source: main/trunk/greenstone2/perllib/plugins/ReadTextFile.pm@ 32096

Last change on this file since 32096 was 32096, checked in by ak19, 3 years ago

Marking all the uses of sysread() with a comment saying they're a candidate to use FileUtils::readUTF8File() in future, if thinking about each case beforehand has confirmed that the contents will indeed be UTF8.

  • Property svn:executable set to *
File size: 19.9 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},@{$CommonUtil::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' => $CommonUtil::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 my $pp_file = &util::prettyprint_file($base_dir,$file,$gli);
125 print $outhandle "$self->{'plugin_type'} processing $pp_file\n"
126 if $self->{'verbosity'} > 1;
127
128 my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
129
130 # Do encoding stuff
131 my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
132 if ($self->{'verbosity'} > 2) {
133 print $outhandle "ReadTextFile: reading $file as ($content_encoding,$language)\n";
134 }
135
136 # create a new document
137 my $doc_obj = new doc ($filename_full_path, "indexed_doc", $self->{'file_rename_method'});
138 my $top_section = $doc_obj->get_top_section();
139
140 # this should look at the plugin option too...
141 $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
142 $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename_full_path));
143
144 my $plugin_filename_encoding = $self->{'filename_encoding'};
145 my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
146 $self->set_Source_metadata($doc_obj, $filename_full_path, $filename_encoding);
147
148 $doc_obj->add_utf8_metadata($top_section, "Language", $language);
149 $doc_obj->add_utf8_metadata($top_section, "Encoding", $content_encoding);
150
151 # read in file ($text will be in perl internal unicode aware format)
152 my $text = "";
153 $self->read_file ($filename_full_path, $content_encoding, $language, \$text);
154
155 if (!length ($text)) {
156 if ($gli) {
157 print STDERR "<ProcessingError n='$file' r='File contains no text'>\n";
158 }
159 gsprintf($outhandle, "$self->{'plugin_type'}: {ReadTextFile.file_has_no_text}\n", $filename_full_path) if $self->{'verbosity'};
160
161 my $failhandle = $self->{'failhandle'};
162 gsprintf($failhandle, "$file: " . ref($self) . ": {ReadTextFile.empty_file}\n");
163 # print $failhandle "$file: " . ref($self) . ": file contains no text\n";
164 $self->{'num_not_processed'} ++;
165
166 return (0,undef); # what should we return here?? error but don't want to pass it on
167 }
168
169 # do plugin specific processing of doc_obj
170 unless (defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli))) {
171 $text = '';
172 undef $text;
173 print STDERR "<ProcessingError n='$file'>\n" if ($gli);
174 return (-1,undef);
175 }
176 $text='';
177 undef $text;
178
179 # include any metadata passed in from previous plugins
180 # note that this metadata is associated with the top level section
181 $self->add_associated_files($doc_obj, $filename_full_path);
182 $self->extra_metadata ($doc_obj, $top_section, $metadata);
183
184 # do any automatic metadata extraction
185 $self->auto_extract_metadata ($doc_obj);
186
187
188 # if we haven't found any Title so far, assign one
189 $self->title_fallback($doc_obj,$top_section,$filename_no_path);
190
191 $self->add_OID($doc_obj);
192
193 return (1,$doc_obj);
194}
195
196# uses the multiread package to read in the entire file pointed to
197# by filename and loads the resulting text into $$textref. Input text
198# may be in any of the encodings handled by multiread, output text
199# will be in utf8
200sub read_file {
201 my $self = shift (@_);
202 my ($filename, $encoding, $language, $textref) = @_;
203
204 if (!-r $filename)
205 {
206 my $outhandle = $self->{'outhandle'};
207 gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
208 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
209 return;
210 }
211 $$textref = "";
212 if (!open (FILE, $filename)) {
213 gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
214 die "\n";
215 }
216
217 if ($encoding eq "ascii") {
218 # The following file reading section is a candidate to use FileUtils::readUTF8File()
219 # in place of calling sysread() directly. But only if we can reason we'd be working with UTF8.
220 # ascii is UTF8, but it may become inefficient to treat ascii as utf8
221
222 # Replace file 'slurp' with faster implementation
223 sysread(FILE, $$textref, -s FILE);
224
225 # The old slow way of reading in a file
226 #undef $/;
227 #$$textref = <FILE>;
228 #$/ = "\n";
229 } else {
230 my $reader = new multiread();
231 $reader->set_handle ('ReadTextFile::FILE');
232 $reader->set_encoding ($encoding);
233 $reader->read_file ($textref);
234 }
235
236 # At this point $$textref is a binary byte string
237 # => turn it into a Unicode aware string, so full
238 # Unicode aware pattern matching can be used.
