[25141] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
| 3 | # ReadTxtFile.pm -- base class for import plugins that have plain text files
|
---|
| 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) 1999-2005 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 ReadTextFile;
|
---|
| 27 |
|
---|
| 28 | use strict;
|
---|
| 29 | no strict 'subs';
|
---|
| 30 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
---|
| 31 |
|
---|
| 32 | use Encode;
|
---|
| 33 |
|
---|
| 34 | use multiread;
|
---|
| 35 | use encodings;
|
---|
| 36 | use unicode;
|
---|
| 37 | use textcat;
|
---|
| 38 | use ngramj;
|
---|
| 39 | use doc;
|
---|
| 40 | use ghtml;
|
---|
| 41 | use gsprintf 'gsprintf';
|
---|
| 42 |
|
---|
| 43 | use AutoExtractMetadata;
|
---|
| 44 |
|
---|
| 45 | sub BEGIN {
|
---|
| 46 | @ReadTextFile::ISA = ( 'AutoExtractMetadata' );
|
---|
| 47 | }
|
---|
| 48 |
|
---|
| 49 | my $encoding_plus_auto_list =
|
---|
| 50 | [ { 'name' => "auto",
|
---|
| 51 | 'desc' => "{ReadTextFile.input_encoding.auto}" } ];
|
---|
| 52 | push(@{$encoding_plus_auto_list},@{$BasePlugin::encoding_list});
|
---|
| 53 |
|
---|
| 54 | my $arguments =
|
---|
| 55 | [ { 'name' => "input_encoding",
|
---|
| 56 | 'desc' => "{ReadTextFile.input_encoding}",
|
---|
| 57 | 'type' => "enum",
|
---|
| 58 | 'list' => $encoding_plus_auto_list,
|
---|
| 59 | 'reqd' => "no" ,
|
---|
| 60 | 'deft' => "auto" } ,
|
---|
| 61 | { 'name' => "default_encoding",
|
---|
| 62 | 'desc' => "{ReadTextFile.default_encoding}",
|
---|
| 63 | 'type' => "enum",
|
---|
| 64 | 'list' => $BasePlugin::encoding_list,
|
---|
| 65 | 'reqd' => "no",
|
---|
| 66 | 'deft' => "utf8" },
|
---|
| 67 | { 'name' => "extract_language",
|
---|
| 68 | 'desc' => "{ReadTextFile.extract_language}",
|
---|
| 69 | 'type' => "flag",
|
---|
| 70 | 'reqd' => "no" },
|
---|
| 71 | { 'name' => "default_language",
|
---|
| 72 | 'desc' => "{ReadTextFile.default_language}",
|
---|
| 73 | 'type' => "string",
|
---|
| 74 | 'deft' => "en",
|
---|
| 75 | 'reqd' => "no" }
|
---|
| 76 | ];
|
---|
| 77 |
|
---|
| 78 |
|
---|
| 79 | my $options = { 'name' => "ReadTextFile",
|
---|
| 80 | 'desc' => "{ReadTextFile.desc}",
|
---|
| 81 | 'abstract' => "yes",
|
---|
| 82 | 'inherits' => "no",
|
---|
| 83 | 'args' => $arguments };
|
---|
| 84 |
|
---|
| 85 |
|
---|
| 86 |
|
---|
| 87 | sub new {
|
---|
| 88 | my $class = shift (@_);
|
---|
| 89 | my ($pluginlist,$inputargs,$hashArgOptLists, $auxiliary) = @_;
|
---|
| 90 | push(@$pluginlist, $class);
|
---|
| 91 |
|
---|
| 92 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
| 93 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
| 94 |
|
---|
| 95 | my $self = new AutoExtractMetadata($pluginlist, $inputargs, $hashArgOptLists, $auxiliary);
|
---|
| 96 |
|
---|
| 97 | return bless $self, $class;
|
---|
| 98 |
|
---|
| 99 | }
|
---|
| 100 |
|
---|
| 101 |
|
---|
| 102 |
|
---|
| 103 | # The ReadTextFile read_into_doc_obj() function. This function does all the
|
---|
| 104 | # right things to make general options work for a given plugin. It reads in
|
---|
| 105 | # a file and sets up a slew of metadata all saved in doc_obj, which
|
---|
| 106 | # it then returns as part of a tuple (process_status,doc_obj)
|
---|
| 107 | #
|
---|
| 108 | # Much of this functionality used to reside in read, but it was broken
|
---|
| 109 | # down into a supporting routine to make the code more flexible.
|
---|
| 110 | #
|
---|
| 111 | # recursive plugins (e.g. RecPlug) and specialized plugins like those
|
---|
| 112 | # capable of processing many documents within a single file (e.g.
|
---|
| 113 | # GMLPlug) will normally want to implement their own version of
|
---|
| 114 | # read_into_doc_obj()
|
---|
| 115 | #
|
---|
| 116 | # Note that $base_dir might be "" and that $file might
|
---|
| 117 | # include directories
|
---|
| 118 | sub read_into_doc_obj {
|
---|
| 119 | my $self = shift (@_);
|
---|
| 120 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
---|
| 121 |
|
---|
| 122 | print STDERR "**** In Pei Jones Local version of ReadTextFile\n";
|
---|
| 123 |
|
---|
| 124 | my $outhandle = $self->{'outhandle'};
|
---|
| 125 | # should we move this to read? What about secondary plugins?
