafe://;
|
|---|
| 603 |
|
|---|
| 604 | my ($select) = $metadata_string =~ /^(parent|sibling)/;
|
|---|
| 605 | $metadata_string =~ s/^(parent|sibling)//;
|
|---|
| 606 | my ($scope, $delim);
|
|---|
| 607 |
|
|---|
| 608 | if ($select) {
|
|---|
| 609 | ($scope) = $metadata_string =~ /^\((Top|All)/;
|
|---|
| 610 | $metadata_string =~ s/^\((Top|All)\)?//;
|
|---|
| 611 | if ($scope) {
|
|---|
| 612 | ($delim) = $metadata_string =~ /^\'([^\']*)\'\)/;
|
|---|
| 613 | $metadata_string =~ s/^\'([^\']*)\'\)//;
|
|---|
| 614 | }
|
|---|
| 615 | }
|
|---|
| 616 | $metadata_string =~ s/^://;
|
|---|
| 617 | # remove ex.
|
|---|
| 618 | $metadata_string =~ s/^ex\.//;
|
|---|
| 619 |
|
|---|
| 620 | #print STDERR "select=$select, scope=$scope, delim=|$delim|, meta = $metadata_string\n";
|
|---|
| 621 |
|
|---|
| 622 | my $new_format = "<gsf:metadata name='$metadata_string' ";
|
|---|
| 623 | if (defined $select) {
|
|---|
| 624 | if ($select eq "sibling") {
|
|---|
| 625 | $new_format .= "multiple='true' ";
|
|---|
| 626 | if (defined $delim) {
|
|---|
| 627 | $new_format .= "separator='$delim' ";
|
|---|
| 628 | }
|
|---|
| 629 | } elsif ($select eq "parent"){
|
|---|
| 630 | if (defined $scope) {
|
|---|
| 631 | if ($scope eq "Top") {
|
|---|
| 632 | $new_format .= "select='root' ";
|
|---|
| 633 | } elsif ($scope eq "All") {
|
|---|
| 634 | $new_format .= "select='ancestors' ";
|
|---|
| 635 | if (defined $delim) {
|
|---|
| 636 | $new_format .= "separator='$delim' ";
|
|---|
| 637 | }
|
|---|
| 638 | }
|
|---|
| 639 | } else {
|
|---|
| 640 | $new_format .= "select='parent' ";
|
|---|
| 641 | }
|
|---|
| 642 | }
|
|---|
| 643 | }
|
|---|
| 644 | $new_format .= "/>";
|
|---|
| 645 | #print STDERR "$new_format\n";
|
|---|
| 646 | return $new_format;
|
|---|
| 647 |
|
|---|
| 648 | }
|
|---|
| 649 |
|
|---|
| 650 | sub format_if {
|
|---|
| 651 |
|
|---|
| 652 | my ($if_string, $node_type) = @_;
|
|---|
| 653 | #print STDERR "if string = $if_string\n";
|
|---|
| 654 |
|
|---|
| 655 | my @parts = split /,/, $if_string;
|
|---|
| 656 | my $test = $parts[0];
|
|---|
| 657 | my $true_option = $parts[1];
|
|---|
| 658 | my $false_option;
|
|---|
| 659 | if (scalar (@parts) == 3) {
|
|---|
| 660 | $false_option = $parts[2];
|
|---|
| 661 | }
|
|---|
| 662 | $test =~ s/^\s*//;
|
|---|
| 663 | $test =~ s/\s*$//;
|
|---|
| 664 | my ($test_meta, $test_type, $test_value);
|
|---|
| 665 | if ($test =~ /^(\[.+\])$/) {
|
|---|
| 666 | $test_meta = $1;
|
|---|
| 667 | $test_type = 'exists';
|
|---|
| 668 | } else {
|
|---|
| 669 | my ($lhs, $exp, $rhs) = $test =~ /^(.+)\s+(eq|ne|lt|gt|le|ge|sw|ew)\s+(.+)$/;
|
|---|
| 670 | #print STDERR "lhs, exp, rhs = $lhs, $exp, $rhs\n";
|
|---|
| 671 | if ($exp eq "eq") {
|
|---|
| 672 | $test_type = "equals";
|
|---|
| 673 | } elsif ($exp eq "sw") {
|
|---|
| 674 | $test_type = "startsWith";
|
|---|
| 675 | } elsif ($exp eq "ew") {
|
|---|
| 676 | $test_type = "endsWith";
|
|---|
| 677 | } elsif ($exp eq "ne") {
|
|---|
| 678 | $test_type = "notEquals";
|
|---|
| 679 | } elsif ($exp eq "lt") {
|
|---|
| 680 | $test_type = "lessThan";
|
|---|
| 681 | }elsif ($exp eq "gt") {
|
|---|
