source: gsdl/trunk/perllib/plugins/HTMLPlugin.pm@ 17863

Last change on this file since 17863 was 17863, checked in by anna, 15 years ago

change the name tidy_html switch in HTMLPlug into use_realistic_book in HTMLPlugin

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 51.4 KB
Line 
1###########################################################################
2#
3# HTMLPlugin.pm -- basic html plugin
4#
5# A component of the Greenstone digital library software
6# from the New Zealand Digital Library Project at the
7# University of Waikato, New Zealand.
8#
9# Copyright (C) 1999 New Zealand Digital Library Project
10#
11# This program is free software; you can redistribute it and/or modify
12# it under the terms of the GNU General Public License as published by
13# the Free Software Foundation; either version 2 of the License, or
14# (at your option) any later version.
15#
16# This program is distributed in the hope that it will be useful,
17# but WITHOUT ANY WARRANTY; without even the implied warranty of
18# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19# GNU General Public License for more details.
20#
21# You should have received a copy of the GNU General Public License
22# along with this program; if not, write to the Free Software
23# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
24#
25###########################################################################
26
27#
28# Note that this plugin handles frames only in a very simple way
29# i.e. each frame is treated as a separate document. This means
30# search results will contain links to individual frames rather
31# than linking to the top level frameset.
32# There may also be some problems caused by the _parent target
33# (it's removed by this plugin)
34#
35
36package HTMLPlugin;
37
38use ReadTextFile;
39use HBPlugin;
40use ghtml;
41use unicode;
42use util;
43use XMLParser;
44
45use Image::Size;
46use File::Copy;
47
48sub BEGIN {
49 @HTMLPlugin::ISA = ('ReadTextFile', 'HBPlugin');
50}
51
52use strict; # every perl program should have this!
53no strict 'refs'; # make an exception so we can use variables as filehandles
54
55my $arguments =
56 [ { 'name' => "process_exp",
57 'desc' => "{BasePlugin.process_exp}",
58 'type' => "regexp",
59 'deft' => &get_default_process_exp() },
60 { 'name' => "block_exp",
61 'desc' => "{BasePlugin.block_exp}",
62 'type' => 'regexp',
63 'deft' => &get_default_block_exp() },
64 { 'name' => "nolinks",
65 'desc' => "{HTMLPlugin.nolinks}",
66 'type' => "flag" },
67 { 'name' => "keep_head",
68 'desc' => "{HTMLPlugin.keep_head}",
69 'type' => "flag" },
70 { 'name' => "no_metadata",
71 'desc' => "{HTMLPlugin.no_metadata}",
72 'type' => "flag" },
73 { 'name' => "metadata_fields",
74 'desc' => "{HTMLPlugin.metadata_fields}",
75 'type' => "string",
76 'deft' => "Title" },
77 { 'name' => "hunt_creator_metadata",
78 'desc' => "{HTMLPlugin.hunt_creator_metadata}",
79 'type' => "flag" },
80 { 'name' => "file_is_url",
81 'desc' => "{HTMLPlugin.file_is_url}",
82 'type' => "flag" },
83 { 'name' => "assoc_files",
84 'desc' => "{HTMLPlugin.assoc_files}",
85 'type' => "regexp",
86 'deft' => &get_default_block_exp() },
87 { 'name' => "rename_assoc_files",
88 'desc' => "{HTMLPlugin.rename_assoc_files}",
89 'type' => "flag" },
90 { 'name' => "title_sub",
91 'desc' => "{HTMLPlugin.title_sub}",
92 'type' => "string",
93 'deft' => "" },
94 { 'name' => "description_tags",
95 'desc' => "{HTMLPlugin.description_tags}",
96 'type' => "flag" },
97 # retain this for backward compatibility (w3mir option was replaced by
98 # file_is_url)
99 { 'name' => "w3mir",
100# 'desc' => "{HTMLPlugin.w3mir}",
101 'type' => "flag",
102 'hiddengli' => "yes"},
103 { 'name' => "no_strip_metadata_html",
104 'desc' => "{HTMLPlugin.no_strip_metadata_html}",
105 'type' => "string",
106 'deft' => "",
107 'reqd' => "no"},
108 { 'name' => "sectionalise_using_h_tags",
109 'desc' => "{HTMLPlugin.sectionalise_using_h_tags}",
110 'type' => "flag" },
111 { 'name' => "use_realistic_book",
112 'desc' => "{HTMLPlugin.tidy_html}",
113 'type' => "flag"},
114 { 'name' => "old_style_HDL",
115 'desc' => "{HTMLPlugin.old_style_HDL}",
116 'type' => "flag"}
117 ];
118
119my $options = { 'name' => "HTMLPlugin",
120 'desc' => "{HTMLPlugin.desc}",
121 'abstract' => "no",
122 'inherits' => "yes",
123 'args' => $arguments };
124
125
126sub HB_read_html_file {
127 my $self = shift (@_);
128 my ($htmlfile, $text) = @_;
129
130 # load in the file
131 if (!open (FILE, $htmlfile)) {
132 print STDERR "ERROR - could not open $htmlfile\n";
133 return;
134 }
135
136 my $foundbody = 0;
137 $self->HB_gettext (\$foundbody, $text, "FILE");
138 close FILE;
139
140 # just in case there was no <body> tag
141 if (!$foundbody) {
142 $foundbody = 1;
143 open (FILE, $htmlfile) || return;
144 $self->HB_gettext (\$foundbody, $text, "FILE");
145 close FILE;
146 }
147 # text is in utf8
148}
149
150# converts the text to utf8, as ghtml does that for &eacute; etc.