|
---|
| 126 | print STDERR "<Processing n='$file' p='$self->{'plugin_type'}'>\n" if ($gli);
|
---|
| 127 | my $pp_file = &util::prettyprint_file($base_dir,$file,$gli);
|
---|
| 128 | print $outhandle "$self->{'plugin_type'} processing $pp_file\n"
|
---|
| 129 | if $self->{'verbosity'} > 1;
|
---|
| 130 |
|
---|
| 131 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
| 132 |
|
---|
| 133 | # Do encoding stuff
|
---|
| 134 | my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
|
---|
| 135 | if ($self->{'verbosity'} > 2) {
|
---|
| 136 | print $outhandle "ReadTextFile: reading $file as ($content_encoding,$language)\n";
|
---|
| 137 | }
|
---|
| 138 |
|
---|
| 139 | # create a new document
|
---|
| 140 | my $doc_obj = new doc ($filename_full_path, "indexed_doc", $self->{'file_rename_method'});
|
---|
| 141 | my $top_section = $doc_obj->get_top_section();
|
---|
| 142 |
|
---|
| 143 | # this should look at the plugin option too...
|
---|
| 144 | $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
|
---|
| 145 | $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename_full_path));
|
---|
| 146 |
|
---|
| 147 | my $plugin_filename_encoding = $self->{'filename_encoding'};
|
---|
| 148 | my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
|
---|
| 149 | $self->set_Source_metadata($doc_obj, $filename_full_path, $filename_encoding);
|
---|
| 150 |
|
---|
| 151 | $doc_obj->add_utf8_metadata($top_section, "Language", $language);
|
---|
| 152 | $doc_obj->add_utf8_metadata($top_section, "Encoding", $content_encoding);
|
---|
| 153 |
|
---|
| 154 | # read in file ($text will be in utf8)
|
---|
| 155 | my $text = "";
|
---|
| 156 | $self->read_file ($filename_full_path, $content_encoding, $language, \$text);
|
---|
| 157 |
|
---|
| 158 | if (!length ($text)) {
|
---|
| 159 | if ($gli) {
|
---|
| 160 | print STDERR "<ProcessingError n='$file' r='File contains no text'>\n";
|
---|
| 161 | }
|
---|
| 162 | gsprintf($outhandle, "$self->{'plugin_type'}: {ReadTextFile.file_has_no_text}\n", $filename_full_path) if $self->{'verbosity'};
|
---|
| 163 |
|
---|
| 164 | my $failhandle = $self->{'failhandle'};
|
---|
| 165 | gsprintf($failhandle, "$file: " . ref($self) . ": {ReadTextFile.empty_file}\n");
|
---|
| 166 | # print $failhandle "$file: " . ref($self) . ": file contains no text\n";
|
---|
| 167 | $self->{'num_not_processed'} ++;
|
---|
| 168 |
|
---|
| 169 | return (0,undef); # what should we return here?? error but don't want to pass it on
|
---|
| 170 | }
|
---|
| 171 |
|
---|
| 172 | # do plugin specific processing of doc_obj
|
---|
| 173 | unless (defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli))) {
|
---|
| 174 | $text = '';
|
---|
| 175 | undef $text;
|
---|
| 176 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
---|
| 177 | return (-1,undef);
|
---|
| 178 | }
|
---|
| 179 | $text='';
|
---|
| 180 | undef $text;
|
---|
| 181 |
|
---|
| 182 | # include any metadata passed in from previous plugins
|
---|
| 183 | # note that this metadata is associated with the top level section
|
---|
| 184 | $self->add_associated_files($doc_obj, $filename_full_path);
|
---|
| 185 | $self->extra_metadata ($doc_obj, $top_section, $metadata);
|
---|
| 186 |
|
---|
| 187 | # do any automatic metadata extraction
|
---|
| 188 | $self->auto_extract_metadata ($doc_obj);
|
---|
| 189 |
|
---|
| 190 |
|
---|
| 191 | # if we haven't found any Title so far, assign one
|
---|
| 192 | $self->title_fallback($doc_obj,$top_section,$filename_no_path);
|
---|
| 193 |
|
---|
| 194 | $self->add_OID($doc_obj);
|
---|
| 195 |
|
---|
| 196 | return (1,$doc_obj);
|
---|
| 197 | }
|
---|
| 198 |
|
---|
| 199 | # uses the multiread package to read in the entire file pointed to
|
---|
| 200 | # by filename and loads the resulting text into $$textref. Input text
|
---|
| 201 | # may be in any of the encodings handled by multiread, output text
|
---|
| 202 | # will be in utf8
|
---|
| 203 | sub read_file {
|
---|
| 204 | my $self = shift (@_);
|
---|
| 205 | my ($filename, $encoding, $language, $textref) = @_;
|
---|
| 206 |
|
---|
| 207 | if (!-r $filename)
|
---|
| 208 | {
|
---|
| 209 | my $outhandle = $self->{'outhandle'};
|
---|
| 210 | gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
|
---|
| 211 | # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
|
---|
| 212 | return;
|
---|
| 213 | }
|
---|
| 214 | $$textref = "";
|
---|
| 215 | if (!open (FILE, $filename)) {
|
---|
| 216 | gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
| 217 | die "\n";
|
---|
| 218 | }
|
---|
| 219 |
|
---|
| 220 | if ($encoding eq "ascii") {
|
---|
| 221 | # Replace file 'slurp' with faster implementation
|
---|
| 222 | sysread(FILE, $$textref, -s FILE);
|
---|
| 223 |
|
---|
| 224 | # The old slow way of reading in a file
|
---|
| 225 | #undef $/;
|
---|
| 226 | #$$textref = <FILE>;
|
---|
| 227 | #$/ = "\n";
|
---|
| 228 | } else {
|
---|
| 229 | my $reader = new multiread();
|
---|
| 230 | $reader->set_handle ('ReadTextFile::FILE');
|
---|
| 231 | $reader->set_encoding ($encoding);
|
---|
| 232 | $reader->read_file ($textref);
|
---|
| 233 | }
|
---|
| 234 |
|
---|
| 235 | # At this point $$textref is a binary byte string
|
---|
| 236 | # => turn it into a Unicode aware string, so full
|
---|
| 237 | # Unicode aware pattern matching can be used.
