| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # buildConfigxml.pm --
|
|---|
| 4 | # A component of the Greenstone digital library software
|
|---|
| 5 | # from the New Zealand Digital Library Project at the
|
|---|
| 6 | # University of Waikato, New Zealand.
|
|---|
| 7 | #
|
|---|
| 8 | # Copyright (C) 1999 New Zealand Digital Library Project
|
|---|
| 9 | #
|
|---|
| 10 | # This program is free software; you can redistribute it and/or modify
|
|---|
| 11 | # it under the terms of the GNU General Public License as published by
|
|---|
| 12 | # the Free Software Foundation; either version 2 of the License, or
|
|---|
| 13 | # (at your option) any later version.
|
|---|
| 14 | #
|
|---|
| 15 | # This program is distributed in the hope that it will be useful,
|
|---|
| 16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 18 | # GNU General Public License for more details.
|
|---|
| 19 | #
|
|---|
| 20 | # You should have received a copy of the GNU General Public License
|
|---|
| 21 | # along with this program; if not, write to the Free Software
|
|---|
| 22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|---|
| 23 | #
|
|---|
| 24 | ###########################################################################
|
|---|
| 25 |
|
|---|
| 26 | # reads in buildConfig.xml
|
|---|
| 27 | # Note, only implemented the bits that are currently used, eg by incremental
|
|---|
| 28 | # build code.
|
|---|
| 29 | # The resulting data is not a full representation on buildConfig.xml.
|
|---|
| 30 |
|
|---|
| 31 | package buildConfigxml;
|
|---|
| 32 |
|
|---|
| 33 | use strict;
|
|---|
| 34 | no strict 'refs';
|
|---|
| 35 | no strict 'subs';
|
|---|
| 36 |
|
|---|
| 37 | use XMLParser;
|
|---|
| 38 |
|
|---|
| 39 |
|
|---|
| 40 | # A mapping hash to resolve name discrepancy between gs2 and gs3.
|
|---|
| 41 | my $nameMap = {"numDocs" => "numdocs",
|
|---|
| 42 | "buildType" => "buildtype",
|
|---|
| 43 | "orthogonalBuildTypes" => "orthogonalbuildtypes"
|
|---|
| 44 | };
|
|---|
| 45 |
|
|---|
| 46 |
|
|---|
| 47 | # A hash structure which is returned by sub read_cfg_file.
|
|---|
| 48 | my $data = {};
|
|---|
| 49 |
|
|---|
| 50 | # use those unique attribute values to locate the text within the elements
|
|---|
| 51 | my $currentLocation = "";
|
|---|
| 52 | my $stringexp = q/^(buildType|numDocs)$/;
|
|---|
| 53 | my $arrayexp = q/^(orthogonalBuildTypes)$/;
|
|---|
| 54 |
|
|---|
| 55 | my $indexmap_name = "";
|
|---|
| 56 | my $haveindexfields = 0;
|
|---|
| 57 |
|
|---|
| 58 | # Reads in the model collection configuration file, collectionConfig.xml,
|
|---|
| 59 | # into a structure which complies with the one used by gs2 (i.e. one read
|
|---|
| 60 | # in by &cfgread::read_cfg_file).
|
|---|
| 61 | sub read_cfg_file {
|
|---|
| 62 | my ($filename) = @_;
|
|---|
| 63 | $data = {};
|
|---|
| 64 | if ($filename !~ /buildConfig\.xml$/ || !-f $filename) {
|
|---|
| 65 | return undef;
|
|---|
| 66 | }
|
|---|
| 67 |
|
|---|
| 68 | # Removed ProtocolEncoding (see MetadataXMLPlugin for details)
|
|---|
| 69 |
|
|---|
| 70 | # create XML::Parser object for parsing metadata.xml files
|
|---|
| 71 | my $parser = new XML::Parser('Style' => 'Stream',
|
|---|
| 72 | 'Pkg' => 'buildConfigxml',
|
|---|
| 73 | 'Handlers' => {'Char' => \&Char,
|
|---|
| 74 | 'Doctype' => \&Doctype
|
|---|
| 75 | });
|
|---|
| 76 |
|
|---|
| 77 | if (!open (COLCFG, $filename)) {
|
|---|
| 78 | print STDERR "buildConfigxml::read_cfg_file couldn't read the cfg file $filename\n";
|
|---|
| 79 | } else {
|
|---|
| 80 |
|
|---|
| 81 | $parser->parsefile ($filename);# (COLCFG);
|
|---|
| 82 | close (COLCFG);
|
|---|
| 83 | }
|
|---|
| 84 |
|
|---|
| 85 | #&Display;
|
|---|
| 86 | return $data;
|
|---|
| 87 | }
|
|---|
| 88 |
|
|---|
| 89 | sub StartTag {
|
|---|
| 90 | # Those marked with #@ will not be executed at the same time when this sub is being called
|
|---|
| 91 | # so that if/elsif is used to avoid unnecessary tests
|
|---|
| 92 | my ($expat, $element) = @_;
|
|---|
| 93 |
|
|---|
| 94 | my $name = $_{'name'};
|
|---|
| 95 | my $shortname = $_{'shortname'};
|
|---|
| 96 |
|
|---|
| 97 |
|
|---|
| 98 | #@ handling block metadataList
|
|---|
| 99 | if (defined $name && (($name =~ m/$stringexp/) || ($name =~ m/$arrayexp/))) {
|
|---|
| 100 | $currentLocation = $name;
|
|---|
| 101 | # the value will be retrieved later in Text sub
|
|---|
| 102 | }
|
|---|
| 103 |
|
|---|
| 104 | #@ handle indexes - store indexmap (mg) or indexfields and indexfieldmap (mgpp/lucene/solr)
|
|---|
| 105 | elsif ($element =~ /^indexList$/) {
|
|---|
| 106 | # set up the data arrays
|
|---|
| 107 | # this assumes that the build type has been read already, which is
|
|---|
| 108 | # currently the order we save the file in.
