source: main/trunk/greenstone2/perllib/plugins/ReadTextFile.pm@ 23348

Last change on this file since 23348 was 23348, checked in by davidb, 13 years ago

Added extra parameter to call to deduce_filename_encoding()
  • Property svn:executable set to *
File size: 19.2 KB
Line 
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
26package ReadTextFile;
27
28use strict;
29no strict 'subs';
30no strict 'refs'; # allow filehandles to be variables and viceversa
31
32use Encode;
33
34use multiread;
35use encodings;
36use unicode;
37use textcat;
38use doc;
39use ghtml;
40use gsprintf 'gsprintf';
41
42use AutoExtractMetadata;
43
44sub BEGIN {
45 @ReadTextFile::ISA = ( 'AutoExtractMetadata' );
46}
47
48my $encoding_plus_auto_list =
49 [ { 'name' => "auto",
50 'desc' => "{ReadTextFile.input_encoding.auto}" } ];
51push(@{$encoding_plus_auto_list},@{$BasePlugin::encoding_list});
52
53my $arguments =
54 [ { 'name' => "input_encoding",
55 'desc' => "{ReadTextFile.input_encoding}",
56 'type' => "enum",
57 'list' => $encoding_plus_auto_list,
58 'reqd' => "no" ,
59 'deft' => "auto" } ,
60 { 'name' => "default_encoding",
61 'desc' => "{ReadTextFile.default_encoding}",
62 'type' => "enum",
63 'list' => $BasePlugin::encoding_list,
64 'reqd' => "no",
65 'deft' => "utf8" },
66 { 'name' => "extract_language",
67 'desc' => "{ReadTextFile.extract_language}",
68 'type' => "flag",
69 'reqd' => "no" },
70 { 'name' => "default_language",
71 'desc' => "{ReadTextFile.default_language}",
72 'type' => "string",
73 'deft' => "en",
74 'reqd' => "no" }
75 ];
76
77
78my $options = { 'name' => "ReadTextFile",
79 'desc' => "{ReadTextFile.desc}",
80 'abstract' => "yes",
81 'inherits' => "no",
82 'args' => $arguments };
83
84
85
86sub new {
87 my $class = shift (@_);
88 my ($pluginlist,$inputargs,$hashArgOptLists, $auxiliary) = @_;
89 push(@$pluginlist, $class);
90
91 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
92 push(@{$hashArgOptLists->{"OptList"}},$options);
93
94 my $self = new AutoExtractMetadata($pluginlist, $inputargs, $hashArgOptLists, $auxiliary);
95
96 return bless $self, $class;
97
98}
99
100
101
102# The ReadTextFile read_into_doc_obj() function. This function does all the
103# right things to make general options work for a given plugin. It reads in
104# a file and sets up a slew of metadata all saved in doc_obj, which
105# it then returns as part of a tuple (process_status,doc_obj)
106#
107# Much of this functionality used to reside in read, but it was broken
108# down into a supporting routine to make the code more flexible.
109#
110# recursive plugins (e.g. RecPlug) and specialized plugins like those
111# capable of processing many documents within a single file (e.g.
112# GMLPlug) will normally want to implement their own version of
113# read_into_doc_obj()
114#
115# Note that $base_dir might be "" and that $file might
116# include directories
117sub read_into_doc_obj {
118 my $self = shift (@_);
119 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
120
121 my $outhandle = $self->{'outhandle'};
122 # should we move this to read? What about secondary plugins?
123 print STDERR "<Processing n='$file' p='$self->{'plugin_type'}'>\n" if ($gli);
124 print $outhandle "$self->{'plugin_type'} processing $file\n"
125 if $self->{'verbosity'} > 1;
126
127 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
128
129 # Do encoding stuff
130 my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
131 if ($self->{'verbosity'} > 2) {
132 print $outhandle "ReadTextFile: reading $file as ($content_encoding,$language)\n";
133 }
134
135 # create a new document
136 my $doc_obj = new doc ($filename_full_path, "indexed_doc", $self->{'file_rename_method'});
137 my $top_section = $doc_obj->get_top_section();
138
139 # this should look at the plugin option too...
