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 | my $verbosity = $self->{'verbosity'};
|
---|
123 | if ($verbosity>=2) {
|
---|
124 | my $outhandle = $self->{'outhandle'};
|
---|
125 | print $outhandle "++ Using Ngram-Java version of ReadTextFile.pm ++\n";
|
---|
126 | }
|
---|
127 |
|
---|
128 | my $outhandle = $self->{'outhandle'};
|
---|
129 | # should we move this to read? What about secondary plugins?
|
---|
130 | print STDERR "<Processing n='$file' p='$self->{'plugin_type'}'>\n" if ($gli);
|
---|
131 | my $pp_file = &util::prettyprint_file($base_dir,$file,$gli);
|
---|
132 | print $outhandle "$self->{'plugin_type'} processing $pp_file\n"
|
---|
133 | if $self->{'verbosity'} > 1;
|
---|
134 |
|
---|
135 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
136 |
|
---|
137 | # Do encoding stuff
|
---|
138 | my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
|
---|
139 | if ($self->{'verbosity'} > 2) {
|
---|
140 | print $outhandle "ReadTextFile: reading $file as ($content_encoding,$language)\n";
|
---|
141 | }
|
---|
142 |
|
---|
143 | # create a new document
|
---|
144 | my $doc_obj = new doc ($filename_full_path, "indexed_doc", $self->{'file_rename_method'});
|
---|
145 | my $top_section = $doc_obj->get_top_section();
|
---|
146 |
|
---|
147 | # this should look at the plugin option too...
|
---|
148 | $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
|
---|
149 | $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename_full_path));
|
---|
150 |
|
---|
151 | my $plugin_filename_encoding = $self->{'filename_encoding'};
|
---|
152 | my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
|
---|
153 | $self->set_Source_metadata($doc_obj, $filename_full_path, $filename_encoding);
|
---|
154 |
|
---|
155 | $doc_obj->add_utf8_metadata($top_section, "Language", $language);
|
---|
156 | $doc_obj->add_utf8_metadata($top_section, "Encoding", $content_encoding);
|
---|
157 |
|
---|
158 | # read in file ($text will be in utf8)
|
---|
159 | my $text = "";
|
---|
160 | $self->read_file ($filename_full_path, $content_encoding, $language, \$text);
|
---|
161 |
|
---|
162 | if (!length ($text)) {
|
---|
163 | if ($gli) {
|
---|
164 | print STDERR "<ProcessingError n='$file' r='File contains no text'>\n";
|
---|
165 | }
|
---|
166 | gsprintf($outhandle, "$self->{'plugin_type'}: {ReadTextFile.file_has_no_text}\n", $filename_full_path) if $self->{'verbosity'};
|
---|
167 |
|
---|
168 | my $failhandle = $self->{'failhandle'};
|
---|
169 | gsprintf($failhandle, "$file: " . ref($self) . ": {ReadTextFile.empty_file}\n");
|
---|
170 | # print $failhandle "$file: " . ref($self) . ": file contains no text\n";
|
---|
171 | $self->{'num_not_processed'} ++;
|
---|
172 |
|
---|
173 | return (0,undef); # what should we return here?? error but don't want to pass it on
|
---|
174 | }
|
---|
175 |
|
---|
176 | # do plugin specific processing of doc_obj
|
---|
177 | unless (defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli))) {
|
---|
178 | $text = '';
|
---|
179 | undef $text;
|
---|
180 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
---|
181 | return (-1,undef);
|
---|
182 | }
|
---|
183 | $text='';
|
---|
184 | undef $text;
|
---|
185 |
|
---|
186 | # include any metadata passed in from previous plugins
|
---|
187 | # note that this metadata is associated with the top level section
|
---|
188 | $self->add_associated_files($doc_obj, $filename_full_path);
|
---|
189 | $self->extra_metadata ($doc_obj, $top_section, $metadata);
|
---|
190 |
|
---|
191 | # do any automatic metadata extraction
|
---|
192 | $self->auto_extract_metadata ($doc_obj);
|
---|
193 |
|
---|
194 |
|
---|
195 | # if we haven't found any Title so far, assign one
|
---|
196 | $self->title_fallback($doc_obj,$top_section,$filename_no_path);
|
---|
197 |
|
---|
198 | $self->add_OID($doc_obj);
|
---|
199 |
|
---|
200 | return (1,$doc_obj);
|
---|
201 | }
|
---|
202 |
|
---|
203 | # uses the multiread package to read in the entire file pointed to
|
---|
204 | # by filename and loads the resulting text into $$textref. Input text
|
---|
205 | # may be in any of the encodings handled by multiread, output text