|
|---|
| 109 | if ($data->{'buildtype'} eq "mg") {
|
|---|
| 110 | $indexmap_name = "indexmap";
|
|---|
| 111 | if (!defined $data->{"indexmap"}) {
|
|---|
| 112 | $data->{"indexmap"} = [];
|
|---|
| 113 | }
|
|---|
| 114 | }
|
|---|
| 115 | else {
|
|---|
| 116 | # mgpp, lucene or solr
|
|---|
| 117 | $indexmap_name = "indexfieldmap";
|
|---|
| 118 | $haveindexfields = 1;
|
|---|
| 119 | if (!defined $data->{"indexfieldmap"}) {
|
|---|
| 120 | $data->{"indexfieldmap"} = [];
|
|---|
| 121 | }
|
|---|
| 122 | if (!defined $data->{"indexfields"}) {
|
|---|
| 123 | $data->{"indexfields"} = [];
|
|---|
| 124 | }
|
|---|
| 125 |
|
|---|
| 126 | }
|
|---|
| 127 |
|
|---|
| 128 | }
|
|---|
| 129 |
|
|---|
| 130 | elsif ($element =~ /index/) {
|
|---|
| 131 | # store each index in the map
|
|---|
| 132 | if (defined $name && defined $shortname) {
|
|---|
| 133 | push @{$data->{$indexmap_name}}, "$name->$shortname";
|
|---|
| 134 | if ($haveindexfields) {
|
|---|
| 135 | push @{$data->{'indexfields'}}, $name;
|
|---|
| 136 | }
|
|---|
| 137 | }
|
|---|
| 138 | }
|
|---|
| 139 |
|
|---|
| 140 |
|
|---|
| 141 | }
|
|---|
| 142 |
|
|---|
| 143 | sub EndTag {
|
|---|
| 144 | my ($expat, $element) = @_;
|
|---|
| 145 | }
|
|---|
| 146 |
|
|---|
| 147 | sub Text {
|
|---|
| 148 | if (defined $currentLocation) {
|
|---|
| 149 | #@ Handling block metadataList(numDocs, buildType)
|
|---|
| 150 | if ($currentLocation =~ /$stringexp/) {
|
|---|
| 151 | #print $currentLocation;
|
|---|
| 152 | my $key = $nameMap->{$currentLocation};
|
|---|
| 153 | $data->{$key} = $_;
|
|---|
| 154 | undef $currentLocation;
|
|---|
| 155 | }
|
|---|
| 156 | elsif ($currentLocation =~ /$arrayexp/) {
|
|---|
| 157 | #print $currentLocation;
|
|---|
| 158 | my $key = $nameMap->{$currentLocation};
|
|---|
| 159 | push(@{$data->{$key}},$_);
|
|---|
| 160 | undef $currentLocation;
|
|---|
| 161 | }
|
|---|
| 162 |
|
|---|
| 163 | }
|
|---|
| 164 | }
|
|---|
| 165 |
|
|---|
| 166 | # This sub is for debugging purposes
|
|---|
| 167 | sub Display {
|
|---|
| 168 |
|
|---|
| 169 | print "NumDocs = ".$data->{'numdocs'}."\n" if (defined $data->{'numdocs'});
|
|---|
| 170 | print "BuildType = ".$data->{'buildtype'}."\n" if (defined $data->{'buildtype'});
|
|---|
| 171 | print "OrthogonalBuildTypes = ".join(",",@{$data->{'orthogonalbuildtypes'}})."\n" if (defined $data->{'orthogonalbuildtypes'});
|
|---|
| 172 | print "IndexMap = ". join(" ",@{$data->{'indexmap'}})."\n" if (defined $data->{'indexmap'});
|
|---|
| 173 | print "IndexFieldMap = ". join(" ",@{$data->{'indexfieldmap'}})."\n" if (defined $data->{'indexfieldmap'});
|
|---|
| 174 | print "IndexFields = ". join(" ",@{$data->{'indexfields'}})."\n" if (defined $data->{'indexfields'});
|
|---|
| 175 |
|
|---|
| 176 | }
|
|---|
| 177 |
|
|---|
| 178 | # is this actually used??
|
|---|
| 179 | sub Doctype {
|
|---|
| 180 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 181 |
|
|---|
| 182 | die if ($name !~ /^buildConfig$/);
|
|---|
| 183 | }
|
|---|
| 184 |
|
|---|
| 185 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
|---|
| 186 | # problem where $expat->{Text} is treated as the return value, slowing
|
|---|
| 187 | # things down significantly in some cases.