140 $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
141 $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename_full_path));
142
143 my $plugin_filename_encoding = $self->{'filename_encoding'};
144 my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
145 $self->set_Source_metadata($doc_obj, $filename_no_path, $filename_encoding);
146
147 $doc_obj->add_utf8_metadata($top_section, "Language", $language);
148 $doc_obj->add_utf8_metadata($top_section, "Encoding", $content_encoding);
149
150 # read in file ($text will be in utf8)
151 my $text = "";
152 $self->read_file ($filename_full_path, $content_encoding, $language, \$text);
153
154 if (!length ($text)) {
155 if ($gli) {
156 print STDERR "<ProcessingError n='$file' r='File contains no text'>\n";
157 }
158 gsprintf($outhandle, "$self->{'plugin_type'}: {ReadTextFile.file_has_no_text}\n", $filename_full_path) if $self->{'verbosity'};
159
160 my $failhandle = $self->{'failhandle'};
161 gsprintf($failhandle, "$file: " . ref($self) . ": {ReadTextFile.empty_file}\n");
162 # print $failhandle "$file: " . ref($self) . ": file contains no text\n";
163 $self->{'num_not_processed'} ++;
164
165 return (0,undef); # what should we return here?? error but don't want to pass it on
166 }
167
168 # do plugin specific processing of doc_obj
169 unless (defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli))) {
170 $text = '';
171 undef $text;
172 print STDERR "<ProcessingError n='$file'>\n" if ($gli);
173 return (-1,undef);
174 }
175 $text='';
176 undef $text;
177
178 # include any metadata passed in from previous plugins
179 # note that this metadata is associated with the top level section
180 $self->add_associated_files($doc_obj, $filename_full_path);
181 $self->extra_metadata ($doc_obj, $top_section, $metadata);
182
183 # do any automatic metadata extraction
184 $self->auto_extract_metadata ($doc_obj);
185
186
187 # if we haven't found any Title so far, assign one
188 $self->title_fallback($doc_obj,$top_section,$filename_no_path);
189
190 $self->add_OID($doc_obj);
191
192 return (1,$doc_obj);
193}
194
195# uses the multiread package to read in the entire file pointed to
196# by filename and loads the resulting text into $$textref. Input text
197# may be in any of the encodings handled by multiread, output text
198# will be in utf8
199sub read_file {
200 my $self = shift (@_);
201 my ($filename, $encoding, $language, $textref) = @_;
202
203 if (!-r $filename)
204 {
205 my $outhandle = $self->{'outhandle'};
206 gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
207 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
208 return;
209 }
210 $$textref = "";
211 if (!open (FILE, $filename)) {
212 gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
213 die "\n";
214 }
215
216 if ($encoding eq "ascii") {
217 undef $/;
218 $$textref = <FILE>;
219 $/ = "\n";
220 } else {
221 my $reader = new multiread();
222 $reader->set_handle ('ReadTextFile::FILE');
223 $reader->set_encoding ($encoding);
224 $reader->read_file ($textref);
225 }
226
227 # At this point $$textref is a binary byte string
228 # => turn it into a Unicode aware string, so full
229 # Unicode aware pattern matching can be used.
230 # For instance: 's/\x{0101}//g' or '[[:upper:]]'
231 #
232
233 $$textref = decode("utf8",$$textref);
234
235 close FILE;
236}
237
238
239# Not currently used
240sub UNUSED_read_file_usingPerlsEncodeModule {
241##sub read_file {
242 my $self = shift (@_);
243 my ($filename, $encoding, $language, $textref) = @_;
244
245 if (!-r $filename)
246 {
247 my $outhandle = $self->{'outhandle'};
248 gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
249 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
250 return;
251 }
252 $$textref = "";
253 if (!open (FILE, $filename)) {
254 gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_f
255or_reading} ($!)\n", $filename);
256 die "\n";
257 }
258
259 my $store_slash = $/;
260 undef $/;
261 my $text = <FILE>;
262 $/ = $store_slash;
263
264 $$textref = decode($encoding,$text);
265
266 close FILE;
267}
268
269
270sub read_file_no_decoding {
271 my $self = shift (@_);
272 my ($filename, $textref) = @_;
273
274 if (!-r $filename)
275 {
276 my $outhandle = $self->{'outhandle'};
277 gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
278 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
279 return;
280 }
281 $$textref = "";
282 if (!open (FILE, $filename)) {
283 gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
284 die "\n";
285 }
286
287 my $reader = new multiread();
288 $reader->set_handle ('ReadTextFile::FILE');
289 $reader->read_file_no_decoding ($textref);
290
291 $self->{'reader'} = $reader;
292
293 close FILE;
294}
295
296
297sub decode_text {
298 my $self = shift (@_);
299 my ($raw_text, $encoding, $language, $textref) = @_;
300
301 my $reader = $self->{'reader'};
302 if (!defined $reader) {
303 gsprintf(STDERR, "ReadTextFile::decode_text needs to call ReadTextFile::read_file_no_decoding first\n");
304 }
305 else {
306 $reader->set_encoding($encoding);
307 $reader->decode_text($raw_text,$textref);
308 }
309}
310
311
312sub textcat_get_language_encoding {
313 my $self = shift (@_);
314 my ($filename) = @_;
315
316 my ($language, $encoding, $extracted_encoding);
317 if ($self->{'input_encoding'} eq "auto") {
318 # use textcat to automatically work out the input encoding and language
319 ($language, $encoding) = $self->get_language_encoding ($filename);
320 } elsif ($self->{'extract_language'}) {
321 # use textcat to get language metadata
322 ($language, $extracted_encoding) = $self->get_language_encoding ($filename);
323 $encoding = $self->{'input_encoding'};
324 # don't print this message for english... english in utf8 is identical
325 # to english in iso-8859-1 (except for some punctuation). We don't have
326 # a language model for en_utf8, so textcat always says iso-8859-1!