151sub HB_gettext {
152 my $self = shift (@_);
153 my ($foundbody, $text, $handle) = @_;
154
155 my $line = "";
156 while (defined ($line = <$handle>)) {
157 # look for body tag
158 if (!$$foundbody) {
159 if ($line =~ s/^.*<body[^>]*>//i) {
160 $$foundbody = 1;
161 } else {
162 next;
163 }
164 }
165
166 # check for symbol fonts
167 if ($line =~ m/<font [^>]*?face\s*=\s*\"?(\w+)\"?/i) {
168 my $font = $1;
169 print STDERR "HBPlug::HB_gettext - warning removed font $font\n"
170 if ($font !~ m/^arial$/i);
171 }
172
173 $$text .= $line;
174 }
175
176 if ($self->{'input_encoding'} eq "iso_8859_1") {
177 # convert to utf-8
178 $$text=&unicode::unicode2utf8(&unicode::convert2unicode("iso_8859_1", $text));
179 }
180 # convert any alphanumeric character entities to their utf-8
181 # equivalent for indexing purposes
182 #&ghtml::convertcharentities ($$text);
183
184 $$text =~ s/\s+/ /g; # remove \n's
185}
186
187sub HB_clean_section {
188 my $self = shift (@_);
189 my ($section) = @_;
190
191 # remove tags without a starting tag from the section
192 my ($tag, $tagstart);
193 while ($section =~ m/<\/([^>]{1,10})>/) {
194 $tag = $1;
195 $tagstart = index($section, "<$tag");
196 last if (($tagstart >= 0) && ($tagstart < index($section, "<\/$tag")));
197 $section =~ s/<\/$tag>//;
198 }
199
200 # remove extra paragraph tags
201 while ($section =~ s/<p\b[^>]*>\s*<p\b/<p/ig) {}
202
203 # remove extra stuff at the end of the section
204 while ($section =~ s/(<u>|<i>|<b>|<p\b[^>]*>|&nbsp;|\s)$//i) {}
205
206 # add a newline at the beginning of each paragraph
207 $section =~ s/(.)\s*<p\b/$1\n\n<p/gi;
208
209 # add a newline every 80 characters at a word boundary
210 # Note: this regular expression puts a line feed before
211 # the last word in each section, even when it is not
212 # needed.
213 $section =~ s/(.{1,80})\s/$1\n/g;
214
215 # fix up the image links
216 $section =~ s/<img[^>]*?src=\"?([^\">]+)\"?[^>]*>/
217 <center><img src=\"$1\" \/><\/center><br\/>/ig;
218 $section =~ s/&lt;&lt;I&gt;&gt;\s*([^\.]+\.(png|jpg|gif))/
219 <center><img src=\"$1\" \/><\/center><br\/>/ig;
220
221 return $section;
222}
223
224# Will convert the oldHDL format to the new HDL format (using the Section tag)
225sub convert_to_newHDLformat
226{
227 my $self = shift (@_);
228 my ($file,$cnfile) = @_;
229 my $input_filename = $file;
230 my $tmp_filename = $cnfile;
231
232 # write HTML tmp file with new HDL format
233 open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
234
235 # read in the file and do basic html cleaning (removing header etc)
236 my $html = "";
237 $self->HB_read_html_file ($input_filename, \$html);
238
239 # process the file one section at a time
240 my $curtoclevel = 1;
241 my $firstsection = 1;
242 my $toclevel = 0;
243 while (length ($html) > 0) {
244 if ($html =~ s/^.*?(?:<p\b[^>]*>)?((<b>|<i>|<u>|\s)*)&lt;&lt;TOC(\d+)&gt;&gt;\s*(.*?)<p\b/<p/i) {
245 $toclevel = $3;
246 my $title = $4;
247 my $sectiontext = "";
248 if ($html =~ s/^(.*?)((?:<p\b[^>]*>)?((<b>|<i>|<u>|\s)*)&lt;&lt;TOC\d+&gt;&gt;)/$2/i) {
249 $sectiontext = $1;
250 } else {
251 $sectiontext = $html;
252 $html = "";
253 }
254
255 # remove tags and extra spaces from the title
256 $title =~ s/<\/?[^>]+>//g;
257 $title =~ s/^\s+|\s+$//g;
258
259 # close any sections below the current level and
260 # create a new section (special case for the firstsection)
261 print PROD "<!--\n";
262 while (($curtoclevel > $toclevel) ||
263 (!$firstsection && $curtoclevel == $toclevel)) {
264 $curtoclevel--;
265 print PROD "</Section>\n";
266 }
267 if ($curtoclevel+1 < $toclevel) {
268 print STDERR "WARNING - jump in toc levels in $input_filename " .
269 "from $curtoclevel to $toclevel\n";
270 }
271 while ($curtoclevel < $toclevel) {
272 $curtoclevel++;
273 }
274
275 if ($curtoclevel == 1) {
276 # add the header tag
277 print PROD "-->\n";
278 print PROD "<HTML>\n<HEAD>\n<TITLE>$title</TITLE>\n</HEAD>\n<BODY>\n";
279 print PROD "<!--\n";
280 }
281
282 print PROD "<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">$title</Metadata>\n\t</Description>\n";
283
284 print PROD "-->\n";
285
286 # clean up the section html
287 $sectiontext = $self->HB_clean_section($sectiontext);
288
289 print PROD "$sectiontext\n";
290
291 } else {
292 print STDERR "WARNING - leftover text\n" , $self->shorten($html),
293 "\nin $input_filename\n";
294 last;
295 }
296 $firstsection = 0;
297 }
298
299 print PROD "<!--\n";
300 while ($curtoclevel > 0) {
301 $curtoclevel--;
302 print PROD "</Section>\n";
303 }
304 print PROD "-->\n";
305
306 close (PROD) || die("Error Closing File: $tmp_filename $!");
307
308 return $tmp_filename;
309}
310
311sub shorten {
312 my $self = shift (@_);
313 my ($text) = @_;
314
315 return "\"$text\"" if (length($text) < 100);
316
317 return "\"" . substr ($text, 0, 50) . "\" ... \"" .
318 substr ($text, length($text)-50) . "\"";
319}
320
321sub convert_tidy_or_oldHDL_file
322{
323 my $self = shift (@_);
324 my ($file) = @_;
325 my $input_filename = $file;
326
327 if (-d $input_filename)
328 {
329 return $input_filename;
330 }
331
332 # get the input filename
333 my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
334 my $base_dirname = $dirname;
335 $suffix = lc($suffix);
336
337 # derive tmp filename from input filename
338 # Remove any white space from filename -- no risk of name collision, and
339 # makes later conversion by utils simpler. Leave spaces in path...
