| 1 | ###############################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # buildcolutils.pm -- index and build the collection. The buildtime counterpart
|
|---|
| 4 | # of inexport.pl
|
|---|
| 5 | #
|
|---|
| 6 | # A component of the Greenstone digital library software
|
|---|
| 7 | # from the New Zealand Digital Library Project at the
|
|---|
| 8 | # University of Waikato, New Zealand.
|
|---|
| 9 | #
|
|---|
| 10 | # Copyright (C) 1999 New Zealand Digital Library Project
|
|---|
| 11 | #
|
|---|
| 12 | # This program is free software; you can redistribute it and/or modify
|
|---|
| 13 | # it under the terms of the GNU General Public License as published by
|
|---|
| 14 | # the Free Software Foundation; either version 2 of the License, or
|
|---|
| 15 | # (at your option) any later version.
|
|---|
| 16 | #
|
|---|
| 17 | # This program is distributed in the hope that it will be useful,
|
|---|
| 18 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 19 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 20 | # GNU General Public License for more details.
|
|---|
| 21 | #
|
|---|
| 22 | # You should have received a copy of the GNU General Public License
|
|---|
| 23 | # along with this program; if not, write to the Free Software
|
|---|
| 24 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|---|
| 25 | #
|
|---|
| 26 | ###############################################################################
|
|---|
| 27 |
|
|---|
| 28 | package buildcolutils;
|
|---|
| 29 |
|
|---|
| 30 | #use strict;
|
|---|
| 31 | #no strict 'refs';
|
|---|
| 32 |
|
|---|
| 33 | use File::Basename;
|
|---|
| 34 |
|
|---|
| 35 | use colcfg;
|
|---|
| 36 | use dbutil;
|
|---|
| 37 | use util;
|
|---|
| 38 | use FileUtils;
|
|---|
| 39 | use scriptutil;
|
|---|
| 40 | use gsprintf;
|
|---|
| 41 | use printusage;
|
|---|
| 42 | use parse2;
|
|---|
| 43 |
|
|---|
| 44 | ## @method new()
|
|---|
| 45 | #
|
|---|
| 46 | # Parses up and validates the arguments to the build process before creating
|
|---|
| 47 | # the appropriate build process to do the actual work
|
|---|
| 48 | #
|
|---|
| 49 | # @note Added true incremental support - John Thompson, DL Consulting Ltd.
|
|---|
| 50 | # @note There were several bugs regarding using directories other than
|
|---|
| 51 | # "import" or "archives" during import and build quashed. - John
|
|---|
| 52 | # Thompson, DL Consulting Ltd.
|
|---|
| 53 | #
|
|---|
| 54 | # @param $incremental If true indicates this build should not regenerate all
|
|---|
| 55 | # the index and metadata files, and should instead just
|
|---|
| 56 | # append the information found in the archives directory
|
|---|
| 57 | # to the existing files. If this requires some complex
|
|---|
| 58 | # work so as to correctly insert into a classifier so be
|
|---|
| 59 | # it. Of course none of this is done here - instead the
|
|---|
| 60 | # incremental argument is passed to the document
|
|---|
| 61 | # processor.
|
|---|
| 62 | #
|
|---|
| 63 | sub new
|
|---|
| 64 | {
|
|---|
| 65 | my $class = shift(@_);
|
|---|
| 66 | my ($argv, $options, $opt_listall_options) = @_;
|
|---|
| 67 |
|
|---|
| 68 | my $self = {'builddir' => undef,
|
|---|
| 69 | 'buildtype' => undef,
|
|---|
| 70 | 'close_faillog' => 0,
|
|---|
| 71 | 'close_out' => 0,
|
|---|
| 72 | 'mode' => '',
|
|---|
| 73 | 'orthogonalbuildtypes' => undef,
|
|---|
| 74 | 'realbuilddir' => undef,
|
|---|
| 75 | 'textindex' => '',
|
|---|
| 76 | 'xml' => 0
|
|---|
| 77 | };
|
|---|
| 78 |
|
|---|
| 79 | # general options available to all plugins
|
|---|
| 80 | my $arguments = $options->{'args'};
|
|---|
| 81 | my $intArgLeftinAfterParsing = &parse2::parse($argv, $arguments, $self, "allow_extra_options");
|
|---|
| 82 | # If parse returns -1 then something has gone wrong
|
|---|
| 83 | if ($intArgLeftinAfterParsing == -1)
|
|---|
| 84 | {
|
|---|
| 85 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
|---|
| 86 | die "\n";
|
|---|
| 87 | }
|
|---|
| 88 |
|
|---|
| 89 | # If $language has been specified, load the appropriate resource bundle
|
|---|
| 90 | # (Otherwise, the default resource bundle will be loaded automatically)
|
|---|
| 91 | if ($self->{'language'} && $self->{'language'} =~ /\S/)
|
|---|
| 92 | {
|
|---|
| 93 | &gsprintf::load_language_specific_resource_bundle($self->{'language'});
|
|---|
| 94 | }
|
|---|
| 95 |
|
|---|
| 96 | # Do we need 'listall' support in buildcol? If so, copy code from inexport
|
|---|
| 97 | # later [jmt12]
|
|---|
| 98 |
|
|---|
| 99 | # <insert explanation here>
|
|---|
| 100 | if ($self->{'xml'})
|
|---|
| 101 | {
|
|---|
| 102 | &PrintUsage::print_xml_usage($options);
|
|---|
| 103 | print "\n";
|
|---|
| 104 | return bless($self, $class);
|
|---|
| 105 | }
|
|---|
| 106 |
|
|---|
| 107 | # the gli wants strings to be in UTF-8
|
|---|
| 108 | if ($gli)
|
|---|
| 109 | {
|
|---|
| 110 | &gsprintf::output_strings_in_UTF8;
|
|---|
| 111 | }
|
|---|
| 112 |
|
|---|
| 113 | # now check that we had exactly one leftover arg, which should be
|
|---|
| 114 | # the collection name. We don't want to do this earlier, cos
|
|---|
| 115 | # -xml arg doesn't need a collection name
|
|---|
| 116 | # Or if the user specified -h, then we output the usage also
|
|---|
| 117 | if ($intArgLeftinAfterParsing != 1 || (@ARGV && $ARGV[0] =~ /^\-+h/))
|
|---|
| 118 | {
|
|---|
| 119 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
|---|
| 120 | die "\n";
|
|---|
| 121 | }
|
|---|
| 122 |
|
|---|
| 123 | my $out = $self->{'out'};
|
|---|
| 124 | if ($out !~ /^(STDERR|STDOUT)$/i)
|
|---|
| 125 | {
|
|---|
| 126 | open (OUT, ">$out") || (&gsprintf::gsprintf(STDERR, "{common.cannot_open_output_file}\n", $out) && die);
|
|---|
| 127 | $out = "buildcolutils::OUT";
|
|---|
| 128 | $self->{'close_out'} = 1;
|
|---|
| 129 | }
|
|---|
| 130 | $out->autoflush(1);
|
|---|
| 131 | $self->{'out'} = $out;
|
|---|
| 132 |
|
|---|
| 133 | # @ARGV should be only one item, the name of the collection
|
|---|
| 134 | $self->{'collection'} = shift(@{$argv});
|
|---|
| 135 |
|
|---|
| 136 | return bless($self, $class);
|
|---|
| 137 | }
|
|---|
| 138 | # new()
|
|---|
| 139 |
|
|---|
| 140 | # newCGI()?