|
---|
| 238 | # For instance: 's/\x{0101}//g' or '[[:upper:]]'
|
---|
| 239 | #
|
---|
| 240 |
|
---|
| 241 | $$textref = decode("utf8",$$textref);
|
---|
| 242 |
|
---|
| 243 | close FILE;
|
---|
| 244 | }
|
---|
| 245 |
|
---|
| 246 |
|
---|
| 247 | # Not currently used
|
---|
| 248 | sub UNUSED_read_file_usingPerlsEncodeModule {
|
---|
| 249 | ##sub read_file {
|
---|
| 250 | my $self = shift (@_);
|
---|
| 251 | my ($filename, $encoding, $language, $textref) = @_;
|
---|
| 252 |
|
---|
| 253 | if (!-r $filename)
|
---|
| 254 | {
|
---|
| 255 | my $outhandle = $self->{'outhandle'};
|
---|
| 256 | gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
|
---|
| 257 | # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
|
---|
| 258 | return;
|
---|
| 259 | }
|
---|
| 260 | $$textref = "";
|
---|
| 261 | if (!open (FILE, $filename)) {
|
---|
| 262 | gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_f
|
---|
| 263 | or_reading} ($!)\n", $filename);
|
---|
| 264 | die "\n";
|
---|
| 265 | }
|
---|
| 266 |
|
---|
| 267 | my $store_slash = $/;
|
---|
| 268 | undef $/;
|
---|
| 269 | my $text = <FILE>;
|
---|
| 270 | $/ = $store_slash;
|
---|
| 271 |
|
---|
| 272 | $$textref = decode($encoding,$text);
|
---|
| 273 |
|
---|
| 274 | close FILE;
|
---|
| 275 | }
|
---|
| 276 |
|
---|
| 277 |
|
---|
| 278 | sub read_file_no_decoding {
|
---|
| 279 | my $self = shift (@_);
|
---|
| 280 | my ($filename, $textref) = @_;
|
---|
| 281 |
|
---|
| 282 | if (!-r $filename)
|
---|
| 283 | {
|
---|
| 284 | my $outhandle = $self->{'outhandle'};
|
---|
| 285 | gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
|
---|
| 286 | # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
|
---|
| 287 | return;
|
---|
| 288 | }
|
---|
| 289 | $$textref = "";
|
---|
| 290 | if (!open (FILE, $filename)) {
|
---|
| 291 | gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
| 292 | die "\n";
|
---|
| 293 | }
|
---|
| 294 |
|
---|
| 295 | my $reader = new multiread();
|
---|
| 296 | $reader->set_handle ('ReadTextFile::FILE');
|
---|
| 297 | $reader->read_file_no_decoding ($textref);
|
---|
| 298 |
|
---|
| 299 | $self->{'reader'} = $reader;
|
---|
| 300 |
|
---|
| 301 | close FILE;
|
---|
| 302 | }
|
---|
| 303 |
|
---|
| 304 |
|
---|
| 305 | sub decode_text {
|
---|
| 306 | my $self = shift (@_);
|
---|
| 307 | my ($raw_text, $encoding, $language, $textref) = @_;
|
---|
| 308 |
|
---|
| 309 | my $reader = $self->{'reader'};
|
---|
| 310 | if (!defined $reader) {
|
---|
| 311 | gsprintf(STDERR, "ReadTextFile::decode_text needs to call ReadTextFile::read_file_no_decoding first\n");
|
---|
| 312 | }
|
---|
| 313 | else {
|
---|
| 314 | $reader->set_encoding($encoding);
|
---|
| 315 | $reader->decode_text($raw_text,$textref);
|
---|
| 316 |
|
---|
| 317 | # At this point $$textref is a binary byte string
|
---|
| 318 | # => turn it into a Unicode aware string, so full
|
---|
| 319 | # Unicode aware pattern matching can be used.
