[932] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
[1852] | 3 | # mgppbuilder.pm -- MGBuilder object
|
---|
[932] | 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 | package mgppbuilder;
|
---|
| 27 |
|
---|
[10468] | 28 | use basebuilder;
|
---|
[932] | 29 | use colcfg;
|
---|
| 30 | use plugin;
|
---|
[15715] | 31 | use strict; no strict 'refs';
|
---|
[932] | 32 | use util;
|
---|
| 33 |
|
---|
[15715] | 34 |
|
---|
[10468] | 35 | sub BEGIN {
|
---|
| 36 | @mgppbuilder::ISA = ('basebuilder');
|
---|
[1694] | 37 | }
|
---|
| 38 |
|
---|
| 39 |
|
---|
| 40 |
|
---|
[9157] | 41 | our %level_map = ('document'=>'Doc',
|
---|
[4811] | 42 | 'section'=>'Sec',
|
---|
| 43 | 'paragraph'=>'Para',
|
---|
| 44 | 'Doc'=>'_textdocument_',
|
---|
| 45 | 'Sec'=>'_textsection_',
|
---|
| 46 | 'Para'=>'_textparagraph_');
|
---|
[1852] | 47 |
|
---|
[9157] | 48 | our %wanted_index_files = ('td'=>1,
|
---|
[932] | 49 | 't'=>1,
|
---|
[1852] | 50 | 'tl'=>1,
|
---|
| 51 | 'ti'=>1,
|
---|
[932] | 52 | 'idb'=>1,
|
---|
| 53 | 'ib1'=>1,
|
---|
| 54 | 'ib2'=>1,
|
---|
| 55 | 'ib3'=>1,
|
---|
[12910] | 56 | 'ib4'=>1,
|
---|
| 57 | 'ib5'=>1,
|
---|
| 58 | 'ib6'=>1,
|
---|
| 59 | 'ib7'=>1,
|
---|
[932] | 60 | 'i'=>1,
|
---|
[1852] | 61 | 'il'=>1,
|
---|
| 62 | 'w'=>1,
|
---|
[932] | 63 | 'wa'=>1);
|
---|
| 64 |
|
---|
[1852] | 65 | # change this so a user can add their own ones in via a file or cfg
|
---|
[4768] | 66 | #add AND, OR, NOT NEAR to this list - these cannot be used as field names
|
---|
[4811] | 67 | #also add the level names (Doc, Sec, Para)
|
---|
[9157] | 68 | our %static_indexfield_map = ('Title'=>'TI',
|
---|
[1852] | 69 | 'TI'=>1,
|
---|
| 70 | 'Subject'=>'SU',
|
---|
| 71 | 'SU'=>1,
|
---|
| 72 | 'Creator'=>'CR',
|
---|
| 73 | 'CR'=>1,
|
---|
[4768] | 74 | 'Organization'=>'ORG',
|
---|
| 75 | 'ORG'=>1,
|
---|
[1852] | 76 | 'Source'=>'SO',
|
---|
| 77 | 'SO'=>1,
|
---|
| 78 | 'Howto'=>'HT',
|
---|
| 79 | 'HT'=>1,
|
---|
| 80 | 'ItemTitle'=>'IT',
|
---|
| 81 | 'IT'=>1,
|
---|
| 82 | 'ProgNumber'=>'PN',
|
---|
| 83 | 'PN'=>1,
|
---|
| 84 | 'People'=>'PE',
|
---|
| 85 | 'PE'=>1,
|
---|
[5643] | 86 | 'Coverage'=>'CO',
|
---|
| 87 | 'CO'=>1,
|
---|
[4794] | 88 | 'allfields'=>'ZZ',
|
---|
[4768] | 89 | 'ZZ'=>1,
|
---|
[4794] | 90 | 'text'=>'TX',
|
---|
[4768] | 91 | 'TX'=>1,
|
---|
| 92 | 'AND'=>1,
|
---|
| 93 | 'OR'=>1,
|
---|
| 94 | 'NOT'=>1,
|
---|
[4811] | 95 | 'NEAR'=>1,
|
---|
| 96 | 'Doc'=>1,
|
---|
| 97 | 'Sec'=>1,
|
---|
| 98 | 'Para'=>1);
|
---|
[932] | 99 |
|
---|
[10468] | 100 | my $maxdocsize = $basebuilder::maxdocsize;
|
---|
| 101 |
|
---|
[932] | 102 | sub new {
|
---|
[7953] | 103 | my $class = shift(@_);
|
---|
| 104 |
|
---|
[10468] | 105 | my $self = new basebuilder (@_);
|
---|
| 106 | $self = bless $self, $class;
|
---|
[932] | 107 |
|
---|
[10468] | 108 | $self->{'indexfieldmap'} = \%static_indexfield_map;
|
---|
[6407] | 109 |
|
---|
[1852] | 110 | # get the levels (Section, Paragraph) for indexing and compression
|
---|
| 111 | $self->{'levels'} = {};
|
---|
[4811] | 112 | $self->{'levelorder'} = ();
|
---|
[1852] | 113 | if (defined $self->{'collect_cfg'}->{'levels'}) {
|
---|
[8716] | 114 | foreach my $level ( @{$self->{'collect_cfg'}->{'levels'}} ){
|
---|
[4811] | 115 | $level =~ tr/A-Z/a-z/;
|
---|
[1852] | 116 | $self->{'levels'}->{$level} = 1;
|
---|
[4811] | 117 | push (@{$self->{'levelorder'}}, $level);
|
---|
[1852] | 118 | }
|
---|
[4811] | 119 | } else { # default to document
|
---|
| 120 | $self->{'levels'}->{'document'} = 1;
|
---|
| 121 | push (@{$self->{'levelorder'}}, 'document');
|
---|
| 122 | }
|
---|
| 123 |
|
---|
[7953] | 124 | $self->{'buildtype'} = "mgpp";
|
---|
[932] | 125 |
|
---|
| 126 | return $self;
|
---|
| 127 | }
|
---|
| 128 |
|
---|
[10468] | 129 | sub generate_index_list {
|
---|
| 130 | my $self = shift (@_);
|
---|
| 131 |
|
---|
| 132 | # sort out the indexes
|
---|
| 133 | #indexes are specified with spaces, but we put them into one index
|
---|
| 134 | my $indexes = $self->{'collect_cfg'}->{'indexes'};
|
---|
| 135 | $self->{'collect_cfg'}->{'indexes'} = [];
|
---|
[13274] | 136 | push (@{$self->{'collect_cfg'}->{'indexes'}}, join(';', @$indexes).";");
|
---|
[932] | 137 | }
|
---|
| 138 |
|
---|
[12910] | 139 | sub generate_index_options {
|
---|
| 140 | my $self = shift (@_);
|
---|
| 141 |
|
---|
[17110] | 142 | $self->SUPER::generate_index_options();
|
---|
| 143 |
|
---|
[12910] | 144 | $self->{'casefold'} = 0;
|
---|
| 145 | $self->{'stem'} = 0;
|
---|
| 146 | $self->{'accentfold'} = 0;
|
---|
| 147 |
|
---|
| 148 | if (!defined($self->{'collect_cfg'}->{'indexoptions'})) {
|
---|
