| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # phind.pm -- the Phind classifier
|
|---|
| 4 | #
|
|---|
| 5 | # Copyright (C) 2000 Gordon W. Paynter
|
|---|
| 6 | # Copyright (C) 2000 New Zealand Digital Library Project
|
|---|
| 7 | #
|
|---|
| 8 | #
|
|---|
| 9 | # A component of the Greenstone digital library software
|
|---|
| 10 | # from the New Zealand Digital Library Project at the
|
|---|
| 11 | # University of Waikato, New Zealand.
|
|---|
| 12 | #
|
|---|
| 13 | # This program is free software; you can redistribute it and/or modify
|
|---|
| 14 | # it under the terms of the GNU General Public License as published by
|
|---|
| 15 | # the Free Software Foundation; either version 2 of the License, or
|
|---|
| 16 | # (at your option) any later version.
|
|---|
| 17 | #
|
|---|
| 18 | # This program is distributed in the hope that it will be useful,
|
|---|
| 19 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 20 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 21 | # GNU General Public License for more details.
|
|---|
| 22 | #
|
|---|
| 23 | # You should have received a copy of the GNU General Public License
|
|---|
| 24 | # along with this program; if not, write to the Free Software
|
|---|
| 25 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|---|
| 26 | #
|
|---|
| 27 | ###########################################################################
|
|---|
| 28 |
|
|---|
| 29 | # The phind clasifier plugin.
|
|---|
| 30 | # Options are dexcribed in the print_usage function.
|
|---|
| 31 | # Type "classinfo.pl phind" at the command line for a summary.
|
|---|
| 32 |
|
|---|
| 33 | package phind;
|
|---|
| 34 |
|
|---|
| 35 | use BasClas;
|
|---|
| 36 | use util;
|
|---|
| 37 | use ghtml;
|
|---|
| 38 | use unicode;
|
|---|
| 39 |
|
|---|
| 40 | sub BEGIN {
|
|---|
| 41 | @ISA = ('BasClas');
|
|---|
| 42 | }
|
|---|
| 43 |
|
|---|
| 44 |
|
|---|
| 45 | sub print_usage {
|
|---|
| 46 | print STDERR "
|
|---|
| 47 | usage: classify phind [options]
|
|---|
| 48 |
|
|---|
| 49 | options:
|
|---|
| 50 | -text Fields The text used to build the phrase hierarchy.
|
|---|
| 51 | (default: 'section:Title,section:text')
|
|---|
| 52 |
|
|---|
| 53 | -title Title The metadata field used to describe each document.
|
|---|
| 54 | (default: 'Title')
|
|---|
| 55 |
|
|---|
| 56 | -button Name The label for the classifier screen and button in
|
|---|
| 57 | navigation bar.
|
|---|
| 58 | (default: 'Phrase')
|
|---|
| 59 |
|
|---|
| 60 | -language Regex Language or languages to use building hierarchy.
|
|---|
| 61 | Languages are identified by two-letter country codes
|
|---|
| 62 | like en (English), es (Spanish), and fr (French).
|
|---|
| 63 | Language is a regular expression, so 'en|fr' (English or
|
|---|
| 64 | French) and '..' (match any language) are valid.
|
|---|
| 65 | (default: 'en'.)
|
|---|
| 66 |
|
|---|
| 67 | -savephrases File If set, the phrase infomation will be stored in
|
|---|
| 68 | the given file as text. It is probably a good idea
|
|---|
| 69 | to use an absolute path.
|
|---|
| 70 | (defualt: not set)
|
|---|
| 71 |
|
|---|
| 72 | -suffixmode N The smode parameter to the phrase extraction program. A
|
|---|
| 73 | value of 0 means that stopwords are ignored, and of 1
|
|---|
| 74 | means that stopwords are used.
|
|---|
| 75 | (default: 1)
|
|---|
| 76 |
|
|---|
| 77 | -thesaurus Name Name of a thesaurus stored in phind format in the
|
|---|
| 78 | collection's etc directory.
|
|---|
| 79 | (default: not set)
|
|---|
| 80 |
|
|---|
| 81 | -untidy Don't remove working files.
|
|---|
| 82 |
|
|---|
| 83 | "; }
|
|---|
| 84 |
|
|---|
| 85 |
|
|---|
| 86 | # Phrase delimiter symbols - these should be abstracted out someplace
|
|---|
| 87 |
|
|---|
| 88 | my $colstart = "COLLECTIONSTART";
|
|---|
| 89 | my $colend = "COLLECTIONEND";
|
|---|
| 90 | my $doclimit = "DOCUMENTLIMIT";
|
|---|
| 91 | my $senlimit = "SENTENCELIMIT";
|
|---|
| 92 | my @delimiters = ($colstart, $colend, $doclimit, $senlimit);
|
|---|
| 93 |
|
|---|
| 94 |
|
|---|
| 95 | # Create a new phind browser based on collect.cfg
|
|---|
| 96 |
|
|---|
| 97 | sub new {
|
|---|
| 98 | my $class = shift (@_);
|
|---|
| 99 | my $self = new BasClas($class, @_);
|
|---|
| 100 |
|
|---|
| 101 | my $out = $self->{'outhandle'};
|
|---|
| 102 |
|
|---|
| 103 |
|
|---|
| 104 | # Phind installation check
|
|---|
| 105 | # The phind phrase browser is research software and is not installed
|
|---|
| 106 | # by defualt. If the user attepts to use it we warn them that it's a
|
|---|
| 107 | # bit dodgy, then tell them how to install it. If they can do that
|
|---|
| 108 | # and get all the files in place, then we let them proceed.
|
|---|
| 109 |
|
|---|
| 110 | print $out "Checking Phind phrase browser requirements...\n";
|
|---|
| 111 |
|
|---|
| 112 | # Make sure we're not in windows
|
|---|
| 113 | if ($ENV{'GSDLOS'} =~ /windows/i) {
|
|---|
| 114 | print STDERR "Sorry - Phind currently only works under Unix";
|
|---|
| 115 | exit(1);
|
|---|
| 116 | }
|
|---|
| 117 |
|
|---|
| 118 | # Ensure the Phind generate scripts are in place
|
|---|
| 119 | my $file1 = &util::filename_cat($ENV{'GSDLHOME'}, "bin", $ENV{'GSDLOS'}, "suffix");
|
|---|
| 120 | my $src = &util::filename_cat($ENV{'GSDLHOME'}, "src", "phind", "generate");
|
|---|
| 121 |
|
|---|
| 122 | if (!(-e $file1)) {
|
|---|
| 123 | print STDERR "The phind \"suffix\" program is not installed. ";
|
|---|
| 124 | print STDERR "To install it, change to the directory\n";
|
|---|
| 125 | print STDERR " $src\n";
|
|---|
| 126 | print STDERR "and type \"make install-phind\".\n\n";
|
|---|
| 127 | exit(1);
|
|---|
| 128 | }
|
|---|
| 129 |
|
|---|
| 130 | # Ensure the Phind CGI script is in place
|
|---|
| 131 | $file1 = &util::filename_cat($ENV{'GSDLHOME'}, "cgi-bin", "phindcgi");
|
|---|
| 132 | $src = &util::filename_cat($ENV{'GSDLHOME'}, "src", "phind", "host");
|
|---|
| 133 |
|
|---|
| 134 | if (!(-e $file1)) {
|
|---|
| 135 | print STDERR "The phind CGI program is not installed. ";
|
|---|
| 136 | print STDERR "To install it, change to the directory\n";
|
|---|
| 137 | print STDERR " $src\n";
|
|---|
| 138 | print STDERR "and type \"make install-phind\".\n\n";
|
|---|
| 139 | exit(1);
|
|---|
| 140 | }
|
|---|
| 141 |
|
|---|
| 142 | # Ensure the Phind Java applet is in place
|
|---|
| 143 | $src = &util::filename_cat($ENV{'GSDLHOME'}, "src", "phind", "client");
|
|---|
| 144 | $file1 = &util::filename_cat($src, "Phind.class");
|
|---|
| 145 |
|
|---|
| 146 | if (!(-e $file1)) {
|
|---|
| 147 | print STDERR "The phind Java classes are not compiled. ";
|
|---|
| 148 | print STDERR "To compile them, change to the directory\n";
|
|---|
| 149 | print STDERR " $src\n";
|
|---|
| 150 | print STDERR "and use your Java compiler to compile Phind.java.\n";
|
|---|
| 151 | print STDERR "(if you have Java 1.2 installed, type \"javac Phind.java\")\n\n";
|
|---|
| 152 | exit(1);
|
|---|
| 153 | }
|
|---|
| 154 |
|
|---|
| 155 | # Parse classifier arguments
|
|---|
| 156 | my $builddir = "";
|
|---|
| 157 | if (!parsargv::parse(\@_,
|
|---|
| 158 | q^text/.*/section:Title,section:text^, \$self->{'indexes'},
|
|---|
| 159 | q^title/.*/Title^, \$self->{'titlefield'},
|
|---|
| 160 | q^button/.*/Phrase^, \$self->{'buttonname'},
|
|---|
| 161 | q^language/.*/en^, \$language,
|
|---|
| 162 | q^builddir/.*/^, \$builddir,
|
|---|
| 163 | q^savephrases/\d/0^, \$self->{'savephrases'},
|
|---|
| 164 | q^suffixmode/\d/1^, \$self->{'suffixmode'},
|
|---|
| 165 | q^thesaurus/.*/^, \$self->{'thesaurus'},
|
|---|
| 166 | q^untidy^, \$self->{'untidy'},
|
|---|
| 167 | "allow_extra_options")) {
|
|---|
| 168 |
|
|---|
| 169 | print STDERR "\nIncorrect options passed to $class, check your collect.cfg file\n";
|
|---|
| 170 | &print_usage();
|
|---|
| 171 | die "\n";
|
|---|
| 172 | }
|
|---|
| 173 |
|
|---|
| 174 | # classifier information
|
|---|
| 175 | $self->{'collection'} = $ENV{'GSDLCOLLECTION'};
|
|---|
| 176 |
|
|---|
| 177 | # limit languages
|
|---|
| 178 | $self->{'language_exp'} = $language;
|
|---|
| 179 |
|
|---|
| 180 | # collection directories
|
|---|
| 181 | $self->{'collectiondir'} = $ENV{'GSDLCOLLECTDIR'};
|
|---|
| 182 | if (!$builddir) {
|
|---|
| 183 | $builddir = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "building");
|
|---|
| 184 | }
|
|---|
| 185 | $self->{'builddir'} = $builddir;
|
|---|
| 186 |
|
|---|
| 187 | return bless $self, $class;
|
|---|
| 188 | }
|
|---|
| 189 |
|
|---|
| 190 |
|
|---|
| 191 | # Initialise the phind classifier
|
|---|
| 192 |
|
|---|
| 193 | sub init {
|
|---|
| 194 | my $self = shift (@_);
|
|---|
| 195 |
|
|---|
| 196 | # ensure we have a build directory
|
|---|
| 197 | my $builddir = $self->{'builddir'};
|
|---|
| 198 | die unless (-e "$builddir");
|
|---|
| 199 |
|
|---|
| 200 | # create phind directory
|
|---|
| 201 | my $phnumber = 1;
|
|---|
| 202 | my $phinddir = &util::filename_cat($builddir, "phind1");
|
|---|
| 203 | while (-e "$phinddir") {
|
|---|
| 204 | $phnumber++;
|
|---|
| 205 | $phinddir = &util::filename_cat($builddir, "phind$phnumber");
|
|---|
| 206 | }
|
|---|
| 207 | &util::mk_dir("$phinddir");
|
|---|
| 208 | $self->{'phinddir'} = $phinddir;
|
|---|
| 209 | $self->{'phindnumber'} = $phnumber;
|
|---|
| 210 |
|
|---|
| 211 | # open filehandles for documents and text
|
|---|
| 212 | my $clausefile = &util::filename_cat("$phinddir", "clauses");
|
|---|
| 213 | &util::rm($clausefile) if (-e $clausefile);
|
|---|
| 214 |
|
|---|
| 215 | my $txthandle = 'TEXT' . $phnumber;
|
|---|
| 216 | open($txthandle, ">$clausefile") || die "Cannot open $clausefile: $!";
|
|---|
| 217 | $self->{'txthandle'} = $txthandle;
|
|---|
| 218 |
|
|---|
| 219 | my $docfile = &util::filename_cat("$phinddir", "docs.txt");
|
|---|
| 220 | &util::rm($docfile) if (-e $docfile);
|
|---|
| 221 |
|
|---|
| 222 | my $dochandle = 'DOC' . $phnumber;
|
|---|
| 223 | open($dochandle, ">$docfile") || die "Cannot open $docfile: $!";
|
|---|
| 224 | $self->{'dochandle'} = $dochandle;
|
|---|
| 225 |
|
|---|
| 226 | }
|
|---|
| 227 |
|
|---|
| 228 |
|
|---|
| 229 | # Classify each document.