340 # tidy up the filename with space, dot, hyphen between
341 $tailname =~ s/\s+//g;
342 $tailname =~ s/\.+//g;
343 $tailname =~ s/\-+//g;
344 # convert to utf-8 otherwise we have problems with the doc.xml file
345 # later on
346 &unicode::ensure_utf8(\$tailname);
347
348 # softlink to collection tmp dir
349 my $tmp_dirname = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "tidytmp");
350 &util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
351
352 my $test_dirname = "";
353 my $f_separator = &util::get_os_dirsep();
354
355 if ($dirname =~ m/import$f_separator/)
356 {
357 $test_dirname = $'; #'
358
359 #print STDERR "init $'\n";
360
361 while ($test_dirname =~ m/[$f_separator]/)
362 {
363 my $folderdirname = $`;
364 $tmp_dirname = &util::filename_cat($tmp_dirname,$folderdirname);
365 &util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
366 $test_dirname = $'; #'
367 }
368 }
369
370 my $tmp_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
371
372 # tidy or convert the input file if it is a HTML-like file or it is accepted by the process_exp
373 if (($suffix eq ".htm") || ($suffix eq ".html") || ($suffix eq ".shtml"))
374 {
375 #convert the input file to a new style HDL
376 my $hdl_output_filename = $input_filename;
377 if ($self->{'old_style_HDL'})
378 {
379 $hdl_output_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
380 $hdl_output_filename = $self->convert_to_newHDLformat($input_filename,$hdl_output_filename);
381 }
382
383 #just for checking copy all other file from the base dir to tmp dir if it is not exists
384 opendir(DIR,$base_dirname) or die "Can't open base directory : $base_dirname!";
385 my @files = grep {!/^\.+$/} readdir(DIR);
386 close(DIR);
387
388 foreach my $file (@files)
389 {
390 my $src_file = &util::filename_cat($base_dirname,$file);
391 my $dest_file = &util::filename_cat($tmp_dirname,$file);
392 if ((!-e $dest_file) && (!-d $src_file))
393 {
394 # just copy the original file back to the tmp directory
395 copy($src_file,$dest_file) or die "Can't copy file $src_file to $dest_file $!";
396 }
397 }
398
399 # tidy the input file
400 my $tidy_output_filename = $hdl_output_filename;
401 if ($self->{'use_realistic_book'})
402 {
403 $tidy_output_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
404 $tidy_output_filename = $self->tmp_tidy_file($hdl_output_filename,$tidy_output_filename);
405 }
406 $tmp_filename = $tidy_output_filename;
407 }
408 else
409 {
410 if (!-e $tmp_filename)
411 {
412 # just copy the original file back to the tmp directory
413 copy($input_filename,$tmp_filename) or die "Can't copy file $input_filename to $tmp_filename $!";
414 }
415 }
416
417 return $tmp_filename;
418}
419
420
421# Will make the html input file as a proper XML file with removed font tag and
422# image size added to the img tag.
423# The tidying process takes place in a collection specific 'tmp' directory so
424# that we don't accidentally damage the input.
425sub tmp_tidy_file
426{
427 my $self = shift (@_);
428 my ($file,$cnfile) = @_;
429 my $input_filename = $file;
430 my $tmp_filename = $cnfile;
431
432 # get the input filename
433 my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
434
435 require HTML::TokeParser::Simple;
436
437 # create HTML parser to decode the input file
438 my $parser = HTML::TokeParser::Simple->new($input_filename);
439
440 # write HTML tmp file without the font tag and image size are added to the img tag
441 open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
442 while (my $token = $parser->get_token())
443 {
444 # is it an img tag
445 if ($token->is_start_tag('img'))
446 {
447 # get the attributes
448 my $attr = $token->return_attr;
449
450 # get the full path to the image
451 my $img_file = &util::filename_cat($dirname,$attr->{src});
452
453 # set the width and height attribute
454 ($attr->{width}, $attr->{height}) = imgsize($img_file);
455
456 # recreate the tag
457 print PROD "<img";
458 print PROD map { qq { $_="$attr->{$_}"} } keys %$attr;
459 print PROD ">";
460 }
461 # is it a font tag
462 else
463 {
464 if (($token->is_start_tag('font')) || ($token->is_end_tag('font')))
465 {
466 # remove font tag
467 print PROD "";
468 }
469 else
470 {
471 # print without changes
472 print PROD $token->as_is;
473 }
474 }
475 }
476 close (PROD) || die("Error Closing File: $tmp_filename $!");
477
478 # run html-tidy on the tmp file to make it a proper XML file
479 my $tidyfile = `tidy -utf8 -wrap 0 -asxml $tmp_filename`;
480
481 # write result back to the tmp file
482 open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
483 print PROD $tidyfile;
484 close (PROD) || die("Error Closing File: $tmp_filename $!");
485
486 # return the output filename
487 return $tmp_filename;
488}
489
490sub read_into_doc_obj
491{
492 my $self = shift (@_);
493 my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
494
495 # get the input file
496 my $input_filename = $file;
497 my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
498 $suffix = lc($suffix);
499
500 if (($self->{'use_realistic_book'}) || ($self->{'old_style_HDL'}))
501 {
502 # because the document has to be sectionalized set the description tags
503 $self->{'description_tags'} = 1;
504
505 # set the file to be tidied
506 $input_filename = &util::filename_cat($base_dir,$file) if $base_dir =~ m/\w/;
507
508 # get the tidied file
509 #my $tidy_filename = $self->tmp_tidy_file($input_filename);
510 my $tidy_filename = $self->convert_tidy_or_oldHDL_file($input_filename);
511
512 # derive tmp filename from input filename
513 my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($tidy_filename, "\\.[^\\.]+\$");
514
515 # set the new input file and base_dir to be from the tidied file
516 $file = "$tailname$suffix";
517 $base_dir = $dirname;
518 }
519
520 # call the parent read_into_doc_obj
521 my ($process_status,$doc_obj) = $self->SUPER::read_into_doc_obj($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli);
522
523 return ($process_status,$doc_obj);
524}
525
526sub new {
527 my ($class) = shift (@_);
528 my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
529 push(@$pluginlist, $class);
530
531 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
532 push(@{$hashArgOptLists->{"OptList"}},$options);
533
534
535 my $self = new ReadTextFile($pluginlist,$inputargs,$hashArgOptLists);
536
537 if ($self->{'w3mir'}) {
538 $self->{'file_is_url'} = 1;
539 }
540 $self->{'aux_files'} = {};
541 $self->{'dir_num'} = 0;
542 $self->{'file_num'} = 0;
543
544 return bless $self, $class;
545}
546
547# may want to use (?i)\.(gif|jpe?g|jpe|png|css|js(?:@.*)?)$
548# if have eg <script language="javascript" src="img/lib.js@123">
549sub get_default_block_exp {
550 my $self = shift (@_);
551
552 #return q^(?i)\.(gif|jpe?g|jpe|jpg|png|css)$^;
553 return "";
554}
555
556sub get_default_process_exp {
557 my $self = shift (@_);
558
559 # the last option is an attempt to encode the concept of an html query ...