| 149 | # just use default options
|
---|
| 150 | $self->{'casefold'} = 1;
|
---|
| 151 | $self->{'stem'} = 1;
|
---|
| 152 | $self->{'accentfold'} = 1;
|
---|
| 153 | } else {
|
---|
| 154 | foreach my $option (@{$self->{'collect_cfg'}->{'indexoptions'}}) {
|
---|
| 155 | if ($option =~ /stem/) {
|
---|
| 156 | $self->{'stem'} = 1;
|
---|
| 157 | } elsif ($option =~ /casefold/) {
|
---|
| 158 | $self->{'casefold'} = 1;
|
---|
| 159 | } elsif ($option =~ /accentfold/) {
|
---|
| 160 | $self->{'accentfold'} = 1;
|
---|
| 161 | }
|
---|
| 162 | }
|
---|
| 163 | }
|
---|
| 164 |
|
---|
| 165 | # now we record this for the build cfg
|
---|
| 166 | $self->{'stemindexes'} = 0;
|
---|
| 167 | if ($self->{'casefold'}) {
|
---|
| 168 | $self->{'stemindexes'} += 1;
|
---|
| 169 | }
|
---|
| 170 | if ($self->{'stem'}) {
|
---|
| 171 | $self->{'stemindexes'} += 2;
|
---|
| 172 | }
|
---|
| 173 | if ($self->{'accentfold'}) {
|
---|
| 174 | $self->{'stemindexes'} += 4;
|
---|
| 175 | }
|
---|
[13341] | 176 |
|
---|
[12910] | 177 | }
|
---|
| 178 |
|
---|
[10468] | 179 | sub default_buildproc {
|
---|
| 180 | my $self = shift (@_);
|
---|
| 181 |
|
---|
| 182 | return "mgppbuildproc";
|
---|
[932] | 183 | }
|
---|
| 184 |
|
---|
| 185 | sub compress_text {
|
---|
| 186 |
|
---|
| 187 | my $self = shift (@_);
|
---|
[10961] | 188 |
|
---|
| 189 | # we don't do anything if we don't want compressed text
|
---|
| 190 | return if $self->{'no_text'};
|
---|
| 191 |
|
---|
[932] | 192 | my ($textindex) = @_;
|
---|
| 193 |
|
---|
[2478] | 194 | my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
|
---|
[932] | 195 | my $exe = &util::get_os_exe ();
|
---|
[2478] | 196 | my $mgpp_passes_exe = &util::filename_cat($exedir, "mgpp_passes$exe");
|
---|
| 197 | my $mgpp_compression_dict_exe = &util::filename_cat($exedir, "mgpp_compression_dict$exe");
|
---|
[1694] | 198 | my $outhandle = $self->{'outhandle'};
|
---|
[932] | 199 |
|
---|
[12340] | 200 | my $maxnumeric = $self->{'maxnumeric'};
|
---|
[12325] | 201 |
|
---|
[932] | 202 | &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, "text"));
|
---|
| 203 |
|
---|
[15003] | 204 | my $collect_tail = &util::get_dirsep_tail($self->{'collection'});
|
---|
| 205 | my $basefilename = &util::filename_cat("text",$collect_tail);
|
---|
[2700] | 206 | my $fulltextprefix = &util::filename_cat ($self->{'build_dir'}, $basefilename);
|
---|
[7904] | 207 |
|
---|
[15003] | 208 | my $osextra = "";
|
---|
[2478] | 209 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
---|
[3115] | 210 | $fulltextprefix =~ s@/@\\@g;
|
---|
[2478] | 211 | }
|
---|
[2700] | 212 | else {
|
---|
| 213 | $osextra = " -d /";
|
---|
| 214 | }
|
---|
[1852] | 215 |
|
---|
| 216 |
|
---|
[4811] | 217 | # define the section names and possibly the doc name for mgpasses
|
---|
[1852] | 218 | # the compressor doesn't need to know about paragraphs - never want to
|
---|
| 219 | # retrieve them
|
---|
[13590] | 220 |
|
---|
| 221 | # always use Doc and Sec levels
|
---|
| 222 | my $mgpp_passes_sections = "-J ". $level_map{"document"} ." -K " . $level_map{"section"} ." ";
|
---|
[12911] | 223 |
|
---|
[1694] | 224 | print $outhandle "\n*** creating the compressed text\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 225 | print STDERR "<Stage name='CompressText'>\n" if $self->{'gli'};
|
---|
[932] | 226 |
|
---|
| 227 | # collect the statistics for the text
|
---|
[1694] | 228 | # -b $maxdocsize sets the maximum document size to be 12 meg
|
---|
[2478] | 229 | print $outhandle "\n collecting text statistics (mgpp_passes -T1)\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 230 | print STDERR "<Phase name='CollectTextStats'/>\n" if $self->{'gli'};
|
---|
[932] | 231 |
|
---|
| 232 | my ($handle);
|
---|
| 233 | if ($self->{'debug'}) {
|
---|
[15715] | 234 | $handle = *STDOUT;
|
---|
| 235 | }
|
---|
| 236 | else {
|
---|
[2478] | 237 | if (!-e "$mgpp_passes_exe" ||
|
---|
[15715] | 238 | !open($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fulltextprefix\" -T1 $osextra")) {
|
---|
[6407] | 239 | print STDERR "<FatalError name='NoRunMGPasses'>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 240 | die "mgppbuilder::compress_text - couldn't run $mgpp_passes_exe\n";
|
---|
[932] | 241 | }
|
---|
| 242 | }
|
---|
[13590] | 243 |
|
---|
[15685] | 244 | my $db_level = "section";
|
---|
[9919] | 245 |
|
---|
[932] | 246 | $self->{'buildproc'}->set_output_handle ($handle);
|
---|
| 247 | $self->{'buildproc'}->set_mode ('text');
|
---|
| 248 | $self->{'buildproc'}->set_index ($textindex);
|
---|
| 249 | $self->{'buildproc'}->set_indexing_text (0);
|
---|