|
---|
| 320 | # For instance: 's/\x{0101}//g' or '[[:upper:]]'
|
---|
| 321 |
|
---|
| 322 | $$textref = decode("utf8",$$textref);
|
---|
| 323 | }
|
---|
| 324 | }
|
---|
| 325 |
|
---|
| 326 |
|
---|
| 327 | sub textcat_get_language_encoding {
|
---|
| 328 | my $self = shift (@_);
|
---|
| 329 | my ($filename) = @_;
|
---|
| 330 |
|
---|
| 331 | $self->{'textcat'} = new ngramj() if (!defined($self->{'textcat'}));
|
---|
| 332 |
|
---|
| 333 | my ($language, $encoding, $extracted_encoding);
|
---|
| 334 | if ($self->{'input_encoding'} eq "auto") {
|
---|
| 335 | # use textcat to automatically work out the input encoding and language
|
---|
| 336 | ($language, $encoding) = $self->get_language_encoding ($filename);
|
---|
| 337 | } elsif ($self->{'extract_language'}) {
|
---|
| 338 | # use textcat to get language metadata
|
---|
| 339 | ($language, $extracted_encoding) = $self->get_language_encoding ($filename);
|
---|
| 340 | $encoding = $self->{'input_encoding'};
|
---|
| 341 | # don't print this message for english... english in utf8 is identical
|
---|
| 342 | # to english in iso-8859-1 (except for some punctuation). We don't have
|
---|
| 343 | # a language model for en_utf8, so textcat always says iso-8859-1!
|
---|
| 344 | if ($extracted_encoding ne $encoding && $language ne "en" && $self->{'verbosity'}) {
|
---|
| 345 | my $plugin_name = ref ($self);
|
---|
| 346 | my $outhandle = $self->{'outhandle'};
|
---|
| 347 | gsprintf($outhandle, "$plugin_name: {ReadTextFile.wrong_encoding}\n", $filename, $encoding, $extracted_encoding);
|
---|
| 348 | }
|
---|
| 349 | } else {
|
---|
| 350 | $language = $self->{'default_language'};
|
---|
| 351 | $encoding = $self->{'input_encoding'};
|
---|
| 352 | }
|
---|
| 353 |
|
---|
| 354 | # print STDERR "**** language encoding of contents of file $filename:\n\t****$language $encoding\n";
|
---|
| 355 |
|
---|
| 356 | return ($language, $encoding);
|
---|
| 357 | }
|
---|
| 358 |
|
---|
| 359 |
|
---|
| 360 | # Uses textcat to work out the encoding and language of the text in
|
---|
| 361 | # $filename. All html tags are removed before processing.
|
---|
| 362 | # returns an array containing "language" and "encoding"
|
---|
| 363 | sub get_language_encoding {
|
---|
| 364 | my $self = shift (@_);
|
---|
| 365 | my ($filename) = @_;
|
---|
| 366 | my $outhandle = $self->{'outhandle'};
|
---|
| 367 | my $unicode_format = "";
|
---|
| 368 | my $best_language = "";
|
---|
| 369 | my $best_encoding = "";
|
---|
| 370 |
|
---|
| 371 |
|
---|
| 372 | # read in file
|
---|
| 373 | if (!open (FILE, $filename)) {
|
---|
| 374 | gsprintf(STDERR, "ReadTextFile::get_language_encoding {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
| 375 | # this is a pretty bad error, but try to continue anyway
|
---|
| 376 | return ($self->{'default_language'}, $self->{'input_encoding'});
|
---|
| 377 | }
|
---|
| 378 | undef $/;
|
---|
| 379 | my $text = <FILE>;
|
---|
| 380 | $/ = "\n";
|
---|
| 381 | close FILE;
|
---|
| 382 |
|
---|
| 383 | # check if first few bytes have a Byte Order Marker
|
---|
| 384 | my $bom=substr($text,0,2); # check 16bit unicode
|
---|
| 385 | if ($bom eq "\xff\xfe") { # little endian 16bit unicode
|
---|
| 386 | $unicode_format="unicode";
|
---|
| 387 | } elsif ($bom eq "\xfe\xff") { # big endian 16bit unicode
|
---|
| 388 | $unicode_format="unicode";
|
---|
| 389 | } else {
|
---|
| 390 | $bom=substr($text,0,3); # check utf-8
|
---|
| 391 | if ($bom eq "\xef\xbb\xbf") { # utf-8 coded FEFF bom
|
---|
| 392 | $unicode_format="utf8";
|
---|
| 393 | # } elsif ($bom eq "\xef\xbf\xbe") { # utf-8 coded FFFE bom. Error!?