560 return q^(?i)(\.html?|\.shtml|\.shm|\.asp|\.php\d?|\.cgi|.+\?.+=.*)$^;
561}
562
563sub store_block_files
564{
565 my $self =shift (@_);
566 my ($filename_full_path, $block_hash) = @_;
567
568 my $html_fname = $filename_full_path;
569 my @file_blocks;
570
571 my ($language, $encoding) = $self->textcat_get_language_encoding ($filename_full_path);
572
573 # read in file ($text will be in utf8)
574 my $raw_text = "";
575 $self->read_file_no_decoding ($filename_full_path, \$raw_text);
576
577 my $textref = \$raw_text;
578 my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
579 my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
580 $$textref =~ s/$opencom(.*?)$closecom//gs;
581
582 my $attval = "\\\"[^\\\"]+\\\"|[^\\s>]+";
583 my @img_matches = ($$textref =~ m/<img[^>]*?src\s*=\s*($attval)[^>]*>/igs);
584 my @usemap_matches = ($$textref =~ m/<img[^>]*?usemap\s*=\s*($attval)[^>]*>/igs);
585 my @link_matches = ($$textref =~ m/<link[^>]*?href\s*=\s*($attval)[^>]*>/igs);
586 my @embed_matches = ($$textref =~ m/<embed[^>]*?src\s*=\s*($attval)[^>]*>/igs);
587 my @tabbg_matches = ($$textref =~ m/<(?:body|table|tr|td)[^>]*?background\s*=\s*($attval)[^>]*>/igs);
588 my @script_matches = ($$textref =~ m/<script[^>]*?src\s*=\s*($attval)[^>]*>/igs);
589
590 if(!defined $self->{'utf8_to_original_filename'}) {
591 # maps from utf8 converted link name -> original filename referrred to by (possibly URL-encoded) src url
592 $self->{'utf8_to_original_filename'} = {};
593 }
594
595 foreach my $link (@img_matches, @usemap_matches, @link_matches, @embed_matches, @tabbg_matches, @script_matches) {
596
597 # remove quotes from link at start and end if necessary
598 if ($link=~/^\"/) {
599 $link=~s/^\"//;
600 $link=~s/\"$//;
601 }
602
603 $link =~ s/\#.*$//s; # remove any anchor names, e.g. foo.html#name becomes foo.html
604 # some links may just be anchor names
605 next unless ($link =~ /\S+/);
606
607 if ($link !~ m@^/@ && $link !~ m/^([A-Z]:?)\\/) {
608 # Turn relative file path into full path
609 my $dirname = &File::Basename::dirname($filename_full_path);
610 $link = &util::filename_cat($dirname, $link);
611 }
612 $link = $self->eval_dir_dots($link);
613
614 # this is the actual filename on the filesystem (that the link refers to)
615 my $url_original_filename = $self->opt_url_decode($link);
616
617 # Convert the url_original_filename into its utf8 version. Store the utf8 link along with the url_original_filename
618 my $utf8_link = "";
619 $self->decode_text($link,$encoding,$language,\$utf8_link);
620
621 $self->{'utf8_to_original_filename'}->{$utf8_link} = $url_original_filename;
622# print STDERR "**** utf8_encoded_link to original src filename:\n\t$utf8_link\n\t".$self->{'utf8_to_original_filename'}->{$utf8_link}."\n";
623
624 if ($url_original_filename ne $utf8_link) {
625 my $outhandle = $self->{'outhandle'};
626
627 print $outhandle "URL Encoding $url_original_filename\n";
628 print $outhandle " ->$utf8_link\n";
629 }
630
631 $block_hash->{'file_blocks'}->{$url_original_filename} = 1;
632 }
633}
634
635# Given a filename in any encoding, will URL decode it to get back the original filename
636# in the original encoding. Because this method is intended to work out the *original*
637# filename*, it not URL decode any filename if a file by the name of the *URL-encoded*
638# string already exists in the local folder.
639# Return the original filename corresponding to the parameter URL-encoded filename, and
640# a decoded flag that is set to true iff URL-decoding had to be applied.
641sub opt_url_decode {
642 my $self = shift (@_);
643 my ($link) = @_;
644
645 # Replace %XX's in URL with decoded value if required.
646 # Note that the filename may include the %XX in some situations
647 if ($link =~ m/\%[A-F0-9]{2}/i) {
648 if (!-e $link) {
649 $link = &unicode::url_decode($link);
650 }
651 }
652
653 return $link;
654}
655
656
657# do plugin specific processing of doc_obj
658sub process {
659 my $self = shift (@_);
660 my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
661 my $outhandle = $self->{'outhandle'};
662
663 if ($ENV{'GSDLOS'} =~ m/^windows/i) {
664 # this makes life so much easier... perl can cope with unix-style '/'s.
665 $base_dir =~ s@(\\)+@/@g;
666 $file =~ s@(\\)+@/@g;
667 }
668
669 # reset per-doc stuff...
670 $self->{'aux_files'} = {};
671 $self->{'dir_num'} = 0;
672 $self->{'file_num'} = 0;
673
674 # process an HTML file where sections are divided by headings tags (H1, H2 ...)
675 # you can also include metadata in the format (X can be any number)
676 # <hX>Title<!--gsdl-metadata
677 # <Metadata name="name1">value1</Metadata>
678 # ...
679 # <Metadata name="nameN">valueN</Metadata>
680 #--></hX>
681 if ($self->{'sectionalise_using_h_tags'}) {
682 # description_tags should allways be activated because we convert headings to description tags
683 $self->{'description_tags'} = 1;
684
685 my $arrSections = [];
686 $$textref =~ s/<h([0-9]+)[^>]*>(.*?)<\/h[0-9]+>/$self->process_heading($1, $2, $arrSections, $file)/isge;
687
688 if (scalar(@$arrSections)) {
689 my $strMetadata = $self->update_section_data($arrSections, -1);
690 if (length($strMetadata)) {
691 $strMetadata = '<!--' . $strMetadata . "\n-->\n</body>";
692 $$textref =~ s/<\/body>/$strMetadata/ig;
693 }
694 }
695 }
696
697 my $cursection = $doc_obj->get_top_section();
698
699 $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection)
700 unless $self->{'no_metadata'} || $self->{'description_tags'};
701
702 # Store URL for page as metadata - this can be used for an
703 # altavista style search interface. The URL won't be valid
704 # unless the file structure contains the domain name (i.e.