|
---|
206 | # will be in utf8
|
---|
207 | sub read_file {
|
---|
208 | my $self = shift (@_);
|
---|
209 | my ($filename, $encoding, $language, $textref) = @_;
|
---|
210 |
|
---|
211 | if (!-r $filename)
|
---|
212 | {
|
---|
213 | my $outhandle = $self->{'outhandle'};
|
---|
214 | gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
|
---|
215 | # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
|
---|
216 | return;
|
---|
217 | }
|
---|
218 | $$textref = "";
|
---|
219 | if (!open (FILE, $filename)) {
|
---|
220 | gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
221 | die "\n";
|
---|
222 | }
|
---|
223 |
|
---|
224 | if ($encoding eq "ascii") {
|
---|
225 | # Replace file 'slurp' with faster implementation
|
---|
226 | sysread(FILE, $$textref, -s FILE);
|
---|
227 |
|
---|
228 | # The old slow way of reading in a file
|
---|
229 | #undef $/;
|
---|
230 | #$$textref = <FILE>;
|
---|
231 | #$/ = "\n";
|
---|
232 | } else {
|
---|
233 | my $reader = new multiread();
|
---|
234 | $reader->set_handle ('ReadTextFile::FILE');
|
---|
235 | $reader->set_encoding ($encoding);
|
---|
236 | $reader->read_file ($textref);
|
---|
237 | }
|
---|
238 |
|
---|
239 | # At this point $$textref is a binary byte string
|
---|
240 | # => turn it into a Unicode aware string, so full
|
---|
241 | # Unicode aware pattern matching can be used.
|
---|
242 | # For instance: 's/\x{0101}//g' or '[[:upper:]]'
|
---|
243 | #
|
---|
244 |
|
---|
245 | $$textref = decode("utf8",$$textref);
|
---|
246 |
|
---|
247 | close FILE;
|
---|
248 | }
|
---|
249 |
|
---|
250 |
|
---|
251 | # Not currently used
|
---|
252 | sub UNUSED_read_file_usingPerlsEncodeModule {
|
---|
253 | ##sub read_file {
|
---|
254 | my $self = shift (@_);
|
---|
255 | my ($filename, $encoding, $language, $textref) = @_;
|
---|
256 |
|
---|
257 | if (!-r $filename)
|
---|
258 | {
|
---|
259 | my $outhandle = $self->{'outhandle'};
|
---|
260 | gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
|
---|
261 | # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
|
---|
262 | return;
|
---|
263 | }
|
---|
264 | $$textref = "";
|
---|
265 | if (!open (FILE, $filename)) {
|
---|
266 | gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_f
|
---|
267 | or_reading} ($!)\n", $filename);
|
---|
268 | die "\n";
|
---|
269 | }
|
---|
270 |
|
---|
271 | my $store_slash = $/;
|
---|
272 | undef $/;
|
---|
273 | my $text = <FILE>;
|
---|
274 | $/ = $store_slash;
|
---|
275 |
|
---|
276 | $$textref = decode($encoding,$text);
|
---|
277 |
|
---|
278 | close FILE;
|
---|
279 | }
|
---|
280 |
|
---|
281 |
|
---|
282 | sub read_file_no_decoding {
|
---|
283 | my $self = shift (@_);
|
---|
284 | my ($filename, $textref) = @_;
|
---|
285 |
|
---|
286 | if (!-r $filename)
|
---|
287 | {
|
---|
288 | my $outhandle = $self->{'outhandle'};
|
---|
289 | gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
|
---|
290 | # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
|
---|
291 | return;
|
---|
292 | }
|
---|
293 | $$textref = "";
|
---|
294 | if (!open (FILE, $filename)) {
|
---|
295 | gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
296 | die "\n";
|
---|
297 | }
|
---|
298 |
|
---|
299 | my $reader = new multiread();
|
---|
300 | $reader->set_handle ('ReadTextFile::FILE');
|
---|
301 | $reader->read_file_no_decoding ($textref);
|
---|
302 |
|
---|
303 | $self->{'reader'} = $reader;
|
---|
304 |
|
---|
305 | close FILE;
|
---|
306 | }
|
---|
307 |
|
---|
308 |
|
---|
309 | sub decode_text {
|
---|
310 | my $self = shift (@_);
|
---|
311 | my ($raw_text, $encoding, $language, $textref) = @_;
|
---|
312 |
|
---|
313 | my $reader = $self->{'reader'};
|
---|
314 | if (!defined $reader) {
|
---|
315 | gsprintf(STDERR, "ReadTextFile::decode_text needs to call ReadTextFile::read_file_no_decoding first\n");
|
---|
316 | }
|
---|
317 | else {
|
---|
318 | $reader->set_encoding($encoding);
|
---|
319 | $reader->decode_text($raw_text,$textref);
|
---|
320 |
|
---|
321 | # At this point $$textref is a binary byte string
|
---|
322 | # => turn it into a Unicode aware string, so full
|
---|
323 | # Unicode aware pattern matching can be used.