|
|---|
| 230 | #
|
|---|
| 231 | # Each document is passed here in turn. The classifier extracts the
|
|---|
| 232 | # text of each and stores it in the clauses file. Document details are
|
|---|
| 233 | # stored in the docs.txt file.
|
|---|
| 234 |
|
|---|
| 235 | sub classify {
|
|---|
| 236 | my $self = shift (@_);
|
|---|
| 237 | my $doc_obj = shift @_;
|
|---|
| 238 |
|
|---|
| 239 | my $verbosity = $self->{'verbosity'};
|
|---|
| 240 | my $top_section = $doc_obj->get_top_section();
|
|---|
| 241 |
|
|---|
| 242 | my $titlefield = $self->{'titlefield'};
|
|---|
| 243 |
|
|---|
| 244 | my $title = $doc_obj->get_metadata_element ($top_section, $titlefield);
|
|---|
| 245 | print "process: $title\n" if ($verbosity > 2);
|
|---|
| 246 |
|
|---|
| 247 | # Only consider the file if it is in the correct language
|
|---|
| 248 | my $doclanguage = $doc_obj->get_metadata_element ($top_section, "Language");
|
|---|
| 249 | my $phrlanguage = $self->{'language_exp'};
|
|---|
| 250 | return if ($doclanguage && ($doclanguage !~ /$phrlanguage/i));
|
|---|
| 251 |
|
|---|
| 252 | # record this file
|
|---|
| 253 | my $total++;
|
|---|
| 254 | print "file $total: $file\n" if ($self->{'$verbosity'});
|
|---|
| 255 |
|
|---|
| 256 |
|
|---|
| 257 | # Store document details
|
|---|
| 258 | my $OID = $doc_obj->get_OID();
|
|---|
| 259 | $OID = "NULL" unless defined $OID;
|
|---|
| 260 | my $dochandle = $self->{'dochandle'};
|
|---|
| 261 | print $dochandle "<Document>\t$OID\t$title\n";
|
|---|
| 262 |
|
|---|
| 263 | # Store the text occuring in this object
|
|---|
| 264 |
|
|---|
| 265 | # output the document delimiter
|
|---|
| 266 | my $txthandle = $self->{'txthandle'};
|
|---|
| 267 | print $txthandle "$doclimit\n";
|
|---|
| 268 |
|
|---|
| 269 | # iterarate over the required indexes and store their text
|
|---|
| 270 | my $indexes = $self->{'indexes'};
|
|---|
| 271 | my $text = "";
|
|---|
| 272 | my ($part, $level, $field, $section, $data, $dataref);
|
|---|
| 273 |
|
|---|
| 274 | foreach $part (split(/,/, $indexes)) {
|
|---|
| 275 |
|
|---|
| 276 | # Each field has a level and a data element ((e.g. document:Title)
|
|---|
| 277 | ($level, $field) = split(/:/, $part);
|
|---|
| 278 | die unless ($level && $field);
|
|---|
| 279 |
|
|---|
| 280 | # Extract the text from every section
|
|---|
| 281 | # (In phind, document:text and section:text are equivalent)
|
|---|
| 282 | if ($field eq "text") {
|
|---|
| 283 | $data = "";
|
|---|
| 284 | $section = $doc_obj->get_top_section();
|
|---|
| 285 | while (defined($section)) {
|
|---|
| 286 | $data .= $doc_obj->get_text($section) . "\n";
|
|---|
| 287 | $section = $doc_obj->get_next_section($section);
|
|---|
| 288 | }
|
|---|
| 289 | $text .= convert_gml_to_tokens($phrlanguage, $data) . "\n";
|
|---|
| 290 | }
|
|---|
| 291 |
|
|---|
| 292 | # Extract a metadata field from a document
|
|---|
| 293 | # (If ther eis more than one element of the given type, get them all.)
|
|---|
| 294 | elsif ($level eq "document") {
|
|---|
| 295 | $dataref = $doc_obj->get_metadata($doc_obj->get_top_section(), $field);
|
|---|
| 296 | foreach $data (@$dataref) {
|
|---|
| 297 | $text .= convert_gml_to_tokens($phrlanguage, $data) . "\n";
|
|---|
| 298 | }
|
|---|
| 299 | }
|
|---|
| 300 |
|
|---|
| 301 | # Extract metadata from every section in a document
|
|---|
| 302 | elsif ($level eq "section") {
|
|---|
| 303 | $data = "";
|
|---|
| 304 | $section = $doc_obj->get_top_section();
|
|---|
| 305 | while (defined($section)) {
|
|---|
| 306 | $dataref = $doc_obj->get_metadata($section, $field);
|
|---|
| 307 | $data .= join("\n", @$dataref) . "\n";
|
|---|
| 308 | $section = $doc_obj->get_next_section($section);
|
|---|
| 309 | }
|
|---|
| 310 | $text .= convert_gml_to_tokens($phrlanguage, $data) . "\n";
|
|---|
| 311 | }
|
|---|
| 312 |
|
|---|
| 313 | # Some sort of specification which I don't understand
|
|---|
| 314 | else {
|
|---|
| 315 | die "Unknown level ($level) in phind index ($part)\n";
|
|---|
| 316 | }
|
|---|
| 317 |
|
|---|
| 318 | }
|
|---|
| 319 |
|
|---|
| 320 | # output the text
|
|---|
| 321 | $text =~ tr/\n//s;
|
|---|
| 322 | print $txthandle "$text";
|
|---|
| 323 | }
|
|---|
| 324 |
|
|---|
| 325 |
|
|---|
| 326 | # Construct the classifier from the information already gathered
|
|---|
| 327 | #
|
|---|
| 328 | # When get_classify_info is called, the clauses and docs.txt files have
|
|---|
| 329 | # already been constructed in the phind directory. This function will
|
|---|
| 330 | # translate them into compressed, indexed MGPP files that can be read by
|
|---|
| 331 | # the phindcgi script. It will also register our classifier so that it
|
|---|
| 332 | # shows up in the navigation bar.