705 # like when w3mir is used to download a website).
706
707 # URL metadata (even invalid ones) are used to support internal
708 # links, so even if 'file_is_url' is off, still need to store info
709
710 my ($tailname,$dirname,$suffix) = &File::Basename::fileparse($file, "\\.[^\\.]+\$");
711 my $utf8_file = $self->filename_to_utf8_metadata($file);
712 my $web_url = "http://";
713 if(defined $dirname) { # local directory
714 $dirname = $self->eval_dir_dots($dirname);
715 $dirname .= "/" if $dirname ne ""; # if there's a directory, it should end on "/"
716 $web_url = $web_url.$dirname.$utf8_file;
717 } else {
718 $web_url = $web_url.$utf8_file;
719 }
720
721 $doc_obj->add_utf8_metadata($cursection, "URL", $web_url);
722
723 if ($self->{'file_is_url'}) {
724 $doc_obj->add_metadata($cursection, "weblink", "<a href=\"$web_url\">");
725 $doc_obj->add_metadata($cursection, "webicon", "_iconworld_");
726 $doc_obj->add_metadata($cursection, "/weblink", "</a>");
727 }
728
729 if ($self->{'description_tags'}) {
730 # remove the html header - note that doing this here means any
731 # sections defined within the header will be lost (so all <Section>
732 # tags must appear within the body of the HTML)
733 my ($head_keep) = ($$textref =~ m/^(.*?)<body[^>]*>/is);
734
735 $$textref =~ s/^.*?<body[^>]*>//is;
736 $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
737
738 my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
739 my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
740
741 my $lt = '(?:<|&lt;)';
742 my $gt = '(?:>|&gt;)';
743 my $quot = '(?:"|&quot;|&rdquo;|&ldquo;)';
744
745 my $dont_strip = '';
746 if ($self->{'no_strip_metadata_html'}) {
747 ($dont_strip = $self->{'no_strip_metadata_html'}) =~ s{,}{|}g;
748 }
749
750 my $found_something = 0; my $top = 1;
751 while ($$textref =~ s/^(.*?)$opencom(.*?)$closecom//s) {
752 my $text = $1;
753 my $comment = $2;
754 if (defined $text) {
755 # text before a comment - note that getting to here
756 # doesn't necessarily mean there are Section tags in
757 # the document
758 $self->process_section(\$text, $base_dir, $file, $doc_obj, $cursection);
759 }
760 while ($comment =~ s/$lt(.*?)$gt//s) {
761 my $tag = $1;
762 if ($tag eq "Section") {
763 $found_something = 1;
764 $cursection = $doc_obj->insert_section($doc_obj->get_end_child($cursection)) unless $top;
765 $top = 0;
766 } elsif ($tag eq "/Section") {
767 $found_something = 1;
768 $cursection = $doc_obj->get_parent_section ($cursection);
769 } elsif ($tag =~ m/^Metadata name=$quot(.*?)$quot/s) {
770 my $metaname = $1;
771 my $accumulate = $tag =~ m/mode=${quot}accumulate${quot}/ ? 1 : 0;
772 $comment =~ s/^(.*?)$lt\/Metadata$gt//s;
773 my $metavalue = $1;
774 $metavalue =~ s/^\s+//;
775 $metavalue =~ s/\s+$//;
776 # assume that no metadata value intentionally includes
777 # carriage returns or HTML tags (if they're there they
778 # were probably introduced when converting to HTML from
779 # some other format).
780 # actually some people want to have html tags in their
781 # metadata.
782 $metavalue =~ s/[\cJ\cM]/ /sg;
783 $metavalue =~ s/<[^>]+>//sg
784 unless $dont_strip && ($dont_strip eq 'all' || $metaname =~ m/^($dont_strip)$/);
785 $metavalue =~ s/\s+/ /sg;
786 if ($accumulate) {
787 $doc_obj->add_utf8_metadata($cursection, $metaname, $metavalue);
788 } else {
789 $doc_obj->set_utf8_metadata_element($cursection, $metaname, $metavalue);
790 }
791 } elsif ($tag eq "Description" || $tag eq "/Description") {
792 # do nothing with containing Description tags
793 } else {
794 # simple HTML tag (probably created by the conversion
795 # to HTML from some other format) - we'll ignore it and
796 # hope for the best ;-)
797 }
798 }
799 }
800 if ($cursection ne "") {
801 print $outhandle "HTMLPlugin: WARNING: $file contains unmatched <Section></Section> tags\n";
802 }
803
804 $$textref =~ s/^.*?<body[^>]*>//is;
805 $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
806 if ($$textref =~ m/\S/) {
807 if (!$found_something) {
808 if ($self->{'verbosity'} > 2) {
809 print $outhandle "HTMLPlugin: WARNING: $file appears to contain no Section tags so\n";
810 print $outhandle " will be processed as a single section document\n";
811 }
812
813 # go ahead and process single-section document
814 $self->process_section($textref, $base_dir, $file, $doc_obj, $cursection);
815
816 # if document contains no Section tags we'll go ahead
817 # and extract metadata (this won't have been done
818 # above as the -description_tags option prevents it)
819 my $complete_text = $head_keep.$doc_obj->get_text($cursection);
820 $self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
821 unless $self->{'no_metadata'};
822
823 } else {
824 print $outhandle "HTMLPlugin: WARNING: $file contains the following text outside\n";
825 print $outhandle " of the final closing </Section> tag. This text will\n";
826 print $outhandle " be ignored.";
827
828 my ($text);
829 if (length($$textref) > 30) {
830 $text = substr($$textref, 0, 30) . "...";
831 } else {
832 $text = $$textref;
833 }
834 $text =~ s/\n/ /isg;
835 print $outhandle " ($text)\n";
836 }
837 } elsif (!$found_something) {
838
839 if ($self->{'verbosity'} > 2) {