|
|---|
| 141 |
|
|---|
| 142 | # @function get_collection
|
|---|
| 143 | #
|
|---|
| 144 | sub get_collection
|
|---|
| 145 | {
|
|---|
| 146 | my $self = shift @_;
|
|---|
| 147 | return $self->{'collection'};
|
|---|
| 148 | }
|
|---|
| 149 | # get_collection()
|
|---|
| 150 |
|
|---|
| 151 | # @function read_collection_cfg
|
|---|
| 152 | #
|
|---|
| 153 | sub read_collection_cfg
|
|---|
| 154 | {
|
|---|
| 155 | my $self = shift(@_);
|
|---|
| 156 | my ($collection, $options) = @_;
|
|---|
| 157 |
|
|---|
| 158 | my $collectdir = $self->{'collectdir'};
|
|---|
| 159 | my $site = $self->{'site'};
|
|---|
| 160 | my $out = $self->{'out'};
|
|---|
| 161 |
|
|---|
| 162 | # get and check the collection
|
|---|
| 163 | if (($collection = &colcfg::use_collection($site, $collection, $collectdir)) eq "")
|
|---|
| 164 | {
|
|---|
| 165 | &PrintUsage::print_txt_usage($options, "{buildcol.params}");
|
|---|
| 166 | die "\n";
|
|---|
| 167 | }
|
|---|
| 168 |
|
|---|
| 169 | # set gs_version 2/3
|
|---|
| 170 | $self->{'gs_version'} = "2";
|
|---|
| 171 | if ((defined $site) && ($site ne ""))
|
|---|
| 172 | {
|
|---|
| 173 | # gs3
|
|---|
| 174 | $self->{'gs_version'} = "3";
|
|---|
| 175 | }
|
|---|
| 176 |
|
|---|
| 177 | # add collection's perllib dir into include path in case we have collection
|
|---|
| 178 | # specific modules
|
|---|
| 179 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib'));
|
|---|
| 180 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib', 'classify'));
|
|---|
| 181 | &util::augmentINC(&FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'perllib', 'plugins'));
|
|---|
| 182 |
|
|---|
| 183 | # check that we can open the faillog
|
|---|
| 184 | my $faillog = $self->{'faillog'};
|
|---|
| 185 | if ($faillog eq "")
|
|---|
| 186 | {
|
|---|
| 187 | $faillog = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
|
|---|
| 188 | }
|
|---|
| 189 | # note that we're appending to the faillog here (import.pl clears it each time)
|
|---|
| 190 | # this could potentially create a situation where the faillog keeps being added
|
|---|
| 191 | # to over multiple builds (if the import process is being skipped)
|
|---|
| 192 | open (FAILLOG, ">>$faillog") || (&gsprintf::gsprintf(STDERR, "{common.cannot_open_fail_log}\n", $faillog) && die);
|
|---|
| 193 | $faillog = 'buildcolutils::FAILLOG';
|
|---|
| 194 | $faillog->autoflush(1);
|
|---|
| 195 | $self->{'faillog'} = $faillog;
|
|---|
| 196 | $self->{'faillogname'} = $faillog;
|
|---|
| 197 | $self->{'close_faillog'} = 1;
|
|---|
| 198 |
|
|---|
| 199 | # Read in the collection configuration file.
|
|---|
| 200 | my $gs_mode = "gs".$self->{'gs_version'}; #gs2 or gs3
|
|---|
| 201 | my $config_filename = &colcfg::get_collect_cfg_name($out, $gs_mode);
|
|---|
| 202 | my $collect_cfg = &colcfg::read_collection_cfg($config_filename, $gs_mode);
|
|---|
| 203 |
|
|---|
| 204 | return ($config_filename, $collect_cfg);
|
|---|
| 205 | }
|
|---|
| 206 | # read_collection_cfg()
|
|---|
| 207 |
|
|---|
| 208 | # @function set_collection_options
|
|---|
| 209 | # This function copies across values for arguments from the collection
|
|---|
| 210 | # configuration file if they are not already provided by the user, then
|
|---|
| 211 | # sets reasonable defaults for any required arguments that remains without
|
|---|
| 212 | # a value.