|
---|
324 | # For instance: 's/\x{0101}//g' or '[[:upper:]]'
|
---|
325 |
|
---|
326 | $$textref = decode("utf8",$$textref);
|
---|
327 | }
|
---|
328 | }
|
---|
329 |
|
---|
330 |
|
---|
331 | sub textcat_get_language_encoding {
|
---|
332 | my $self = shift (@_);
|
---|
333 | my ($filename) = @_;
|
---|
334 |
|
---|
335 | $self->{'textcat'} = new ngramj($self->{'verbosity'},$self->{'outhandle'}) if (!defined($self->{'textcat'}));
|
---|
336 |
|
---|
337 | my ($language, $encoding, $extracted_encoding);
|
---|
338 | if ($self->{'input_encoding'} eq "auto") {
|
---|
339 | # use textcat to automatically work out the input encoding and language
|
---|
340 | ($language, $encoding) = $self->get_language_encoding ($filename);
|
---|
341 | } elsif ($self->{'extract_language'}) {
|
---|
342 | # use textcat to get language metadata
|
---|
343 | ($language, $extracted_encoding) = $self->get_language_encoding ($filename);
|
---|
344 | $encoding = $self->{'input_encoding'};
|
---|
345 | # don't print this message for english... english in utf8 is identical
|
---|
346 | # to english in iso-8859-1 (except for some punctuation). We don't have
|
---|
347 | # a language model for en_utf8, so textcat always says iso-8859-1!
|
---|
348 | if ($extracted_encoding ne $encoding && $language ne "en" && $self->{'verbosity'}) {
|
---|
349 | my $plugin_name = ref ($self);
|
---|
350 | my $outhandle = $self->{'outhandle'};
|
---|
351 | gsprintf($outhandle, "$plugin_name: {ReadTextFile.wrong_encoding}\n", $filename, $encoding, $extracted_encoding);
|
---|
352 | }
|
---|
353 | } else {
|
---|
354 | $language = $self->{'default_language'};
|
---|
355 | $encoding = $self->{'input_encoding'};
|
---|
356 | }
|
---|
357 |
|
---|
358 | # print STDERR "**** language encoding of contents of file $filename:\n\t****$language $encoding\n";
|
---|
359 |
|
---|
360 | return ($language, $encoding);
|
---|
361 | }
|
---|
362 |
|
---|
363 |
|
---|
364 | # Uses textcat to work out the encoding and language of the text in
|
---|
365 | # $filename. All html tags are removed before processing.