| 682 | $test_type = "greaterThan";
|
|---|
| 683 | }elsif ($exp eq "le") {
|
|---|
| 684 | $test_type = "lessThanOrEquals";
|
|---|
| 685 | }elsif ($exp eq "ge") {
|
|---|
| 686 | $test_type = "greaterThanOrEquals";
|
|---|
| 687 | }
|
|---|
| 688 | if ($lhs =~ /^\[.+\]$/) {
|
|---|
| 689 | $test_meta = $lhs;
|
|---|
| 690 | $test_value = $rhs;
|
|---|
| 691 | } else {
|
|---|
| 692 | # assume rhs has meta
|
|---|
| 693 | $test_meta = $rhs;
|
|---|
| 694 | $test_value = $lhs;
|
|---|
| 695 | }
|
|---|
| 696 |
|
|---|
| 697 | #remove beginning and end quotes
|
|---|
| 698 | $test_value =~ s/^[\'\"]//;
|
|---|
| 699 | $test_value =~ s/[\'\"]$//;
|
|---|
| 700 | }
|
|---|
| 701 | my $test_atts = "test='$test_type' ";
|
|---|
| 702 | if (defined $test_value) {
|
|---|
| 703 | $test_atts .= "test-value='$test_value' ";
|
|---|
| 704 | }
|
|---|
| 705 | #print STDERR "test, true, false = $test, $true_option, $false_option\n";
|
|---|
| 706 | my $new_format = "<gsf:switch>$test_meta";
|
|---|
| 707 | $new_format .= "<gsf:when $test_atts>$true_option</gsf:when>";
|
|---|
| 708 | if (defined $false_option) {
|
|---|
| 709 | $new_format .="<gsf:otherwise>$false_option</gsf:otherwise>";
|
|---|
| 710 | }
|
|---|
| 711 | $new_format .= "</gsf:switch>";
|
|---|
| 712 |
|
|---|
| 713 | #print STDERR "new format = $new_format\n";
|
|---|
| 714 | return $new_format;
|
|---|
| 715 | }
|
|---|
| 716 |
|
|---|
| 717 | sub format_or {
|
|---|
| 718 | my ($or_string) = @_;
|
|---|
| 719 | my @meta_list = split (',', $or_string);
|
|---|
| 720 | return "" unless scalar (@meta_list);
|
|---|
| 721 | my $new_format = "<gsf:choose-metadata>";
|
|---|
| 722 | foreach my $m (@meta_list) {
|
|---|
| 723 | if ($m =~ /^\[(.*)\]$/) {
|
|---|
| 724 | $new_format .= &format_metadata($1);
|
|---|
| 725 | } else {
|
|---|
| 726 | # a default value
|
|---|
| 727 | $new_format .= "<gsf:default>$m</gsf:default>";
|
|---|
| 728 | last;
|
|---|
| 729 | }
|
|---|
| 730 | }
|
|---|
| 731 | $new_format .= "</gsf:choose-metadata>";
|
|---|
| 732 | return $new_format;
|
|---|
| 733 | }
|
|---|
| 734 |
|
|---|
| 735 | sub open_database {
|
|---|
| 736 | my ($db_file) = @_;
|
|---|
| 737 |
|
|---|
| 738 | my $database = ();
|
|---|
| 739 | # tie (%$database, 'GDBM_File', $db_file, GDBM_READER, 0400) ||
|
|---|
| 740 | # die "Couldn't open database $db_file\n";
|
|---|
| 741 |
|
|---|
| 742 | return $database;
|
|---|
| 743 | }
|
|---|
| 744 |
|
|---|
| 745 | sub close_database {
|
|---|
| 746 | my ($database) = @_;
|
|---|
| 747 | untie %$database;
|
|---|
| 748 | }
|
|---|
| 749 | sub isHorizontalClassifier {
|
|---|
| 750 | my ($database, $name) = @_;
|
|---|
| 751 |
|
|---|
| 752 | return 0; # can't get this to work for windows
|
|---|
| 753 | my $record = $database->{$name};
|
|---|
| 754 | my ($childtype) = $record =~ /<childtype>(\w*)/;
|
|---|
| 755 | if ($childtype eq "HList") {
|
|---|
| 756 | return 1;
|
|---|
| 757 | }
|
|---|
| 758 | return 0;
|
|---|
| 759 | }
|
|---|
| 760 | #$writer->startTag('');
|
|---|
| 761 | #$writer->endTag('');
|
|---|
| 762 | #$writer->characters();
|
|---|
| 763 | #$writer->emptyTag('');
|
|---|
| 764 |
|
|---|
| 765 | 1;
|
|---|