[1852] | 250 | $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
| 251 | $self->{'buildproc'}->set_levels ($self->{'levels'});
|
---|
[15685] | 252 | $self->{'buildproc'}->set_db_level ($db_level);
|
---|
[932] | 253 | $self->{'buildproc'}->reset();
|
---|
| 254 | &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
| 255 | $self->{'buildproc'}, $self->{'maxdocs'});
|
---|
| 256 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
[16379] | 257 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
[932] | 258 | &plugin::end($self->{'pluginfo'});
|
---|
| 259 |
|
---|
| 260 | close ($handle) unless $self->{'debug'};
|
---|
| 261 |
|
---|
[2478] | 262 | $self->print_stats();
|
---|
| 263 |
|
---|
[932] | 264 | # create the compression dictionary
|
---|
| 265 | # the compression dictionary is built by assuming the stats are from a seed
|
---|
| 266 | # dictionary (-S), if a novel word is encountered it is spelled out (-H),
|
---|
| 267 | # and the resulting dictionary must be less than 5 meg with the most
|
---|
| 268 | # frequent words being put into the dictionary first (-2 -k 5120)
|
---|
[1852] | 269 | # note: these options are left over from mg version
|
---|
[932] | 270 | if (!$self->{'debug'}) {
|
---|
[1694] | 271 | print $outhandle "\n creating the compression dictionary\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 272 | print STDERR "<Phase name='CreatingCompress'/>\n" if $self->{'gli'};
|
---|
[2478] | 273 | if (!-e "$mgpp_compression_dict_exe") {
|
---|
[6407] | 274 | print STDERR "<FatalError name='NoRunMGCompress'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 275 | die "mgppbuilder::compress_text - couldn't run $mgpp_compression_dict_exe\n";
|
---|
[932] | 276 | }
|
---|
[2700] | 277 | system ("mgpp_compression_dict$exe -f \"$fulltextprefix\" -S -H -2 -k 5120 $osextra");
|
---|
[932] | 278 |
|
---|
| 279 | if (!$self->{'debug'}) {
|
---|
[2478] | 280 | if (!-e "$mgpp_passes_exe" ||
|
---|
[12325] | 281 | !open ($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fulltextprefix\" -T2 $osextra")) {
|
---|
[6407] | 282 | print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 283 | die "mgppbuilder::compress_text - couldn't run $mgpp_passes_exe\n";
|
---|
[932] | 284 | }
|
---|
| 285 | }
|
---|
| 286 | }
|
---|
[6407] | 287 | else {
|
---|
| 288 | print STDERR "<Phase name='SkipCreatingComp'/>\n" if $self->{'gli'};
|
---|
| 289 | }
|
---|
[932] | 290 |
|
---|
| 291 | $self->{'buildproc'}->reset();
|
---|
| 292 | # compress the text
|
---|
[2478] | 293 | print $outhandle "\n compressing the text (mgpp_passes -T2)\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 294 | print STDERR "<Phase name='CompressingText'/>\n" if $self->{'gli'};
|
---|
| 295 |
|
---|
[932] | 296 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
[16379] | 297 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
[932] | 298 | close ($handle) unless $self->{'debug'};
|
---|
[1694] | 299 |
|
---|
| 300 | $self->print_stats();
|
---|
[6407] | 301 | print STDERR "</Stage>\n" if $self->{'gli'};
|
---|
[932] | 302 | }
|
---|
| 303 |
|
---|
| 304 |
|
---|
[10468] | 305 | sub build_indexes_extra {
|
---|
| 306 | my $self = shift(@_);
|
---|
[5617] | 307 | #define the final field lists
|
---|
| 308 | $self->make_final_field_list();
|
---|
[10468] | 309 | }
|
---|
[5617] | 310 |
|
---|
[932] | 311 | # creates directory names for each of the index descriptions
|
---|
| 312 | sub create_index_mapping {
|
---|
| 313 | my $self = shift (@_);
|
---|
| 314 | my ($indexes) = @_;
|
---|
| 315 |
|
---|
| 316 | my %mapping = ();
|
---|
[5935] | 317 |
|
---|
[932] | 318 | $mapping{'indexmaporder'} = [];
|
---|
| 319 | $mapping{'subcollectionmaporder'} = [];
|
---|
| 320 | $mapping{'languagemaporder'} = [];
|
---|
| 321 |
|
---|
| 322 | # dirnames is used to check for collisions. Start this off
|
---|
| 323 | # with the manditory directory names
|
---|
| 324 | my %dirnames = ('text'=>'text',
|
---|
| 325 | 'extra'=>'extra');
|
---|
[8716] | 326 | my %pnames = ('index' => {}, 'subcollection' => {}, 'languages' => {});
|
---|
[932] | 327 |
|
---|
[8716] | 328 | foreach my $index (@$indexes) {
|
---|
[932] | 329 | my ($fields, $subcollection, $languages) = split (":", $index);
|
---|
[13590] | 330 |
|
---|
| 331 | # we only ever have one index, and its called 'idx'
|
---|
[8716] | 332 | my $pindex = 'idx';
|
---|
[4768] | 333 |
|
---|
[932] | 334 | # next comes a processed version of the subcollection if there is one.
|
---|
| 335 | my $psub = $self->process_field ($subcollection);
|
---|
| 336 | $psub = lc ($psub);
|
---|
| 337 |
|
---|
| 338 | # next comes a processed version of the language if there is one.