|
---|
| 394 | # $unicode_format="utf8";
|
---|
| 395 | }
|
---|
| 396 | }
|
---|
| 397 |
|
---|
| 398 | my $found_html_encoding = 0;
|
---|
| 399 | # handle html files specially
|
---|
| 400 | # XXX this doesn't match plugins derived from HTMLPlug (except ConvertTo)
|
---|
| 401 | if (ref($self) eq 'HTMLPlugin' ||
|
---|
| 402 | (exists $self->{'converted_to'} && $self->{'converted_to'} eq 'HTML')){
|
---|
| 403 |
|
---|
| 404 | # remove comments in head, including multiline ones, so that we don't match on
|
---|
| 405 | # inactive tags (those that are nested inside comments)
|
---|
| 406 | my ($head) = ($text =~ m/<head>(.*)<\/head>/si);
|
---|
| 407 | $head = "" unless defined $head; # some files are not proper HTML eg php files
|
---|
| 408 | $head =~ s/<!--.*?-->//sg;
|
---|
| 409 |
|
---|
| 410 | # remove <title>stuff</title> -- as titles tend often to be in English
|
---|
| 411 | # for foreign language documents
|
---|
| 412 | $text =~ s!<title>.*?</title>!!si;
|
---|
| 413 |
|
---|
| 414 | # see if this html file specifies its encoding
|
---|
| 415 | if ($text =~ /^<\?xml.*encoding="(.+?)"/) {
|
---|
| 416 | $best_encoding = $1;
|
---|
| 417 | }
|
---|
| 418 | # check the meta http-equiv charset tag
|
---|
| 419 | elsif ($head =~ m/<meta http-equiv.*content-type.*charset=(.+?)\"/si) {
|
---|
| 420 | $best_encoding = $1;
|
---|
| 421 | }
|
---|
| 422 | if ($best_encoding) { # we extracted an encoding
|
---|
| 423 | $best_encoding =~ s/-+/_/g;
|
---|
| 424 | $best_encoding = lc($best_encoding); # lowercase
|
---|
| 425 | if ($best_encoding eq "utf_8") { $best_encoding = "utf8" }
|
---|
| 426 | $found_html_encoding = 1;
|
---|
| 427 | # We shouldn't be modifying this here!!
|
---|
| 428 | #$self->{'input_encoding'} = $best_encoding;
|
---|
| 429 | }
|
---|
| 430 |
|
---|
| 431 | # remove all HTML tags
|
---|
| 432 | $text =~ s/<[^>]*>//sg;
|
---|
| 433 | }
|
---|
| 434 |
|
---|
| 435 | # don't need to do textcat if we know the encoding now AND don't need to extract language
|
---|
| 436 | if($found_html_encoding && !$self->{'extract_language'}) { # encoding specified in html file
|
---|
| 437 | $best_language = $self->{'default_language'};
|
---|
| 438 | }
|
---|
| 439 |
|
---|
| 440 | else { # need to use textcat to get either the language, or get both language and encoding
|
---|
| 441 | $self->{'textcat'} = new textcat() if (!defined($self->{'textcat'}));
|
---|
| 442 |
|
---|
| 443 | if($found_html_encoding) { # know encoding, find language by limiting search to known encoding
|
---|
| 444 | my $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
| 445 |
|
---|
| 446 | my $language;
|
---|
| 447 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(?:.*))?$/ if (scalar @$results > 0);
|
---|
| 448 |
|
---|
| 449 | if (!defined $language || scalar @$results > 3) {
|
---|
| 450 | # if there were too many results even when restricting results by encoding,
|
---|
| 451 | # or if there were no results, use default language with the known encoding
|
---|
| 452 | $best_language = $self->use_default_language($filename);
|
---|
| 453 | }
|
---|
| 454 | else { # fewer than 3 results means textcat is more certain, use the first result
|
---|
| 455 | $best_language = $language;
|
---|
| 456 | }
|
---|
| 457 | }
|
---|
| 458 | else { # don't know encoding or language yet, therefore we use textcat
|
---|
| 459 | my $results = $self->{'textcat'}->classify_contents(\$text, $filename);
|
---|
| 460 |
|
---|
| 461 | # if textcat returns 3 or less possibilities we'll use the first one in the list
|
---|
| 462 | if (scalar @$results <= 3) { # results will be > 0 when we don't constrain textcat by an encoding
|
---|
| 463 | my ($language, $encoding) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/;
|
---|
| 464 |
|
---|
| 465 | $language = $self->use_default_language($filename) unless defined $language;
|
---|
| 466 | $encoding = $self->use_default_encoding($filename) unless defined $encoding;
|
---|
| 467 |
|
---|
| 468 | $best_language = $language;
|
---|
| 469 | $best_encoding = $encoding;
|
---|
| 470 | }
|
---|
| 471 | else { # if (scalar @$results > 3) {
|
---|
| 472 | if ($unicode_format) { # in case the first had a BOM
|
---|
| 473 | $best_encoding=$unicode_format;
|
---|
| 474 | }
|
---|
| 475 | else {
|
---|
| 476 | # Find the most frequent encoding in the textcat results returned
|
---|
| 477 | # Returns "" if there's no encoding more frequent than another
|
---|
| 478 | $best_encoding = $self->{'textcat'}->most_frequent_encoding($results);
|
---|
| 479 | }
|
---|
| 480 |
|
---|
| 481 | if ($best_encoding eq "") { # encoding still not set, use defaults
|
---|
| 482 | $best_language = $self->use_default_language($filename);
|
---|
| 483 | $best_encoding = $self->use_default_encoding($filename);
|
---|
| 484 | }
|
---|
| 485 | elsif (!$self->{'extract_language'}) { # know encoding but don't need to discover language
|
---|
| 486 | $best_language = $self->use_default_language($filename);
|
---|
| 487 | }
|
---|
| 488 | else { # textcat again using the most frequent encoding or the $unicode_format set above
|
---|
| 489 | $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
| 490 | my $language;
|
---|
| 491 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/ if (scalar @$results > 0);
|
---|
| 492 | if (!defined $language || scalar @$results > 3) {
|
---|
| 493 | # if no result or too many results, use default language for the encoding previously found
|
---|
| 494 | $best_language = $self->use_default_language($filename);
|
---|
| 495 | }
|
---|
| 496 | else { # fewer than 3 results, use the language of the first result
|
---|
| 497 | $best_language = $language;
|
---|
| 498 | }
|
---|
| 499 | }
|
---|
| 500 | }
|
---|
| 501 | }
|
---|
| 502 | }
|
---|
| 503 |
|
---|
| 504 | if($best_encoding eq "" || $best_language eq "") {
|
---|
| 505 | print STDERR "****Shouldn't happen: encoding and/or language still not set. Using defaults.\n";
|
---|
| 506 | $best_encoding = $self->use_default_encoding($filename) if $best_encoding eq "";
|
---|
| 507 | $best_language = $self->use_default_language($filename) if $best_language eq "";
|
---|
| 508 | }
|
---|
| 509 | # print STDERR "****Content language: $best_language; Encoding: $best_encoding.\n";
|
---|
| 510 |
|
---|
| 511 |
|
---|
| 512 | if ($best_encoding =~ /^iso_8859/ && &unicode::check_is_utf8($text)) {
|
---|
| 513 | # the text is valid utf8, so assume that's the real encoding
|
---|
| 514 | # (since textcat is based on probabilities)
|
---|
| 515 | $best_encoding = 'utf8';
|
---|
| 516 | }
|
---|
| 517 |
|
---|
| 518 | # check for equivalents where textcat doesn't have some encodings...