|
|---|
| 333 |
|
|---|
| 334 | sub get_classify_info {
|
|---|
| 335 | my $self = shift (@_);
|
|---|
| 336 |
|
|---|
| 337 | close $self->{'txthandle'};
|
|---|
| 338 | my $verbosity = $self->{'verbosity'};
|
|---|
| 339 | my $out = $self->{'outhandle'};
|
|---|
| 340 | my $phinddir = $self->{'phinddir'};
|
|---|
| 341 | my $exe = &util::get_os_exe ();
|
|---|
| 342 |
|
|---|
| 343 | if ($verbosity) {
|
|---|
| 344 | print $out "\n*** phind.pm generating indexes for ", $self->{'indexes'}, "\n";
|
|---|
| 345 | print $out "*** in", $self->{'phinddir'}, "\n";
|
|---|
| 346 | }
|
|---|
| 347 |
|
|---|
| 348 | # Construct phind indexes
|
|---|
| 349 | my $suffixmode = $self->{'suffixmode'};
|
|---|
| 350 | my ($command, $status);
|
|---|
| 351 |
|
|---|
| 352 | # Generate the vocabulary, symbol statistics, and numbers file
|
|---|
| 353 | # from the clauses file
|
|---|
| 354 | print $out "\nExtracting vocabulary and statistics\n" if $verbosity;
|
|---|
| 355 | &extract_vocabulary($self);
|
|---|
| 356 |
|
|---|
| 357 | # Use the suffix program to generate the phind/phrases file
|
|---|
| 358 | print $out "\nExtracting phrases from processed text (with suffix)\n" if $verbosity;
|
|---|
| 359 | &execute("suffix $phinddir $suffixmode $verbosity", $verbosity, $out);
|
|---|
| 360 |
|
|---|
| 361 | # Create the phrase file and put phrase numbers in phind/phrases
|
|---|
| 362 | print $out "\nSorting and renumbering phrases for input to mgpp\n" if $verbosity;
|
|---|
| 363 | &renumber_phrases($self);
|
|---|
| 364 |
|
|---|
| 365 | # Create the mg phrase database
|
|---|
| 366 |
|
|---|
| 367 | print $out "\nCreating phrase databases\n";
|
|---|
| 368 | my $mg_input = &util::filename_cat($phinddir, "pdata.txt");
|
|---|
| 369 | my $mg_stem = "pdata";
|
|---|
| 370 |
|
|---|
| 371 | &execute("mgpp_passes$exe -d $phinddir -f $mg_stem -T1 $mg_input", $verbosity, $out);
|
|---|
| 372 | &execute("mgpp_compression_dict$exe -d $phinddir -f $mg_stem", $verbosity, $out);
|
|---|
| 373 | &execute("mgpp_passes$exe -d $phinddir -f $mg_stem -T2 $mg_input", $verbosity, $out);
|
|---|
| 374 |
|
|---|
| 375 | # create the mg index of words
|
|---|
| 376 | print $out "\nCreating word-level search indexes\n";
|
|---|
| 377 | $mg_input = &util::filename_cat($phinddir, "pword.txt");
|
|---|
| 378 | $mg_stem = "pword";
|
|---|
| 379 |
|
|---|
| 380 | &execute("mgpp_passes$exe -d $phinddir -f $mg_stem -T1 -I1 $mg_input", $verbosity, $out);
|
|---|
| 381 | &execute("mgpp_compression_dict$exe -d $phinddir -f $mg_stem", $verbosity, $out);
|
|---|
| 382 | &execute("mgpp_perf_hash_build$exe -d $phinddir -f $mg_stem", $verbosity, $out);
|
|---|
| 383 | &execute("mgpp_passes$exe -d $phinddir -f $mg_stem -T2 -I2 $mg_input", $verbosity, $out);
|
|---|
| 384 | &execute("mgpp_weights_build$exe -d $phinddir -f $mg_stem", $verbosity, $out);
|
|---|
| 385 | &execute("mgpp_invf_dict$exe -d $phinddir -f $mg_stem", $verbosity, $out);
|
|---|
| 386 |
|
|---|
| 387 | &execute("mgpp_stem_idx$exe -d $phinddir -f $mg_stem -s 1", $verbosity, $out);
|
|---|
| 388 | &execute("mgpp_stem_idx$exe -d $phinddir -f $mg_stem -s 2", $verbosity, $out);
|
|---|
| 389 | &execute("mgpp_stem_idx$exe -d $phinddir -f $mg_stem -s 3", $verbosity, $out);
|
|---|
| 390 |
|
|---|
| 391 | # create the mg document information database
|
|---|
| 392 | print $out "\nCreating document information databases\n";
|
|---|
| 393 | $mg_input = &util::filename_cat($phinddir, "docs.txt");
|
|---|
| 394 | $mg_stem = "docs";
|
|---|
| 395 |
|
|---|
| 396 | &execute("mgpp_passes$exe -d $phinddir -f $mg_stem -T1 $mg_input", $verbosity, $out);
|
|---|
| 397 | &execute("mgpp_compression_dict$exe -d $phinddir -f $mg_stem", $verbosity, $out);
|
|---|
| 398 | &execute("mgpp_passes$exe -d $phinddir -f $mg_stem -T2 $mg_input", $verbosity, $out);
|
|---|
| 399 |
|
|---|
| 400 |
|
|---|
| 401 | # Tidy up stray files
|
|---|
| 402 | if (!$self->{'untidy'}) {
|
|---|
| 403 | print $out "\nCleaning up\n" if ($verbosity > 2);
|
|---|
| 404 | &util::rm("$phinddir/clauses", "$phinddir/clauses.numbers",
|
|---|
| 405 | "$phinddir/clauses.vocab", "$phinddir/clauses.stats",
|
|---|
| 406 | "$phinddir/phrases", "$phinddir/phrases.3", "$phinddir/docs.txt",
|
|---|
| 407 | "$phinddir/pdata.txt", "$phinddir/pword.txt");
|
|---|
| 408 | my $outfile = 1;
|
|---|
| 409 | while (-e "$phinddir/outPhrase.$outfile") {
|
|---|
| 410 | &util::rm("$phinddir/outPhrase.$outfile");
|
|---|
| 411 | $outfile++;
|
|---|
| 412 | }
|
|---|
| 413 | }
|
|---|
| 414 |
|
|---|
| 415 |
|
|---|
| 416 | # Return the information about the classifier that we'll later want to
|
|---|
| 417 | # use to create macros when the Phind classifier document is displayed.
|
|---|
| 418 | my %classifyinfo = ('thistype'=>'Invisible',
|
|---|
| 419 | 'childtype'=>'Phind',
|
|---|
| 420 | 'Title'=>$self->{'buttonname'},
|
|---|
| 421 | 'parameters'=>"phindnumber=$self->{'phindnumber'}",
|
|---|
| 422 | 'contains'=>[]);
|
|---|
| 423 |
|
|---|
| 424 | my $collection = $self->{'collection'};
|
|---|
| 425 | my $url = "library?a=p&p=phind&c=$collection";
|
|---|
| 426 | push (@{$classifyinfo{'contains'}}, {'OID'=>$url});
|
|---|
| 427 |
|
|---|
| 428 | return \%classifyinfo;
|
|---|
| 429 | }
|
|---|
| 430 |
|
|---|
| 431 |
|
|---|
| 432 |
|
|---|
| 433 | sub convert_gml_to_tokens {
|
|---|
| 434 |
|
|---|
| 435 | my ($language_exp, $text) = @_;
|
|---|
| 436 |
|
|---|
| 437 | if ($language_exp =~ /en/) {
|
|---|
| 438 | return &convert_gml_to_tokens_EN($text);
|
|---|
| 439 | }
|
|---|
| 440 |
|
|---|
| 441 | $_ = $text;
|
|---|
| 442 |
|
|---|
| 443 | # 1. remove GML tags
|
|---|
| 444 |
|
|---|
| 445 | # Remove everything that is in a tag
|
|---|
| 446 | s/\s*<p>\s*/ PARAGRAPHBREAK /isgo;
|
|---|
| 447 | s/\s*<br>\s*/ LINEBREAK /isgo;
|
|---|
| 448 | s/<[^>]*>/ /sgo;
|
|---|
| 449 |
|
|---|
| 450 | # Now we have the text, but it may contain HTML
|
|---|
| 451 | # elements coded as > etc. Remove these tags.
|
|---|
| 452 | s/&/&/sgo;
|
|---|
| 453 | s/</</sgo;
|
|---|
| 454 | s/>/>/sgo;
|
|---|
| 455 | s/\s*<p>\s*/ PARAGRAPHBREAK /isgo;
|
|---|
| 456 | s/\s*<br>\s*/ LINEBREAK /isgo;
|
|---|
| 457 | s/<[^>]*>/ /sgo;
|
|---|
| 458 |
|
|---|
| 459 | # replace<p> and <br> placeholders with clause break symbol (\n)
|
|---|
| 460 | s/\s+/ /gso;
|
|---|
| 461 | s/PARAGRAPHBREAK/\n/sgo;
|
|---|
| 462 | s/LINEBREAK/\n/sgo;
|
|---|
| 463 |
|
|---|
| 464 |
|
|---|
| 465 |
|
|---|
| 466 |
|
|---|
| 467 | # 2. Split the remaining text into space-delimited tokens
|
|---|
| 468 |
|
|---|
| 469 | # Convert any HTML special characters (like ") to their UTF8 equivalent
|
|---|
| 470 | s/&([^;]+);/&unicode::ascii2utf8(\&ghtml::getcharequiv($1,1))/gse;
|
|---|
| 471 |
|
|---|
| 472 | # Split text at word boundaries
|
|---|
| 473 | s/\b/ /go;
|
|---|
| 474 |
|
|---|
| 475 | # 3. Convert the remaining text to "clause format"
|
|---|
| 476 |
|
|---|
| 477 | # Insert newline if the end of a sentence is detected
|
|---|
| 478 | # (delimter is: "[\.\?\!]\s")
|
|---|
| 479 | # s/\s*[\.\?\!]\s+/\n/go;
|
|---|
| 480 |
|
|---|
| 481 | # remove unnecessary punctuation and replace with clause break symbol (\n)
|
|---|
| 482 | s/[^\w ]/\n/go;
|
|---|
| 483 |
|
|---|
| 484 | # remove extraneous whitespace
|
|---|
| 485 | s/ +/ /sgo;
|
|---|
| 486 | s/^\s+//mgo;
|
|---|
| 487 | s/\s*$/\n/mgo;
|
|---|
| 488 |
|
|---|
| 489 | # remove lines that contain one word or less
|
|---|
| 490 | s/^\S*$//mgo;
|
|---|
| 491 | s/^\s*$//mgo;
|
|---|
| 492 | tr/\n//s;
|
|---|
| 493 |
|
|---|
| 494 | return $_;
|
|---|
| 495 | }
|
|---|
| 496 |
|
|---|
| 497 | # A version of convert_gml_to_tokens that is fine-tuned to the English language.
|
|---|
| 498 |
|
|---|
| 499 | sub convert_gml_to_tokens_EN {
|
|---|
| 500 | $_ = shift @_;
|
|---|
| 501 |
|
|---|
| 502 | # FIRST, remove GML tags
|
|---|
| 503 |
|
|---|
| 504 | # Replace all whitespace with a simple space
|
|---|
| 505 | s/\s+/ /gs;
|
|---|
| 506 |
|
|---|
| 507 | # Remove everything that is in a tag
|
|---|
| 508 | s/\s*<p>\s*/ PARAGRAPHBREAK /isg;
|
|---|
| 509 | s/\s*<br>\s*/ LINEBREAK /isg;
|
|---|
| 510 | s/<[^>]*>/ /sg;
|
|---|
| 511 |
|
|---|
| 512 | # Now we have the text, but it may contain HTML
|
|---|
| 513 | # elements coded as > etc. Remove these tags.