|
|---|
| 188 | sub Char {
|
|---|
| 189 | if ($]<5.008) {
|
|---|
| 190 | use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+ and Perl 5.6
|
|---|
| 191 | }
|
|---|
| 192 | $_[0]->{'Text'} .= $_[1];
|
|---|
| 193 | return undef;
|
|---|
| 194 | }
|
|---|
| 195 |
|
|---|
| 196 |
|
|---|
| 197 |
|
|---|
| 198 | sub write_line {
|
|---|
| 199 | my ($filehandle, $line) = @_;
|
|---|
| 200 | print $filehandle join ("", @$line), "\n";
|
|---|
| 201 | }
|
|---|
| 202 |
|
|---|
| 203 | sub search_and_retrieve_settings
|
|---|
| 204 | {
|
|---|
| 205 | my ($buildcfg,$collectcfg) = @_;
|
|---|
| 206 |
|
|---|
| 207 | my $settings = {};
|
|---|
| 208 |
|
|---|
| 209 | if (defined $buildcfg->{"buildtype"}) {
|
|---|
| 210 | $settings->{'buildtype'} = $buildcfg->{"buildtype"};
|
|---|
| 211 | } else {
|
|---|
| 212 | $settings->{'buildtype'} = "mgpp";
|
|---|
| 213 | }
|
|---|
| 214 | my $buildtype = $settings->{'buildtype'};
|
|---|
| 215 |
|
|---|
| 216 | if (defined $collectcfg->{"orthogonalbuildtypes"}) {
|
|---|
| 217 | # Note the use of collectcfg, not bulidcfg
|
|---|
| 218 | $settings->{'orthogonalbuildtypes'} = $collectcfg->{"orthogonalbuildtypes"};
|
|---|
| 219 | } else {
|
|---|
| 220 | $settings->{'orthogonalbuildtypes '}= [];
|
|---|
| 221 | }
|
|---|
| 222 |
|
|---|
| 223 | if (defined $buildcfg->{"numdocs"}) {
|
|---|
| 224 | $settings->{'numdocs'} = $buildcfg->{"numdocs"};
|
|---|
| 225 | }
|
|---|
| 226 | else {
|
|---|
| 227 | $settings->{'numdocs'} = 0;
|
|---|
| 228 | }
|
|---|
| 229 |
|
|---|
| 230 | my $service_type = "MGPP";
|
|---|
| 231 | if ($buildtype eq "mg") {
|
|---|
| 232 | $service_type = "MG";
|
|---|
| 233 | } elsif ($buildtype eq "lucene") {
|
|---|
| 234 | $service_type = "Lucene";
|
|---|
| 235 | } elsif ($buildtype eq "solr") {
|
|---|
| 236 | $service_type = "Solr";
|
|---|
| 237 | }
|
|---|
| 238 | $settings->{'service_type'} = $service_type;
|
|---|
| 239 |
|
|---|
| 240 |
|
|---|
| 241 | if (defined $buildcfg->{"infodbtype"}) {
|
|---|
| 242 | $settings->{'infodbtype'} = $buildcfg->{'infodbtype'};
|
|---|
| 243 | }
|
|---|
| 244 | else {
|
|---|
| 245 | $settings->{'infodbtype'} = "gdbm";
|
|---|
| 246 | }
|
|---|
| 247 |
|
|---|
| 248 |
|
|---|
| 249 | #--
|
|---|
| 250 | # indexes
|
|---|
| 251 | #--
|
|---|
| 252 |
|
|---|
| 253 | my $indexmap = {}; # maps index name to shortname
|
|---|
| 254 | my $indexlist = []; # keeps the order for indexes
|
|---|
| 255 | my $defaultindex = "";
|
|---|
| 256 | my $maptype = ($buildtype eq "mg") ? "indexmap" : "indexfieldmap";
|
|---|
| 257 |
|
|---|
| 258 | if (defined $buildcfg->{$maptype}) {
|
|---|
| 259 | my $first = 1;
|
|---|
| 260 | my $indexmap_t = $buildcfg->{$maptype};
|
|---|
| 261 | foreach my $i (@$indexmap_t) {
|
|---|
| 262 | my ($k, $v) = $i =~ /^(.*)\-\>(.*)$/;
|
|---|
| 263 | $indexmap->{$k} = $v;
|
|---|
| 264 | push @$indexlist, $k;
|
|---|
| 265 | if ($first) {
|
|---|
| 266 | $defaultindex = $v;
|
|---|
| 267 | $first = 0;
|
|---|
| 268 | }
|
|---|
| 269 | }
|
|---|
| 270 | # now if the user has assigned a default index, we use it
|
|---|
| 271 | if (defined $collectcfg->{"defaultindex"}) {
|
|---|
| 272 | $defaultindex = $indexmap->{$collectcfg->{"defaultindex"}};
|
|---|
| 273 | }
|
|---|
| 274 | } else {
|
|---|
| 275 | print STDERR "$maptype not defined";
|
|---|
| 276 | }
|
|---|
| 277 |
|
|---|
| 278 | $settings->{'defaultindex'} = $defaultindex;
|
|---|
| 279 | $settings->{'indexmap'} = $indexmap;
|
|---|
| 280 | $settings->{'indexlist'} = $indexlist;
|
|---|
| 281 |
|
|---|
| 282 | #--
|
|---|
| 283 | # default lang
|
|---|
| 284 | #--
|
|---|
| 285 | $settings->{'default_lang'} = "";
|
|---|
| 286 | $settings->{'default_lang_short'} = "";
|
|---|
| 287 |
|
|---|
| 288 | if (defined $buildcfg->{"languagemap"}) {
|
|---|
| 289 | my $langmap_t = $buildcfg->{"languagemap"};
|
|---|
| 290 | if ((defined $langmap_t) && (scalar(@$langmap_t)>=1)) {
|
|---|
| 291 | my $l = $langmap_t->[0];
|
|---|
| 292 | my ($k, $v) = $l =~ m/^(.*)\-\>(.*)$/;
|
|---|
| 293 | $settings->{'default_lang'} = $k; #name
|
|---|
| 294 | $settings->{'default_lang_short'} = $v; #short name
|
|---|
| 295 | }
|
|---|
| 296 |
|
|---|
| 297 | # now if the user has assigned a default language (as "en", "ru" etc.)