239 # For instance: 's/\x{0101}//g' or '[[:upper:]]'
240 #
241
242 $$textref = decode("utf8",$$textref);
243
244 close FILE;
245}
246
247
248# Not currently used
249sub UNUSED_read_file_usingPerlsEncodeModule {
250##sub read_file {
251 my $self = shift (@_);
252 my ($filename, $encoding, $language, $textref) = @_;
253
254 if (!-r $filename)
255 {
256 my $outhandle = $self->{'outhandle'};
257 gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
258 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
259 return;
260 }
261 $$textref = "";
262 if (!open (FILE, $filename)) {
263 gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_f
264or_reading} ($!)\n", $filename);
265 die "\n";
266 }
267
268 my $store_slash = $/;
269 undef $/;
270 my $text = <FILE>;
271 $/ = $store_slash;
272
273 $$textref = decode($encoding,$text);
274
275 close FILE;
276}
277
278
279sub read_file_no_decoding {
280 my $self = shift (@_);
281 my ($filename, $textref) = @_;
282
283 if (!-r $filename)
284 {
285 my $outhandle = $self->{'outhandle'};
286 gsprintf($outhandle, "{ReadTextFile.read_denied}\n", $filename) if $self->{'verbosity'};
287 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'};
288 return;
289 }
290 $$textref = "";
291 if (!open (FILE, $filename)) {
292 gsprintf(STDERR, "ReadTextFile::read_file {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
293 die "\n";
294 }
295
296 my $reader = new multiread();
297 $reader->set_handle ('ReadTextFile::FILE');
298 $reader->read_file_no_decoding ($textref);
299
300 $self->{'reader'} = $reader;
301
302 close FILE;
303}
304
305
306sub decode_text {
307 my $self = shift (@_);
308 my ($raw_text, $encoding, $language, $textref) = @_;
309
310 my $reader = $self->{'reader'};
311 if (!defined $reader) {
312 gsprintf(STDERR, "ReadTextFile::decode_text needs to call ReadTextFile::read_file_no_decoding first\n");
313 }
314 else {
315 $reader->set_encoding($encoding);
316 $reader->decode_text($raw_text,$textref);
317
318 # At this point $$textref is a binary byte string
319 # => turn it into a Unicode aware string, so full
320 # Unicode aware pattern matching can be used.
321 # For instance: 's/\x{0101}//g' or '[[:upper:]]'
322
323 $$textref = decode("utf8",$$textref);
324 }
325}
326
327
328sub textcat_get_language_encoding {
329 my $self = shift (@_);
330 my ($filename) = @_;
331
332 my ($language, $encoding, $extracted_encoding);
333 if ($self->{'input_encoding'} eq "auto") {
334 # use textcat to automatically work out the input encoding and language
335 ($language, $encoding) = $self->get_language_encoding ($filename);
336 } elsif ($self->{'extract_language'}) {
337 # use textcat to get language metadata
338 ($language, $extracted_encoding) = $self->get_language_encoding ($filename);
339 $encoding = $self->{'input_encoding'};
340 # don't print this message for english... english in utf8 is identical
341 # to english in iso-8859-1 (except for some punctuation). We don't have
342 # a language model for en_utf8, so textcat always says iso-8859-1!
343 if ($extracted_encoding ne $encoding && $language ne "en" && $self->{'verbosity'}) {
344 my $plugin_name = ref ($self);
345 my $outhandle = $self->{'outhandle'};
346 gsprintf($outhandle, "$plugin_name: {ReadTextFile.wrong_encoding}\n", $filename, $encoding, $extracted_encoding);
347 }
348 } else {
349 $language = $self->{'default_language'};
350 $encoding = $self->{'input_encoding'};
351 }
352
353# print STDERR "**** language encoding of contents of file $filename:\n\t****$language $encoding\n";
354
355 return ($language, $encoding);
356}
357
358
359# Uses textcat to work out the encoding and language of the text in
360# $filename. All html tags are removed before processing.