|
|---|
| 514 | s/</</sg;
|
|---|
| 515 | s/>/>/sg;
|
|---|
| 516 |
|
|---|
| 517 | s/\s+/ /sg;
|
|---|
| 518 | s/\s*<p>\s*/ PARAGRAPHBREAK /isg;
|
|---|
| 519 | s/\s*<br>\s*/ LINEBREAK /isg;
|
|---|
| 520 | s/<[^>]*>/ /sg;
|
|---|
| 521 |
|
|---|
| 522 | # remove & and other miscellaneous markup tags
|
|---|
| 523 | s/&/&/sg;
|
|---|
| 524 | s/</</sg;
|
|---|
| 525 | s/>/>/sg;
|
|---|
| 526 | s/&/&/sg;
|
|---|
| 527 |
|
|---|
| 528 | # replace<p> and <br> placeholders with carriage returns
|
|---|
| 529 | s/PARAGRAPHBREAK/\n/sg;
|
|---|
| 530 | s/LINEBREAK/\n/sg;
|
|---|
| 531 |
|
|---|
| 532 |
|
|---|
| 533 | # Exceptional punctuation
|
|---|
| 534 | #
|
|---|
| 535 | # We make special cases of some punctuation
|
|---|
| 536 |
|
|---|
| 537 | # remove any apostrophe that indicates omitted letters
|
|---|
| 538 | s/(\w+)\'(\w*\s)/ $1$2 /g;
|
|---|
| 539 |
|
|---|
| 540 | # remove period that appears in a person's initals
|
|---|
| 541 | s/\s([A-Z])\./ $1 /g;
|
|---|
| 542 |
|
|---|
| 543 | # replace hyphens in hypheanted words and names with a space
|
|---|
| 544 | s/([A-Za-z])-\s*([A-Za-z])/$1 $2/g;
|
|---|
| 545 |
|
|---|
| 546 | # Convert the remaining text to "clause format",
|
|---|
| 547 | # This means removing all excess punctuation and garbage text,
|
|---|
| 548 | # normalising valid punctuation to fullstops and commas,
|
|---|
| 549 | # then putting one cluse on each line.
|
|---|
| 550 |
|
|---|
| 551 | # Insert newline when the end of a sentence is detected
|
|---|
| 552 | # (delimter is: "[\.\?\!]\s")
|
|---|
| 553 | s/\s*[\.\?\!]\s+/\n/g;
|
|---|
| 554 |
|
|---|
| 555 | # split numbers after four digits
|
|---|
| 556 | s/(\d\d\d\d)/$1 /g;
|
|---|
| 557 |
|
|---|
| 558 | # split words after 32 characters
|
|---|
| 559 |
|
|---|
| 560 | # squash repeated punctuation
|
|---|
| 561 | tr/A-Za-z0-9 //cs;
|
|---|
| 562 |
|
|---|
| 563 | # save email addresses
|
|---|
| 564 | # s/\w+@\w+\.[\w\.]+/EMAIL/g;
|
|---|
| 565 |
|
|---|
| 566 | # normalise clause breaks (mostly punctuation symbols) to commas
|
|---|
| 567 | s/[^A-Za-z0-9 \n]+/ , /g;
|
|---|
| 568 |
|
|---|
| 569 | # Remove repeated commas, and replace with newline
|
|---|
| 570 | s/\s*,[, ]+/\n/g;
|
|---|
| 571 |
|
|---|
| 572 | # remove extra whitespace
|
|---|
| 573 | s/ +/ /sg;
|
|---|
| 574 | s/^\s+//mg;
|
|---|
| 575 | s/\s*$/\n/mg;
|
|---|
| 576 |
|
|---|
| 577 | # remove lines that contain one word or less
|
|---|
| 578 | s/^\w*$//mg;
|
|---|
| 579 | s/^\s*$//mg;
|
|---|
| 580 | tr/\n//s;
|
|---|
| 581 |
|
|---|
| 582 | return $_;
|
|---|
| 583 |
|
|---|
| 584 | }
|
|---|
| 585 |
|
|---|
| 586 |
|
|---|
| 587 |
|
|---|
| 588 | # Execute a system command
|
|---|
| 589 |
|
|---|
| 590 | sub execute {
|
|---|
| 591 | my ($command, $verbosity, $outhandle) = @_;
|
|---|
| 592 | print $outhandle "Executing: $command\n" if ($verbosity > 2);
|
|---|
| 593 | my $status = system($command);
|
|---|
| 594 | if ($status != 0) {
|
|---|
| 595 | print STDERR "phind - Error executing $command: $!\n";
|
|---|
| 596 | exit($status);
|
|---|
| 597 | }
|
|---|
| 598 | }
|
|---|
| 599 |
|
|---|
| 600 |
|
|---|
| 601 | # Generate the vocabulary, symbol statistics, and numbers file from the
|
|---|
| 602 | # clauses file. This is legacy code, so is a bit messy and probably wont
|
|---|
| 603 | # run under windows.
|
|---|
| 604 |
|
|---|
| 605 | sub extract_vocabulary {
|
|---|
| 606 | my ($self) = @_;
|
|---|
| 607 |
|
|---|
| 608 | my $verbosity = $self->{'verbosity'};
|
|---|
| 609 | my $out = $self->{'outhandle'};
|
|---|
| 610 |
|
|---|
| 611 | my $collectiondir = $self->{'collectiondir'};
|
|---|
| 612 | my $phinddir = $self->{'phinddir'};
|
|---|
| 613 |
|
|---|
| 614 | my $language_exp = $self->{'language_exp'};
|
|---|
| 615 |
|
|---|
| 616 | my ($w, $l, $line, $word);
|
|---|
| 617 |
|
|---|
| 618 | my ($first_delimiter, $last_delimiter,
|
|---|
| 619 | $first_stopword, $last_stopword,
|
|---|
| 620 | $first_extractword, $last_extractword,
|
|---|
| 621 | $first_contentword, $last_contentword,
|
|---|
| 622 | $phrasedelimiter);
|
|---|
| 623 |
|
|---|
| 624 | my $thesaurus = $self->{'thesaurus'};
|
|---|
| 625 | my ($thesaurus_links, $thesaurus_terms,
|
|---|
| 626 | %thesaurus, $first_thesaurusword, $last_thesaurusword);
|
|---|
| 627 |
|
|---|
| 628 | my %symbol;
|
|---|
| 629 | my (%freq);
|
|---|
| 630 |
|
|---|
| 631 | print $out "Calculating vocabulary\n" if ($verbosity > 1);
|
|---|
| 632 |
|
|---|
| 633 | # Read and store the stopwords
|
|---|
| 634 | my $stopdir = &util::filename_cat($ENV{'GSDLHOME'}, "etc", "stopwords");
|
|---|
| 635 | my $stopword_files = ();
|
|---|
| 636 | my ($language, $language_dir, $file, $file_name);
|
|---|
| 637 | my %stopwords;
|
|---|
| 638 |
|
|---|
| 639 | # Examine each directory in the stopword directory
|
|---|
| 640 | opendir(STOPDIR, $stopdir);
|
|---|
| 641 | foreach $language (readdir STOPDIR) {
|
|---|
| 642 |
|
|---|
| 643 | # Ignore entries that do not match the classifier's language
|
|---|
| 644 | next unless ($language =~ /$language_exp/);
|
|---|
| 645 | $language_dir = &util::filename_cat($stopdir, $language);
|
|---|
| 646 | next unless (-d "$language_dir");
|
|---|
| 647 |
|
|---|
| 648 | opendir(LANGDIR, $language_dir);
|
|---|
| 649 | foreach $file (readdir LANGDIR) {
|
|---|
| 650 |
|
|---|
| 651 | # Ignore entries that are not stopword files
|
|---|
| 652 | next unless ($file =~ /sw$/);
|
|---|
| 653 | $file_name = &util::filename_cat($language_dir, $file);
|
|---|
| 654 | next unless (-f "$file_name");
|
|---|
| 655 |
|
|---|
| 656 | # Read the stopwords
|
|---|
| 657 | open(STOPFILE, "<$file_name");
|
|---|
| 658 | while (<STOPFILE>) {
|
|---|
| 659 | s/^\s+//;
|
|---|
| 660 | s/\s.*//;
|
|---|
| 661 | $word = $_;
|
|---|
| 662 | $l = lc($word);
|
|---|
| 663 | $stopwords{$l} = $word;
|
|---|
| 664 | }
|
|---|
| 665 | close STOPFILE;
|
|---|
| 666 |
|
|---|
| 667 | }
|
|---|
| 668 | closedir LANGDIR;
|
|---|
| 669 | }
|
|---|
| 670 | closedir STOPDIR;
|
|---|
| 671 |
|
|---|
| 672 | # Read thesaurus information
|
|---|
| 673 | if ($thesaurus) {
|
|---|
| 674 |
|
|---|
| 675 | # link file exists
|
|---|
| 676 | $thesaurus_links = &util::filename_cat($collectiondir, "etc", "$thesaurus.lnk");
|
|---|
| 677 | die "Cannot find thesaurus link file" unless (-e "$thesaurus_links");
|
|---|
| 678 |
|
|---|
| 679 | # ensure term file exists in the correct language
|
|---|
| 680 | if ($language_exp =~ /^([a-z][a-z])/) {
|
|---|
| 681 | $language = $1;
|
|---|
| 682 | } else {
|
|---|
| 683 | $language = 'en';
|
|---|
| 684 | }
|
|---|
| 685 | $thesaurus_terms = &util::filename_cat($collectiondir, "etc", "$thesaurus.$language");
|
|---|
| 686 | die "Cannot find thesaurus term file" unless (-e "$thesaurus_terms");
|
|---|
| 687 |
|
|---|
| 688 |
|
|---|
| 689 | # Read the thesaurus terms
|
|---|
| 690 | open(TH, "<$thesaurus_terms");
|
|---|
| 691 | while(<TH>) {
|
|---|
| 692 | s/^\d+ //;
|
|---|
| 693 | s/\(.*\)//;
|
|---|
| 694 | foreach $w (split(/\s+/, $_)) {
|
|---|
| 695 | $thesaurus{lc($w)} = $w;
|
|---|
| 696 | }
|
|---|
| 697 | }
|
|---|
| 698 | close TH;
|
|---|
| 699 | }
|
|---|
| 700 |
|
|---|
| 701 | # Read words in the text and count occurences
|
|---|
| 702 | open(TXT, "<$phinddir/clauses");
|
|---|
| 703 | my @words;
|
|---|
| 704 |
|
|---|
| 705 | while(<TXT>) {
|
|---|
| 706 | $line = $_;
|
|---|
| 707 | next unless ($line =~ /./);
|
|---|
| 708 |
|
|---|
| 709 | @words = split(/\s+/, $line);
|
|---|
| 710 | foreach $w (@words) {