|
---|
| 339 | my $plang = $self->process_field ($languages);
|
---|
| 340 | $plang = lc ($plang);
|
---|
| 341 |
|
---|
| 342 | my $dirname = $pindex . $psub . $plang;
|
---|
| 343 |
|
---|
| 344 | # check to be sure all index names are unique
|
---|
| 345 | while (defined ($dirnames{$dirname})) {
|
---|
| 346 | $dirname = $self->make_unique (\%pnames, $index, \$pindex, \$psub, \$plang);
|
---|
| 347 | }
|
---|
| 348 |
|
---|
[2478] | 349 | $mapping{$index} = $dirname;
|
---|
| 350 |
|
---|
[932] | 351 | # store the mapping orders as well as the maps
|
---|
[2478] | 352 | # also put index, subcollection and language fields into the mapping thing -
|
---|
[4794] | 353 | # (the full index name (eg text:subcol:lang) is not used on
|
---|
[2478] | 354 | # the query page) -these are used for collectionmeta later on
|
---|
[932] | 355 | if (!defined $mapping{'indexmap'}{"$fields"}) {
|
---|
| 356 | $mapping{'indexmap'}{"$fields"} = $pindex;
|
---|
| 357 | push (@{$mapping{'indexmaporder'}}, "$fields");
|
---|
[2478] | 358 | if (!defined $mapping{"$fields"}) {
|
---|
| 359 | $mapping{"$fields"} = $pindex;
|
---|
| 360 | }
|
---|
[932] | 361 | }
|
---|
| 362 | if ($psub =~ /\w/ && !defined ($mapping{'subcollectionmap'}{$subcollection})) {
|
---|
| 363 | $mapping{'subcollectionmap'}{$subcollection} = $psub;
|
---|
| 364 | push (@{$mapping{'subcollectionmaporder'}}, $subcollection);
|
---|
[2478] | 365 | $mapping{$subcollection} = $psub;
|
---|
[932] | 366 | }
|
---|
| 367 | if ($plang =~ /\w/ && !defined ($mapping{'languagemap'}{$languages})) {
|
---|
| 368 | $mapping{'languagemap'}{$languages} = $plang;
|
---|
[6544] | 369 | push (@{$mapping{'languagemaporder'}}, $languages);
|
---|
[2478] | 370 | $mapping{$languages} = $plang;
|
---|
[932] | 371 | }
|
---|
| 372 | $dirnames{$dirname} = $index;
|
---|
[8716] | 373 | $pnames{'index'}->{$pindex} = "$fields";
|
---|
| 374 | $pnames{'subcollection'}->{$psub} = $subcollection;
|
---|
| 375 | $pnames{'languages'}->{$plang} = $languages;
|
---|
[932] | 376 | }
|
---|
| 377 |
|
---|
| 378 | return \%mapping;
|
---|
| 379 | }
|
---|
| 380 |
|
---|
| 381 | sub make_unique {
|
---|
| 382 | my $self = shift (@_);
|
---|
| 383 | my ($namehash, $index, $indexref, $subref, $langref) = @_;
|
---|
| 384 | my ($fields, $subcollection, $languages) = split (":", $index);
|
---|
| 385 |
|
---|
| 386 | if ($namehash->{'index'}->{$$indexref} ne "$fields") {
|
---|
| 387 | $self->get_next_version ($indexref);
|
---|
| 388 | } elsif ($namehash->{'subcollection'}->{$$subref} ne $subcollection) {
|
---|
| 389 | $self->get_next_version ($subref);
|
---|
| 390 | } elsif ($namehash->{'languages'}->{$$langref} ne $languages) {
|
---|
| 391 | $self->get_next_version ($langref);
|
---|
| 392 | }
|
---|
| 393 | return "$$indexref$$subref$$langref";
|
---|
| 394 | }
|
---|
| 395 |
|
---|
| 396 |
|
---|
| 397 | sub build_index {
|
---|
| 398 | my $self = shift (@_);
|
---|
| 399 | my ($index) = @_;
|
---|
[1694] | 400 | my $outhandle = $self->{'outhandle'};
|
---|
[932] | 401 |
|
---|
| 402 | # get the full index directory path and make sure it exists
|
---|
| 403 | my $indexdir = $self->{'index_mapping'}->{$index};
|
---|
| 404 | &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, $indexdir));
|
---|
[15003] | 405 |
|
---|
| 406 | my $collect_tail = &util::get_dirsep_tail($self->{'collection'});
|
---|
[2700] | 407 | my $fullindexprefix = &util::filename_cat ($self->{'build_dir'},
|
---|
| 408 | $indexdir,
|
---|
[15003] | 409 | $collect_tail);
|
---|
[2700] | 410 | my $fulltextprefix = &util::filename_cat ($self->{'build_dir'}, "text",
|
---|
[15003] | 411 | $collect_tail);
|
---|
[932] | 412 |
|
---|
| 413 | # get any os specific stuff
|
---|
[2478] | 414 | my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
|
---|
[932] | 415 |
|
---|
| 416 | my $exe = &util::get_os_exe ();
|
---|
[2478] | 417 | my $mgpp_passes_exe = &util::filename_cat($exedir, "mgpp_passes$exe");
|
---|
[1852] | 418 |
|
---|
| 419 | # define the section names for mgpasses
|
---|
[13590] | 420 | my $mgpp_passes_sections = "-J ". $level_map{"document"} ." -K " . $level_map{"section"} ." ";
|
---|
| 421 | if ($self->{'levels'}->{'paragraph'}) {
|
---|
| 422 | $mgpp_passes_sections .= "-K " . $level_map{'paragraph'}. " ";
|
---|
[1852] | 423 | }
|
---|
| 424 |
|
---|
[2478] | 425 | my $mgpp_perf_hash_build_exe =
|
---|
| 426 | &util::filename_cat($exedir, "mgpp_perf_hash_build$exe");
|
---|
| 427 | my $mgpp_weights_build_exe =
|
---|
| 428 | &util::filename_cat ($exedir, "mgpp_weights_build$exe");
|
---|
| 429 | my $mgpp_invf_dict_exe =
|
---|
| 430 | &util::filename_cat ($exedir, "mgpp_invf_dict$exe");
|
---|
| 431 | my $mgpp_stem_idx_exe =
|
---|
| 432 | &util::filename_cat ($exedir, "mgpp_stem_idx$exe");
|
---|
[932] | 433 |
|
---|
[12340] | 434 | my $maxnumeric = $self->{'maxnumeric'};
|
---|
[12325] | 435 |
|
---|
| 436 | my $osextra = "";
|
---|
[2700] | 437 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
---|
[3115] | 438 | $fullindexprefix =~ s@/@\\@g;
|
---|
[2700] | 439 | } else {
|
---|
| 440 | $osextra = " -d /";
|
---|
[3115] | 441 | if ($outhandle ne "STDERR") {
|
---|
| 442 | # so mgpp_passes doesn't print to stderr if we redirect output
|
---|
| 443 | $osextra .= " 2>/dev/null";
|
---|
| 444 | }
|
---|
[2478] | 445 | }
|
---|
[2700] | 446 |
|
---|
[932] | 447 | # get the index expression if this index belongs
|
---|
| 448 | # to a subcollection
|
---|
| 449 | my $indexexparr = [];
|
---|
[9669] | 450 | my $langarr = [];
|
---|
[2478] | 451 | # there may be subcollection info, and language info.