840 # may get to here if document contained no valid Section
841 # tags but did contain some comments. The text will have
842 # been processed already but we should print the warning
843 # as above and extract metadata
844 print $outhandle "HTMLPlugin: WARNING: $file appears to contain no Section tags and\n";
845 print $outhandle " is blank or empty. Metadata will be assigned if present.\n";
846 }
847
848 my $complete_text = $head_keep.$doc_obj->get_text($cursection);
849 $self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
850 unless $self->{'no_metadata'};
851 }
852
853 } else {
854
855 # remove header and footer
856 if (!$self->{'keep_head'} || $self->{'description_tags'}) {
857 $$textref =~ s/^.*?<body[^>]*>//is;
858 $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
859 }
860
861 # single section document
862 $self->process_section($textref, $base_dir, $file, $doc_obj, $cursection);
863 }
864 return 1;
865}
866
867
868sub process_heading
869{
870 my ($self, $nHeadNo, $strHeadingText, $arrSections, $file) = @_;
871 $strHeadingText = '' if (!defined($strHeadingText));
872
873 my $strMetadata = $self->update_section_data($arrSections, int($nHeadNo));
874
875 my $strSecMetadata = '';
876 while ($strHeadingText =~ s/<!--gsdl-metadata(.*?)-->//is)
877 {
878 $strSecMetadata .= $1;
879 }
880
881 $strHeadingText =~ s/^\s+//g;
882 $strHeadingText =~ s/\s+$//g;
883 $strSecMetadata =~ s/^\s+//g;
884 $strSecMetadata =~ s/\s+$//g;
885
886 $strMetadata .= "\n<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">" . $strHeadingText . "</Metadata>\n";
887
888 if (length($strSecMetadata)) {
889 $strMetadata .= "\t\t" . $strSecMetadata . "\n";
890 }
891
892 $strMetadata .= "\t</Description>\n";
893
894 return "<!--" . $strMetadata . "-->";
895}
896
897
898sub update_section_data
899{
900 my ($self, $arrSections, $nCurTocNo) = @_;
901 my ($strBuffer, $nLast, $nSections) = ('', 0, scalar(@$arrSections));
902
903 if ($nSections == 0) {
904 push @$arrSections, $nCurTocNo;
905 return $strBuffer;
906 }
907 $nLast = $arrSections->[$nSections - 1];
908 if ($nCurTocNo > $nLast) {
909 push @$arrSections, $nCurTocNo;
910 return $strBuffer;
911 }
912 for(my $i = $nSections - 1; $i >= 0; $i--) {
913 if ($nCurTocNo <= $arrSections->[$i]) {
914 $strBuffer .= "\n</Section>";
915 pop @$arrSections;
916 }
917 }
918 push @$arrSections, $nCurTocNo;
919 return $strBuffer;
920}
921
922
923# note that process_section may be called multiple times for a single
924# section (relying on the fact that add_utf8_text appends the text to any
925# that may exist already).
926sub process_section {
927 my $self = shift (@_);
928 my ($textref, $base_dir, $file, $doc_obj, $cursection) = @_;
929 # trap links
930 if (!$self->{'nolinks'}) {
931 # usemap="./#index" not handled correctly => change to "#index"
932## $$textref =~ s/(<img[^>]*?usemap\s*=\s*[\"\']?)([^\"\'>\s]+)([\"\']?[^>]*>)/
933
934 $$textref =~ s/(<img[^>]*?usemap\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
935 $self->replace_usemap_links($1, $2, $3)/isge;
936
937## $$textref =~ s/(<(?:a|area|frame|link|script)\s+[^>]*?\s*(?:href|src)\s*=\s*[\"\']?)([^\"\'>\s]+)([\"\']?[^>]*>)/
938
939 $$textref =~ s/(<(?:a|area|frame|link|script)\s+[^>]*?\s*(?:href|src)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
940 $self->replace_href_links ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
941 }
942
943 # trap images
944
945 # Previously, by default, HTMLPlugin would embed <img> tags inside anchor tags
946 # i.e. <a href="image><img src="image"></a> in order to overcome a problem that
947 # turned regular text succeeding images into links. That is, by embedding <imgs>
948 # inside <a href=""></a>, the text following images were no longer misbehaving.
949 # However, there would be many occasions whereby images were not meant to link
950 # to their source images but where the images would link to another web page.
951 # To allow this, the no_image_links option was introduced: it would prevent
952 # the behaviour of embedding images into links that referenced the source images.
953
954 # Somewhere along the line, the problem of normal text turning into links when
955 # such text followed images which were not embedded in <a href=""></a> ceased
956 # to occur. This is why the following lines have been commented out (as well as
957 # two lines in replace_images). They appear to no longer apply.
958
959 # If at any time, there is a need for having images embedded in <a> anchor tags,
960 # then it might be better to turn that into an HTMLPlugin option rather than make
961 # it the default behaviour. Also, eventually, no_image_links needs to become
962 # a deprecated option for HTMLPlugin as it has now become the default behaviour.
963
964 #if(!$self->{'no_image_links'}){
965 $$textref =~ s/(<(?:img|embed|table|tr|td)[^>]*?(?:src|background)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
966 $self->replace_images ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
967 #}
968
969 # add text to document object
970 # turn \ into \\ so that the rest of greenstone doesn't think there
971 # is an escape code following. (Macro parsing loses them...)