|
|---|
| 711 | $l = lc($w);
|
|---|
| 712 | $w = $l if ((defined $stopwords{$l}) || (defined $thesaurus{$l}));
|
|---|
| 713 | $freq{$w}++;
|
|---|
| 714 | }
|
|---|
| 715 | $freq{$senlimit}++;
|
|---|
| 716 | }
|
|---|
| 717 |
|
|---|
| 718 | close TXT;
|
|---|
| 719 |
|
|---|
| 720 | # Calculate the "best" form of each word
|
|---|
| 721 | my (%bestform, %totalfreq, %bestfreq);
|
|---|
| 722 |
|
|---|
| 723 | foreach $w (sort (keys %freq)) {
|
|---|
| 724 | $l = lc($w);
|
|---|
| 725 |
|
|---|
| 726 | # totalfreq is the number of times a term appears in any form
|
|---|
| 727 | $totalfreq{$l} += $freq{$w};
|
|---|
| 728 |
|
|---|
| 729 | if (defined $stopwords{$l}) {
|
|---|
| 730 | $bestform{$l} = $stopwords{$l};
|
|---|
| 731 |
|
|---|
| 732 | } elsif (defined $thesaurus{$l}) {
|
|---|
| 733 | $bestform{$l} = $thesaurus{$l};
|
|---|
| 734 |
|
|---|
| 735 | } elsif (!$bestform{$l} || ($freq{$w} > $bestfreq{$l})) {
|
|---|
| 736 | $bestfreq{$l} = $freq{$w};
|
|---|
| 737 | $bestform{$l} = $w;
|
|---|
| 738 | }
|
|---|
| 739 | }
|
|---|
| 740 |
|
|---|
| 741 | undef %freq;
|
|---|
| 742 | undef %bestfreq;
|
|---|
| 743 |
|
|---|
| 744 |
|
|---|
| 745 | # Assign symbol numbers to tokens
|
|---|
| 746 | my $nextsymbol = 1;
|
|---|
| 747 | my (@vocab);
|
|---|
| 748 |
|
|---|
| 749 | # Delimiters
|
|---|
| 750 | $first_delimiter = 1;
|
|---|
| 751 |
|
|---|
| 752 | foreach $word (@delimiters) {
|
|---|
| 753 |
|
|---|
| 754 | $word = lc($word);
|
|---|
| 755 | $bestform{$word} = uc($word);
|
|---|
| 756 | $vocab[$nextsymbol] = $word;
|
|---|
| 757 | $symbol{$word} = $nextsymbol;
|
|---|
| 758 | $nextsymbol++;
|
|---|
| 759 | }
|
|---|
| 760 | $last_delimiter = $nextsymbol - 1;
|
|---|
| 761 |
|
|---|
| 762 | # Stopwords
|
|---|
| 763 | $first_stopword = $nextsymbol;
|
|---|
| 764 |
|
|---|
| 765 | foreach my $word (sort keys %stopwords) {
|
|---|
| 766 |
|
|---|
| 767 | # don't incluse stopword unless it occurs in the text
|
|---|
| 768 | $word = lc($word);
|
|---|
| 769 | next unless ($totalfreq{$word});
|
|---|
| 770 | next if ($symbol{$word});
|
|---|
| 771 |
|
|---|
| 772 | $vocab[$nextsymbol] = $word;
|
|---|
| 773 | $symbol{$word} = $nextsymbol;
|
|---|
| 774 | $nextsymbol++;
|
|---|
| 775 | }
|
|---|
| 776 | $last_stopword = $nextsymbol - 1;
|
|---|
| 777 | $first_contentword = $nextsymbol;
|
|---|
| 778 |
|
|---|
| 779 | # Thesaurus terms
|
|---|
| 780 | if ($thesaurus) {
|
|---|
| 781 | $first_thesaurusword = $nextsymbol;
|
|---|
| 782 |
|
|---|
| 783 | foreach my $word (sort keys %thesaurus) {
|
|---|
| 784 |
|
|---|
| 785 | $word = lc($word);
|
|---|
| 786 | next if ($symbol{$word});
|
|---|
| 787 | $bestform{$word} = $thesaurus{$word};
|
|---|
| 788 |
|
|---|
| 789 | $vocab[$nextsymbol] = $word;
|
|---|
| 790 | $symbol{$word} = $nextsymbol;
|
|---|
| 791 | $nextsymbol++;
|
|---|
| 792 |
|
|---|
| 793 | }
|
|---|
| 794 | $last_thesaurusword = $nextsymbol - 1;
|
|---|
| 795 | }
|
|---|
| 796 |
|
|---|
| 797 | # Other content words
|
|---|
| 798 | $first_extractword = $nextsymbol;
|
|---|
| 799 |
|
|---|
| 800 | foreach my $word (sort (keys %bestform)) {
|
|---|
| 801 |
|
|---|
| 802 | next if ($symbol{$word});
|
|---|
| 803 |
|
|---|
| 804 | $vocab[$nextsymbol] = $word;
|
|---|
| 805 | $symbol{$word} = $nextsymbol;
|
|---|
| 806 | $nextsymbol++;
|
|---|
| 807 | }
|
|---|
| 808 | $last_extractword = $nextsymbol - 1;
|
|---|
| 809 | $last_contentword = $nextsymbol - 1;
|
|---|
| 810 |
|
|---|
| 811 |
|
|---|
| 812 | # Outut the words
|
|---|
| 813 | print $out "Saving vocabulary in $phinddir/clauses.vocab\n" if ($verbosity > 1);
|
|---|
| 814 | open(VOC, ">$phinddir/clauses.vocab");
|
|---|
| 815 |
|
|---|
| 816 | for (my $i = 1; $i < $nextsymbol; $i++) {
|
|---|
| 817 | $w = $vocab[$i];
|
|---|
| 818 |
|
|---|
| 819 | print VOC "$bestform{$w}\n";
|
|---|
| 820 | $totalfreq{$w} = 0 unless ($totalfreq{$w});
|
|---|
| 821 | }
|
|---|
| 822 | close VOC;
|
|---|
| 823 |
|
|---|
| 824 |
|
|---|
| 825 | # Create statistics file
|
|---|
| 826 | # Output statistics about the vocablary
|
|---|
| 827 | print $out "Saving statistics in $phinddir/clauses.stats\n" if ($verbosity > 1);
|
|---|
| 828 | &util::rm("$phinddir/clauses.stats") if (-e "$phinddir/clauses.stats");
|
|---|
| 829 |
|
|---|
| 830 | open(STAT, ">$phinddir/clauses.stats")
|
|---|
| 831 | || die "Cannot open $phinddir/clauses.stats: $!";
|
|---|
| 832 |
|
|---|
| 833 | print STAT "first_delimiter $first_delimiter\n";
|
|---|
| 834 | print STAT "last_delimiter $last_delimiter\n";
|
|---|
| 835 | print STAT "first_stopword $first_stopword\n";
|
|---|
| 836 | print STAT "last_stopword $last_stopword\n";
|
|---|
| 837 | if ($thesaurus) {
|
|---|
| 838 | print STAT "first_thesaurusword $first_thesaurusword\n";
|
|---|
| 839 | print STAT "last_thesaurusword $last_thesaurusword\n";
|
|---|
| 840 | }
|
|---|
| 841 | print STAT "first_extractword $first_extractword\n";
|
|---|
| 842 | print STAT "last_extractword $last_extractword\n";
|
|---|
| 843 | print STAT "first_contentword $first_contentword\n";
|
|---|
| 844 | print STAT "last_contentword $last_contentword\n";
|
|---|
| 845 | print STAT "first_symbol $first_delimiter\n";
|
|---|
| 846 | print STAT "last_symbol $last_contentword\n";
|
|---|
| 847 | print STAT "first_word $first_stopword\n";
|
|---|
| 848 | print STAT "last_word $last_contentword\n";
|
|---|
| 849 | close STAT;
|
|---|
| 850 |
|
|---|
| 851 | undef @vocab;
|
|---|
| 852 |
|
|---|
| 853 |
|
|---|
| 854 | # Create numbers file
|
|---|
| 855 | # Save text as symbol numbers
|
|---|
| 856 | print $out "Saving text as numbers in $phinddir/clauses.numbers\n" if ($verbosity > 1);
|
|---|
| 857 |
|
|---|
| 858 | open(TXT, "<$phinddir/clauses");
|
|---|
| 859 | open(NUM, ">$phinddir/clauses.numbers");
|
|---|
| 860 |
|
|---|
| 861 | $phrasedelimiter = $symbol{lc($senlimit)};
|
|---|
| 862 | print NUM "$symbol{lc($colstart)}\n";
|
|---|
| 863 |
|
|---|
| 864 | # set up the special symbols that delimit documents and sentences
|
|---|
| 865 | while(<TXT>) {
|
|---|
| 866 |
|
|---|
| 867 | # split sentence into a list of tokens
|
|---|
| 868 | $line = $_;
|
|---|
| 869 | next unless ($line =~ /./);
|
|---|
| 870 | @words = split(/\s+/, $line);
|
|---|
| 871 |
|
|---|
| 872 | # output one token at a time
|
|---|
| 873 | foreach $word (@words) {
|
|---|
| 874 | $word = lc($word);
|
|---|
| 875 | print NUM "$symbol{$word}\n";
|
|---|
| 876 | }
|
|---|
| 877 |
|
|---|
| 878 | # output phrase delimiter
|
|---|
| 879 | print NUM "$phrasedelimiter\n";
|
|---|
| 880 | }
|
|---|
| 881 |
|
|---|
| 882 | close TXT;
|
|---|
| 883 | print NUM "$symbol{lc($colend)}\n";
|
|---|
| 884 | close NUM;
|
|---|
| 885 |
|
|---|
| 886 | # Save thesaurus data in one convienient file
|
|---|
| 887 | if ($thesaurus) {
|
|---|
| 888 |
|
|---|
| 889 | my $thesaurusfile = &util::filename_cat($phinddir, "$thesaurus.numbers");
|
|---|
| 890 |
|
|---|
| 891 |
|
|---|
| 892 | print $out "Saving thesaurus as numbers in $thesaurusfile\n"
|
|---|
| 893 | if ($verbosity > 1);
|
|---|
| 894 |
|
|---|
| 895 | # Read the thesaurus terms
|
|---|
| 896 | my ($num, $text, %thes_symbols);
|
|---|
| 897 |
|
|---|
| 898 | open(TH, "<$thesaurus_terms");
|
|---|
| 899 | while(<TH>) {
|
|---|
| 900 | chomp;
|
|---|
| 901 | @words = split(/\s+/, $_);
|
|---|
| 902 | $num = shift @words;
|
|---|
| 903 | $text = "";
|
|---|
| 904 |
|
|---|
| 905 | # translate words into symbol numbers
|
|---|
| 906 | foreach $word (@words) {
|
|---|