|
---|
| 452 | my ($fields, $subcollection, $language) = split (":", $index);
|
---|
[932] | 453 | my @subcollections = ();
|
---|
| 454 | @subcollections = split /,/, $subcollection if (defined $subcollection);
|
---|
| 455 |
|
---|
| 456 | foreach $subcollection (@subcollections) {
|
---|
| 457 | if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
|
---|
| 458 | push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
|
---|
| 459 | }
|
---|
| 460 | }
|
---|
| 461 |
|
---|
| 462 | # add expressions for languages if this index belongs to
|
---|
[2478] | 463 | # a language subcollection - only put languages expressions for the
|
---|
| 464 | # ones we want in the index
|
---|
[6544] | 465 |
|
---|
[2478] | 466 | my @languages = ();
|
---|
[9548] | 467 | my $language_metadata = "Language";
|
---|
| 468 | if (defined ($self->{'collect_cfg'}->{'language_metadata'})) {
|
---|
| 469 | $language_metadata = $self->{'collect_cfg'}->{'language_metadata'};
|
---|
| 470 | }
|
---|
[2478] | 471 | @languages = split /,/, $language if (defined $language);
|
---|
[9548] | 472 | foreach my $language (@languages) {
|
---|
[2478] | 473 | my $not=0;
|
---|
[932] | 474 | if ($language =~ s/^\!//) {
|
---|
[2478] | 475 | $not = 1;
|
---|
[932] | 476 | }
|
---|
[9548] | 477 | if($not) {
|
---|
[9669] | 478 | push (@$langarr, "!$language");
|
---|
[6544] | 479 | } else {
|
---|
[9669] | 480 | push (@$langarr, "$language");
|
---|
[2478] | 481 | }
|
---|
[932] | 482 | }
|
---|
| 483 |
|
---|
| 484 | # Build index dictionary. Uses verbatim stem method
|
---|
[2478] | 485 | print $outhandle "\n creating index dictionary (mgpp_passes -I1)\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 486 | print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
|
---|
[932] | 487 | my ($handle);
|
---|
| 488 | if ($self->{'debug'}) {
|
---|
[15715] | 489 | $handle = *STDOUT;
|
---|
| 490 | }
|
---|
| 491 | else {
|
---|
[2478] | 492 | if (!-e "$mgpp_passes_exe" ||
|
---|
[15715] | 493 | !open($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fullindexprefix\" -I1 $osextra")) {
|
---|
[6407] | 494 | print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 495 | die "mgppbuilder::build_index - couldn't run $mgpp_passes_exe\n";
|
---|
[932] | 496 | }
|
---|
| 497 | }
|
---|
[9919] | 498 |
|
---|
[15685] | 499 | # db_level is always section
|
---|
| 500 | my $db_level = "section";
|
---|
[9919] | 501 |
|
---|
[4794] | 502 | # set up the document processr
|
---|
[932] | 503 | $self->{'buildproc'}->set_output_handle ($handle);
|
---|
| 504 | $self->{'buildproc'}->set_mode ('text');
|
---|
| 505 | $self->{'buildproc'}->set_index ($index, $indexexparr);
|
---|
[9669] | 506 | $self->{'buildproc'}->set_index_languages ($language_metadata, $langarr) if (defined $language);
|
---|
[932] | 507 | $self->{'buildproc'}->set_indexing_text (1);
|
---|
[1852] | 508 | $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
[9919] | 509 | $self->{'buildproc'}->set_levels ($self->{'levels'});
|
---|
[15685] | 510 | $self->{'buildproc'}->set_db_level ($db_level);
|
---|
[9919] | 511 |
|
---|
[932] | 512 | $self->{'buildproc'}->reset();
|
---|
| 513 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
[16379] | 514 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
[932] | 515 | close ($handle) unless $self->{'debug'};
|
---|
| 516 |
|
---|
[1694] | 517 | $self->print_stats();
|
---|
| 518 |
|
---|
[5768] | 519 | # now we check to see if the required files have been produced - if not we quit building this index so the whole process doesn't crap out.
|
---|
| 520 | # we check on the .id file - index dictionary
|
---|
| 521 | my $dict_file = "$fullindexprefix.id";
|
---|
| 522 | if (!-e $dict_file) {
|
---|
| 523 | print $outhandle "mgppbuilder::build_index - Couldn't create index $index\n";
|
---|
[6407] | 524 | print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[5768] | 525 | $self->{'notbuilt'}->{$index}=1;
|
---|
| 526 | return;
|
---|
| 527 | }
|
---|
| 528 |
|
---|
[932] | 529 | if (!$self->{'debug'}) {
|
---|
| 530 | # create the perfect hash function
|
---|
[2478] | 531 | if (!-e "$mgpp_perf_hash_build_exe") {
|
---|
[6407] | 532 | print STDERR "<FatalError name='NoRunMGHash'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 533 | die "mgppbuilder::build_index - couldn't run $mgpp_perf_hash_build_exe\n";
|
---|
[932] | 534 | }
|
---|
[2700] | 535 | system ("mgpp_perf_hash_build$exe -f \"$fullindexprefix\" $osextra");
|
---|
[932] | 536 |
|
---|
[2478] | 537 | if (!-e "$mgpp_passes_exe" ||
|
---|
[12325] | 538 | !open ($handle, "| mgpp_passes$exe -M $maxnumeric $mgpp_passes_sections -f \"$fullindexprefix\" -I2 $osextra")) {
|
---|
[6407] | 539 | print STDERR "<FatalError name='NoRunMGPasses'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 540 | die "mgppbuilder::build_index - couldn't run $mgpp_passes_exe\n";
|
---|
[932] | 541 | }
|
---|
| 542 | }
|
---|
| 543 |
|
---|
| 544 | # invert the text
|
---|
[2478] | 545 | print $outhandle "\n inverting the text (mgpp_passes -I2)\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 546 | print STDERR "<Phase name='InvertingText'/>\n" if $self->{'gli'};
