| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # lucenebuilder.pm -- perl wrapper for building index with Lucene
|
|---|
| 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 lucenebuilder;
|
|---|
| 27 |
|
|---|
| 28 | # Use same basic XML structure setup by mgppbuilder/mgppbuildproc
|
|---|
| 29 |
|
|---|
| 30 | use mgppbuilder;
|
|---|
| 31 |
|
|---|
| 32 | sub BEGIN {
|
|---|
| 33 | @lucenebuilder::ISA = ('mgppbuilder');
|
|---|
| 34 | }
|
|---|
| 35 |
|
|---|
| 36 |
|
|---|
| 37 | sub new {
|
|---|
| 38 | my $class = shift(@_);
|
|---|
| 39 | my ($collection, $source_dir, $build_dir, $verbosity,
|
|---|
| 40 | $maxdocs, $debug, $keepold, $allclassifications,
|
|---|
| 41 | $outhandle, $no_text, $gli) = @_;
|
|---|
| 42 |
|
|---|
| 43 | my $self = new mgppbuilder (@_);
|
|---|
| 44 | $self = bless $self, $class;
|
|---|
| 45 |
|
|---|
| 46 | $self->{'buildtype'} = "lucene";
|
|---|
| 47 |
|
|---|
| 48 | return $self;
|
|---|
| 49 | }
|
|---|
| 50 |
|
|---|
| 51 | sub default_buildproc {
|
|---|
| 52 | my $self = shift (@_);
|
|---|
| 53 |
|
|---|
| 54 | return "lucenebuildproc";
|
|---|
| 55 | }
|
|---|
| 56 |
|
|---|
| 57 | # this writes a nice version of the text docs
|
|---|
| 58 | sub compress_text {
|
|---|
| 59 |
|
|---|
| 60 | my $self = shift (@_);
|
|---|
| 61 | my ($textindex) = @_;
|
|---|
| 62 | my $outhandle = $self->{'outhandle'};
|
|---|
| 63 | print STDERR "Saving the document text\n";
|
|---|
| 64 | # the text directory
|
|---|
| 65 | my $text_dir = &util::filename_cat($self->{'build_dir'}, "text");
|
|---|
| 66 | my $build_dir = &util::filename_cat($self->{'build_dir'},"");
|
|---|
| 67 | &util::mk_all_dir ($text_dir);
|
|---|
| 68 |
|
|---|
| 69 | my $osextra = "";
|
|---|
| 70 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
|---|
| 71 | $text_dir =~ s@/@\\@g;
|
|---|
| 72 | } else {
|
|---|
| 73 | if ($outhandle ne "STDERR") {
|
|---|
| 74 | # so lucene_passes doesn't print to stderr if we redirect output
|
|---|
| 75 | $osextra .= " 2>/dev/null";
|
|---|
| 76 | }
|
|---|
| 77 | }
|
|---|
| 78 |
|
|---|
| 79 |
|
|---|
| 80 | # get any os specific stuff
|
|---|
| 81 | my $scriptdir = "$ENV{'GSDLHOME'}/bin/script";
|
|---|
| 82 |
|
|---|
| 83 | my $lucene_passes_exe = &util::filename_cat($scriptdir, "lucene_passes.pl");
|
|---|
| 84 | my $full_lucene_passes_exe = $lucene_passes_exe;
|
|---|
| 85 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
|---|
| 86 | $full_lucene_passes_exe = "perl.exe -S \"$lucene_passes_exe\"";
|
|---|
| 87 | }
|
|---|
| 88 | my $lucene_passes_sections = "Doc";
|
|---|
| 89 |
|
|---|
| 90 | my ($handle);
|
|---|
| 91 |
|
|---|
| 92 | if ($self->{'debug'}) {
|
|---|
| 93 | $handle = STDOUT;
|
|---|
| 94 | } else {
|
|---|
| 95 | if (!-e "$lucene_passes_exe" ||
|
|---|
| 96 | !open (PIPEOUT, "| $full_lucene_passes_exe text $lucene_passes_sections \"$build_dir\" \"dummy\" $osextra")) {
|
|---|
| 97 | print STDERR "<FatalError name='NoRunLucenePasses'/>\n</Stage>\n" if $self->{'gli'};
|
|---|
| 98 | die "lucenebuilder::build_index - couldn't run $lucene_passes_exe\n";
|
|---|
| 99 | }
|
|---|
| 100 | $handle = lucenebuilder::PIPEOUT;
|
|---|
| 101 | }
|
|---|
| 102 | my $levels = $self->{'levels'};
|
|---|
| 103 | my $gdbm_level = "document";
|
|---|
| 104 | if ($levels->{'section'}) {
|
|---|
| 105 | $gdbm_level = "section";
|
|---|
| 106 | }
|
|---|
| 107 |
|
|---|
| 108 | undef $levels->{'paragraph'}; # get rid of para if we had it.