|
|---|
| 298 | if (defined $collectcfg->{"defaultlanguage"}) {
|
|---|
| 299 | $settings->{'default_lang'} = $collectcfg->{"defaultlanguage"};
|
|---|
| 300 | # what about default_lang_short ?? ####
|
|---|
| 301 | }
|
|---|
| 302 | }
|
|---|
| 303 |
|
|---|
| 304 | # default subcol
|
|---|
| 305 | $settings->{'default_subcol'} = "";
|
|---|
| 306 | if (defined $buildcfg->{'subcollectionmap'}) {
|
|---|
| 307 | my $subcolmap_t = $buildcfg->{'subcollectionmap'};
|
|---|
| 308 | if ((defined $subcolmap_t) && (scalar(@$subcolmap_t)>=1)) {
|
|---|
| 309 | my $l = $subcolmap_t->[0];
|
|---|
| 310 | my ($k, $v) = $l =~ m/^(.*)\-\>(.*)$/;
|
|---|
| 311 |
|
|---|
| 312 | $settings->{'default_subcol'} = $v;
|
|---|
| 313 | }
|
|---|
| 314 | }
|
|---|
| 315 |
|
|---|
| 316 |
|
|---|
| 317 | #--
|
|---|
| 318 | # indexstem
|
|---|
| 319 | #--
|
|---|
| 320 | if (defined $buildcfg->{'indexstem'}) {
|
|---|
| 321 | $settings->{'indexstem'} = $buildcfg->{'indexstem'};
|
|---|
| 322 | }
|
|---|
| 323 |
|
|---|
| 324 | #--
|
|---|
| 325 | # levelList
|
|---|
| 326 | #--
|
|---|
| 327 |
|
|---|
| 328 | my $levelmap = {};
|
|---|
| 329 | my $levellist = [];
|
|---|
| 330 | my $default_search_level = "Doc";
|
|---|
| 331 | my $default_retrieve_level = "Doc";
|
|---|
| 332 | my $default_db_level = "Doc";
|
|---|
| 333 |
|
|---|
| 334 | if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
|
|---|
| 335 | if (defined $buildcfg->{'levelmap'}) {
|
|---|
| 336 | my $first = 1;
|
|---|
| 337 |
|
|---|
| 338 | my $levelmap_t = $buildcfg->{'levelmap'};
|
|---|
| 339 | foreach my $l (@$levelmap_t) {
|
|---|
| 340 | my ($key, $val) = $l =~ /^(.*)\-\>(.*)$/;
|
|---|
| 341 | $levelmap->{$key} = $val;
|
|---|
| 342 | push @$levellist, $key;
|
|---|
| 343 | if ($first) {
|
|---|
| 344 | # let default search level follow the first level in the level list
|
|---|
| 345 | $default_search_level = $val;
|
|---|
| 346 | # retrieve/database levels may get modified later if text level is defined
|
|---|
| 347 | $default_retrieve_level = $val;
|
|---|
| 348 | $default_db_level = $val;
|
|---|
| 349 | $first = 0;
|
|---|
| 350 | }
|
|---|
| 351 | }
|
|---|
| 352 | }
|
|---|
| 353 | # the default level assigned by the user is no longer ignored [Shaoqun], but the retrievel level stays the same.
|
|---|
| 354 | #if (defined $collectcfg->{"defaultlevel"}) {
|
|---|
| 355 | $default_search_level = $levelmap->{$collectcfg->{"defaultlevel"}};
|
|---|
| 356 | # $default_retrieve_level = $default_search_level;
|
|---|
| 357 | #}
|
|---|
| 358 |
|
|---|
| 359 | if (defined $buildcfg->{'textlevel'}) {
|
|---|
| 360 | # let the retrieve/database levels always follow the textlevel
|
|---|
| 361 | $default_retrieve_level = $buildcfg->{'textlevel'};
|
|---|
| 362 | $default_db_level = $buildcfg->{'textlevel'};
|
|---|
| 363 |
|
|---|
| 364 | }
|
|---|
| 365 | }
|
|---|
| 366 | $settings->{'levelmap'} = $levelmap;
|
|---|
| 367 | $settings->{'levellist'} = $levellist;
|
|---|
| 368 | $settings->{'default_search_level'} = $default_search_level;
|
|---|
| 369 | $settings->{'default_retrieve_level'} = $default_retrieve_level;
|
|---|
| 370 | $settings->{'default_db_level'} = $default_db_level;
|
|---|
| 371 |
|
|---|
| 372 |
|
|---|
| 373 | return $settings;
|
|---|
| 374 | }
|
|---|
| 375 |
|
|---|
| 376 |
|
|---|
| 377 | sub write_search_servicerack
|
|---|
| 378 | {
|
|---|
| 379 | my ($buildcfg,$settings) = @_;
|
|---|
| 380 |
|
|---|
| 381 | my $buildtype = $settings->{'buildtype'};
|
|---|
| 382 | my $infodbtype = $settings->{'infodbtype'};
|
|---|
| 383 | my $service_type = $settings->{'service_type'};
|
|---|
| 384 |
|
|---|
| 385 | # do the search service
|
|---|
| 386 | &write_line('COLCFG', ["<serviceRack name=\"GS2", $service_type, "Search\">"]);
|
|---|