|
|---|
| 213 | sub set_collection_options
|
|---|
| 214 | {
|
|---|
| 215 | my $self = shift @_;
|
|---|
| 216 | my ($collectcfg) = @_;
|
|---|
| 217 | my ($buildtype, $orthogonalbuildtypes);
|
|---|
| 218 |
|
|---|
| 219 | # If the infodbtype value wasn't defined in the collect.cfg file, use the default
|
|---|
| 220 | if (!defined($collectcfg->{'infodbtype'}))
|
|---|
| 221 | {
|
|---|
| 222 | $collectcfg->{'infodbtype'} = &dbutil::get_default_infodb_type();
|
|---|
| 223 | }
|
|---|
| 224 | # - just so I don't have to pass collectcfg around as well
|
|---|
| 225 | $self->{'infodbtype'} = $collectcfg->{'infodbtype'};
|
|---|
| 226 |
|
|---|
| 227 | if ($self->{'verbosity'} !~ /\d+/)
|
|---|
| 228 | {
|
|---|
| 229 | if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/)
|
|---|
| 230 | {
|
|---|
| 231 | $self->{'verbosity'} = $collectcfg->{'verbosity'};
|
|---|
| 232 | }
|
|---|
| 233 | else
|
|---|
| 234 | {
|
|---|
| 235 | $self->{'verbosity'} = 2; # the default
|
|---|
| 236 | }
|
|---|
| 237 | }
|
|---|
| 238 |
|
|---|
| 239 | # we use searchtype for determining buildtype, but for old versions, use buildtype
|
|---|
| 240 | if (defined $collectcfg->{'buildtype'})
|
|---|
| 241 | {
|
|---|
| 242 | $self->{'buildtype'} = $collectcfg->{'buildtype'};
|
|---|
| 243 | }
|
|---|
| 244 | elsif (defined $collectcfg->{'searchtypes'} || defined $collectcfg->{'searchtype'})
|
|---|
| 245 | {
|
|---|
| 246 | $self->{'buildtype'} = "mgpp";
|
|---|
| 247 | }
|
|---|
| 248 | else
|
|---|
| 249 | {
|
|---|
| 250 | $self->{'buildtype'} = "mg"; #mg is the default
|
|---|
| 251 | }
|
|---|
| 252 |
|
|---|
| 253 | if ($self->{'buildtype'} eq "mgpp" && defined $collectcfg->{'textcompress'})
|
|---|
| 254 | {
|
|---|
| 255 | $self->{'textindex'} = $collectcfg->{'textcompress'};
|
|---|
| 256 | }
|
|---|
| 257 |
|
|---|
| 258 | # is it okay to always clobber or possible remain undefined? [jmt12]
|
|---|
| 259 | if (defined $collectcfg->{'orthogonalbuildtypes'})
|
|---|
| 260 | {
|
|---|
| 261 | $self->{'orthogonalbuildtypes'} = $collectcfg->{'orthogonalbuildtypes'};
|
|---|
| 262 | }
|
|---|
| 263 |
|
|---|
| 264 | # - resolve (and possibly set to default) builddir
|
|---|
| 265 | if (defined $collectcfg->{'archivedir'} && $self->{'archivedir'} eq "")
|
|---|
| 266 | {
|
|---|
| 267 | $self->{'archivedir'} = $collectcfg->{'archivedir'};
|
|---|
| 268 | }
|
|---|
| 269 | # Modified so that the archivedir, if provided as an argument, is made
|
|---|
| 270 | # absolute if it isn't already
|
|---|
| 271 | if ($self->{'archivedir'} eq "")
|
|---|
| 272 | {
|
|---|
| 273 | $self->{'archivedir'} = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "archives");
|
|---|
| 274 | }
|
|---|
| 275 | else
|
|---|
| 276 | {
|
|---|
| 277 | $self->{'archivedir'} = &util::make_absolute($ENV{'GSDLCOLLECTDIR'}, $self->{'archivedir'});
|
|---|
| 278 | }
|
|---|
| 279 | # End Mod
|
|---|
| 280 | $self->{'archivedir'} = &FileUtils::sanitizePath($self->{'archivedir'});
|
|---|
| 281 | #$self->{'archivedir'} =~ s/[\\\/]+/\//g;
|
|---|
| 282 | #$self->{'archivedir'} =~ s/\/$//;
|
|---|
| 283 |
|
|---|
| 284 | # - resolve (and possibly set to default) builddir
|
|---|
| 285 | if (defined $collectcfg->{'builddir'} && $self->{'builddir'} eq "")
|
|---|
| 286 | {
|
|---|
| 287 | $self->{'builddir'} = $collectcfg->{'builddir'};
|
|---|
| 288 | }
|
|---|
| 289 | if ($self->{'builddir'} eq "")
|
|---|
| 290 | {
|
|---|
| 291 | $self->{'builddir'} = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, 'building');
|
|---|
| 292 | if ($incremental)
|
|---|
| 293 | {
|
|---|
| 294 | &gsprintf::gsprintf($out, "{buildcol.incremental_default_builddir}\n");
|
|---|
| 295 | }
|
|---|
| 296 | } else {
|
|---|
| 297 | # make absolute if not already
|
|---|
| 298 | $self->{'builddir'} = &util::make_absolute($ENV{'GSDLCOLLECTDIR'}, $self->{'builddir'});
|
|---|
| 299 | }
|
|---|
| 300 |
|
|---|
| 301 | # - why don't we make builddir absolute similar to archivedir?