|
---|
366 | # returns an array containing "language" and "encoding"
|
---|
367 | sub get_language_encoding {
|
---|
368 | my $self = shift (@_);
|
---|
369 | my ($filename) = @_;
|
---|
370 | my $outhandle = $self->{'outhandle'};
|
---|
371 | my $unicode_format = "";
|
---|
372 | my $best_language = "";
|
---|
373 | my $best_encoding = "";
|
---|
374 |
|
---|
375 |
|
---|
376 | # read in file
|
---|
377 | if (!open (FILE, $filename)) {
|
---|
378 | gsprintf(STDERR, "ReadTextFile::get_language_encoding {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
379 | # this is a pretty bad error, but try to continue anyway
|
---|
380 | return ($self->{'default_language'}, $self->{'input_encoding'});
|
---|
381 | }
|
---|
382 | undef $/;
|
---|
383 | my $text = <FILE>;
|
---|
384 | $/ = "\n";
|
---|
385 | close FILE;
|
---|
386 |
|
---|
387 | # check if first few bytes have a Byte Order Marker
|
---|
388 | my $bom=substr($text,0,2); # check 16bit unicode
|
---|
389 | if ($bom eq "\xff\xfe") { # little endian 16bit unicode
|
---|
390 | $unicode_format="unicode";
|
---|
391 | } elsif ($bom eq "\xfe\xff") { # big endian 16bit unicode
|
---|
392 | $unicode_format="unicode";
|
---|
393 | } else {
|
---|
394 | $bom=substr($text,0,3); # check utf-8
|
---|
395 | if ($bom eq "\xef\xbb\xbf") { # utf-8 coded FEFF bom
|
---|
396 | $unicode_format="utf8";
|
---|
397 | # } elsif ($bom eq "\xef\xbf\xbe") { # utf-8 coded FFFE bom. Error!?
|
---|
398 | # $unicode_format="utf8";
|
---|
399 | }
|
---|
400 | }
|
---|
401 |
|
---|
402 | my $found_html_encoding = 0;
|
---|
403 | # handle html files specially
|
---|
404 | # XXX this doesn't match plugins derived from HTMLPlug (except ConvertTo)
|
---|
405 | if (ref($self) eq 'HTMLPlugin' ||
|
---|
406 | (exists $self->{'converted_to'} && $self->{'converted_to'} eq 'HTML')){
|
---|
407 |
|
---|
408 | # remove comments in head, including multiline ones, so that we don't match on
|
---|
409 | # inactive tags (those that are nested inside comments)
|
---|
410 | my ($head) = ($text =~ m/<head>(.*)<\/head>/si);
|
---|
411 | $head = "" unless defined $head; # some files are not proper HTML eg php files
|
---|
412 | $head =~ s/<!--.*?-->//sg;
|
---|
413 |
|
---|
414 | # remove <title>stuff</title> -- as titles tend often to be in English
|
---|
415 | # for foreign language documents
|
---|
416 | $text =~ s!<title>.*?</title>!!si;
|
---|
417 |
|
---|
418 | # see if this html file specifies its encoding
|
---|
419 | if ($text =~ /^<\?xml.*encoding="(.+?)"/) {
|
---|
420 | $best_encoding = $1;
|
---|
421 | }
|
---|
422 | # check the meta http-equiv charset tag
|
---|
423 | elsif ($head =~ m/<meta http-equiv.*content-type.*charset=(.+?)\"/si) {
|
---|
424 | $best_encoding = $1;
|
---|
425 | }
|
---|
426 | if ($best_encoding) { # we extracted an encoding
|
---|
427 | $best_encoding =~ s/-+/_/g;
|
---|
428 | $best_encoding = lc($best_encoding); # lowercase
|
---|
429 | if ($best_encoding eq "utf_8") { $best_encoding = "utf8" }
|
---|
430 | $found_html_encoding = 1;
|
---|
431 | # We shouldn't be modifying this here!!