|
---|
| 519 | # eg MS versions of standard encodings
|
---|
| 520 | if ($best_encoding =~ /^iso_8859_(\d+)/) {
|
---|
| 521 | my $iso = $1; # which variant of the iso standard?
|
---|
| 522 | # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
|
---|
| 523 | if ($text =~ /[\x80-\x9f]/) {
|
---|
| 524 | # Western Europe
|
---|
| 525 | if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
|
---|
| 526 | elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
|
---|
| 527 | elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
|
---|
| 528 | elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
|
---|
| 529 | elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
|
---|
| 530 | elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
|
---|
| 531 | elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
|
---|
| 532 | }
|
---|
| 533 | }
|
---|
| 534 |
|
---|
| 535 | if ($best_encoding !~ /^(ascii|utf8|unicode)$/ &&
|
---|
| 536 | !defined $encodings::encodings->{$best_encoding}) {
|
---|
| 537 | if ($self->{'verbosity'}) {
|
---|
| 538 | gsprintf($outhandle, "ReadTextFile: {ReadTextFile.unsupported_encoding}\n",
|
---|
| 539 | $filename, $best_encoding, $self->{'default_encoding'});
|
---|
| 540 | }
|
---|
| 541 | $best_encoding = $self->{'default_encoding'};
|
---|
| 542 | }
|
---|
| 543 |
|
---|
| 544 | return ($best_language, $best_encoding);
|
---|
| 545 | }
|
---|
| 546 |
|
---|
| 547 |
|
---|
| 548 |
|
---|
| 549 | # Uses java ngram solution to work out the language of the text in
|
---|
| 550 | # $filename. Fall back on textcat to to the character encoding
|
---|
| 551 | # All html tags are removed before processing.
|
---|
| 552 | # returns an array containing "language" and "encoding"
|
---|
| 553 | sub get_language_encoding {
|
---|
| 554 | my $self = shift (@_);
|
---|
| 555 | my ($filename) = @_;
|
---|
| 556 | my $outhandle = $self->{'outhandle'};
|
---|
| 557 | my $unicode_format = "";
|
---|
| 558 | my $best_language = "";
|
---|
| 559 | my $best_encoding = "";
|
---|
| 560 |
|
---|
| 561 |
|
---|
| 562 | # read in file
|
---|
| 563 | if (!open (FILE, $filename)) {
|
---|
| 564 | gsprintf(STDERR, "ReadTextFile::java_ngram_get_language_encoding {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
| 565 | # this is a pretty bad error, but try to continue anyway
|
---|
| 566 | return ($self->{'default_language'}, $self->{'input_encoding'});
|
---|
| 567 | }
|
---|
| 568 | undef $/;
|
---|
| 569 | my $text = <FILE>;
|
---|
| 570 | $/ = "\n";
|
---|
| 571 | close FILE;
|
---|
| 572 |
|
---|
| 573 | # check if first few bytes have a Byte Order Marker
|
---|
| 574 | my $bom=substr($text,0,2); # check 16bit unicode
|
---|
| 575 | if ($bom eq "\xff\xfe") { # little endian 16bit unicode
|
---|
| 576 | $unicode_format="unicode";
|
---|
| 577 | } elsif ($bom eq "\xfe\xff") { # big endian 16bit unicode
|
---|
| 578 | $unicode_format="unicode";
|
---|
| 579 | } else {
|
---|
| 580 | $bom=substr($text,0,3); # check utf-8
|
---|
| 581 | if ($bom eq "\xef\xbb\xbf") { # utf-8 coded FEFF bom
|
---|
| 582 | $unicode_format="utf8";
|
---|
| 583 | # } elsif ($bom eq "\xef\xbf\xbe") { # utf-8 coded FFFE bom. Error!?