|
|---|
| 302 | $self->{'builddir'} = &FileUtils::sanitizePath($self->{'builddir'});
|
|---|
| 303 | #$self->{'builddir'} =~ s/[\\\/]+/\//g;
|
|---|
| 304 | #$self->{'builddir'} =~ s/\/$//;
|
|---|
| 305 |
|
|---|
| 306 | if (defined $collectcfg->{'cachedir'} && $self->{'cachedir'} eq "")
|
|---|
| 307 | {
|
|---|
| 308 | $self->{'cachedir'} = $collectcfg->{'cachedir'};
|
|---|
| 309 | }
|
|---|
| 310 |
|
|---|
| 311 | if ($self->{'maxdocs'} !~ /\-?\d+/)
|
|---|
| 312 | {
|
|---|
| 313 | if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/)
|
|---|
| 314 | {
|
|---|
| 315 | $self->{'maxdocs'} = $collectcfg->{'maxdocs'};
|
|---|
| 316 | }
|
|---|
| 317 | else
|
|---|
| 318 | {
|
|---|
| 319 | $self->{'maxdocs'} = -1; # the default
|
|---|
| 320 | }
|
|---|
| 321 | }
|
|---|
| 322 |
|
|---|
| 323 | # always clobbers? [jmt12]
|
|---|
| 324 | if (defined $collectcfg->{'maxnumeric'} && $collectcfg->{'maxnumeric'} =~ /\d+/)
|
|---|
| 325 | {
|
|---|
| 326 | $self->{'maxnumeric'} = $collectcfg->{'maxnumeric'};
|
|---|
| 327 | }
|
|---|
| 328 | if ($self->{'maxnumeric'} < 4 || $self->{'maxnumeric'} > 512)
|
|---|
| 329 | {
|
|---|
| 330 | $self->{'maxnumeric'} = 4;
|
|---|
| 331 | }
|
|---|
| 332 |
|
|---|
| 333 | if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i)
|
|---|
| 334 | {
|
|---|
| 335 | $self->{'debug'} = 1;
|
|---|
| 336 | }
|
|---|
| 337 |
|
|---|
| 338 | if ($self->{'mode'} !~ /^(all|compress_text|build_index|infodb|extra)$/)
|
|---|
| 339 | {
|
|---|
| 340 | if (defined $collectcfg->{'mode'} && $collectcfg->{'mode'} =~ /^(all|compress_text|build_index|infodb|extra)$/)
|
|---|
| 341 | {
|
|---|
| 342 | $self->{'mode'} = $collectcfg->{'mode'};
|
|---|
| 343 | }
|
|---|
| 344 | else
|
|---|
| 345 | {
|
|---|
| 346 | $self->{'mode'} = "all"; # the default
|
|---|
| 347 | }
|
|---|
| 348 | }
|
|---|
| 349 |
|
|---|
| 350 | # Presumably 'index' from the collect.cfg still works [jmt12]
|
|---|
| 351 | if (defined $collectcfg->{'index'} && $self->{'indexname'} eq "")
|
|---|
| 352 | {
|
|---|
| 353 | $self->{'indexname'} = $collectcfg->{'index'};
|
|---|
| 354 | }
|
|---|
| 355 | # - 'index' from the command line doesn't make it through parsing so I
|
|---|
| 356 | # renamed this option 'indexname' [jmt12]
|
|---|
| 357 | if (defined $collectcfg->{'indexname'} && $self->{'indexname'} eq "")
|
|---|
| 358 | {
|
|---|
| 359 | $self->{'indexname'} = $collectcfg->{'indexname'};
|
|---|
| 360 | }
|
|---|
| 361 | # - we may also define the index level to build now [jmt12]
|
|---|
| 362 | if (defined $collectcfg->{'indexlevel'} && $self->{'indexlevel'} eq "")
|
|---|
| 363 | {
|
|---|
| 364 | $self->{'indexlevel'} = $collectcfg->{'indexlevel'};
|
|---|
| 365 | }
|
|---|
| 366 |
|
|---|
| 367 | if (defined $collectcfg->{'no_text'} && $self->{'no_text'} == 0)
|
|---|
| 368 | {
|
|---|
| 369 | if ($collectcfg->{'no_text'} =~ /^true$/i)
|
|---|
| 370 | {
|
|---|
| 371 | $self->{'no_text'} = 1;
|
|---|
| 372 | }
|
|---|
| 373 | }
|
|---|
| 374 |
|
|---|
| 375 | if (defined $collectcfg->{'no_strip_html'} && $self->{'no_strip_html'} == 0)
|
|---|
| 376 | {
|
|---|
| 377 | if ($collectcfg->{'no_strip_html'} =~ /^true$/i)
|
|---|
| 378 | {
|
|---|
| 379 | $self->{'no_strip_html'} = 1;
|
|---|
| 380 | }
|
|---|
| 381 | }
|
|---|
| 382 |
|
|---|
| 383 | if (defined $collectcfg->{'store_metadata_coverage'} && $self->{'store_metadata_coverage'} == 0)
|
|---|
| 384 | {
|
|---|
| 385 | if ($collectcfg->{'store_metadata_coverage'} =~ /^true$/i)
|
|---|
| 386 | {
|
|---|
| 387 | $self->{'store_metadata_coverage'} = 1;
|
|---|
| 388 | }
|
|---|
| 389 | }
|
|---|
| 390 |
|
|---|
| 391 | if (defined $collectcfg->{'remove_empty_classifications'} && $self->{'remove_empty_classifications'} == 0)
|
|---|
| 392 | {
|
|---|
| 393 | if ($collectcfg->{'remove_empty_classifications'} =~ /^true$/i)
|
|---|
| 394 | {
|
|---|
| 395 | $self->{'remove_empty_classifications'} = 1;
|
|---|
| 396 | }
|
|---|
| 397 | }
|
|---|
| 398 |
|
|---|
| 399 | if (defined $collectcfg->{'gli'} && $collectcfg->{'gli'} =~ /^true$/i)
|
|---|
| 400 | {
|
|---|
| 401 | $self->{'gli'} = 1;
|
|---|
| 402 | }
|
|---|
| 403 | if (!defined $self->{'gli'})
|
|---|
| 404 | {
|
|---|
| 405 | $self->{'gli'} = 0;
|
|---|
| 406 | }
|
|---|
| 407 |
|
|---|
| 408 | if ($self->{'sections_index_document_metadata'} !~ /\S/ && defined $collectcfg->{'sections_index_document_metadata'})
|
|---|