|
|---|
| 109 | # set up the document processr
|
|---|
| 110 | $self->{'buildproc'}->set_output_handle ($handle);
|
|---|
| 111 | $self->{'buildproc'}->set_mode ('text');
|
|---|
| 112 | $self->{'buildproc'}->set_index ($textindex);
|
|---|
| 113 | $self->{'buildproc'}->set_indexing_text (0);
|
|---|
| 114 | $self->{'buildproc'}->set_store_text(1);
|
|---|
| 115 | $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
|---|
| 116 | $self->{'buildproc'}->set_levels ($levels);
|
|---|
| 117 | $self->{'buildproc'}->set_gdbm_level ($gdbm_level);
|
|---|
| 118 | $self->{'buildproc'}->reset();
|
|---|
| 119 | &plugin::begin($self->{'pluginfo'}, $self->{'source_dir'},
|
|---|
| 120 | $self->{'buildproc'}, $self->{'maxdocs'});
|
|---|
| 121 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
|---|
| 122 | "", {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
|---|
| 123 | &plugin::end($self->{'pluginfo'});
|
|---|
| 124 | close ($handle) unless $self->{'debug'};
|
|---|
| 125 | close PIPEOUT;
|
|---|
| 126 | $self->print_stats();
|
|---|
| 127 |
|
|---|
| 128 | print STDERR "</Stage>\n" if $self->{'gli'};
|
|---|
| 129 |
|
|---|
| 130 | }
|
|---|
| 131 |
|
|---|
| 132 | sub build_indexes {
|
|---|
| 133 | my $self = shift (@_);
|
|---|
| 134 | my ($indexname) = @_;
|
|---|
| 135 | my $outhandle = $self->{'outhandle'};
|
|---|
| 136 |
|
|---|
| 137 | my $indexes = [];
|
|---|
| 138 | if (defined $indexname && $indexname =~ /\w/) {
|
|---|
| 139 | push @$indexes, $indexname;
|
|---|
| 140 | } else {
|
|---|
| 141 | $indexes = $self->{'collect_cfg'}->{'indexes'};
|
|---|
| 142 | }
|
|---|
| 143 |
|
|---|
| 144 | # create the mapping between the index descriptions
|
|---|
| 145 | # and their directory names (includes subcolls and langs)
|
|---|
| 146 | $self->{'index_mapping'} = $self->create_index_mapping ($indexes);
|
|---|
| 147 |
|
|---|
| 148 | # build each of the indexes
|
|---|
| 149 | foreach $index (@$indexes) {
|
|---|
| 150 | if ($self->want_built($index)) {
|
|---|
| 151 |
|
|---|
| 152 | my $idx = $self->{'index_mapping'}->{$index};
|
|---|
| 153 | foreach my $level (keys %{$self->{'levels'}}) {
|
|---|
| 154 | next if $level =~ /paragraph/; # we don't do para indexing
|
|---|
| 155 | my ($pindex) = $level =~ /^(.)/;
|
|---|
| 156 | # should probably check that new name with level
|
|---|
| 157 | # is unique ... but currently (with doc sec and para)
|
|---|
| 158 | # each has unique first letter.
|
|---|
| 159 | $self->{'index_mapping'}->{$index} = $pindex.$idx;
|
|---|
| 160 |
|
|---|
| 161 | my $llevel = $mgppbuilder::level_map{$level};
|
|---|
| 162 | print $outhandle "\n*** building index $index at level $llevel in subdirectory " .