| 387 | if (defined $buildcfg->{'indexstem'}) {
|
|---|
| 388 | my $indexstem = $buildcfg->{'indexstem'};
|
|---|
| 389 | &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
|
|---|
| 390 | }
|
|---|
| 391 | if (defined $buildcfg->{'infodbtype'}) {
|
|---|
| 392 | &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
|
|---|
| 393 | }
|
|---|
| 394 |
|
|---|
| 395 | #indexes
|
|---|
| 396 | my $indexmap = $settings->{'indexmap'};
|
|---|
| 397 | my $indexlist = $settings->{'indexlist'};
|
|---|
| 398 | my $defaultindex = $settings->{'defaultindex'};
|
|---|
| 399 |
|
|---|
| 400 | #for each index in indexList, write them out
|
|---|
| 401 | &write_line('COLCFG', ["<indexList>"]);
|
|---|
| 402 | foreach my $i (@$indexlist) {
|
|---|
| 403 | my $index = $indexmap->{$i};
|
|---|
| 404 | &write_line('COLCFG', ["<index name=\"", $i, "\" ", "shortname=\"", $index, "\" />"]);
|
|---|
| 405 | }
|
|---|
| 406 | &write_line('COLCFG', ["</indexList>"]);
|
|---|
| 407 |
|
|---|
| 408 |
|
|---|
| 409 | #$defaultindex = "ZZ" if (!$defaultindex); # index allfields by default
|
|---|
| 410 | if ($defaultindex) {
|
|---|
| 411 | &write_line('COLCFG', ["<defaultIndex shortname=\"", $defaultindex, "\" />"]);
|
|---|
| 412 | }
|
|---|
| 413 |
|
|---|
| 414 |
|
|---|
| 415 | # do indexOptionList
|
|---|
| 416 | if ($buildtype eq "mg" || $buildtype eq "mgpp") {
|
|---|
| 417 | &write_line('COLCFG', ["<indexOptionList>"]);
|
|---|
| 418 | my $stemindexes = 3; # default is stem and casefold
|
|---|
| 419 | if (defined $buildcfg->{'stemindexes'} && $buildcfg->{'stemindexes'} =~ /^\d+$/ ) {
|
|---|
| 420 | $stemindexes = $buildcfg->{'stemindexes'};
|
|---|
| 421 | }
|
|---|
| 422 | &write_line('COLCFG', ["<indexOption name=\"stemIndexes\" value=\"", $stemindexes, "\" />"]);
|
|---|
| 423 |
|
|---|
| 424 | my $maxnumeric = 4; # default
|
|---|
| 425 | if (defined $buildcfg->{'maxnumeric'} && $buildcfg->{'maxnumeric'} =~ /^\d+$/) {
|
|---|
| 426 | $maxnumeric = $buildcfg->{'maxnumeric'};
|
|---|
| 427 | }
|
|---|
| 428 | &write_line('COLCFG', ["<indexOption name=\"maxnumeric\" value=\"", $maxnumeric, "\" />"]);
|
|---|
| 429 | &write_line('COLCFG', ["</indexOptionList>"]);
|
|---|
| 430 | }
|
|---|
| 431 |
|
|---|
| 432 | #--
|
|---|
| 433 | # levelList
|
|---|
| 434 | #--
|
|---|
| 435 | my $levelmap = $settings->{'levelmap'};
|
|---|
| 436 | my $levellist = $settings->{'levellist'};
|
|---|
| 437 | my $default_search_level = $settings->{'default_search_level'};
|
|---|
| 438 | my $default_retrieve_level = $settings->{'default_retrieve_level'};
|
|---|
| 439 | my $default_db_level = $settings->{'default_db_level'};
|
|---|
| 440 |
|
|---|
| 441 | #for each level in levelList, write them out
|
|---|
| 442 | if ($buildtype ne "mg") {
|
|---|
| 443 | &write_line('COLCFG', ["<levelList>"]);
|
|---|
| 444 | foreach my $lv (@$levellist) {
|
|---|
| 445 | my $level = $levelmap->{$lv};
|
|---|
| 446 | &write_line('COLCFG', ["<level name=\"", $lv, "\" shortname=\"", $level, "\" />"]);
|
|---|
| 447 | }
|
|---|
| 448 | &write_line('COLCFG', ["</levelList>"]);
|
|---|
| 449 | }
|
|---|
| 450 | # add in defaultLevel as the same level as indexLevelList, making the reading job easier
|
|---|
| 451 | if ($buildtype eq "lucene" || $buildtype eq "mgpp" || $buildtype eq "solr") {
|
|---|
| 452 | &write_line('COLCFG', ["<defaultLevel shortname=\"", $default_search_level, "\" />"]);
|
|---|
| 453 | }
|
|---|
| 454 | if ($buildtype eq "lucene" || $buildtype eq "mgpp" || $buildtype eq "solr") {
|
|---|
| 455 | &write_line('COLCFG', ["<defaultDBLevel shortname=\"", $default_db_level, "\" />"]);
|
|---|
| 456 | }
|
|---|
| 457 | # do searchTypeList
|
|---|
| 458 | if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
|
|---|
| 459 | &write_line('COLCFG', ["<searchTypeList>"]);