|
---|
[932] | 547 | $self->{'buildproc'}->reset();
|
---|
| 548 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
---|
[16379] | 549 | "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
---|
[1694] | 550 |
|
---|
| 551 | $self->print_stats ();
|
---|
[932] | 552 |
|
---|
| 553 | if (!$self->{'debug'}) {
|
---|
| 554 |
|
---|
| 555 | close ($handle);
|
---|
| 556 |
|
---|
| 557 | # create the weights file
|
---|
[1694] | 558 | print $outhandle "\n create the weights file\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 559 | print STDERR "<Phase name='CreateTheWeights'/>\n" if $self->{'gli'};
|
---|
[2478] | 560 | if (!-e "$mgpp_weights_build_exe") {
|
---|
[6407] | 561 | print STDERR "<FatalError name='NoRunMGWeights'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 562 | die "mgppbuilder::build_index - couldn't run $mgpp_weights_build_exe\n";
|
---|
[932] | 563 | }
|
---|
[2700] | 564 | system ("mgpp_weights_build$exe -f \"$fullindexprefix\" $osextra");
|
---|
[932] | 565 |
|
---|
| 566 | # create 'on-disk' stemmed dictionary
|
---|
[1694] | 567 | print $outhandle "\n creating 'on-disk' stemmed dictionary\n" if ($self->{'verbosity'} >= 1);
|
---|
[2478] | 568 | if (!-e "$mgpp_invf_dict_exe") {
|
---|
[6407] | 569 | print STDERR "<FatalError name='NoRunMGInvf'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 570 | die "mgppbuilder::build_index - couldn't run $mgpp_invf_dict_exe\n";
|
---|
[932] | 571 | }
|
---|
[2700] | 572 | system ("mgpp_invf_dict$exe -f \"$fullindexprefix\" $osextra" );
|
---|
[932] | 573 |
|
---|
| 574 |
|
---|
| 575 | # creates stem index files for the various stemming methods
|
---|
[1694] | 576 | print $outhandle "\n creating stem indexes\n" if ($self->{'verbosity'} >= 1);
|
---|
[6407] | 577 | print STDERR "<Phase name='CreatingStemIndx'/>\n" if $self->{'gli'};
|
---|
[2478] | 578 | if (!-e "$mgpp_stem_idx_exe") {
|
---|
[6407] | 579 | print STDERR "<FatalError name='NoRunMGStem'/>\n</Stage>\n" if $self->{'gli'};
|
---|
[2478] | 580 | die "mgppbuilder::build_index - couldn't run $mgpp_stem_idx_exe\n";
|
---|
[932] | 581 | }
|
---|
[12910] | 582 | my $accent_folding_enabled = 1;
|
---|
| 583 | if ($self->{'accentfold'}) {
|
---|
| 584 | # the first time we do this, we test for accent folding enabled
|
---|
[13813] | 585 | if (system ("mgpp_stem_idx$exe -b 4096 -s4 -f \"$fullindexprefix\" $osextra") == 2) {
|
---|
[12910] | 586 | # accent folding has not been enabled in mgpp
|
---|
| 587 | $accent_folding_enabled = 0;
|
---|
| 588 | $self->{'stemindexes'} -= 4;
|
---|
| 589 | }
|
---|
| 590 | }
|
---|
| 591 | if ($self->{'casefold'}) {
|
---|
| 592 | system ("mgpp_stem_idx$exe -b 4096 -s1 -f \"$fullindexprefix\" $osextra");
|
---|
| 593 | if ($accent_folding_enabled && $self->{'accentfold'}) {
|
---|
| 594 | system ("mgpp_stem_idx$exe -b 4096 -s5 -f \"$fullindexprefix\" $osextra");
|
---|
| 595 | }
|
---|
| 596 | }
|
---|
| 597 | if ($self->{'stem'}) {
|
---|
| 598 | system ("mgpp_stem_idx$exe -b 4096 -s2 -f \"$fullindexprefix\" $osextra");
|
---|
| 599 | if ($accent_folding_enabled && $self->{'accentfold'}) {
|
---|
| 600 | system ("mgpp_stem_idx$exe -b 4096 -s6 -f \"$fullindexprefix\" $osextra");
|
---|
| 601 | }
|
---|
| 602 | }
|
---|
| 603 | if ($self->{'casefold'} && $self->{'stem'}) {
|
---|
| 604 | system ("mgpp_stem_idx$exe -b 4096 -s3 -f \"$fullindexprefix\" $osextra");
|
---|
| 605 | if ($accent_folding_enabled && $self->{'accentfold'}) {
|
---|
| 606 | system ("mgpp_stem_idx$exe -b 4096 -s7 -f \"$fullindexprefix\" $osextra");
|
---|
| 607 | }
|
---|
| 608 | }
|
---|
| 609 |
|
---|
[932] | 610 | # remove unwanted files
|
---|
[1852] | 611 | my $tmpdir = &util::filename_cat ($self->{'build_dir'}, $indexdir);
|
---|
| 612 | opendir (DIR, $tmpdir) || die
|
---|
| 613 | "mgppbuilder::build_index - couldn't read directory $tmpdir\n";
|
---|
[8716] | 614 | foreach my $file (readdir(DIR)) {
|
---|
[1852] | 615 | next if $file =~ /^\./;
|
---|
| 616 | my ($suffix) = $file =~ /\.([^\.]+)$/;
|
---|
| 617 | if (defined $suffix && !defined $wanted_index_files{$suffix}) {
|
---|
[932] | 618 | # delete it!
|
---|
[1852] | 619 | print $outhandle "deleting $file\n" if $self->{'verbosity'} > 2;
|
---|
[2772] | 620 | #&util::rm (&util::filename_cat ($tmpdir, $file));
|
---|
[1852] | 621 | }
|
---|
| 622 | }
|
---|
| 623 | closedir (DIR);
|
---|
[4794] | 624 | }
|
---|
[6407] | 625 | print STDERR "</Stage>\n" if $self->{'gli'};
|
---|
[932] | 626 | }
|
---|
| 627 |
|
---|
[15709] | 628 |
|
---|
| 629 | sub get_collection_meta_indexes
|
---|
| 630 | {
|
---|
[10468] | 631 | my $self = shift(@_);
|
---|
[15709] | 632 | my $collection_infodb = shift(@_);
|
---|
[1694] | 633 |
|
---|
[10477] | 634 | # define the indexed field mapping if not already done so (ie if infodb called separately from build_index)
|
---|
| 635 | if (!defined $self->{'build_cfg'}) {
|
---|
| 636 | $self->read_final_field_list();
|
---|
| 637 | }
|
---|
| 638 |
|
---|
[4794] | 639 | # first do the collection meta stuff - everything without a dot
|
---|