|
---|
| 584 | # $unicode_format="utf8";
|
---|
| 585 | }
|
---|
| 586 | }
|
---|
| 587 |
|
---|
| 588 | my $found_html_encoding = 0;
|
---|
| 589 | # handle html files specially
|
---|
| 590 | # XXX this doesn't match plugins derived from HTMLPlug (except ConvertTo)
|
---|
| 591 | if (ref($self) eq 'HTMLPlugin' ||
|
---|
| 592 | (exists $self->{'converted_to'} && $self->{'converted_to'} eq 'HTML')){
|
---|
| 593 |
|
---|
| 594 | # remove comments in head, including multiline ones, so that we don't match on
|
---|
| 595 | # inactive tags (those that are nested inside comments)
|
---|
| 596 | my ($head) = ($text =~ m/<head>(.*)<\/head>/si);
|
---|
| 597 | $head = "" unless defined $head; # some files are not proper HTML eg php files
|
---|
| 598 | $head =~ s/<!--.*?-->//sg;
|
---|
| 599 |
|
---|
| 600 | # remove <title>stuff</title> -- as titles tend often to be in English
|
---|
| 601 | # for foreign language documents
|
---|
| 602 | $text =~ s!<title>.*?</title>!!si;
|
---|
| 603 |
|
---|
| 604 | # see if this html file specifies its encoding
|
---|
| 605 | if ($text =~ /^<\?xml.*encoding="(.+?)"/) {
|
---|
| 606 | $best_encoding = $1;
|
---|
| 607 | }
|
---|
| 608 | # check the meta http-equiv charset tag
|
---|
| 609 | elsif ($head =~ m/<meta http-equiv.*content-type.*charset=(.+?)\"/si) {
|
---|
| 610 | $best_encoding = $1;
|
---|
| 611 | }
|
---|
| 612 | if ($best_encoding) { # we extracted an encoding
|
---|
| 613 | $best_encoding =~ s/-+/_/g;
|
---|
| 614 | $best_encoding = lc($best_encoding); # lowercase
|
---|
| 615 | if ($best_encoding eq "utf_8") { $best_encoding = "utf8" }
|
---|
| 616 | $found_html_encoding = 1;
|
---|
| 617 | # We shouldn't be modifying this here!!
|
---|
| 618 | #$self->{'input_encoding'} = $best_encoding;
|
---|
| 619 | }
|
---|
| 620 |
|
---|
| 621 | # remove all HTML tags
|
---|
| 622 | $text =~ s/<[^>]*>//sg;
|
---|
| 623 | }
|
---|
| 624 |
|
---|
| 625 | # don't need to do textcat if we know the encoding now AND don't need to extract language
|
---|
| 626 | if($found_html_encoding && !$self->{'extract_language'}) { # encoding specified in html file
|
---|
| 627 | $best_language = $self->{'default_language'};
|
---|
| 628 | }
|
---|
| 629 |
|
---|
| 630 | else { # need to use textcat to get either the language, or get both language and encoding
|
---|
| 631 | $self->{'ngramj'} = new ngramj() if (!defined($self->{'ngramj'}));
|
---|
| 632 |
|
---|
| 633 | if($found_html_encoding) { # know encoding, find language by limiting search to known encoding
|
---|
| 634 | my $results = $self->{'ngramj'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
| 635 |
|
---|
| 636 | my $language;
|
---|
| 637 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(?:.*))?$/ if (scalar @$results > 0);
|
---|
| 638 |
|
---|
| 639 | if (!defined $language || scalar @$results > 3) {
|
---|
| 640 | # if there were too many results even when restricting results by encoding,
|
---|
| 641 | # or if there were no results, use default language with the known encoding
|
---|
| 642 | $best_language = $self->use_default_language($filename);
|
---|
| 643 | }
|
---|
| 644 | else { # fewer than 3 results means textcat is more certain, use the first result
|
---|
| 645 | $best_language = $language;
|
---|
| 646 | }
|
---|
| 647 | }
|
---|
| 648 | else { # don't know encoding or language yet, therefore we use textcat
|
---|
| 649 | my $results = $self->{'ngramj'}->classify_contents(\$text, $filename);
|
---|
| 650 |
|
---|
| 651 | # if textcat returns 3 or less possibilities we'll use the first one in the list
|
---|
| 652 | if (scalar @$results <= 3) { # results will be > 0 when we don't constrain textcat by an encoding
|
---|
| 653 | my ($language, $encoding) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/;
|
---|
| 654 |
|
---|
| 655 | $language = $self->use_default_language($filename) unless defined $language;
|
---|
| 656 | $encoding = $self->use_default_encoding($filename) unless defined $encoding;
|
---|
| 657 |
|
---|
| 658 | $best_language = $language;
|
---|
| 659 | $best_encoding = $encoding;
|
---|
| 660 | }
|
---|
| 661 | else { # if (scalar @$results > 3) {
|
---|
| 662 | if ($unicode_format) { # in case the first had a BOM
|
---|
| 663 | $best_encoding=$unicode_format;
|
---|
| 664 | }
|
---|
| 665 | else {
|
---|
| 666 | # Find the most frequent encoding in the textcat results returned
|
---|
| 667 | # Returns "" if there's no encoding more frequent than another
|
---|
| 668 | $best_encoding = $self->{'textcat'}->most_frequent_encoding($results);
|
---|
| 669 | }
|
---|
| 670 |
|
---|
| 671 | if ($best_encoding eq "") { # encoding still not set, use defaults