|
---|
432 | #$self->{'input_encoding'} = $best_encoding;
|
---|
433 | }
|
---|
434 |
|
---|
435 | # remove all HTML tags
|
---|
436 | $text =~ s/<[^>]*>//sg;
|
---|
437 | }
|
---|
438 |
|
---|
439 | # don't need to do textcat if we know the encoding now AND don't need to extract language
|
---|
440 | if($found_html_encoding && !$self->{'extract_language'}) { # encoding specified in html file
|
---|
441 | $best_language = $self->{'default_language'};
|
---|
442 | }
|
---|
443 |
|
---|
444 | else { # need to use textcat to get either the language, or get both language and encoding
|
---|
445 | $self->{'textcat'} = new textcat() if (!defined($self->{'textcat'}));
|
---|
446 |
|
---|
447 | if($found_html_encoding) { # know encoding, find language by limiting search to known encoding
|
---|
448 | my $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
449 |
|
---|
450 | my $language;
|
---|
451 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(?:.*))?$/ if (scalar @$results > 0);
|
---|
452 |
|
---|
453 | if (!defined $language || scalar @$results > 3) {
|
---|
454 | # if there were too many results even when restricting results by encoding,
|
---|
455 | # or if there were no results, use default language with the known encoding
|
---|
456 | $best_language = $self->use_default_language($filename);
|
---|
457 | }
|
---|
458 | else { # fewer than 3 results means textcat is more certain, use the first result
|
---|
459 | $best_language = $language;
|
---|
460 | }
|
---|
461 | }
|
---|
462 | else { # don't know encoding or language yet, therefore we use textcat
|
---|
463 | my $results = $self->{'textcat'}->classify_contents(\$text, $filename);
|
---|
464 |
|
---|
465 | # if textcat returns 3 or less possibilities we'll use the first one in the list
|
---|
466 | if (scalar @$results <= 3) { # results will be > 0 when we don't constrain textcat by an encoding
|
---|
467 | my ($language, $encoding) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/;
|
---|
468 |
|
---|
469 | $language = $self->use_default_language($filename) unless defined $language;
|
---|
470 | $encoding = $self->use_default_encoding($filename) unless defined $encoding;
|
---|
471 |
|
---|
472 | $best_language = $language;
|
---|
473 | $best_encoding = $encoding;
|
---|
474 | }
|
---|
475 | else { # if (scalar @$results > 3) {
|
---|
476 | if ($unicode_format) { # in case the first had a BOM
|
---|
477 | $best_encoding=$unicode_format;
|
---|
478 | }
|
---|
479 | else {
|
---|
480 | # Find the most frequent encoding in the textcat results returned
|
---|
481 | # Returns "" if there's no encoding more frequent than another
|
---|
482 | $best_encoding = $self->{'textcat'}->most_frequent_encoding($results);
|
---|
483 | }
|
---|
484 |
|
---|
485 | if ($best_encoding eq "") { # encoding still not set, use defaults
|
---|
486 | $best_language = $self->use_default_language($filename);
|
---|
487 | $best_encoding = $self->use_default_encoding($filename);
|
---|
488 | }
|
---|
489 | elsif (!$self->{'extract_language'}) { # know encoding but don't need to discover language
|
---|
490 | $best_language = $self->use_default_language($filename);
|
---|
491 | }
|
---|
492 | else { # textcat again using the most frequent encoding or the $unicode_format set above
|
---|
493 | $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
494 | my $language;
|
---|
495 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/ if (scalar @$results > 0);
|
---|
496 | if (!defined $language || scalar @$results > 3) {
|
---|
497 | # if no result or too many results, use default language for the encoding previously found
|
---|
498 | $best_language = $self->use_default_language($filename);
|
---|
499 | }
|
---|
500 | else { # fewer than 3 results, use the language of the first result
|
---|
501 | $best_language = $language;
|
---|
502 | }
|
---|
503 | }
|
---|
504 | }
|
---|
505 | }
|
---|
506 | }
|
---|
507 |
|
---|
508 | if($best_encoding eq "" || $best_language eq "") {
|
---|
509 | print STDERR "****Shouldn't happen: encoding and/or language still not set. Using defaults.\n";
|
---|
510 | $best_encoding = $self->use_default_encoding($filename) if $best_encoding eq "";
|
---|
511 | $best_language = $self->use_default_language($filename) if $best_language eq "";
|
---|
512 | }
|
---|
513 | # print STDERR "****Content language: $best_language; Encoding: $best_encoding.\n";
|
---|
514 |
|
---|
515 |
|
---|
516 | if ($best_encoding =~ /^iso_8859/ && &unicode::check_is_utf8($text)) {
|
---|
517 | # the text is valid utf8, so assume that's the real encoding
|
---|
518 | # (since textcat is based on probabilities)
|
---|
519 | $best_encoding = 'utf8';
|
---|
520 | }
|
---|
521 |
|
---|
522 | # check for equivalents where textcat doesn't have some encodings...
|
---|
523 | # eg MS versions of standard encodings
|
---|
524 | if ($best_encoding =~ /^iso_8859_(\d+)/) {
|
---|
525 | my $iso = $1; # which variant of the iso standard?