361# returns an array containing "language" and "encoding"
362sub get_language_encoding {
363 my $self = shift (@_);
364 my ($filename) = @_;
365 my $outhandle = $self->{'outhandle'};
366 my $unicode_format = "";
367 my $best_language = "";
368 my $best_encoding = "";
369
370
371 # read in file
372 if (!open (FILE, $filename)) {
373 gsprintf(STDERR, "ReadTextFile::get_language_encoding {ReadTextFile.could_not_open_for_reading} ($!)\n", $filename);
374 # this is a pretty bad error, but try to continue anyway
375 return ($self->{'default_language'}, $self->{'input_encoding'});
376 }
377 undef $/;
378 my $text = <FILE>;
379 $/ = "\n";
380 close FILE;
381
382 # check if first few bytes have a Byte Order Marker
383 my $bom=substr($text,0,2); # check 16bit unicode
384 if ($bom eq "\xff\xfe") { # little endian 16bit unicode
385 $unicode_format="unicode";
386 } elsif ($bom eq "\xfe\xff") { # big endian 16bit unicode
387 $unicode_format="unicode";
388 } else {
389 $bom=substr($text,0,3); # check utf-8
390 if ($bom eq "\xef\xbb\xbf") { # utf-8 coded FEFF bom
391 $unicode_format="utf8";
392# } elsif ($bom eq "\xef\xbf\xbe") { # utf-8 coded FFFE bom. Error!?
393# $unicode_format="utf8";
394 }
395 }
396
397 my $found_html_encoding = 0;
398 # handle html files specially
399 # XXX this doesn't match plugins derived from HTMLPlug (except ConvertTo)
400 if (ref($self) eq 'HTMLPlugin' ||
401 (exists $self->{'converted_to'} && $self->{'converted_to'} eq 'HTML')){
402
403 # remove comments in head, including multiline ones, so that we don't match on
404 # inactive tags (those that are nested inside comments)
405 my ($head) = ($text =~ m/<head>(.*)<\/head>/si);
406 $head = "" unless defined $head; # some files are not proper HTML eg php files
407 $head =~ s/<!--.*?-->//sg;
408
409 # remove <title>stuff</title> -- as titles tend often to be in English
410 # for foreign language documents
411 $text =~ s!<title>.*?</title>!!si;
412
413 # see if this html file specifies its encoding
414 if ($text =~ /^<\?xml.*encoding="(.+?)"/) {
415 $best_encoding = $1;
416 }
417 # check the meta http-equiv charset tag
418 elsif ($head =~ m/<meta http-equiv.*content-type.*charset=(.+?)\"/si) {
419 $best_encoding = $1;
420 }
421 if ($best_encoding) { # we extracted an encoding
422 $best_encoding =~ s/-+/_/g;
423 $best_encoding = lc($best_encoding); # lowercase
424 if ($best_encoding eq "utf_8") { $best_encoding = "utf8" }
425 $found_html_encoding = 1;
426 # We shouldn't be modifying this here!!
427 #$self->{'input_encoding'} = $best_encoding;
428 }
429
430 # remove all HTML tags
431 $text =~ s/<[^>]*>//sg;
432 }
433
434 # don't need to do textcat if we know the encoding now AND don't need to extract language
435 if($found_html_encoding && !$self->{'extract_language'}) { # encoding specified in html file
436 $best_language = $self->{'default_language'};
437 }
438
439 else { # need to use textcat to get either the language, or get both language and encoding
440 $self->{'textcat'} = new textcat() if (!defined($self->{'textcat'}));
441
442 if($found_html_encoding) { # know encoding, find language by limiting search to known encoding
443 my $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
444
445 my $language;
446 ($language) = $results->[0] =~ m/^([^-]*)(?:-(?:.*))?$/ if (scalar @$results > 0);
447
448 if (!defined $language || scalar @$results > 3) {
449 # if there were too many results even when restricting results by encoding,
450 # or if there were no results, use default language with the known encoding
451 $best_language = $self->use_default_language($filename);
452 }
453 else { # fewer than 3 results means textcat is more certain, use the first result
454 $best_language = $language;
455 }
456 }
457 else { # don't know encoding or language yet, therefore we use textcat
458 my $results = $self->{'textcat'}->classify_contents(\$text, $filename);
459
460 # if textcat returns 3 or less possibilities we'll use the first one in the list
461 if (scalar @$results <= 3) { # results will be > 0 when we don't constrain textcat by an encoding
462 my ($language, $encoding) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/;
463
464 $language = $self->use_default_language($filename) unless defined $language;