| 409 | {
|
|---|
| 410 | $self->{'sections_index_document_metadata'} = $collectcfg->{'sections_index_document_metadata'};
|
|---|
| 411 | }
|
|---|
| 412 |
|
|---|
| 413 | if ($self->{'sections_index_document_metadata'} !~ /^(never|always|unless_section_metadata_exists)$/) {
|
|---|
| 414 | $self->{'sections_index_document_metadata'} = 'never';
|
|---|
| 415 | }
|
|---|
| 416 |
|
|---|
| 417 | if ($self->{'sections_sort_on_document_metadata'} !~ /\S/ && defined $collectcfg->{'sections_sort_on_document_metadata'})
|
|---|
| 418 | {
|
|---|
| 419 | $self->{'sections_sort_on_document_metadata'} = $collectcfg->{'sections_sort_on_document_metadata'};
|
|---|
| 420 | }
|
|---|
| 421 |
|
|---|
| 422 | if ($self->{'sections_sort_on_document_metadata'} !~ /^(never|always|unless_section_metadata_exists)$/) {
|
|---|
| 423 | $self->{'sections_sort_on_document_metadata'} = 'never';
|
|---|
| 424 | }
|
|---|
| 425 |
|
|---|
| 426 | my ($removeold, $keepold, $incremental, $incremental_mode)
|
|---|
| 427 | = &scriptutil::check_removeold_and_keepold($self->{'removeold'}, $self->{'keepold'},
|
|---|
| 428 | $self->{'incremental'}, 'building',
|
|---|
| 429 | $collectcfg);
|
|---|
| 430 | $self->{'removeold'} = $removeold;
|
|---|
| 431 | $self->{'keepold'} = $keepold;
|
|---|
| 432 | $self->{'incremental'} = $incremental;
|
|---|
| 433 | $self->{'incremental_mode'} = $incremental_mode;
|
|---|
| 434 |
|
|---|
| 435 | # New argument to track whether build is incremental
|
|---|
| 436 | if (!defined $self->{'incremental'})
|
|---|
| 437 | {
|
|---|
| 438 | $self->{'incremental'} = 0;
|
|---|
| 439 | }
|
|---|
| 440 |
|
|---|
| 441 | #set the text index
|
|---|
| 442 | if (($self->{'buildtype'} eq 'mgpp') || ($self->{'buildtype'} eq 'lucene') || ($self->{'buildtype'} eq 'solr'))
|
|---|
| 443 | {
|
|---|
| 444 | if ($self->{'textindex'} eq '')
|
|---|
| 445 | {
|
|---|
| 446 | $self->{'textindex'} = 'text';
|
|---|
| 447 | }
|
|---|
| 448 | }
|
|---|
| 449 | else
|
|---|
| 450 | {
|
|---|
| 451 | $self->{'textindex'} = 'section:text';
|
|---|
| 452 | }
|
|---|
| 453 | }
|
|---|
| 454 | # set_collection_options()
|
|---|
| 455 |
|
|---|
| 456 | # @function prepare_builders
|
|---|
| 457 | #
|
|---|
| 458 | sub prepare_builders
|
|---|
| 459 | {
|
|---|
| 460 | my $self = shift @_;
|
|---|
| 461 | my ($config_filename,$collectcfg) = @_;
|
|---|
| 462 |
|
|---|
| 463 | my $archivedir = $self->{'archivedir'};
|
|---|
| 464 | my $builddir = $self->{'builddir'};
|
|---|
| 465 | my $buildtype = $self->{'buildtype'};
|
|---|
| 466 | my $cachedir = $self->{'cachedir'};
|
|---|
| 467 | my $collectdir = $self->{'collectdir'};
|
|---|
| 468 | my $collection = $self->{'collection'};
|
|---|
| 469 | my $debug = $self->{'debug'};
|
|---|
| 470 | my $faillog = $self->{'faillog'};
|
|---|
| 471 | my $gli = $self->{'gli'};
|
|---|
| 472 | my $incremental = $self->{'incremental'};
|
|---|
| 473 | my $incremental_mode = $self->{'incremental_mode'};
|
|---|
| 474 | my $keepold = $self->{'keepold'};
|
|---|
| 475 | my $maxdocs = $self->{'maxdocs'};
|
|---|
| 476 | my $maxnumeric = $self->{'maxnumeric'};
|
|---|
| 477 | my $no_strip_html = $self->{'no_strip_html'};
|
|---|
| 478 | my $no_text = $self->{'no_text'};
|
|---|
| 479 | my $orthogonalbuildtypes = $self->{'orthogonalbuildtypes'};
|
|---|
| 480 | my $out = $self->{'out'};
|
|---|
| 481 | my $remove_empty_classifications = $self->{'remove_empty_classifications'};
|
|---|
| 482 | my $sections_index_document_metadata = $self->{'sections_index_document_metadata'};
|
|---|
| 483 | my $sections_sort_on_document_metadata = $self->{'sections_sort_on_document_metadata'};
|
|---|
| 484 | my $site = $self->{'site'};
|
|---|
| 485 | my $store_metadata_coverage = $self->{'store_metadata_coverage'};
|
|---|
| 486 | my $verbosity = $self->{'verbosity'};
|
|---|
| 487 |
|
|---|
| 488 | if ($gli)
|
|---|
| 489 | {
|
|---|
| 490 | print STDERR "<Build>\n";
|
|---|
| 491 | }
|
|---|
| 492 |
|
|---|
| 493 | # fill in the default archives and building directories if none
|
|---|
| 494 | # were supplied, turn all \ into / and remove trailing /
|
|---|
| 495 |
|
|---|
| 496 | my ($realarchivedir, $realbuilddir);
|
|---|
| 497 | # update the archive cache if needed
|
|---|
| 498 | if ($cachedir)
|
|---|
| 499 | {
|
|---|
| 500 | if ($verbosity >= 1)
|
|---|
| 501 | {
|
|---|
| 502 | &gsprintf::gsprintf($out, "{buildcol.updating_archive_cache}\n")