|
---|
526 | # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
|
---|
527 | if ($text =~ /[\x80-\x9f]/) {
|
---|
528 | # Western Europe
|
---|
529 | if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
|
---|
530 | elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
|
---|
531 | elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
|
---|
532 | elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
|
---|
533 | elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
|
---|
534 | elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
|
---|
535 | elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
|
---|
536 | }
|
---|
537 | }
|
---|
538 |
|
---|
539 | if ($best_encoding !~ /^(ascii|utf8|unicode)$/ &&
|
---|
540 | !defined $encodings::encodings->{$best_encoding}) {
|
---|
541 | if ($self->{'verbosity'}) {
|
---|
542 | gsprintf($outhandle, "ReadTextFile: {ReadTextFile.unsupported_encoding}\n",
|
---|
543 | $filename, $best_encoding, $self->{'default_encoding'});
|
---|
544 | }
|
---|
545 | $best_encoding = $self->{'default_encoding'};
|
---|
546 | }
|
---|
547 |
|
---|
548 | return ($best_language, $best_encoding);
|
---|
549 | }
|
---|
550 |
|
---|
551 |
|
---|
552 |
|
---|
553 | # Uses java ngram solution to work out the language of the text in
|
---|
554 | # $filename. Fall back on textcat to to the character encoding
|
---|
555 | # All html tags are removed before processing.
|
---|
556 | # returns an array containing "language" and "encoding"
|
---|
557 | sub get_language_encoding {
|
---|
558 | my $self = shift (@_);
|
---|
559 | my ($filename) = @_;
|
---|
560 | my $outhandle = $self->{'outhandle'};
|
---|
561 | my $unicode_format = "";
|
---|
562 | my $best_language = "";
|
---|
563 | my $best_encoding = "";
|
---|
564 |
|
---|
565 |
|
---|
566 | # read in file
|
---|
567 | if (!open (FILE, $filename)) {
|
---|
568 | gsprintf(STDERR, "ReadTextFile::java_ngram_get_language_encoding {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
|
---|
569 | # this is a pretty bad error, but try to continue anyway
|
---|
570 | return ($self->{'default_language'}, $self->{'input_encoding'});
|
---|
571 | }
|
---|
572 | undef $/;
|
---|
573 | my $text = <FILE>;
|
---|
574 | $/ = "\n";
|
---|
575 | close FILE;
|
---|
576 |
|
---|
577 | # check if first few bytes have a Byte Order Marker
|
---|
578 | my $bom=substr($text,0,2); # check 16bit unicode
|
---|
579 | if ($bom eq "\xff\xfe") { # little endian 16bit unicode
|
---|
580 | $unicode_format="unicode";
|
---|
581 | } elsif ($bom eq "\xfe\xff") { # big endian 16bit unicode
|
---|
582 | $unicode_format="unicode";
|
---|
583 | } else {
|
---|
584 | $bom=substr($text,0,3); # check utf-8
|
---|
585 | if ($bom eq "\xef\xbb\xbf") { # utf-8 coded FEFF bom
|
---|
586 | $unicode_format="utf8";
|
---|
587 | # } elsif ($bom eq "\xef\xbf\xbe") { # utf-8 coded FFFE bom. Error!?
|
---|
588 | # $unicode_format="utf8";
|
---|
589 | }
|
---|
590 | }
|
---|
591 |
|
---|
592 | my $found_html_encoding = 0;
|
---|
593 | # handle html files specially
|
---|
594 | # XXX this doesn't match plugins derived from HTMLPlug (except ConvertTo)
|
---|
595 | if (ref($self) eq 'HTMLPlugin' ||
|
---|
596 | (exists $self->{'converted_to'} && $self->{'converted_to'} eq 'HTML')){
|
---|
597 |
|
---|
598 | # remove comments in head, including multiline ones, so that we don't match on
|
---|
599 | # inactive tags (those that are nested inside comments)
|
---|
600 | my ($head) = ($text =~ m/<head>(.*)<\/head>/si);
|
---|
601 | $head = "" unless defined $head; # some files are not proper HTML eg php files
|
---|
602 | $head =~ s/<!--.*?-->//sg;
|
---|
603 |
|
---|
604 | # remove <title>stuff</title> -- as titles tend often to be in English
|
---|
605 | # for foreign language documents
|
---|
606 | $text =~ s!<title>.*?</title>!!si;
|
---|
607 |
|
---|
608 | # see if this html file specifies its encoding
|
---|
609 | if ($text =~ /^<\?xml.*encoding="(.+?)"/) {
|
---|
610 | $best_encoding = $1;
|
---|
611 | }
|
---|
612 | # check the meta http-equiv charset tag
|
---|
613 | elsif ($head =~ m/<meta http-equiv.*content-type.*charset=(.+?)\"/si) {
|
---|
614 | $best_encoding = $1;
|
---|
615 | }
|
---|
616 | if ($best_encoding) { # we extracted an encoding
|
---|
617 | $best_encoding =~ s/-+/_/g;
|
---|
618 | $best_encoding = lc($best_encoding); # lowercase
|
---|
619 | if ($best_encoding eq "utf_8") { $best_encoding = "utf8" }
|
---|
620 | $found_html_encoding = 1;
|
---|
621 | # We shouldn't be modifying this here!!