327 if ($extracted_encoding ne $encoding && $language ne "en" && $self->{'verbosity'}) {
328 my $plugin_name = ref ($self);
329 my $outhandle = $self->{'outhandle'};
330 gsprintf($outhandle, "$plugin_name: {ReadTextFile.wrong_encoding}\n", $filename, $encoding, $extracted_encoding);
331 }
332 } else {
333 $language = $self->{'default_language'};
334 $encoding = $self->{'input_encoding'};
335 }
336
337# print STDERR "**** language encoding of contents of file $filename:\n\t****$language $encoding\n";
338
339 return ($language, $encoding);
340}
341
342
343# Uses textcat to work out the encoding and language of the text in
344# $filename. All html tags are removed before processing.
345# returns an array containing "language" and "encoding"
346sub get_language_encoding {
347 my $self = shift (@_);
348 my ($filename) = @_;
349 my $outhandle = $self->{'outhandle'};
350 my $unicode_format = "";
351 my $best_language = "";
352 my $best_encoding = "";
353
354
355 # read in file
356 if (!open (FILE, $filename)) {
357 gsprintf(STDERR, "ReadTextFile::get_language_encoding {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
358 # this is a pretty bad error, but try to continue anyway
359 return ($self->{'default_language'}, $self->{'input_encoding'});
360 }
361 undef $/;
362 my $text = <FILE>;
363 $/ = "\n";
364 close FILE;
365
366 # check if first few bytes have a Byte Order Marker
367 my $bom=substr($text,0,2); # check 16bit unicode
368 if ($bom eq "\xff\xfe") { # little endian 16bit unicode
369 $unicode_format="unicode";
370 } elsif ($bom eq "\xfe\xff") { # big endian 16bit unicode
371 $unicode_format="unicode";
372 } else {
373 $bom=substr($text,0,3); # check utf-8
374 if ($bom eq "\xef\xbb\xbf") { # utf-8 coded FEFF bom
375 $unicode_format="utf8";
376# } elsif ($bom eq "\xef\xbf\xbe") { # utf-8 coded FFFE bom. Error!?
377# $unicode_format="utf8";
378 }
379 }
380
381 my $found_html_encoding = 0;
382 # handle html files specially
383 # XXX this doesn't match plugins derived from HTMLPlug (except ConvertTo)
384 if (ref($self) eq 'HTMLPlugin' ||
385 (exists $self->{'converted_to'} && $self->{'converted_to'} eq 'HTML')){
386
387 # remove comments in head, including multiline ones, so that we don't match on
388 # inactive tags (those that are nested inside comments)
389 my ($head) = ($text =~ m/<head>(.*)<\/head>/si);
390 $head = "" unless defined $head; # some files are not proper HTML eg php files
391 $head =~ s/<!--.*?-->//sg;
392
393 # remove <title>stuff</title> -- as titles tend often to be in English
394 # for foreign language documents
395 $text =~ s!<title>.*?</title>!!si;
396
397 # see if this html file specifies its encoding
398 if ($text =~ /^<\?xml.*encoding="(.+?)"/) {
399 $best_encoding = $1;
400 }
401 # check the meta http-equiv charset tag
402 elsif ($head =~ m/<meta http-equiv.*content-type.*charset=(.+?)\"/si) {
403 $best_encoding = $1;
404 }
405 if ($best_encoding) { # we extracted an encoding
406 $best_encoding =~ s/-+/_/g;
407 $best_encoding = lc($best_encoding); # lowercase
408 if ($best_encoding eq "utf_8") { $best_encoding = "utf8" }
409 $found_html_encoding = 1;
410 # We shouldn't be modifying this here!!