|
|---|
| 503 | }
|
|---|
| 504 |
|
|---|
| 505 | $cachedir =~ s/[\\\/]+$//;
|
|---|
| 506 | if ($cachedir !~ /collect[\/\\]$collection/)
|
|---|
| 507 | {
|
|---|
| 508 | $cachedir = &FileUtils::filenameConcatenate($cachedir, 'collect', $collection);
|
|---|
| 509 | }
|
|---|
| 510 |
|
|---|
| 511 | $realarchivedir = &FileUtils::filenameConcatenate($cachedir, 'archives');
|
|---|
| 512 | $realbuilddir = &FileUtils::filenameConcatenate($cachedir, 'building');
|
|---|
| 513 | &FileUtils::makeAllDirectories($realarchivedir);
|
|---|
| 514 | &FileUtils::makeAllDirectories($realbuilddir);
|
|---|
| 515 | &FileUtils::synchronizeDirectory($archivedir, $realarchivedir, $verbosity);
|
|---|
| 516 | }
|
|---|
| 517 | else
|
|---|
| 518 | {
|
|---|
| 519 | $realarchivedir = $archivedir;
|
|---|
| 520 | $realbuilddir = $builddir;
|
|---|
| 521 | }
|
|---|
| 522 | $self->{'realarchivedir'} = $realarchivedir;
|
|---|
| 523 | $self->{'realbuilddir'} = $realbuilddir;
|
|---|
| 524 |
|
|---|
| 525 | # build it in realbuilddir
|
|---|
| 526 | &FileUtils::makeAllDirectories($realbuilddir);
|
|---|
| 527 |
|
|---|
| 528 | my ($buildertype, $builderdir, $builder);
|
|---|
| 529 | # if a builder class has been created for this collection, use it
|
|---|
| 530 | # otherwise, use the mg or mgpp builder
|
|---|
| 531 | if (-e "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib/custombuilder.pm")
|
|---|
| 532 | {
|
|---|
| 533 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/custom/${collection}/perllib";
|
|---|
| 534 | $buildertype = "custombuilder";
|
|---|
| 535 | }
|
|---|
| 536 | elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/custombuilder.pm")
|
|---|
| 537 | {
|
|---|
| 538 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
|
|---|
| 539 | $buildertype = "custombuilder";
|
|---|
| 540 | }
|
|---|
| 541 | elsif (-e "$ENV{'GSDLCOLLECTDIR'}/perllib/${collection}builder.pm")
|
|---|
| 542 | {
|
|---|
| 543 | $builderdir = "$ENV{'GSDLCOLLECTDIR'}/perllib";
|
|---|
| 544 | $buildertype = $collection . 'builder';
|
|---|
| 545 | }
|
|---|
| 546 | else
|
|---|
| 547 | {
|
|---|
| 548 | $builderdir = undef;
|
|---|
| 549 | if ($buildtype ne '')
|
|---|
| 550 | {
|
|---|
| 551 | # caters for extension-based build types, such as 'solr'
|
|---|
| 552 | $buildertype = $buildtype . 'builder';
|
|---|
| 553 | }
|
|---|
| 554 | else
|
|---|
| 555 | {
|
|---|
| 556 | # Default to mgpp
|
|---|
| 557 | $buildertype = 'mgppbuilder';
|
|---|
| 558 | }
|
|---|
| 559 | }
|
|---|
| 560 | # check for extension specific builders
|
|---|
| 561 | # (that will then be run after main builder.pm
|
|---|
| 562 | my @builderdir_list = ($builderdir);
|
|---|
| 563 | my @buildertype_list = ($buildertype);
|
|---|
| 564 |
|
|---|
| 565 | my $mode = $self->{'mode'};
|
|---|
| 566 |
|
|---|
| 567 | if ($mode eq "extra") {
|
|---|
| 568 | # knock out the main builder type, by reseting the lists to be empty
|
|---|
| 569 | @builderdir_list = ();
|
|---|
| 570 | @buildertype_list = ();
|
|---|
| 571 | }
|
|---|
| 572 |
|
|---|
| 573 | if (defined $orthogonalbuildtypes)
|
|---|
| 574 | {
|
|---|
| 575 | foreach my $obt (@$orthogonalbuildtypes)
|
|---|
| 576 | {
|
|---|
| 577 | push(@builderdir_list,undef); # rely on @INC to find it
|
|---|
| 578 | push(@buildertype_list,$obt."Builder");
|
|---|
| 579 | }
|
|---|
| 580 | }
|
|---|
| 581 |
|
|---|
| 582 | # Set up array of the main builder.pm, followed by any ones
|
|---|
| 583 | # from the extension folders
|
|---|
| 584 |
|
|---|
| 585 | my $num_builders = scalar(@buildertype_list);
|
|---|
| 586 | my @builders = ();
|
|---|
| 587 |
|
|---|
| 588 | for (my $i=0; $i<$num_builders; $i++)
|
|---|
| 589 | {
|
|---|
| 590 | my $this_builder;
|
|---|
| 591 | my $this_buildertype = $buildertype_list[$i];
|
|---|
| 592 | my $this_builderdir = $builderdir_list[$i];
|
|---|
| 593 |
|
|---|
| 594 | if ((defined $this_builderdir) && ($this_builderdir ne ""))
|
|---|
| 595 | {
|
|---|
| 596 | require "$this_builderdir/$this_buildertype.pm";
|
|---|
| 597 | }
|
|---|
| 598 | else
|
|---|
| 599 | {
|
|---|
| 600 | require "$this_buildertype.pm";
|
|---|
| 601 | }
|
|---|
| 602 |
|
|---|
| 603 | eval("\$this_builder = new $this_buildertype(\$site, \$collection, " .
|
|---|
| 604 | "\$realarchivedir, \$realbuilddir, \$verbosity, " .
|
|---|
| 605 | "\$maxdocs, \$debug, \$keepold, \$incremental, \$incremental_mode, " .