|
|---|
| 163 | "$self->{'index_mapping'}->{$index}\n" if ($self->{'verbosity'} >= 1);
|
|---|
| 164 | print STDERR "<Stage name='Index' source='$index' level=$llevel>\n" if $self->{'gli'};
|
|---|
| 165 |
|
|---|
| 166 | $self->build_index($index,$llevel);
|
|---|
| 167 | }
|
|---|
| 168 | $self->{'index_mapping'}->{$index} = $idx;
|
|---|
| 169 |
|
|---|
| 170 | } else {
|
|---|
| 171 | print $outhandle "\n*** ignoring index $index\n" if ($self->{'verbosity'} >= 1);
|
|---|
| 172 | }
|
|---|
| 173 | }
|
|---|
| 174 |
|
|---|
| 175 | #define the final field lists
|
|---|
| 176 | $self->make_final_field_list();
|
|---|
| 177 | }
|
|---|
| 178 |
|
|---|
| 179 | sub build_index {
|
|---|
| 180 | my $self = shift (@_);
|
|---|
| 181 | my ($index,$llevel) = @_;
|
|---|
| 182 | my $outhandle = $self->{'outhandle'};
|
|---|
| 183 | my $build_dir = $self->{'build_dir'};
|
|---|
| 184 |
|
|---|
| 185 | # get the full index directory path and make sure it exists
|
|---|
| 186 | my $indexdir = $self->{'index_mapping'}->{$index};
|
|---|
| 187 | &util::mk_all_dir (&util::filename_cat($build_dir, $indexdir));
|
|---|
| 188 |
|
|---|
| 189 | # get any os specific stuff
|
|---|
| 190 | my $exedir = "$ENV{'GSDLHOME'}/bin/$ENV{'GSDLOS'}";
|
|---|
| 191 | my $scriptdir = "$ENV{'GSDLHOME'}/bin/script";
|
|---|
| 192 |
|
|---|
| 193 | my $exe = &util::get_os_exe ();
|
|---|
| 194 | my $lucene_passes_exe = &util::filename_cat($scriptdir, "lucene_passes.pl");
|
|---|
| 195 | my $full_lucene_passes_exe = $lucene_passes_exe;
|
|---|
| 196 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
|---|
| 197 | $full_lucene_passes_exe = "perl.exe -S \"$lucene_passes_exe\"";
|
|---|
| 198 | }
|
|---|
| 199 |
|
|---|
| 200 | # define the section names for lucenepasses
|
|---|
| 201 | # define the section names and possibly the doc name for lucenepasses
|
|---|
| 202 | my $lucene_passes_sections = $llevel;
|
|---|
| 203 |
|
|---|
| 204 | my $opt_create_index = ($self->{'keepold'}) ? "" : "-create";
|
|---|
| 205 |
|
|---|
| 206 | my $osextra = "";
|
|---|
| 207 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
|---|
| 208 | $build_dir =~ s@/@\\@g;
|
|---|
| 209 | } else {
|
|---|
| 210 | if ($outhandle ne "STDERR") {
|
|---|
| 211 | # so lucene_passes doesn't print to stderr if we redirect output
|
|---|
| 212 | $osextra .= " 2>/dev/null";
|
|---|
| 213 | }
|
|---|
| 214 | }
|
|---|
| 215 |
|
|---|
| 216 | # get the index expression if this index belongs
|
|---|
| 217 | # to a subcollection
|
|---|
| 218 | my $indexexparr = [];
|
|---|
| 219 | my $langarr = [];
|
|---|
| 220 |
|
|---|
| 221 | # there may be subcollection info, and language info.
|
|---|
| 222 | my ($fields, $subcollection, $language) = split (":", $index);
|
|---|
| 223 | my @subcollections = ();
|
|---|
| 224 | @subcollections = split /,/, $subcollection if (defined $subcollection);
|
|---|
| 225 |
|
|---|
| 226 | foreach $subcollection (@subcollections) {
|
|---|
| 227 | if (defined ($self->{'collect_cfg'}->{'subcollection'}->{$subcollection})) {
|
|---|
| 228 | push (@$indexexparr, $self->{'collect_cfg'}->{'subcollection'}->{$subcollection});
|
|---|
| 229 | }
|
|---|
| 230 | }
|
|---|
| 231 |
|
|---|
| 232 | # add expressions for languages if this index belongs to
|
|---|
| 233 | # a language subcollection - only put languages expressions for the
|
|---|
| 234 | # ones we want in the index
|
|---|
| 235 |
|
|---|
| 236 | my @languages = ();
|
|---|
| 237 | my $language_metadata = "Language";