| 907 | $word = lc($word);
|
|---|
| 908 | if ($symbol{$word}) {
|
|---|
| 909 | $text .= "s$symbol{$word} ";
|
|---|
| 910 | } elsif ($verbosity) {
|
|---|
| 911 | print $out "phind: No thesaurus symbol, ignoring \"$word\"\n";
|
|---|
| 912 | }
|
|---|
| 913 | }
|
|---|
| 914 | $text =~ s/ $//;
|
|---|
| 915 | $thes_symbols{$num} = $text;
|
|---|
| 916 | }
|
|---|
| 917 | close TH;
|
|---|
| 918 |
|
|---|
| 919 | # Read the thesaurus links and write the corresponding data
|
|---|
| 920 | open(TH, "<$thesaurus_links");
|
|---|
| 921 | open(THOUT, ">$thesaurusfile");
|
|---|
| 922 |
|
|---|
| 923 | while(<TH>) {
|
|---|
| 924 | chomp;
|
|---|
| 925 | ($num, $text) = split(/:/, $_);
|
|---|
| 926 |
|
|---|
| 927 | if (defined($thes_symbols{$num})) {
|
|---|
| 928 | print THOUT "$num:$thes_symbols{$num}:$text\n";
|
|---|
| 929 | } else {
|
|---|
| 930 | print THOUT "$num:untranslated:$text\n";
|
|---|
| 931 | }
|
|---|
| 932 | }
|
|---|
| 933 | close TH;
|
|---|
| 934 | close THOUT;
|
|---|
| 935 | }
|
|---|
| 936 |
|
|---|
| 937 |
|
|---|
| 938 |
|
|---|
| 939 |
|
|---|
| 940 | }
|
|---|
| 941 |
|
|---|
| 942 |
|
|---|
| 943 | # renumber_phrases
|
|---|
| 944 | #
|
|---|
| 945 | # Prepare the phrases file to be input to mgpp. The biggest problem is
|
|---|
| 946 | # reconciling the phrase identifiers used by the suffix program (which
|
|---|
| 947 | # we'll call suffix-id numbers) with the numbers used in the thesaurus
|
|---|
| 948 | # (theesaurus-id) to create a ciommon set of phind id numbers (phind-id).
|
|---|
| 949 | # Phind-id numbers must be sorted by frequency of occurance.
|
|---|
| 950 | #
|
|---|
| 951 | # Start creating a set of phind-id numbers from the sorted suffix-id
|
|---|
| 952 | # numbers and (if required) the thesaurus-id numbers. Then add any other
|
|---|
| 953 | # phrases occuring in the thesaurus.
|
|---|
| 954 | #
|
|---|
| 955 | # The last thing we have to do is restore the vocabulary information to the
|
|---|
| 956 | # phrase file so that the phrases are stored as words, not as symbol
|
|---|
| 957 | # numbers.
|
|---|
| 958 |
|
|---|
| 959 | # The original phrases file looks something like this:
|
|---|
| 960 | # 159396-1:s5175:4:1:116149-2:3:d2240,2;d2253;d2254
|
|---|
| 961 | # 159409-1:s5263:6:1:159410-2:6:d2122;d2128;d2129;d2130;d2215;d2380
|
|---|
| 962 | # 159415-1:s5267:9:1:159418-2:8:d3,2;d632;d633;d668;d1934;d2010;d2281;d2374
|
|---|
| 963 | # 159426-1:s5273:5:2:159429-2,115168-17:5:d252;d815;d938;d939;d2361
|
|---|
| 964 |
|
|---|
| 965 |
|
|---|
| 966 | sub renumber_phrases {
|
|---|
| 967 | my ($self) = @_;
|
|---|
| 968 |
|
|---|
| 969 | renumber_suffix_data($self);
|
|---|
| 970 | renumber_thesaurus_data($self);
|
|---|
| 971 | restore_vocabulary_data($self);
|
|---|
| 972 |
|
|---|
| 973 | }
|
|---|
| 974 |
|
|---|
| 975 |
|
|---|
| 976 |
|
|---|
| 977 | # renumber_suffix_data
|
|---|
| 978 | #
|
|---|
| 979 | # Translate phrases file to phrases.2 using phind keys instead
|
|---|
| 980 | # of suffix keys and sorting the expansion data.
|
|---|
| 981 |
|
|---|
| 982 | sub renumber_suffix_data {
|
|---|
| 983 | my ($self) = @_;
|
|---|
| 984 |
|
|---|
| 985 | my $verbosity = $self->{'verbosity'};
|
|---|
| 986 | my $out = $self->{'outhandle'};
|
|---|
| 987 | print $out "Translate phrases: suffix-ids become phind-id's\n"
|
|---|
| 988 | if ($verbosity);
|
|---|
| 989 |
|
|---|
| 990 | my $phinddir = $self->{'phinddir'};
|
|---|
| 991 | my $infile = &util::filename_cat($phinddir, 'phrases');
|
|---|
| 992 | my $outfile = &util::filename_cat($phinddir, 'phrases.2');
|
|---|
| 993 |
|
|---|
| 994 | # Read the phrase file. Calculate initial set of phind-id
|
|---|
| 995 | # numbers and store (suffixid -> frequency) relation.
|
|---|
| 996 |
|
|---|
| 997 | my %suffixtophind;
|
|---|
| 998 | my @phindfrequency;
|
|---|
| 999 | my (@fields, $suffixid);
|
|---|
| 1000 | my $nextphind = 1;
|
|---|
| 1001 |
|
|---|
| 1002 | open(IN, "<$infile");
|
|---|
| 1003 | while(<IN>) {
|
|---|
| 1004 |
|
|---|
| 1005 | chomp;
|
|---|
| 1006 | @fields = split(/:/, $_);
|
|---|
| 1007 |
|
|---|
| 1008 | # get next suffixid and phindid
|
|---|
| 1009 | $suffixid = shift @fields;
|
|---|
| 1010 | $suffixtophind{$suffixid} = $nextphind;
|
|---|
| 1011 |
|
|---|
| 1012 | # store total frequency
|
|---|
| 1013 | shift @fields;
|
|---|
| 1014 | $totalfrequency[$nextphind] = shift @fields;
|
|---|
| 1015 |
|
|---|
| 1016 | $nextphind++;
|
|---|
| 1017 | }
|
|---|
| 1018 | close IN;
|
|---|
| 1019 |
|
|---|
| 1020 |
|
|---|
| 1021 | # Translate phrases file to phrases.2. Use phind keys (not suffix
|
|---|
| 1022 | # keys), sort expansion and document occurance data in order of
|
|---|
| 1023 | # descending frequency..
|
|---|
| 1024 | open(IN, "<$infile");
|
|---|
| 1025 | open(OUT, ">$outfile");
|
|---|
| 1026 |
|
|---|
| 1027 | my ($phindid, $text, $tf, $countexp, $expansions, $countdocs, $documents);
|
|---|
| 1028 | my (@documwents, @newexp, $k, $n);
|
|---|
| 1029 | my $linenumber = 0;
|
|---|
| 1030 |
|
|---|
| 1031 | while(<IN>) {
|
|---|
| 1032 |
|
|---|
| 1033 | # read the line
|
|---|
| 1034 | chomp;
|
|---|
| 1035 | @fields = split(/:/, $_);
|
|---|
| 1036 |
|
|---|
| 1037 | # get a phrase number for this line
|
|---|
| 1038 | $suffixid = shift @fields;
|
|---|
| 1039 | die unless (defined($suffixtophind{$suffixid}));
|
|---|
| 1040 | $phindid = $suffixtophind{$suffixid};
|
|---|
| 1041 |
|
|---|
| 1042 | # get the symbols in the phrase
|
|---|
| 1043 | $text = shift @fields;
|
|---|
| 1044 |
|
|---|
| 1045 | # output status information
|
|---|
| 1046 | $linenumber++;
|
|---|
| 1047 | if ($verbosity > 2) {
|
|---|
| 1048 | if ($linenumber % 1000 == 0) {
|
|---|
| 1049 | print $out "line $linenumber:\t$phindid\t$suffixid\t($text)\n";
|
|---|
| 1050 | }
|
|---|
| 1051 | print $out "$num: $key\t($text)\n" if ($verbosity > 3);
|
|---|
| 1052 | }
|
|---|
| 1053 |
|
|---|
| 1054 | # get the phrase frequency
|
|---|
| 1055 | $tf = shift @fields;
|
|---|
| 1056 |
|
|---|
| 1057 | # get the number of expansions
|
|---|
| 1058 | $countexp = shift @fields;
|
|---|
| 1059 |
|
|---|
| 1060 | # get the expansions, convert them into phind-id numbers, and sort them
|
|---|
| 1061 | $expansions = shift @fields;
|
|---|
| 1062 | @newexp = ();
|
|---|
| 1063 | foreach $k (split(/,/, $expansions)) {
|
|---|
| 1064 | die "ERROR - no phindid for: $k" unless (defined($suffixtophind{$k}));
|
|---|
| 1065 | $n = $suffixtophind{$k};
|
|---|
| 1066 | push @newexp, $n;
|
|---|
| 1067 | }
|
|---|
| 1068 | @newexp = sort {$totalfrequency[$b] <=> $totalfrequency[$a]} @newexp;
|
|---|
| 1069 |
|
|---|
| 1070 | # get the number of documents
|
|---|
| 1071 | $countdocs = shift @fields;
|
|---|
| 1072 |
|
|---|
| 1073 | # get the documents and sort them
|
|---|
| 1074 | $documents = shift @fields;
|
|---|
| 1075 | $documents =~ s/d//g;
|
|---|
| 1076 | @documents = split(/;/, $documents);
|
|---|
| 1077 | @documents = sort by_doc_frequency @documents;
|
|---|
| 1078 |
|
|---|
| 1079 | # output the phrase data
|
|---|
| 1080 | print OUT "$phindid:$text:$tf:$countexp:$countdocs:";
|
|---|
| 1081 | print OUT join(",", @newexp), ",:", join(";", @documents), ";\n";
|
|---|
| 1082 |
|
|---|
| 1083 | }
|
|---|
| 1084 |
|
|---|
| 1085 | close IN;
|
|---|
| 1086 | close OUT;
|
|---|
| 1087 | }
|
|---|
| 1088 |
|
|---|
| 1089 |
|
|---|
| 1090 | # renumber_thesaurus_data
|
|---|
| 1091 | #
|
|---|
| 1092 | # Translate phrases.2 to phrases.3, adding thesaurus data if available.