972 $$textref =~ s/\\/\\\\/go;
973
974 $doc_obj->add_utf8_text($cursection, $$textref);
975}
976
977sub replace_images {
978 my $self = shift (@_);
979 my ($front, $link, $back, $base_dir,
980 $file, $doc_obj, $section) = @_;
981
982 # remove quotes from link at start and end if necessary
983 if ($link=~/^[\"\']/) {
984 $link=~s/^[\"\']//;
985 $link=~s/[\"\']$//;
986 $front.='"';
987 $back="\"$back";
988 }
989
990 $link =~ s/\n/ /g;
991
992 # Hack to overcome Windows wv 0.7.1 bug that causes embedded images to be broken
993 # If the Word file path has spaces in it, wv messes up and you end up with
994 # absolute paths for the images, and without the "file://" prefix
995 # So check for this special case and massage the data to be correct
996 if ($ENV{'GSDLOS'} =~ m/^windows/i && $self->{'plugin_type'} eq "WordPlug" && $link =~ m/^[A-Za-z]\:\\/) {
997 $link =~ s/^.*\\([^\\]+)$/$1/;
998 }
999
1000 my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
1001
1002 my $img_file = $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section);
1003
1004# print STDERR "**** link = $link\n**** href = $href\n**** img_file = $img_file, rl = $rl\n";
1005
1006 my $anchor_name = $img_file;
1007 #$anchor_name =~ s/^.*\///;
1008 #$anchor_name = "<a name=\"$anchor_name\" ></a>";
1009
1010 my $image_link = $front . $img_file .$back;
1011 return $image_link;
1012
1013 # The reasons for why the following two lines are no longer necessary can be
1014 # found in subroutine process_section
1015 #my $anchor_link = "<a href=\"$img_file\" >".$image_link."</a>";
1016 #return $anchor_link;
1017
1018 #return $front . $img_file . $back . $anchor_name;
1019}
1020
1021sub replace_href_links {
1022 my $self = shift (@_);
1023 my ($front, $link, $back, $base_dir, $file, $doc_obj, $section) = @_;
1024
1025 # remove quotes from link at start and end if necessary
1026 if ($link=~/^[\"\']/) {
1027 $link=~s/^[\"\']//;
1028 $link=~s/[\"\']$//;
1029 $front.='"';
1030 $back="\"$back";
1031 }
1032
1033 # attempt to sort out targets - frames are not handled
1034 # well in this plugin and some cases will screw things
1035 # up - e.g. the _parent target (so we'll just remove
1036 # them all ;-)
1037 $front =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
1038 $back =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
1039 $front =~ s/target=\"?_parent\"?//is;
1040 $back =~ s/target=\"?_parent\"?//is;
1041
1042 return $front . $link . $back if $link =~ m/^\#/s;
1043 $link =~ s/\n/ /g;
1044
1045 # Find file referred to by $link on file system
1046 # This is more complicated than it sounds when char encodings
1047 # is taken in to account
1048 my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
1049
1050 # href may use '\'s where '/'s should be on Windows
1051 $href =~ s/\\/\//g;
1052 my ($filename) = $href =~ m/^(?:.*?):(?:\/\/)?(.*)/;
1053
1054
1055 ##### leave all these links alone (they won't be picked up by intermediate
1056 ##### pages). I think that's safest when dealing with frames, targets etc.
1057 ##### (at least until I think of a better way to do it). Problems occur with
1058 ##### mailto links from within small frames, the intermediate page is displayed
1059 ##### within that frame and can't be seen. There is still potential for this to
1060 ##### happen even with html pages - the solution seems to be to somehow tell
1061 ##### the browser from the server side to display the page being sent (i.e.
1062 ##### the intermediate page) in the top level window - I'm not sure if that's
1063 ##### possible - the following line should probably be deleted if that can be done
1064 return $front . $link . $back if $href =~ m/^(mailto|news|gopher|nntp|telnet|javascript):/is;
1065
1066
1067 if (($rl == 0) || ($filename =~ m/$self->{'process_exp'}/) ||
1068 ($href =~ m/\/$/) || ($href =~ m/^(mailto|news|gopher|nntp|telnet|javascript):/i)) {
1069 &ghtml::urlsafe ($href);
1070 return $front . "_httpextlink_&rl=" . $rl . "&href=" . $href . $hash_part . $back;
1071 } else {
1072 # link is to some other type of file (eg image) so we'll
1073 # need to associate that file
1074 return $front . $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) . $back;
1075 }
1076}
1077
1078sub add_file {
1079 my $self = shift (@_);
1080 my ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) = @_;
1081 my ($newname);
1082
1083 my $filename = $href;
1084 if ($base_dir eq "") {
1085 # remove http:/ thereby leaving one slash at the start
1086 $filename =~ s/^[^:]*:\///;
1087 }
1088 else {
1089 # remove http://
1090 $filename =~ s/^[^:]*:\/\///;
1091 }
1092
1093 $filename = &util::filename_cat($base_dir, $filename);
1094 # Replace %XX's in URL with decoded value if required. Note that the filename may include the %XX in some
1095 # situations. If the *original* file's name was in URL encoding, the following method will not decode it.
1096 my $utf8_filename = $filename;
1097 $filename = $self->opt_url_decode($utf8_filename);
1098
1099 # some special processing if the intended filename was converted to utf8, but
1100 # the actual file still needs to be renamed
1101 if (!-e $filename) {
1102 # try the original filename stored in map
1103 my $original_filename = $self->{'utf8_to_original_filename'}->{$filename};
1104 if (defined $original_filename && -e $original_filename) {
1105 $filename = $original_filename;
1106 }
1107 }
1108
1109 my ($ext) = $filename =~ m/(\.[^\.]*)$/;
1110
1111 if ($rl == 0) {
1112 if ((!defined $ext) || ($ext !~ m/$self->{'assoc_files'}/)) {
1113 return "_httpextlink_&rl=0&el=prompt&href=" . $href . $hash_part;
1114 }
1115 else {
1116 return "_httpextlink_&rl=0&el=direct&href=" . $href . $hash_part;
1117 }
1118 }
1119
1120 if ((!defined $ext) || ($ext !~ m/$self->{'assoc_files'}/)) {
1121 return "_httpextlink_&rl=" . $rl . "&href=" . $href . $hash_part;
1122 }
1123 if ($self->{'rename_assoc_files'}) {
1124 if (defined $self->{'aux_files'}->{$href}) {
1125 $newname = $self->{'aux_files'}->{$href}->{'dir_num'} . "/" .