|
|---|
| 238 | if (defined ($self->{'collect_cfg'}->{'language_metadata'})) {
|
|---|
| 239 | $language_metadata = $self->{'collect_cfg'}->{'language_metadata'};
|
|---|
| 240 | }
|
|---|
| 241 | @languages = split /,/, $language if (defined $language);
|
|---|
| 242 | foreach my $language (@languages) {
|
|---|
| 243 | my $not=0;
|
|---|
| 244 | if ($language =~ s/^\!//) {
|
|---|
| 245 | $not = 1;
|
|---|
| 246 | }
|
|---|
| 247 | if($not) {
|
|---|
| 248 | push (@$langarr, "!$language");
|
|---|
| 249 | } else {
|
|---|
| 250 | push (@$langarr, "$language");
|
|---|
| 251 | }
|
|---|
| 252 | }
|
|---|
| 253 |
|
|---|
| 254 | # Build index dictionary. Uses verbatim stem method
|
|---|
| 255 | print $outhandle "\n creating index dictionary (lucene_passes -I1)\n" if ($self->{'verbosity'} >= 1);
|
|---|
| 256 | print STDERR "<Phase name='CreatingIndexDic'/>\n" if $self->{'gli'};
|
|---|
| 257 | my ($handle);
|
|---|
| 258 |
|
|---|
| 259 | if ($self->{'debug'}) {
|
|---|
| 260 | $handle = STDOUT;
|
|---|
| 261 | } else {
|
|---|
| 262 | if (!-e "$lucene_passes_exe" ||
|
|---|
| 263 | !open (PIPEOUT, "| $full_lucene_passes_exe $opt_create_index index $lucene_passes_sections \"$build_dir\" \"$indexdir\" $osextra")) {
|
|---|
| 264 | print STDERR "<FatalError name='NoRunLucenePasses'/>\n</Stage>\n" if $self->{'gli'};
|
|---|
| 265 | die "lucenebuilder::build_index - couldn't run $lucene_passes_exe\n";
|
|---|
| 266 | }
|
|---|
| 267 | $handle = lucenebuilder::PIPEOUT;
|
|---|
| 268 | }
|
|---|
| 269 |
|
|---|
| 270 | my $store_levels = $self->{'levels'};
|
|---|
| 271 | my $gdbm_level = "document";
|
|---|
| 272 | if ($store_levels->{'section'}) {
|
|---|
| 273 | $gdbm_level = "section";
|
|---|
| 274 | }
|
|---|
| 275 |
|
|---|
| 276 | my $dom_level = "";
|
|---|
| 277 | foreach my $key (keys %$store_levels) {
|
|---|
| 278 | if ($mgppbuilder::level_map{$key} eq $llevel) {
|
|---|
| 279 | $dom_level = $key;
|
|---|
| 280 | }
|
|---|
| 281 | }
|
|---|
| 282 | if ($dom_level eq "") {
|
|---|
| 283 | print STDERR "Warning: unrecognized tag level $llevel\n";
|
|---|
| 284 | $dom_level = "document";
|
|---|
| 285 | }
|
|---|
| 286 |
|
|---|
| 287 | my $local_levels = { $dom_level => 1 }; # work on one level at a time
|
|---|
| 288 |
|
|---|
| 289 | # set up the document processr
|
|---|
| 290 | $self->{'buildproc'}->set_output_handle ($handle);
|
|---|
| 291 | $self->{'buildproc'}->set_mode ('text');
|
|---|
| 292 | $self->{'buildproc'}->set_index ($index, $indexexparr);
|
|---|
| 293 | $self->{'buildproc'}->set_index_languages ($language_metadata, $langarr) if (defined $language);
|
|---|
| 294 | $self->{'buildproc'}->set_indexing_text (1);
|
|---|
| 295 | $self->{'buildproc'}->set_store_text(1);
|
|---|
| 296 | $self->{'buildproc'}->set_indexfieldmap ($self->{'indexfieldmap'});
|
|---|
| 297 | $self->{'buildproc'}->set_levels ($local_levels);
|
|---|
| 298 | $self->{'buildproc'}->set_gdbm_level($gdbm_level);
|
|---|
| 299 | $self->{'buildproc'}->reset();
|
|---|
| 300 | &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
|
|---|
| 301 | "", {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
|
|---|
| 302 | close ($handle) unless $self->{'debug'};
|
|---|
| 303 |
|
|---|
| 304 | $self->print_stats();
|
|---|
| 305 |
|
|---|
| 306 | $self->{'buildproc'}->set_levels ($store_levels);
|
|---|
| 307 | print STDERR "</Stage>\n" if $self->{'gli'};
|
|---|
| 308 | }
|
|---|
| 309 |
|
|---|
| 310 | 1;
|
|---|
| 311 |
|
|---|
| 312 |
|
|---|