|
|---|
| 460 |
|
|---|
| 461 | if (defined $buildcfg->{"searchtype"}) {
|
|---|
| 462 | my $searchtype_t = $buildcfg->{"searchtype"};
|
|---|
| 463 | foreach my $s (@$searchtype_t) {
|
|---|
| 464 | &write_line('COLCFG', ["<searchType name=\"", $s, "\" />"]);
|
|---|
| 465 | }
|
|---|
| 466 | } else {
|
|---|
| 467 | &write_line('COLCFG', ["<searchType name=\"plain\" />"]);
|
|---|
| 468 | &write_line('COLCFG', ["<searchType name=\"form\" />"]);
|
|---|
| 469 | }
|
|---|
| 470 | &write_line('COLCFG', ["</searchTypeList>"]);
|
|---|
| 471 | }
|
|---|
| 472 |
|
|---|
| 473 | # do indexLanguageList [in collect.cfg: languages; in build.cfg: languagemap]
|
|---|
| 474 | my $default_lang = $settings->{'default_lang'};
|
|---|
| 475 | my $default_lang_short = $settings->{'default_lang_short'};
|
|---|
| 476 | if (defined $buildcfg->{"languagemap"}) {
|
|---|
| 477 | &write_line('COLCFG', ["<indexLanguageList>"]);
|
|---|
| 478 |
|
|---|
| 479 | my $langmap_t = $buildcfg->{"languagemap"};
|
|---|
| 480 | foreach my $l (@$langmap_t) {
|
|---|
| 481 | my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
|
|---|
| 482 |
|
|---|
| 483 | &write_line('COLCFG', ["<indexLanguage name=\"", $k, "\" shortname=\"", $v, "\" />"]);
|
|---|
| 484 | }
|
|---|
| 485 |
|
|---|
| 486 | &write_line('COLCFG', ["</indexLanguageList>"]);
|
|---|
| 487 |
|
|---|
| 488 | &write_line('COLCFG', ["<defaultIndexLanguage name=\"", $default_lang, "\" shortname=\"", $default_lang_short, "\" />"]);
|
|---|
| 489 | }
|
|---|
| 490 |
|
|---|
| 491 | # do indexSubcollectionList
|
|---|
| 492 | my $default_subcol = $settings->{'default_subcol'};
|
|---|
| 493 |
|
|---|
| 494 | if (defined $buildcfg->{'subcollectionmap'}) {
|
|---|
| 495 | &write_line('COLCFG', ["<indexSubcollectionList>"]);
|
|---|
| 496 | my $subcolmap = {};
|
|---|
| 497 | my @subcollist = ();
|
|---|
| 498 |
|
|---|
| 499 | my $subcolmap_t = $buildcfg->{'subcollectionmap'};
|
|---|
| 500 | foreach my $l (@$subcolmap_t) {
|
|---|
| 501 | my ($k, $v) = $l =~ /^(.*)\-\>(.*)$/;
|
|---|
| 502 | $subcolmap->{$k} = $v;
|
|---|
| 503 | push @subcollist, $k;
|
|---|
| 504 | }
|
|---|
| 505 |
|
|---|
| 506 | foreach my $sl (@subcollist) {
|
|---|
| 507 | my $subcol = $subcolmap->{$sl};
|
|---|
| 508 | &write_line('COLCFG', ["<indexSubcollection name=\"", $sl, "\" shortname=\"", $subcol, "\" />"]);
|
|---|
| 509 | }
|
|---|
| 510 |
|
|---|
| 511 | &write_line('COLCFG', ["</indexSubcollectionList>"]);
|
|---|
| 512 | &write_line('COLCFG', ["<defaultIndexSubcollection shortname=\"", $default_subcol, "\" />"]);
|
|---|
| 513 | }
|
|---|
| 514 |
|
|---|
| 515 | # close off search service
|
|---|
| 516 | &write_line('COLCFG', ["</serviceRack>"]);
|
|---|
| 517 |
|
|---|
| 518 | }
|
|---|
| 519 |
|
|---|
| 520 |
|
|---|
| 521 | sub write_orthogonalsearch_serviceracks
|
|---|
| 522 | {
|
|---|
| 523 | my ($buildcfg,$settings) = @_;
|
|---|
| 524 |
|
|---|
| 525 | my $infodbtype = $settings->{'infodbtype'};
|
|---|
| 526 |
|
|---|
| 527 | my $orthogonalbuildtypes = $settings->{'orthogonalbuildtypes'};
|
|---|
| 528 |
|
|---|
| 529 | foreach my $obt (@$orthogonalbuildtypes) {
|
|---|
| 530 | $obt =~ s/^(.)/\u$1/; # capitialize initial letter
|
|---|
| 531 | $obt =~ s/-(.)/\u$1/g; # change any hyphenated words to cap next letter
|
|---|
| 532 |
|
|---|
| 533 | &write_line('COLCFG', ["<serviceRack name=\"GS2", $obt, "Search\">"]);
|
|---|
| 534 |
|
|---|
| 535 | &write_line('COLCFG',["<databaseType name=\"",$infodbtype,"\" />"]);
|
|---|
| 536 | &write_line('COLCFG', ["</serviceRack>"]);
|
|---|
| 537 | }
|
|---|
| 538 | }
|
|---|
| 539 |
|
|---|
| 540 |
|
|---|
| 541 |
|
|---|
| 542 | sub write_retrieve_servicerack
|
|---|
| 543 | {
|
|---|
| 544 | my ($buildcfg,$settings) = @_;
|
|---|
| 545 |
|
|---|