|
|---|
| 606 | "\$remove_empty_classifications, " .
|
|---|
| 607 | "\$out, \$no_text, \$faillog, \$gli)");
|
|---|
| 608 | die "$@" if $@;
|
|---|
| 609 |
|
|---|
| 610 | push(@builders,$this_builder);
|
|---|
| 611 | }
|
|---|
| 612 |
|
|---|
| 613 | # Init phase for builders
|
|---|
| 614 | for (my $i=0; $i<$num_builders; $i++)
|
|---|
| 615 | {
|
|---|
| 616 | my $this_buildertype = $buildertype_list[$i];
|
|---|
| 617 | my $this_builderdir = $builderdir_list[$i];
|
|---|
| 618 | my $this_builder = $builders[$i];
|
|---|
| 619 |
|
|---|
| 620 | $this_builder->init();
|
|---|
| 621 | $this_builder->set_maxnumeric($maxnumeric);
|
|---|
| 622 |
|
|---|
| 623 | if (($this_buildertype eq "mgppbuilder") && $no_strip_html)
|
|---|
| 624 | {
|
|---|
| 625 | $this_builder->set_strip_html(0);
|
|---|
| 626 | }
|
|---|
| 627 |
|
|---|
| 628 | if ($sections_index_document_metadata ne "never")
|
|---|
| 629 | {
|
|---|
| 630 | $this_builder->set_sections_index_document_metadata($sections_index_document_metadata);
|
|---|
| 631 | }
|
|---|
| 632 | if (($this_buildertype eq "lucenebuilder" || $this_buildertype eq "solrbuilder") && $sections_sort_on_document_metadata ne "never")
|
|---|
| 633 | {
|
|---|
| 634 | $this_builder->set_sections_sort_on_document_metadata($sections_sort_on_document_metadata);
|
|---|
| 635 | }
|
|---|
| 636 |
|
|---|
| 637 | if ($store_metadata_coverage)
|
|---|
| 638 | {
|
|---|
| 639 | $this_builder->set_store_metadata_coverage(1);
|
|---|
| 640 | }
|
|---|
| 641 | }
|
|---|
| 642 | return \@builders;
|
|---|
| 643 | }
|
|---|
| 644 |
|
|---|
| 645 | sub build_collection
|
|---|
| 646 | {
|
|---|
| 647 | my $self = shift(@_);
|
|---|
| 648 | my @builders = @{shift(@_)};
|
|---|
| 649 |
|
|---|
| 650 | my $indexlevel = $self->{'indexlevel'};
|
|---|
| 651 | my $indexname = $self->{'indexname'};
|
|---|
| 652 | my $mode = $self->{'mode'};
|
|---|
| 653 | my $textindex = $self->{'textindex'};
|
|---|
| 654 |
|
|---|
| 655 | # Run the requested passes
|
|---|
| 656 | if ($mode =~ /^(all|extra)$/i)
|
|---|
| 657 | {
|
|---|
| 658 | # 'map' modifies the elements of the original array, so calling
|
|---|
| 659 | # methods -- as done below -- will cause (by default) @builders
|
|---|
| 660 | # to be changed to whatever these functions return (which is *not*
|
|---|
| 661 | # what we want -- we want to leave the values unchanged)
|
|---|
| 662 | # => Use 'local' (dynamic scoping) to give each 'map' call its
|
|---|
| 663 | # own local copy This could also be done with:
|
|---|
| 664 | # (my $new =$_)->method(); $new
|
|---|
| 665 | # but is a bit more cumbersome to write
|
|---|
| 666 | map { local $_=$_; $_->compress_text($textindex); } @builders;
|
|---|
| 667 | # - we pass the required indexname and indexlevel (if specified) to the
|
|---|
| 668 | # processor [jmt12]
|
|---|
| 669 | map { local $_=$_; $_->build_indexes($indexname, $indexlevel); } @builders;
|
|---|
| 670 | map { local $_=$_; $_->make_infodatabase(); } @builders;
|
|---|
| 671 | map { local $_=$_; $_->collect_specific(); } @builders;
|
|---|
| 672 | }
|
|---|
| 673 | elsif ($mode =~ /^compress_text$/i)
|
|---|
| 674 | {
|
|---|
| 675 | map { local $_=$_; $_->compress_text($textindex); } @builders;
|
|---|
| 676 | }
|
|---|
| 677 | elsif ($mode =~ /^build_index$/i)
|
|---|
| 678 | {
|
|---|
| 679 | map { local $_=$_; $_->build_indexes($indexname, $indexlevel); } @builders;
|
|---|
| 680 | }
|
|---|
| 681 | elsif ($mode =~ /^infodb$/i)
|
|---|
| 682 | {
|
|---|
| 683 | map { local $_=$_; $_->make_infodatabase(); } @builders;
|
|---|
| 684 | }
|
|---|
| 685 | else
|
|---|
| 686 | {
|
|---|
| 687 | (&gsprintf::gsprintf(STDERR, "{buildcol.unknown_mode}\n", $mode) && die);
|
|---|
| 688 | }
|
|---|
| 689 | }
|
|---|
| 690 | # build_collection()
|
|---|
| 691 |
|
|---|
| 692 | # @function build_auxiliary_files
|
|---|
| 693 | #
|
|---|
| 694 | sub build_auxiliary_files
|
|---|
| 695 | {
|
|---|
| 696 | my $self = shift(@_);
|
|---|
| 697 | my @builders = @{shift(@_)};
|
|---|
| 698 | if (!$self->{'debug'})
|
|---|
| 699 | {
|
|---|
| 700 | map {local $_=$_; $_->make_auxiliary_files(); } @builders;
|
|---|
| 701 | }
|
|---|
| 702 | }
|
|---|
| 703 | # build_auxiliary_files()
|
|---|
| 704 |
|
|---|
| 705 | # @function complete_builders
|
|---|