| 640 | my $collmetadefined = 0;
|
---|
[10468] | 641 | my $metadata_entry;
|
---|
[932] | 642 | if (defined $self->{'collect_cfg'}->{'collectionmeta'}) {
|
---|
[4794] | 643 | $collmetadefined = 1;
|
---|
| 644 | }
|
---|
[11965] | 645 |
|
---|
[4811] | 646 | #add the index field macros to [collection]
|
---|
[4794] | 647 | # eg <TI>Title
|
---|
| 648 | # <SU>Subject
|
---|
| 649 | # these now come from collection meta. if that is not defined, usses the metadata name
|
---|
[8716] | 650 | my $collmeta = "";
|
---|
| 651 | foreach my $longfield (@{$self->{'build_cfg'}->{'indexfields'}}){
|
---|
| 652 | my $shortfield = $self->{'buildproc'}->{'indexfieldmap'}->{$longfield};
|
---|
[4794] | 653 | next if $shortfield eq 1;
|
---|
| 654 |
|
---|
[11965] | 655 | # we need to check if some coll meta has been defined - don't output
|
---|
| 656 | # any that have
|
---|
[8716] | 657 | $collmeta = ".$longfield";
|
---|
[11965] | 658 | if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{$collmeta}) {
|
---|
[4794] | 659 | if ($longfield eq "allfields") {
|
---|
[15709] | 660 | $collection_infodb->{$shortfield} = [ "_query:textallfields_" ];
|
---|
[4794] | 661 | } elsif ($longfield eq "text") {
|
---|
[15709] | 662 | $collection_infodb->{$shortfield} = [ "_query:texttextonly_" ];
|
---|
[4794] | 663 | } else {
|
---|
[15709] | 664 | $collection_infodb->{$shortfield} = [ $longfield ];
|
---|
[4794] | 665 | }
|
---|
[932] | 666 | }
|
---|
| 667 | }
|
---|
[2772] | 668 |
|
---|
[4811] | 669 | # now add the level names
|
---|
[8716] | 670 | my $level_entry = "";
|
---|
| 671 | foreach my $level (@{$self->{'collect_cfg'}->{'levels'}}) {
|
---|
| 672 | $collmeta = ".$level"; # based on the original specification
|
---|
[4811] | 673 | $level =~ tr/A-Z/a-z/; # make it lower case
|
---|
[7090] | 674 | my $levelid = $level_map{$level}; # find the actual value we used in the index
|
---|
[11965] | 675 | if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{$collmeta}) {
|
---|
[4811] | 676 | # use the default macro
|
---|
[15709] | 677 | $collection_infodb->{$levelid} = [ $level_map{$levelid} ];
|
---|
[4811] | 678 | }
|
---|
| 679 | }
|
---|
[5935] | 680 |
|
---|
| 681 | # now add subcoll meta
|
---|
[8716] | 682 | my $subcoll_entry = "";
|
---|
| 683 | my $shortname = "";
|
---|
| 684 | my $one_entry = "";
|
---|
| 685 | foreach my $subcoll (@{$self->{'index_mapping'}->{'subcollectionmaporder'}}) {
|
---|
[11965] | 686 | $shortname = $self->{'index_mapping'}->{$subcoll};
|
---|
| 687 | if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{".$subcoll"}) {
|
---|
[15709] | 688 | $collection_infodb->{$shortname} = [ $subcoll ];
|
---|
[5935] | 689 | }
|
---|
| 690 | }
|
---|
[10158] | 691 |
|
---|
| 692 | # now add language meta
|
---|
[8716] | 693 | my $lang_entry = "";
|
---|
| 694 | foreach my $lang (@{$self->{'index_mapping'}->{'languagemaporder'}}) {
|
---|
[11965] | 695 | $shortname = $self->{'index_mapping'}->{$lang};
|
---|
| 696 | if (!$collmetadefined || !defined $self->{'collect_cfg'}->{'collectionmeta'}->{".$lang"}) {
|
---|
[15709] | 697 | $collection_infodb->{$shortname} = [ $lang ];
|
---|
[6544] | 698 | }
|
---|
| 699 | }
|
---|
[15709] | 700 | }
|
---|
[932] | 701 |
|
---|
[15709] | 702 |
|
---|
| 703 | # default is to output the metadata sets (prefixes) used in collection
|
---|
| 704 | sub output_collection_meta
|
---|
| 705 | {
|
---|
| 706 | my $self = shift(@_);
|
---|
| 707 | my $infodb_handle = shift(@_);
|
---|
| 708 |
|
---|
| 709 | my %collection_infodb = ();
|
---|
| 710 | $self->get_collection_meta_sets(\%collection_infodb);
|
---|
| 711 | $self->get_collection_meta_indexes(\%collection_infodb);
|
---|
[15725] | 712 | &dbutil::write_infodb_entry($self->{'infodbtype'}, $infodb_handle, "collection", \%collection_infodb);
|
---|
[932] | 713 | }
|
---|
[7150] | 714 |
|
---|
[15709] | 715 |
|
---|
[10158] | 716 | # at the end of building, we have an indexfieldmap with all the mappings,
|
---|
| 717 | # plus some extras, and indexmap with any indexes in it that weren't
|
---|
| 718 | # specified in the index definition. we want to make an ordered list of
|
---|
| 719 | # fields that are indexed, and a list of mappings that are used. this will
|
---|
| 720 | # be used for the build.cfg file, and for collection meta definition we
|
---|
| 721 | # store these in a build.cfg bit
|
---|
[4794] | 722 | sub make_final_field_list {
|
---|
| 723 | my $self = shift (@_);
|
---|
| 724 |
|
---|
| 725 | $self->{'build_cfg'} = {};
|
---|
[10158] | 726 |
|
---|
[4794] | 727 | # store the indexfieldmap information
|
---|
| 728 | my @indexfieldmap = ();
|
---|
| 729 | my @indexfields = ();
|
---|
| 730 | my $specifiedfields = {};
|
---|
| 731 | my @specifiedfieldorder = ();
|
---|
[10158] | 732 |
|
---|
| 733 | # go through the index definition and add each thing to a map, so we
|
---|
| 734 | # can easily check if it is already specified - when doing the
|
---|
| 735 | # metadata, we print out all the individual fields, but some may
|
---|
| 736 | # already be specified in the index definition, so we dont want to add
|
---|
| 737 | # those again.