|
---|
622 | #$self->{'input_encoding'} = $best_encoding;
|
---|
623 | }
|
---|
624 |
|
---|
625 | # remove all HTML tags
|
---|
626 | $text =~ s/<[^>]*>//sg;
|
---|
627 | }
|
---|
628 |
|
---|
629 | # don't need to do textcat if we know the encoding now AND don't need to extract language
|
---|
630 | if($found_html_encoding && !$self->{'extract_language'}) { # encoding specified in html file
|
---|
631 | $best_language = $self->{'default_language'};
|
---|
632 | }
|
---|
633 |
|
---|
634 | else { # need to use textcat to get either the language, or get both language and encoding
|
---|
635 | $self->{'ngramj'} = new ngramj($self->{'verbosity'},$self->{'outhandle'}) if (!defined($self->{'ngramj'}));
|
---|
636 |
|
---|
637 | if($found_html_encoding) { # know encoding, find language by limiting search to known encoding
|
---|
638 | my $results = $self->{'ngramj'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
639 |
|
---|
640 | my $language;
|
---|
641 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(?:.*))?$/ if (scalar @$results > 0);
|
---|
642 |
|
---|
643 | if (!defined $language || scalar @$results > 3) {
|
---|
644 | # if there were too many results even when restricting results by encoding,
|
---|
645 | # or if there were no results, use default language with the known encoding
|
---|
646 | $best_language = $self->use_default_language($filename);
|
---|
647 | }
|
---|
648 | else { # fewer than 3 results means textcat is more certain, use the first result
|
---|
649 | $best_language = $language;
|
---|
650 | }
|
---|
651 | }
|
---|
652 | else { # don't know encoding or language yet, therefore we use textcat
|
---|
653 | my $results = $self->{'ngramj'}->classify_contents(\$text, $filename);
|
---|
654 |
|
---|
655 | # if textcat returns 3 or less possibilities we'll use the first one in the list
|
---|
656 | if (scalar @$results <= 3) { # results will be > 0 when we don't constrain textcat by an encoding
|
---|
657 | my ($language, $encoding) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/;
|
---|
658 |
|
---|
659 | $language = $self->use_default_language($filename) unless defined $language;
|
---|
660 | $encoding = $self->use_default_encoding($filename) unless defined $encoding;
|
---|
661 |
|
---|
662 | $best_language = $language;
|
---|
663 | $best_encoding = $encoding;
|
---|
664 | }
|
---|
665 | else { # if (scalar @$results > 3) {
|
---|
666 | if ($unicode_format) { # in case the first had a BOM
|
---|
667 | $best_encoding=$unicode_format;
|
---|
668 | }
|
---|
669 | else {
|
---|
670 | # Find the most frequent encoding in the textcat results returned
|
---|
671 | # Returns "" if there's no encoding more frequent than another
|
---|
672 | $best_encoding = $self->{'textcat'}->most_frequent_encoding($results);
|
---|
673 | }
|
---|
674 |
|
---|
675 | if ($best_encoding eq "") { # encoding still not set, use defaults
|
---|
676 | $best_language = $self->use_default_language($filename);
|
---|
677 | $best_encoding = $self->use_default_encoding($filename);
|
---|
678 | }
|
---|
679 | elsif (!$self->{'extract_language'}) { # know encoding but don't need to discover language
|
---|
680 | $best_language = $self->use_default_language($filename);
|
---|
681 | }
|
---|
682 | else { # textcat again using the most frequent encoding or the $unicode_format set above
|
---|
683 | $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
|
---|
684 | my $language;
|
---|
685 | ($language) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/ if (scalar @$results > 0);
|
---|
686 | if (!defined $language || scalar @$results > 3) {
|
---|
687 | # if no result or too many results, use default language for the encoding previously found