|
|---|
| 1093 |
|
|---|
| 1094 | sub renumber_thesaurus_data {
|
|---|
| 1095 | my ($self) = @_;
|
|---|
| 1096 |
|
|---|
| 1097 | my $out = $self->{'outhandle'};
|
|---|
| 1098 | my $verbosity = $self->{'verbosity'};
|
|---|
| 1099 | my $thesaurus = $self->{'thesaurus'};
|
|---|
| 1100 |
|
|---|
| 1101 | my $phinddir = $self->{'phinddir'};
|
|---|
| 1102 | my $infile = &util::filename_cat($phinddir, "phrases.2");
|
|---|
| 1103 | my $outfile = &util::filename_cat($phinddir, "phrases.3");
|
|---|
| 1104 |
|
|---|
| 1105 |
|
|---|
| 1106 | # If no thesaurus is defined, simply move the phrases file.
|
|---|
| 1107 | if (!$thesaurus) {
|
|---|
| 1108 | print $out "Translate phrases.2: no thesaurus data\n"
|
|---|
| 1109 | if ($verbosity);
|
|---|
| 1110 | &util::mv($infile, $outfile);
|
|---|
| 1111 | return;
|
|---|
| 1112 | }
|
|---|
| 1113 |
|
|---|
| 1114 | print $out "Translate phrases.2: add thesaurus data\n"
|
|---|
| 1115 | if ($verbosity);
|
|---|
| 1116 |
|
|---|
| 1117 | # 1.
|
|---|
| 1118 | # Read thesaurus file and store (symbols->thesaurusid) mapping
|
|---|
| 1119 | my $thesaurusfile = &util::filename_cat($phinddir, "$thesaurus.numbers");
|
|---|
| 1120 | my %symbolstothesid;
|
|---|
| 1121 | my (@fields, $thesid, $symbols);
|
|---|
| 1122 |
|
|---|
| 1123 | open(TH, "<$thesaurusfile");
|
|---|
| 1124 |
|
|---|
| 1125 | while (<TH>) {
|
|---|
| 1126 |
|
|---|
| 1127 | chomp;
|
|---|
| 1128 | @fields = split(/:/, $_);
|
|---|
| 1129 |
|
|---|
| 1130 | # get id and text
|
|---|
| 1131 | $thesid = shift @fields;
|
|---|
| 1132 | $symbols = shift @fields;
|
|---|
| 1133 | $symbolstothesid{$symbols} = $thesid;
|
|---|
| 1134 | }
|
|---|
| 1135 | close TH;
|
|---|
| 1136 |
|
|---|
| 1137 | # 2.
|
|---|
| 1138 | # Read phrases file to find thesaurus entries that already
|
|---|
| 1139 | # have a phindid. Store their phind-ids for later translation,
|
|---|
| 1140 | # and store their frequency for later sorting.
|
|---|
| 1141 | my %thesaurustophindid;
|
|---|
| 1142 | my %phindidtofrequency;
|
|---|
| 1143 | my ($phindid, $freq);
|
|---|
| 1144 |
|
|---|
| 1145 | open(IN, "<$infile");
|
|---|
| 1146 |
|
|---|
| 1147 | while(<IN>) {
|
|---|
| 1148 |
|
|---|
| 1149 | chomp;
|
|---|
| 1150 | @fields = split(/:/, $_);
|
|---|
| 1151 |
|
|---|
| 1152 | # phindid and symbols for this line
|
|---|
| 1153 | $phindid = shift @fields;
|
|---|
| 1154 | $symbols = shift @fields;
|
|---|
| 1155 | $freq = shift @fields;
|
|---|
| 1156 |
|
|---|
| 1157 | # do we have a thesaurus id corresponding to this phrase?
|
|---|
| 1158 | if (defined($symbolstothesid{$symbols})) {
|
|---|
| 1159 | $thesid = $symbolstothesid{$symbols};
|
|---|
| 1160 | $thesaurustophindid{$thesid} = $phindid;
|
|---|
| 1161 | $phindidtofrequency{$phindid} = $freq;
|
|---|
| 1162 | }
|
|---|
| 1163 | }
|
|---|
| 1164 | close IN;
|
|---|
| 1165 |
|
|---|
| 1166 | undef %symbolstothesid;
|
|---|
| 1167 |
|
|---|
| 1168 | # 3.
|
|---|
| 1169 | # Create phind-id numbers for remaining thesaurus entries,
|
|---|
| 1170 | # and note that their frequency is 0 for later sorting.
|
|---|
| 1171 | my $nextphindid = $phindid + 1;
|
|---|
| 1172 |
|
|---|
| 1173 | open(TH, "<$thesaurusfile");
|
|---|
| 1174 | while(<TH>) {
|
|---|
| 1175 |
|
|---|
| 1176 | chomp;
|
|---|
| 1177 | @fields = split(/:/, $_);
|
|---|
| 1178 |
|
|---|
| 1179 | # read thesaurus-id and ensure it has a corresponding phind-id
|
|---|
| 1180 | $thesid = shift @fields;
|
|---|
| 1181 | if (!defined($thesaurustophindid{$thesid})) {
|
|---|
| 1182 | $thesaurustophindid{$thesid} = $nextphindid;
|
|---|
| 1183 | $phindidtofrequency{$nextphindid} = 0;
|
|---|
| 1184 | $nextphindid++;
|
|---|
| 1185 | }
|
|---|
| 1186 | }
|
|---|
| 1187 | close TH;
|
|---|
| 1188 |
|
|---|
| 1189 | # 4.
|
|---|
| 1190 | # Translate thesaurus file, replacing thesaurus-id numbers with
|
|---|
| 1191 | # phind-id numbers.
|
|---|
| 1192 | my $newthesaurusfile = &util::filename_cat($phinddir, "$thesaurus.phindid");
|
|---|
| 1193 | my ($relations, $linkcounter, $linktext, $linktype, @linkdata);
|
|---|
| 1194 | my (@links, $linkid, %linkidtotype, $newrelation);
|
|---|
| 1195 |
|
|---|
| 1196 | open(TH, "<$thesaurusfile");
|
|---|
| 1197 | open(TO, ">$newthesaurusfile");
|
|---|
| 1198 | while(<TH>) {
|
|---|
| 1199 |
|
|---|
| 1200 | chomp;
|
|---|
| 1201 | @fields = split(/:/, $_);
|
|---|
| 1202 |
|
|---|
| 1203 | # phindid and symbols for this line
|
|---|
| 1204 | ($thesid, $symbols, $relations) = @fields;
|
|---|
| 1205 |
|
|---|
| 1206 | die unless ($thesid && $symbols);
|
|---|
| 1207 | die unless $thesaurustophindid{$thesid};
|
|---|
| 1208 | $phindid = $thesaurustophindid{$thesid};
|
|---|
| 1209 |
|
|---|
| 1210 | # convert each part of the relation string to use phind-id numbers
|
|---|
| 1211 | # at the same time, we want to sort the list by frequency.
|
|---|
| 1212 | undef %linkidtotype;
|
|---|
| 1213 |
|
|---|
| 1214 | foreach $linktext (split(/;/, $relations)) {
|
|---|
| 1215 | @linkdata = split(/,/, $linktext);
|
|---|
| 1216 |
|
|---|
| 1217 | # remember the linktype (e.g. BT, NT)
|
|---|
| 1218 | $linktype = shift @linkdata;
|
|---|
| 1219 |
|
|---|
| 1220 | # store the type of each link
|
|---|
| 1221 | foreach $thesid (@linkdata) {
|
|---|
| 1222 | die unless (defined($thesaurustophindid{$thesid}));
|
|---|
| 1223 | $linkidtotype{$thesaurustophindid{$thesid}} = $linktype;
|
|---|
| 1224 | }
|
|---|
| 1225 | }
|
|---|
| 1226 |
|
|---|
| 1227 | # sort the list of links, first by frequency, then by type.
|
|---|
| 1228 | @links = sort { ($phindidtofrequency{$b} <=> $phindidtofrequency{$a})
|
|---|
| 1229 | or ($linkidtotype{$a} cmp $linkidtotype{$b}) } (keys %linkidtotype);
|
|---|
| 1230 | $linkcounter = (scalar @links);
|
|---|
| 1231 |
|
|---|
| 1232 | # create a string describing the link information
|
|---|
| 1233 | $linktype = $linkidtotype{$links[0]};
|
|---|
| 1234 | $newrelation = $linktype;
|
|---|
| 1235 | foreach $linkid (@links) {
|
|---|
| 1236 | if ($linkidtotype{$linkid} ne $linktype) {
|
|---|
| 1237 | $linktype = $linkidtotype{$linkid};
|
|---|
| 1238 | $newrelation .= ";" . $linktype;
|
|---|
| 1239 | }
|
|---|
| 1240 | $newrelation .= "," . $linkid;
|
|---|
| 1241 | }
|
|---|
| 1242 | $newrelation .= ";";
|
|---|
| 1243 |
|
|---|
| 1244 |
|
|---|
| 1245 | # output the new line
|
|---|
| 1246 | print TO "$phindid:$symbols:$linkcounter:$newrelation:\n";
|
|---|
| 1247 | }
|
|---|
| 1248 | close TH;
|
|---|
| 1249 | close TO;
|
|---|
| 1250 |
|
|---|
| 1251 | undef %thesaurustophindid;
|
|---|
| 1252 | undef %linkidtotype;
|
|---|
| 1253 | undef %phindidtofrequency;
|
|---|
| 1254 |
|
|---|
| 1255 | # 5.