|
---|
| 738 |
|
---|
[10468] | 739 | my $field;
|
---|
| 740 | foreach $field (@{$self->{'collect_cfg'}->{'indexes'}}) {
|
---|
[5617] | 741 | # remove subcoll stuff
|
---|
| 742 | my $parts = $field;
|
---|
| 743 | $parts =~ s/:.*$//;
|
---|
[10961] | 744 | # *************
|
---|
| 745 | my @fs = split(';', $parts);
|
---|
[8716] | 746 | foreach my $f(@fs) {
|
---|
[5617] | 747 | if (!defined $specifiedfields->{$f}) {
|
---|
| 748 | $specifiedfields->{$f}=1;
|
---|
| 749 | push (@specifiedfieldorder, "$f");
|
---|
| 750 | }
|
---|
[4794] | 751 | }
|
---|
| 752 | }
|
---|
[5643] | 753 |
|
---|
[4794] | 754 | #add all fields bit
|
---|
[10468] | 755 | foreach $field (@specifiedfieldorder) {
|
---|
[4794] | 756 | if ($field eq "metadata") {
|
---|
[8716] | 757 | foreach my $newfield (keys %{$self->{'buildproc'}->{'indexfields'}}) {
|
---|
[4794] | 758 | if (!defined $specifiedfields->{$newfield}) {
|
---|
| 759 | push (@indexfieldmap, "$newfield\-\>$self->{'buildproc'}->{'indexfieldmap'}->{$newfield}");
|
---|
| 760 | push (@indexfields, "$newfield");
|
---|
| 761 | }
|
---|
| 762 | }
|
---|
| 763 |
|
---|
| 764 | } elsif ($field eq 'text') {
|
---|
| 765 | push (@indexfieldmap, "text\-\>TX");
|
---|
| 766 | push (@indexfields, "text");
|
---|
| 767 | } elsif ($field eq 'allfields') {
|
---|
| 768 | push (@indexfieldmap, "allfields\-\>ZZ");
|
---|
| 769 | push (@indexfields, "allfields");
|
---|
| 770 | } else {
|
---|
[11996] | 771 |
|
---|
| 772 | my $ifm = $self->{'buildproc'}->{'indexfieldmap'};
|
---|
| 773 |
|
---|
| 774 | if (defined $ifm->{$field}) {
|
---|
| 775 | push (@indexfieldmap, "$field\-\>$ifm->{$field}");
|
---|
| 776 | push (@indexfields, "$field");
|
---|
| 777 | }
|
---|
| 778 |
|
---|
[4794] | 779 |
|
---|
| 780 | }
|
---|
| 781 | }
|
---|
[10158] | 782 |
|
---|
[4794] | 783 | $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
|
---|
| 784 | $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
|
---|
[10961] | 785 |
|
---|
[4794] | 786 | }
|
---|
| 787 |
|
---|
| 788 |
|
---|
[10158] | 789 | # recreate the field list from the build.cfg file, look first in building,
|
---|
| 790 | # then in index to find it. if there is no build.cfg, we can't do the field
|
---|
| 791 | # list (there is unlikely to be any index anyway.)
|
---|
[4794] | 792 | sub read_final_field_list {
|
---|
| 793 | my $self = shift (@_);
|
---|
| 794 | $self->{'build_cfg'} = {};
|
---|
| 795 | my @indexfieldmap = ();
|
---|
| 796 | my @indexfields = ();
|
---|
[14666] | 797 | my @indexmap = ();
|
---|
| 798 |
|
---|
[4794] | 799 | if (scalar(keys %{$self->{'buildproc'}->{'indexfieldmap'}}) == 0) {
|
---|
| 800 | # set the default mapping
|
---|
| 801 | $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
---|
| 802 | }
|
---|
| 803 | # we read the stuff in from the build.cfg file - if its there
|
---|
[8716] | 804 | my $buildconfigfile = &util::filename_cat($self->{'build_dir'}, "build.cfg");
|
---|
[4794] | 805 |
|
---|
| 806 | if (!-e $buildconfigfile) {
|
---|
| 807 | # try the index dir - but do we know where it is?? try here
|
---|
| 808 | $buildconfigfile = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "index", "build.cfg");
|
---|
| 809 | if (!-e $buildconfigfile) {
|
---|
| 810 | #we cant find a config file - just ignore the field list
|
---|
| 811 | return;
|
---|
| 812 | }
|
---|
| 813 | }
|
---|
[10158] | 814 |
|
---|
[8716] | 815 | my $buildcfg = &colcfg::read_build_cfg( $buildconfigfile);
|
---|
[10468] | 816 | my $field;
|
---|
[4794] | 817 | if (defined $buildcfg->{'indexfields'}) {
|
---|
[10468] | 818 | foreach $field (@{$buildcfg->{'indexfields'}}) {
|
---|
[4794] | 819 | push (@indexfields, "$field");
|
---|
| 820 | }
|
---|
| 821 | }
|
---|
[10158] | 822 |
|
---|
[4794] | 823 | if (defined $buildcfg->{'indexfieldmap'}) {
|
---|
[10468] | 824 | foreach $field (@{$buildcfg->{'indexfieldmap'}}) {
|
---|
[4794] | 825 | push (@indexfieldmap, "$field");
|
---|
[8716] | 826 | my ($f, $v) = $field =~ /^(.*)\-\>(.*)$/;
|
---|
[4794] | 827 | $self->{'buildproc'}->{'indexfieldmap'}->{$f} = $v;
|
---|
| 828 | }
|
---|
| 829 | }
|
---|
[10158] | 830 |
|
---|
[14666] | 831 | if (defined $buildcfg->{'indexmap'}) {
|
---|
| 832 | foreach $field (@{$buildcfg->{'indexmap'}}) {
|
---|
| 833 | push (@indexmap, "$field");
|
---|
| 834 | }
|
---|
| 835 | }
|
---|
| 836 |
|
---|
[4794] | 837 | $self->{'build_cfg'}->{'indexfieldmap'} = \@indexfieldmap;
|
---|
| 838 | $self->{'build_cfg'}->{'indexfields'} = \@indexfields;
|
---|
[14666] | 839 | $self->{'build_cfg'}->{'indexmap'} = \@indexmap;
|
---|
[4794] | 840 | }
|
---|
[10158] | 841 |
|
---|
[10468] | 842 |
|
---|
| 843 | sub build_cfg_extra {
|
---|
[932] | 844 | my $self = shift (@_);
|
---|
[10468] | 845 | my ($build_cfg) = @_;
|
---|
| 846 |
|
---|
| 847 | $build_cfg->{'numsections'} = $self->{'buildproc'}->get_num_sections();
|
---|
[4794] | 848 |
|
---|
[4811] | 849 | # store the level info
|
---|
| 850 | my @indexlevels = ();
|
---|
[9936] | 851 | my @levelmap = ();
|
---|
[8716] | 852 | foreach my $l (@{$self->{'levelorder'}}) {
|
---|
[7090] | 853 | push (@indexlevels, $level_map{$l});
|
---|
[9936] | 854 | push (@levelmap, "$l\-\>$level_map{$l}");
|
---|
[4811] | 855 | }
|
---|
| 856 | $build_cfg->{'indexlevels'} = \@indexlevels;
|
---|
[9936] | 857 | $build_cfg->{'levelmap'} = \@levelmap;
|
---|
| 858 |
|
---|
[15687] | 859 | # text level (and database level) is always section
|
---|
[13590] | 860 | $build_cfg->{'textlevel'} = $level_map{'section'};
|
---|
[10468] | 861 |
|
---|
[932] | 862 | }
|
---|
| 863 |
|
---|
| 864 | 1;
|
---|
| 865 |
|
---|
| 866 |
|
---|