|
---|
688 | $best_language = $self->use_default_language($filename);
|
---|
689 | }
|
---|
690 | else { # fewer than 3 results, use the language of the first result
|
---|
691 | $best_language = $language;
|
---|
692 | }
|
---|
693 | }
|
---|
694 | }
|
---|
695 | }
|
---|
696 | }
|
---|
697 |
|
---|
698 | if($best_encoding eq "" || $best_language eq "") {
|
---|
699 | print STDERR "****Shouldn't happen: encoding and/or language still not set. Using defaults.\n";
|
---|
700 | $best_encoding = $self->use_default_encoding($filename) if $best_encoding eq "";
|
---|
701 | $best_language = $self->use_default_language($filename) if $best_language eq "";
|
---|
702 | }
|
---|
703 | # print STDERR "****Content language: $best_language; Encoding: $best_encoding.\n";
|
---|
704 |
|
---|
705 |
|
---|
706 | if ($best_encoding =~ /^iso_8859/ && &unicode::check_is_utf8($text)) {
|
---|
707 | # the text is valid utf8, so assume that's the real encoding
|
---|
708 | # (since textcat is based on probabilities)
|
---|
709 | $best_encoding = 'utf8';
|
---|
710 | }
|
---|
711 |
|
---|
712 | # check for equivalents where textcat doesn't have some encodings...
|
---|
713 | # eg MS versions of standard encodings
|
---|
714 | if ($best_encoding =~ /^iso_8859_(\d+)/) {
|
---|
715 | my $iso = $1; # which variant of the iso standard?
|
---|
716 | # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
|
---|
717 | if ($text =~ /[\x80-\x9f]/) {
|
---|
718 | # Western Europe
|
---|
719 | if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
|
---|
720 | elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
|
---|
721 | elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
|
---|
722 | elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
|
---|
723 | elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
|
---|
724 | elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
|
---|
725 | elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
|
---|
726 | }
|
---|
727 | }
|
---|
728 |
|
---|
729 | if ($best_encoding !~ /^(ascii|utf8|unicode)$/ &&
|
---|
730 | !defined $encodings::encodings->{$best_encoding}) {
|
---|
731 | if ($self->{'verbosity'}) {
|
---|
732 | gsprintf($outhandle, "ReadTextFile: {ReadTextFile.unsupported_encoding}\n",
|
---|
733 | $filename, $best_encoding, $self->{'default_encoding'});
|
---|
734 | }
|
---|
735 | $best_encoding = $self->{'default_encoding'};
|
---|
736 | }
|
---|
737 |
|
---|
738 | return ($best_language, $best_encoding);
|
---|
739 | }
|
---|
740 |
|
---|
741 |
|
---|
742 |
|
---|
743 |
|
---|
744 | sub use_default_language {
|
---|
745 | my $self = shift (@_);
|
---|
746 | my ($filename) = @_;
|
---|
747 |
|
---|
748 | if ($self->{'verbosity'}>2) {
|
---|
749 | gsprintf($self->{'outhandle'},
|
---|
750 | "ReadTextFile: {ReadTextFile.could_not_extract_language}\n",
|
---|
751 | $filename, $self->{'default_language'});
|
---|
752 | }
|
---|
753 | return $self->{'default_language'};
|
---|
754 | }
|
---|
755 |
|
---|
756 | sub use_default_encoding {
|
---|
757 | my $self = shift (@_);
|
---|
758 | my ($filename) = @_;
|
---|
759 |
|
---|
760 | if ($self->{'verbosity'}>2) {
|
---|
761 | gsprintf($self->{'outhandle'},
|
---|
762 | "ReadTextFile: {ReadTextFile.could_not_extract_encoding}\n",
|
---|
763 | $filename, $self->{'default_encoding'});
|
---|
764 | }
|
---|
765 | return $self->{'default_encoding'};
|
---|
766 | }
|
---|
767 |
|
---|
768 | # Overridden by exploding plugins (eg. ISISPlug)
|
---|
769 | sub clean_up_after_exploding
|
---|
770 | {
|
---|
771 | my $self = shift(@_);
|
---|
772 | }
|
---|
773 |
|
---|
774 |
|
---|
775 | 1;
|
---|