465 $encoding = $self->use_default_encoding($filename) unless defined $encoding;
466
467 $best_language = $language;
468 $best_encoding = $encoding;
469 }
470 else { # if (scalar @$results > 3) {
471 if ($unicode_format) { # in case the first had a BOM
472 $best_encoding=$unicode_format;
473 }
474 else {
475 # Find the most frequent encoding in the textcat results returned
476 # Returns "" if there's no encoding more frequent than another
477 $best_encoding = $self->{'textcat'}->most_frequent_encoding($results);
478 }
479
480 if ($best_encoding eq "") { # encoding still not set, use defaults
481 $best_language = $self->use_default_language($filename);
482 $best_encoding = $self->use_default_encoding($filename);
483 }
484 elsif (!$self->{'extract_language'}) { # know encoding but don't need to discover language
485 $best_language = $self->use_default_language($filename);
486 }
487 else { # textcat again using the most frequent encoding or the $unicode_format set above
488 $results = $self->{'textcat'}->classify_contents_for_encoding(\$text, $filename, $best_encoding);
489 my $language;
490 ($language) = $results->[0] =~ m/^([^-]*)(?:-(.*))?$/ if (scalar @$results > 0);
491 if (!defined $language || scalar @$results > 3) {
492 # if no result or too many results, use default language for the encoding previously found
493 $best_language = $self->use_default_language($filename);
494 }
495 else { # fewer than 3 results, use the language of the first result
496 $best_language = $language;
497 }
498 }
499 }
500 }
501 }
502
503 if($best_encoding eq "" || $best_language eq "") {
504 print STDERR "****Shouldn't happen: encoding and/or language still not set. Using defaults.\n";
505 $best_encoding = $self->use_default_encoding($filename) if $best_encoding eq "";
506 $best_language = $self->use_default_language($filename) if $best_language eq "";
507 }
508# print STDERR "****Content language: $best_language; Encoding: $best_encoding.\n";
509
510
511 if ($best_encoding =~ /^iso_8859/ && &unicode::check_is_utf8($text)) {
512 # the text is valid utf8, so assume that's the real encoding
513 # (since textcat is based on probabilities)
514 $best_encoding = 'utf8';
515 }
516
517 # check for equivalents where textcat doesn't have some encodings...
518 # eg MS versions of standard encodings
519 if ($best_encoding =~ /^iso_8859_(\d+)/) {
520 my $iso = $1; # which variant of the iso standard?
521 # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
522 if ($text =~ /[\x80-\x9f]/) {
523 # Western Europe
524 if ($iso == 1 or $iso == 15) { $best_encoding = 'windows_1252' }
525 elsif ($iso == 2) {$best_encoding = 'windows_1250'} # Central Europe
526 elsif ($iso == 5) {$best_encoding = 'windows_1251'} # Cyrillic
527 elsif ($iso == 6) {$best_encoding = 'windows_1256'} # Arabic
528 elsif ($iso == 7) {$best_encoding = 'windows_1253'} # Greek
529 elsif ($iso == 8) {$best_encoding = 'windows_1255'} # Hebrew
530 elsif ($iso == 9) {$best_encoding = 'windows_1254'} # Turkish
531 }
532 }
533
534 if ($best_encoding !~ /^(ascii|utf8|unicode)$/ &&
535 !defined $encodings::encodings->{$best_encoding}) {
536 if ($self->{'verbosity'}) {
537 gsprintf($outhandle, "ReadTextFile: {ReadTextFile.unsupported_encoding}\n",
538 $filename, $best_encoding, $self->{'default_encoding'});
539 }
540 $best_encoding = $self->{'default_encoding'};
541 }
542
543 return ($best_language, $best_encoding);
544}
545
546
547sub use_default_language {
548 my $self = shift (@_);
549 my ($filename) = @_;
550
551 if ($self->{'verbosity'}>2) {
552 gsprintf($self->{'outhandle'},
553 "ReadTextFile: {ReadTextFile.could_not_extract_language}\n",
554 $filename, $self->{'default_language'});
555 }
556 return $self->{'default_language'};
557}
558
559sub use_default_encoding {
560 my $self = shift (@_);
561 my ($filename) = @_;
562
563 if ($self->{'verbosity'}>2) {
564 gsprintf($self->{'outhandle'},
565 "ReadTextFile: {ReadTextFile.could_not_extract_encoding}\n",
566 $filename, $self->{'default_encoding'});
567 }
568 return $self->{'default_encoding'};
569}
570
571# Overridden by exploding plugins (eg. ISISPlug)
572sub clean_up_after_exploding
573{
574 my $self = shift(@_);
575}
576
577
5781;
Note: See TracBrowser for help on using the repository browser.