| 706 | #
|
|---|
| 707 | sub complete_builders
|
|---|
| 708 | {
|
|---|
| 709 | my $self = shift(@_);
|
|---|
| 710 | my @builders = @{shift(@_)};
|
|---|
| 711 |
|
|---|
| 712 | map {local $_=$_; $_->deinit(); } @builders;
|
|---|
| 713 |
|
|---|
| 714 | if (($self->{'realbuilddir'} ne $self->{'builddir'}) && !$self->{'debug'})
|
|---|
| 715 | {
|
|---|
| 716 | if ($self->{'verbosity'} >= 1)
|
|---|
| 717 | {
|
|---|
| 718 | &gsprintf::gsprintf($out, "{buildcol.copying_back_cached_build}\n");
|
|---|
| 719 | }
|
|---|
| 720 | &FileUtils::removeFilesRecursive($self->{'builddir'});
|
|---|
| 721 | &FileUtils::copyFilesRecursive($self->{'realbuilddir'}, $self->{'builddir'});
|
|---|
| 722 | }
|
|---|
| 723 |
|
|---|
| 724 | # for RSS support: Need rss-items.rdf file in index folder
|
|---|
| 725 | # check if a file called rss-items.rdf exists in archives, then copy it into the building folder
|
|---|
| 726 | # so that when building is moved to index, this file will then also be in index as desired
|
|---|
| 727 | my $collection_dir = &util::resolve_collection_dir($self->{'collectdir'},
|
|---|
| 728 | $self->{'collection'},
|
|---|
| 729 | $self->{'site'});
|
|---|
| 730 | my $rss_items_rdf_file = &FileUtils::filenameConcatenate($self->{'archivedir'}, 'rss-items.rdf');
|
|---|
| 731 | # @todo FileUtils
|
|---|
| 732 | if(defined $self->{'builddir'} && &FileUtils::directoryExists($self->{'builddir'}) && &FileUtils::fileExists($rss_items_rdf_file))
|
|---|
| 733 | {
|
|---|
| 734 | if ($self->{'verbosity'} >= 1)
|
|---|
| 735 | {
|
|---|
| 736 | my $archivedir_tail = "'".basename($self->{'archivedir'})."'";
|
|---|
| 737 | my $builddir_tail = "'".basename($self->{'builddir'})."'";
|
|---|
| 738 |
|
|---|
| 739 | &gsprintf::gsprintf($self->{'out'}, "{buildcol.copying_rss_items_rdf}\n", $archivedir_tail, $builddir_tail);
|
|---|
| 740 | }
|
|---|
| 741 | &FileUtils::copyFiles($rss_items_rdf_file, $self->{'builddir'});
|
|---|
| 742 | }
|
|---|
| 743 |
|
|---|
| 744 | if ($self->{'gli'})
|
|---|
| 745 | {
|
|---|
| 746 | print STDERR "</Build>\n";
|
|---|
| 747 | }
|
|---|
| 748 | }
|
|---|
| 749 | # complete_builders()
|
|---|
| 750 |
|
|---|
| 751 | # @function activate_collection
|
|---|
| 752 | #
|
|---|
| 753 | sub activate_collection
|
|---|
| 754 | {
|
|---|
| 755 | my $self = shift(@_);
|
|---|
| 756 | # if buildcol.pl was run with -activate, need to run activate.pl
|
|---|
| 757 | # now that building's complete
|
|---|
| 758 | if ($self->{'activate'})
|
|---|
| 759 | {
|
|---|
| 760 | #my $quoted_argv = join(" ", map { "\"$_\"" } @ARGV);
|
|---|
| 761 | my @activate_argv = ();
|
|---|
| 762 | push(@activate_argv, '-collectdir', $self->{'collectdir'}) if ($self->{'collectdir'});
|
|---|
| 763 | push(@activate_argv, '-builddir', $self->{'builddir'}) if ($self->{'builddir'});
|
|---|
| 764 | push(@activate_argv, '-site', $self->{'site'}) if ($self->{'site'});
|
|---|
| 765 | push(@activate_argv, '-verbosity', $self->{'verbosity'}) if ($self->{'verbosity'});
|
|---|
| 766 | push(@activate_argv, '-removeold') if ($self->{'removeold'});
|
|---|
| 767 | push(@activate_argv, '-keepold') if ($self->{'keepold'});
|
|---|
| 768 | push(@activate_argv, '-incremental') if ($self->{'incremental'});
|
|---|
| 769 | my $quoted_argv = join(' ', map { "\"$_\"" } @activate_argv);
|
|---|
| 770 | my $activatecol_cmd = '"' . &util::get_perl_exec(). '" -S activate.pl ' . $quoted_argv . ' "' . $self->get_collection() . '"';
|
|---|
| 771 | my $activatecol_status = system($activatecol_cmd)/256;
|
|---|
| 772 |
|
|---|
| 773 | if ($activatecol_status != 0)
|
|---|
| 774 | {
|
|---|
| 775 | print STDERR "Error: Failed to run: $activatecol_cmd\n";
|
|---|
| 776 | print STDERR " $!\n" if ($! ne '');
|
|---|
| 777 | exit(-1);
|
|---|
| 778 | }
|
|---|
| 779 | }
|
|---|
| 780 | }
|
|---|
| 781 |
|
|---|
| 782 | # @function deinit()
|
|---|
| 783 | #
|
|---|
| 784 | sub deinit
|
|---|
| 785 | {
|
|---|
| 786 | my $self = shift(@_);
|
|---|
| 787 |
|
|---|
| 788 | if ($self->{'close_out'})
|
|---|
| 789 | {
|
|---|
| 790 | close OUT;
|
|---|
| 791 | }
|
|---|
| 792 | if ($self->{'close_faillog'})
|
|---|
| 793 | {
|
|---|
| 794 | close FAILLOG;
|
|---|
| 795 | }
|
|---|
| 796 | }
|
|---|
| 797 | # deinit()
|
|---|
| 798 |
|
|---|
| 799 | 1;
|
|---|