411 #$self->{'input_encoding'} = $best_encoding;
412 }
413
414 # remove all HTML tags
415 $text =~ s/<[^>]*>//sg;
416 }
417
418 # don't need to do textcat if we know the encoding now AND don't need to extract language
419 if($found_html_encoding && !$self->{'extract_language'}) { # encoding specified in html file
420 $best_language = $self->{'default_language'};
421 }
422
423 else { # need to use textcat to get either the language, or get both language and encoding
424 $self->{'textcat'} = new textcat() if (!defined($self->{'textcat'}));
425
426 if($found_html_encoding) { # know encoding, find language by limiting search to known encoding
427 my $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
428
429 my $language;
430 ($language) = $results->[0] =~ m/^([^-]*)(?:-(?:.*))?$/ if (scalar @$results > 0);
431
432 if (!defined $language || scalar @$results > 3) {
433 # if there were too many results even when restricting results by encoding,
434 # or if there were no results, use default language with the known encoding
435 $best_language = $self->use_default_language($filename);
436 }
437 else { # fewer than 3 results means textcat is more certain, use the first result
438 $best_language = $language;
439 }
440 }
441 else { # don't know encoding or language yet, therefore we use textcat
442 my $results = $self->{'textcat'}->classify_contents(\$text, $filename);
443
444 # if textcat returns 3 or less possibilities we'll use the first one in the list
445 if (scalar @$results <= 3) { # results will be > 0 when we don't constrain textcat by an encoding
446 my ($language, $encoding) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/;
447
448 $language = $self->use_default_language($filename) unless defined $language;
449 $encoding = $self->use_default_encoding($filename) unless defined $encoding;
450
451 $best_language = $language;
452 $best_encoding = $encoding;
453 }
454 else { # if (scalar @$results > 3) {
455 if ($unicode_format) { # in case the first had a BOM
456 $best_encoding=$unicode_format;
457 }
458 else {
459 # Find the most frequent encoding in the textcat results returned
460 # Returns "" if there's no encoding more frequent than another
461 $best_encoding = $self->{'textcat'}->most_frequent_encoding($results);
462 }
463
464 if ($best_encoding eq "") { # encoding still not set, use defaults
465 $best_language = $self->use_default_language($filename);
466 $best_encoding = $self->use_default_encoding($filename);
467 }
468 elsif (!$self->{'extract_language'}) { # know encoding but don't need to discover language
469 $best_language = $self->use_default_language($filename);
470 }
471 else { # textcat again using the most frequent encoding or the $unicode_format set above
472 $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
473 my $language;
474 ($language) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/ if (scalar @$results > 0);
475 if (!defined $language || scalar @$results > 3) {
476 # if no result or too many results, use default language for the encoding previously found
477 $best_language = $self->use_default_language($filename);
478 }
479 else { # fewer than 3 results, use the language of the first result
480 $best_language = $language;
481 }
482 }
483 }
484 }
485 }
486
487 if($best_encoding eq "" || $best_language eq "") {
488 print STDERR "****Shouldn't happen: encoding and/or language still not set. Using defaults.\n";
489 $best_encoding = $self->use_default_encoding($filename) if $best_encoding eq "";
490 $best_language = $self->use_default_language($filename) if $best_language eq "";
491 }
492# print STDERR "****Content language: $best_language; Encoding: $best_encoding.\n";
493
494
495 if ($best_encoding =~ /^iso_8859/ && &unicode::check_is_utf8($text)) {
496 # the text is valid utf8, so assume that's the real encoding
497 # (since textcat is based on probabilities)
498 $best_encoding = 'utf8';
499 }
500
501 # check for equivalents where textcat doesn't have some encodings...
502 # eg MS versions of standard encodings
503 if ($best_encoding =~ /^iso_8859_(\d+)/) {
504 my $iso = $1; # which variant of the iso standard?
505 # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
506 if ($text =~ /[\x80-\x9f]/) {
507 # Western Europe
508 if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
509 elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
510 elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
511 elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
512 elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
513 elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
514 elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
515 }
516 }
517
518 if ($best_encoding !~ /^(ascii|utf8|unicode)$/ &&
519 !defined $encodings::encodings->{$best_encoding}) {
520 if ($self->{'verbosity'}) {
521 gsprintf($outhandle, "ReadTextFile: {ReadTextFile.unsupported_encoding}\n",
522 $filename, $best_encoding, $self->{'default_encoding'});
523 }
524 $best_encoding = $self->{'default_encoding'};
525 }
526
527 return ($best_language, $best_encoding);
528}
529
530
531sub use_default_language {
532 my $self = shift (@_);
533 my ($filename) = @_;
534
535 if ($self->{'verbosity'}>2) {
536 gsprintf($self->{'outhandle'},
537 "ReadTextFile: {ReadTextFile.could_not_extract_language}\n",
538 $filename, $self->{'default_language'});
539 }
540 return $self->{'default_language'};
541}
542
543sub use_default_encoding {
544 my $self = shift (@_);
545 my ($filename) = @_;
546
547 if ($self->{'verbosity'}>2) {
548 gsprintf($self->{'outhandle'},
549 "ReadTextFile: {ReadTextFile.could_not_extract_encoding}\n",
550 $filename, $self->{'default_encoding'});
551 }
552 return $self->{'default_encoding'};
553}
554
555# Overridden by exploding plugins (eg. ISISPlug)
556sub clean_up_after_exploding
557{
558 my $self = shift(@_);
559}
560
561
5621;
Note: See TracBrowser for help on using the repository browser.