| 546 | my $buildtype = $settings->{'buildtype'};
|
|---|
| 547 | my $infodbtype = $settings->{'infodbtype'};
|
|---|
| 548 |
|
|---|
| 549 | my $service_type = $settings->{'service_type'};
|
|---|
| 550 |
|
|---|
| 551 | # do the retrieve service
|
|---|
| 552 | &write_line('COLCFG', ["<serviceRack name=\"GS2", $service_type, "Retrieve\">"]);
|
|---|
| 553 |
|
|---|
| 554 | # do default index
|
|---|
| 555 | if (defined $buildcfg->{"languagemap"}) {
|
|---|
| 556 | my $default_lang = $settings->{'default_lang'};
|
|---|
| 557 | &write_line('COLCFG', ["<defaultIndexLanguage shortname=\"", $default_lang, "\" />"]);
|
|---|
| 558 | }
|
|---|
| 559 | if (defined $buildcfg->{'subcollectionmap'}) {
|
|---|
| 560 | my $default_subcol = $settings->{'default_subcol'};
|
|---|
| 561 | &write_line('COLCFG', ["<defaultIndexSubcollection shortname=\"", $default_subcol, "\" />"]);
|
|---|
| 562 | }
|
|---|
| 563 | if ($buildtype eq "mg") {
|
|---|
| 564 | my $defaultindex = $settings->{'defaultindex'};
|
|---|
| 565 | &write_line('COLCFG', ["<defaultIndex shortname=\"", $defaultindex, "\" />"]);
|
|---|
| 566 | }
|
|---|
| 567 |
|
|---|
| 568 | if (defined $buildcfg->{'indexstem'}) {
|
|---|
| 569 | my $indexstem = $buildcfg->{'indexstem'};
|
|---|
| 570 | &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
|
|---|
| 571 | }
|
|---|
| 572 | if ($buildtype eq "mgpp" || $buildtype eq "lucene" || $buildtype eq "solr") {
|
|---|
| 573 | my $default_retrieve_level = $settings->{'default_retrieve_level'};
|
|---|
| 574 | &write_line('COLCFG', ["<defaultLevel shortname=\"", $default_retrieve_level, "\" />"]);
|
|---|
| 575 | }
|
|---|
| 576 | if (defined $buildcfg->{'infodbtype'}) {
|
|---|
| 577 | &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
|
|---|
| 578 | }
|
|---|
| 579 |
|
|---|
| 580 | &write_line('COLCFG', ["</serviceRack>"]);
|
|---|
| 581 |
|
|---|
| 582 | }
|
|---|
| 583 |
|
|---|
| 584 |
|
|---|
| 585 | # Create the buildConfig.xml file for a specific collection
|
|---|
| 586 | sub write_cfg_file {
|
|---|
| 587 | # this sub is called in make_auxiliary_files() in basebuilder.pm
|
|---|
| 588 | # the received args: $buildoutfile - destination file: buildConfig.xml
|
|---|
| 589 | # $buildcfg - all build options,
|
|---|
| 590 | # $collectcfg - contents of collectionConfig.xml read in by read_cfg_file sub in buildConfigxml.pm.
|
|---|
| 591 | my ($buildoutfile, $buildcfg, $collectcfg) = @_;
|
|---|
| 592 | my $line = [];
|
|---|
| 593 |
|
|---|
| 594 | if (!open (COLCFG, ">$buildoutfile")) {
|
|---|
| 595 | print STDERR "buildConfigxml::write_cfg_file couldn't write the build config file $buildoutfile\n";
|
|---|
| 596 | die;
|
|---|
| 597 | }
|
|---|
| 598 |
|
|---|
| 599 | my $settings = search_and_retrieve_settings($buildcfg,$collectcfg);
|
|---|
| 600 |
|
|---|
| 601 | my $buildtype = $settings->{'buildtype'};
|
|---|
| 602 | my $orthogonalbuildtypes = $settings->{'orthogonalbuildtypes'};
|
|---|
| 603 | my $numdocs = $settings->{'numdocs'};
|
|---|
| 604 |
|
|---|
| 605 | &write_line('COLCFG', ["<buildConfig xmlns:gsf=\"http://www.greenstone.org/greenstone3/schema/ConfigFormat\">"]);
|
|---|
| 606 |
|
|---|
| 607 | # output building metadata to build config file
|
|---|
| 608 | &write_line('COLCFG', ["<metadataList>"]);
|
|---|
| 609 | &write_line('COLCFG', ["<metadata name=\"numDocs\">", $numdocs, "</metadata>"]);
|
|---|
| 610 | &write_line('COLCFG', ["<metadata name=\"buildType\">", $buildtype, "</metadata>"]);
|
|---|
| 611 | foreach my $obt (@$orthogonalbuildtypes) {
|
|---|
| 612 | &write_line('COLCFG', ["<metadata name=\"orthogonalBuildTypes\">", $obt, "</metadata>"]);
|
|---|
| 613 | }
|
|---|
| 614 |
|
|---|
| 615 | if (defined $buildcfg->{'indexstem'}) {
|
|---|