|
|---|
| 1256 | # Read thesaurus data (in phind-id format) into memory
|
|---|
| 1257 | my %thesaurusdata;
|
|---|
| 1258 |
|
|---|
| 1259 | open(TH, "<$newthesaurusfile");
|
|---|
| 1260 | while(<TH>) {
|
|---|
| 1261 | chomp;
|
|---|
| 1262 | ($phindid, $symbols, $linkcounter, $relations) = split(/:/, $_);
|
|---|
| 1263 | die unless ($phindid && $symbols);
|
|---|
| 1264 | $thesaurusdata{$phindid} = "$symbols:$linkcounter:$relations";
|
|---|
| 1265 | }
|
|---|
| 1266 |
|
|---|
| 1267 | # 6.
|
|---|
| 1268 | # Add thesaurus data to phrases file
|
|---|
| 1269 | my ($text, $tf, $countexp, $expansions, $countdocs, $documents);
|
|---|
| 1270 | my (@documwents, @newexp, $k, $n);
|
|---|
| 1271 | my $linenumber = 0;
|
|---|
| 1272 |
|
|---|
| 1273 | open(IN, "<$infile");
|
|---|
| 1274 | open(OUT, ">$outfile");
|
|---|
| 1275 |
|
|---|
| 1276 | # Update existing phrases
|
|---|
| 1277 | while(<IN>) {
|
|---|
| 1278 |
|
|---|
| 1279 | chomp;
|
|---|
| 1280 | @fields = split(/:/, $_);
|
|---|
| 1281 |
|
|---|
| 1282 | # get data for this line
|
|---|
| 1283 | $phindid = shift @fields;
|
|---|
| 1284 |
|
|---|
| 1285 | # output the phrase data, with thesaurus information
|
|---|
| 1286 | print OUT "$phindid:", join(":", @fields);
|
|---|
| 1287 |
|
|---|
| 1288 | # add thesaurus data
|
|---|
| 1289 | if (defined($thesaurusdata{$phindid})) {
|
|---|
| 1290 | @fields = split(/:/, $thesaurusdata{$phindid});
|
|---|
| 1291 | shift @fields;
|
|---|
| 1292 | $linkcounter = shift @fields;
|
|---|
| 1293 | $relations = shift @fields;
|
|---|
| 1294 |
|
|---|
| 1295 | print OUT ":$linkcounter:$relations";
|
|---|
| 1296 | $thesaurusdata{$phindid} = "";
|
|---|
| 1297 | }
|
|---|
| 1298 | print OUT "\n";
|
|---|
| 1299 | }
|
|---|
| 1300 | close IN;
|
|---|
| 1301 |
|
|---|
| 1302 | # Add phrases that aren't already in the file
|
|---|
| 1303 | foreach $phindid (sort numerically keys %thesaurusdata) {
|
|---|
| 1304 | next unless ($thesaurusdata{$phindid});
|
|---|
| 1305 |
|
|---|
| 1306 | @fields = split(/:/, $thesaurusdata{$phindid});
|
|---|
| 1307 | $symbols = shift @fields;
|
|---|
| 1308 | $linkcounter = shift @fields;
|
|---|
| 1309 | $relations = shift @fields;
|
|---|
| 1310 |
|
|---|
| 1311 | print OUT "$phindid:$symbols:0:0:0:::$linkcounter:$relations\n";
|
|---|
| 1312 | }
|
|---|
| 1313 | close OUT;
|
|---|
| 1314 |
|
|---|
| 1315 | }
|
|---|
| 1316 |
|
|---|
| 1317 | # restore_vocabulary_data
|
|---|
| 1318 | #
|
|---|
| 1319 | # Read phrases.3 and restore vocabulary information. Then write
|
|---|
| 1320 | # this data to the MGPP input files (pwrod.txt and pdata.txt) and
|
|---|
| 1321 | # (if requested) to the saved phrases file.
|
|---|
| 1322 |
|
|---|
| 1323 | sub restore_vocabulary_data {
|
|---|
| 1324 | my ($self) = @_;
|
|---|
| 1325 |
|
|---|
| 1326 | my $out = $self->{'outhandle'};
|
|---|
| 1327 | my $verbosity = $self->{'verbosity'};
|
|---|
| 1328 | print $out "Translate phrases.3: restore vocabulary\n" if ($verbosity);
|
|---|
| 1329 |
|
|---|
| 1330 | my $phinddir = $self->{'phinddir'};
|
|---|
| 1331 | my $infile = &util::filename_cat($phinddir, 'phrases.3');
|
|---|
| 1332 | my $vocabfile = &util::filename_cat($phinddir, 'clauses.vocab');
|
|---|
| 1333 | my $datafile = &util::filename_cat($phinddir, 'pdata.txt');
|
|---|
| 1334 | my $wordfile = &util::filename_cat($phinddir, 'pword.txt');
|
|---|
| 1335 |
|
|---|
| 1336 | my $savephrases = $self->{'savephrases'};
|
|---|
| 1337 |
|
|---|
| 1338 | # 1.
|
|---|
| 1339 | # Read the vocabulary file
|
|---|
| 1340 | open(V, "<$vocabfile")
|
|---|
| 1341 | || die "Cannot open $vocabfile: $!";
|
|---|
| 1342 | my @symbol;
|
|---|
| 1343 | my $i = 1;
|
|---|
| 1344 | while(<V>) {
|
|---|
| 1345 | chomp;
|
|---|
| 1346 | $symbol[$i++] = $_;
|
|---|
| 1347 | }
|
|---|
| 1348 |
|
|---|
| 1349 |
|
|---|
| 1350 | # 2.
|
|---|
| 1351 | # Translate phrases.3 to MGPP input files
|
|---|
| 1352 | my ($key, $text, $word, $isThesaurus);
|
|---|
| 1353 | my @fields;
|
|---|
| 1354 | my $linenumber = 0;
|
|---|
| 1355 |
|
|---|
| 1356 | open(IN, "<$infile");
|
|---|
| 1357 | open(DATA, ">$datafile");
|
|---|
| 1358 | open(WORD, ">$wordfile");
|
|---|
| 1359 |
|
|---|
| 1360 | # Save the phrases in a separate text file
|
|---|
| 1361 | if ($savephrases) {
|
|---|
| 1362 | print $out "Saving phrases in $savephrases\n" if ($verbosity);
|
|---|
| 1363 | open(SAVE, ">$savephrases");
|
|---|
| 1364 | }
|
|---|
| 1365 |
|
|---|
| 1366 | while(<IN>) {
|
|---|
| 1367 |
|
|---|
| 1368 | # read the line
|
|---|
| 1369 | chomp;
|
|---|
| 1370 | $line = $_;
|
|---|
| 1371 | @fields = split(/:/, $line);
|
|---|
| 1372 |
|
|---|
| 1373 | # get a phrase number for this line
|
|---|
| 1374 | $key = shift @fields;
|
|---|
| 1375 |
|
|---|
| 1376 | # restore the text of the phrase
|
|---|
| 1377 | $text = shift @fields;
|
|---|
| 1378 | $text =~ s/s(\d+)/$symbol[$1]/g;
|
|---|
| 1379 | if ($text =~ / /) {
|
|---|
| 1380 | $word = "";
|
|---|
| 1381 | } elsif ($text ne 'untranslated') {
|
|---|
| 1382 | $word = $text;
|
|---|
| 1383 | }
|
|---|
| 1384 |
|
|---|
| 1385 | # output the phrase data
|
|---|
| 1386 | print DATA "<Document>";
|
|---|
| 1387 | print DATA "$key:$text:", join(":", @fields), ":\n";
|
|---|
| 1388 |
|
|---|
| 1389 | # output the word index search data
|
|---|
| 1390 | print WORD "<Document>$word\n";
|
|---|
| 1391 |
|
|---|
| 1392 | # output the phrases to a text file
|
|---|
| 1393 | if ($savephrases) {
|
|---|
| 1394 | if ((scalar @fields) == 7) {
|
|---|
| 1395 | $isThesaurus = 1;
|
|---|
| 1396 | } else {
|
|---|
| 1397 | $isThesaurus = 0;
|
|---|
| 1398 | }
|
|---|
| 1399 | print SAVE $fields[0], "\t", $fields[2], "\t$isThesaurus\t$text\n";
|
|---|
| 1400 | }
|
|---|
| 1401 | }
|
|---|
| 1402 | close IN;
|
|---|
| 1403 | close WORD;
|
|---|
| 1404 | close DATA;
|
|---|
| 1405 | close SAVE if ($savephrases);
|
|---|
| 1406 |
|
|---|
| 1407 | }
|
|---|
| 1408 |
|
|---|
| 1409 |
|
|---|
| 1410 |
|
|---|
| 1411 | # sort routines used to renumber phrases
|
|---|
| 1412 |
|
|---|
| 1413 | sub numerically { $a <=> $b }
|
|---|
| 1414 |
|
|---|
| 1415 | sub by_doc_frequency {
|
|---|
| 1416 | my $fa = 1;
|
|---|
| 1417 | if ($a =~ /,/) {
|
|---|
| 1418 | $fa = $a;
|
|---|
| 1419 | $fa =~ s/\d+,//;
|
|---|
| 1420 | }
|
|---|
| 1421 | my $fb = 1;
|
|---|
| 1422 | if ($b =~ /,/) {
|
|---|
| 1423 | $fb = $b;
|
|---|
| 1424 | $fb =~ s/\d+,//;
|
|---|
| 1425 | }
|
|---|
| 1426 |
|
|---|
| 1427 | return ($fb <=> $fa);
|
|---|
| 1428 | }
|
|---|
| 1429 |
|
|---|
| 1430 | 1;
|
|---|