|
---|
| 672 | $best_language = $self->use_default_language($filename);
|
---|
| 673 | $best_encoding = $self->use_default_encoding($filename);
|
---|
| 674 | }
|
---|
| 675 | elsif (!$self->{'extract_language'}) { # know encoding but don't need to discover language
|
---|
| 676 | $best_language = $self->use_default_language($filename);
|
---|
| 677 | }
|
---|
| 678 | else { # textcat again using the most frequent encoding or the $unicode_format set above
|
---|
| 679 | $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
| 680 | my $language;
|
---|
| 681 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/ if (scalar @$results > 0);
|
---|
| 682 | if (!defined $language || scalar @$results > 3) {
|
---|
| 683 | # if no result or too many results, use default language for the encoding previously found
|
---|
| 684 | $best_language = $self->use_default_language($filename);
|
---|
| 685 | }
|
---|
| 686 | else { # fewer than 3 results, use the language of the first result
|
---|
| 687 | $best_language = $language;
|
---|
| 688 | }
|
---|
| 689 | }
|
---|
| 690 | }
|
---|
| 691 | }
|
---|
| 692 | }
|
---|
| 693 |
|
---|
| 694 | if($best_encoding eq "" || $best_language eq "") {
|
---|
| 695 | print STDERR "****Shouldn't happen: encoding and/or language still not set. Using defaults.\n";
|
---|
| 696 | $best_encoding = $self->use_default_encoding($filename) if $best_encoding eq "";
|
---|
| 697 | $best_language = $self->use_default_language($filename) if $best_language eq "";
|
---|
| 698 | }
|
---|
| 699 | # print STDERR "****Content language: $best_language; Encoding: $best_encoding.\n";
|
---|
| 700 |
|
---|
| 701 |
|
---|
| 702 | if ($best_encoding =~ /^iso_8859/ && &unicode::check_is_utf8($text)) {
|
---|
| 703 | # the text is valid utf8, so assume that's the real encoding
|
---|
| 704 | # (since textcat is based on probabilities)
|
---|
| 705 | $best_encoding = 'utf8';
|
---|
| 706 | }
|
---|
| 707 |
|
---|
| 708 | # check for equivalents where textcat doesn't have some encodings...
|
---|
| 709 | # eg MS versions of standard encodings
|
---|
| 710 | if ($best_encoding =~ /^iso_8859_(\d+)/) {
|
---|
| 711 | my $iso = $1; # which variant of the iso standard?
|
---|
| 712 | # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
|
---|
| 713 | if ($text =~ /[\x80-\x9f]/) {
|
---|
| 714 | # Western Europe
|
---|
| 715 | if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
|
---|
| 716 | elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
|
---|
| 717 | elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
|
---|
| 718 | elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
|
---|
| 719 | elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
|
---|
| 720 | elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
|
---|
| 721 | elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
|
---|
| 722 | }
|
---|
| 723 | }
|
---|
| 724 |
|
---|
| 725 | if ($best_encoding !~ /^(ascii|utf8|unicode)$/ &&
|
---|
| 726 | !defined $encodings::encodings->{$best_encoding}) {
|
---|
| 727 | if ($self->{'verbosity'}) {
|
---|
| 728 | gsprintf($outhandle, "ReadTextFile: {ReadTextFile.unsupported_encoding}\n",
|
---|
| 729 | $filename, $best_encoding, $self->{'default_encoding'});
|
---|
| 730 | }
|
---|
| 731 | $best_encoding = $self->{'default_encoding'};
|
---|
| 732 | }
|
---|
| 733 |
|
---|
| 734 | return ($best_language, $best_encoding);
|
---|
| 735 | }
|
---|
| 736 |
|
---|
| 737 |
|
---|
| 738 |
|
---|
| 739 |
|
---|
| 740 | sub use_default_language {
|
---|
| 741 | my $self = shift (@_);
|
---|
| 742 | my ($filename) = @_;
|
---|
| 743 |
|
---|
| 744 | if ($self->{'verbosity'}>2) {
|
---|
| 745 | gsprintf($self->{'outhandle'},
|
---|
| 746 | "ReadTextFile: {ReadTextFile.could_not_extract_language}\n",
|
---|
| 747 | $filename, $self->{'default_language'});
|
---|
| 748 | }
|
---|
| 749 | return $self->{'default_language'};
|
---|
| 750 | }
|
---|
| 751 |
|
---|
| 752 | sub use_default_encoding {
|
---|
| 753 | my $self = shift (@_);
|
---|
| 754 | my ($filename) = @_;
|
---|
| 755 |
|
---|
| 756 | if ($self->{'verbosity'}>2) {
|
---|
| 757 | gsprintf($self->{'outhandle'},
|
---|
| 758 | "ReadTextFile: {ReadTextFile.could_not_extract_encoding}\n",
|
---|
| 759 | $filename, $self->{'default_encoding'});
|
---|
| 760 | }
|
---|
| 761 | return $self->{'default_encoding'};
|
---|
| 762 | }
|
---|
| 763 |
|
---|
| 764 | # Overridden by exploding plugins (eg. ISISPlug)
|
---|
| 765 | sub clean_up_after_exploding
|
---|
| 766 | {
|
---|
| 767 | my $self = shift(@_);
|
---|
| 768 | }
|
---|
| 769 |
|
---|
| 770 |
|
---|
| 771 | 1;
|
---|