| 616 | &write_line('COLCFG', ["<metadata name=\"indexStem\">", $buildcfg->{"indexstem"}, "</metadata>"]);
|
|---|
| 617 | }
|
|---|
| 618 | if (defined $buildcfg->{'infodbtype'}) {
|
|---|
| 619 | &write_line('COLCFG', ["<metadata name=\"infodbType\">", $buildcfg->{"infodbtype"}, "</metadata>"]);
|
|---|
| 620 | }
|
|---|
| 621 | if (defined $buildcfg->{'earliestdatestamp'}) {
|
|---|
| 622 | &write_line('COLCFG', ["<metadata name=\"earliestdatestamp\">", $buildcfg->{"earliestdatestamp"}, "</metadata>"]);
|
|---|
| 623 | }
|
|---|
| 624 |
|
|---|
| 625 | &write_line('COLCFG', ["</metadataList>"]);
|
|---|
| 626 |
|
|---|
| 627 | # output serviceRackList
|
|---|
| 628 | &write_line('COLCFG', ["<serviceRackList>"]);
|
|---|
| 629 |
|
|---|
| 630 | write_search_servicerack($buildcfg,$settings);
|
|---|
| 631 |
|
|---|
| 632 | # add in orthogonalbuildtypes
|
|---|
| 633 | write_orthogonalsearch_serviceracks($buildcfg,$settings);
|
|---|
| 634 |
|
|---|
| 635 | write_retrieve_servicerack($buildcfg,$settings);
|
|---|
| 636 |
|
|---|
| 637 | # do the browse service
|
|---|
| 638 | my $count = 1;
|
|---|
| 639 | my $phind = 0;
|
|---|
| 640 | my $started_classifiers = 0;
|
|---|
| 641 |
|
|---|
| 642 | my $classifiers = $collectcfg->{"classify"};
|
|---|
| 643 | foreach my $cl (@$classifiers) {
|
|---|
| 644 | my $name = "CL$count";
|
|---|
| 645 | $count++;
|
|---|
| 646 | my ($classname) = @$cl[0];
|
|---|
| 647 | if ($classname =~ /^phind$/i) {
|
|---|
| 648 | $phind=1;
|
|---|
| 649 | #should add it into coll config classifiers
|
|---|
| 650 | next;
|
|---|
| 651 | }
|
|---|
| 652 |
|
|---|
| 653 | if (not $started_classifiers) {
|
|---|
| 654 | &write_line('COLCFG', ["<serviceRack name=\"GS2Browse\">"]);
|
|---|
| 655 | if (defined $buildcfg->{'indexstem'}) {
|
|---|
| 656 | my $indexstem = $buildcfg->{'indexstem'};
|
|---|
| 657 | &write_line('COLCFG', ["<indexStem name=\"", $indexstem, "\" />"]);
|
|---|
| 658 | }
|
|---|
| 659 | if (defined $buildcfg->{'infodbtype'}) {
|
|---|
| 660 | my $infodbtype = $buildcfg->{'infodbtype'};
|
|---|
| 661 | &write_line('COLCFG', ["<databaseType name=\"", $infodbtype, "\" />"]);
|
|---|
| 662 | }
|
|---|
| 663 | &write_line('COLCFG', ["<classifierList>"]);
|
|---|
| 664 | $started_classifiers = 1;
|
|---|
| 665 | }
|
|---|
| 666 | my $content = ''; #use buttonname first, then metadata
|
|---|
| 667 | if ($classname eq "DateList") {
|
|---|
| 668 | $content = "Date";
|
|---|
| 669 | } else {
|
|---|
| 670 | for (my $j=0; $j<scalar(@$cl); $j++) {
|
|---|
| 671 | my $arg = @$cl[$j];
|
|---|
| 672 | if ($arg eq "-buttonname"){
|
|---|
| 673 | $content = @$cl[$j+1];
|
|---|
| 674 | last;
|
|---|
| 675 | } elsif ($arg eq "-metadata") {
|
|---|
| 676 | $content = @$cl[$j+1];
|
|---|
| 677 | }
|
|---|
| 678 |
|
|---|
| 679 | }
|
|---|
| 680 | }
|
|---|
| 681 | &write_line('COLCFG', ["<classifier name=\"", $name, "\" content=\"", $content, "\" />"]);
|
|---|
| 682 | }
|
|---|
| 683 | if ($started_classifiers) {
|
|---|
| 684 | # end the classifiers
|
|---|
| 685 | &write_line('COLCFG', ["</classifierList>"]);
|
|---|
| 686 | # close off the Browse service
|
|---|
| 687 | &write_line('COLCFG', ["</serviceRack>"]);
|
|---|
| 688 | }
|
|---|
| 689 |
|
|---|
| 690 | # the phind classifier is a separate service
|
|---|
| 691 | if ($phind) {
|
|---|
| 692 | # if phind classifier
|
|---|
| 693 | &write_line('COLCFG', ["<serviceRack name=\"PhindPhraseBrowse\" />"]);
|
|---|
| 694 | }
|
|---|
| 695 |
|
|---|
| 696 |
|
|---|
| 697 | &write_line('COLCFG', ["</serviceRackList>"]);
|
|---|
| 698 | &write_line('COLCFG', ["</buildConfig>"]);
|
|---|
| 699 |
|
|---|
| 700 | close (COLCFG);
|
|---|
| 701 | }
|
|---|
| 702 |
|
|---|
| 703 |
|
|---|
| 704 | #########################################################
|
|---|
| 705 |
|
|---|
| 706 | 1;
|
|---|