1126 $self->{'aux_files'}->{$href}->{'file_num'} . $ext;
1127 } else {
1128 $newname = $self->{'dir_num'} . "/" . $self->{'file_num'} . $ext;
1129 $self->{'aux_files'}->{$href} = {'dir_num' => $self->{'dir_num'}, 'file_num' => $self->{'file_num'}};
1130 $self->inc_filecount ();
1131 }
1132 $doc_obj->associate_file($filename, $newname, undef, $section);
1133 return "_httpdocimg_/$newname";
1134 } else {
1135 if(&unicode::is_url_encoded($utf8_filename)) {
1136 # use the possibly-decoded filename instead to avoid double URL encoding
1137 ($newname) = $filename =~ m/([^\/\\]*)$/;
1138 } else {
1139 ($newname) = $utf8_filename =~ m/([^\/\\]*)$/;
1140 }
1141
1142 # Make sure this name uses only ASCII characters
1143 # => use URL encoding, as it preserves original encoding
1144 $newname = &unicode::url_encode($newname);
1145
1146 $doc_obj->associate_file($filename, $newname, undef, $section);
1147
1148 # Since the generated image will be URL-encoded to avoid file-system/browser mess-ups
1149 # of filenames, URL-encode the additional percent signs of the URL-encoded filename
1150 my $newname_url = $newname;
1151 $newname_url = &unicode::filename_to_url($newname_url); #$newname_url =~ s/%/%25/g if $newname_url !~ m/%25/;
1152 return "_httpdocimg_/$newname_url";
1153 }
1154}
1155
1156
1157sub format_link {
1158 my $self = shift (@_);
1159 my ($link, $base_dir, $file) = @_;
1160
1161 my ($before_hash, $hash_part) = $link =~ m/^([^\#]*)(\#?.*)$/;
1162
1163 $hash_part = "" if !defined $hash_part;
1164 if (!defined $before_hash || $before_hash !~ m/[\w\.\/]/) {
1165 my $outhandle = $self->{'outhandle'};
1166 print $outhandle "HTMLPlugin: ERROR - badly formatted tag ignored ($link)\n"
1167 if $self->{'verbosity'};
1168 return ($link, "", 0);
1169 }
1170
1171 if ($before_hash =~ s@^((?:http|ftp|file)://)@@i) {
1172 my $type = $1;
1173
1174 if ($link =~ m/^(http|ftp):/i) {
1175 # Turn url (using /) into file name (possibly using \ on windows)
1176 my @http_dir_split = split('/', $before_hash);
1177 $before_hash = &util::filename_cat(@http_dir_split);
1178 }
1179
1180 $before_hash = $self->eval_dir_dots($before_hash);
1181
1182 my $linkfilename = &util::filename_cat ($base_dir, $before_hash);
1183
1184 my $rl = 0;
1185 $rl = 1 if (-e $linkfilename);
1186
1187 # make sure there's a slash on the end if it's a directory
1188 if ($before_hash !~ m/\/$/) {
1189 $before_hash .= "/" if (-d $linkfilename);
1190 }
1191 return ($type . $before_hash, $hash_part, $rl);
1192
1193 } elsif ($link !~ m/^(mailto|news|gopher|nntp|telnet|javascript):/i && $link !~ m/^\//) {
1194
1195 if ($before_hash =~ s@^/@@ || $before_hash =~ m/\\/) {
1196
1197 # the first directory will be the domain name if file_is_url
1198 # to generate archives, otherwise we'll assume all files are
1199 # from the same site and base_dir is the root
1200
1201 if ($self->{'file_is_url'}) {
1202 my @dirs = split /[\/\\]/, $file;
1203 my $domname = shift (@dirs);
1204 $before_hash = &util::filename_cat($domname, $before_hash);
1205 $before_hash =~ s@\\@/@g; # for windows
1206 }
1207 else
1208 {
1209 # see if link shares directory with source document
1210 # => turn into relative link if this is so!
1211
1212 if ($ENV{'GSDLOS'} =~ m/^windows/i) {
1213 # too difficult doing a pattern match with embedded '\'s...
1214 my $win_before_hash=$before_hash;
1215 $win_before_hash =~ s@(\\)+@/@g;
1216 # $base_dir is already similarly "converted" on windows.
1217 if ($win_before_hash =~ s@^$base_dir/@@o) {
1218 # if this is true, we removed a prefix
1219 $before_hash=$win_before_hash;
1220 }
1221 }
1222 else {
1223 # before_hash has lost leading slash by this point,
1224 # -> add back in prior to substitution with $base_dir
1225 $before_hash = "/$before_hash";
1226
1227 $before_hash = &util::filename_cat("",$before_hash);
1228 $before_hash =~ s@^$base_dir/@@;
1229 }
1230 }
1231 } else {
1232 # Turn relative file path into full path
1233 my $dirname = &File::Basename::dirname($file);
1234 $before_hash = &util::filename_cat($dirname, $before_hash);
1235 $before_hash = $self->eval_dir_dots($before_hash);
1236 }
1237
1238 my $linkfilename = &util::filename_cat ($base_dir, $before_hash);
1239 # make sure there's a slash on the end if it's a directory
1240 if ($before_hash !~ m/\/$/) {
1241 $before_hash .= "/" if (-d $linkfilename);
1242 }
1243 return ("http://" . $before_hash, $hash_part, 1);
1244 } else {
1245 # mailto, news, nntp, telnet, javascript or gopher link
1246 return ($before_hash, "", 0);
1247 }
1248}
1249
1250sub extract_first_NNNN_characters {
1251 my $self = shift (@_);
1252 my ($textref, $doc_obj, $thissection) = @_;
1253
1254 foreach my $size (split /,/, $self->{'first'}) {
1255 my $tmptext = $$textref;
1256 # skip to the body
1257 $tmptext =~ s/.*<body[^>]*>//i;
1258 # remove javascript
1259 $tmptext =~ s@<script.*?</script>@ @sig;
1260 $tmptext =~ s/<[^>]*>/ /g;
1261 $tmptext =~ s/&nbsp;/ /g;
1262 $tmptext =~ s/^\s+//;
1263 $tmptext =~ s/\s+$//;
1264 $tmptext =~ s/\s+/ /gs;
1265 $tmptext = &unicode::substr ($tmptext, 0, $size);
1266 $tmptext =~ s/\s\S*$/&#8230;/; # adds an ellipse (...)
1267 $doc_obj->add_utf8_metadata ($thissection, "First$size", $tmptext);
1268 }
1269}
1270
1271
1272sub extract_metadata {
1273 my $self = shift (@_);
1274 my ($textref, $metadata, $doc_obj, $section) = @_;
1275 my $outhandle = $self->{'outhandle'};
1276 # if we don't want metadata, we may as well not be here ...
1277 return if (!defined $self->{'metadata_fields'});
1278
1279 # metadata fields to extract/save. 'key' is the (lowercase) name of the
1280 # html meta, 'value' is the metadata name for greenstone to use
1281 my %find_fields = ();
1282
1283 my %creator_fields = (); # short-cut for lookups
1284
1285
1286 foreach my $field (split /,/, $self->{'metadata_fields'}) {
1287 $field =~ s/^\s+//; # remove leading whitespace
1288 $field =~ s/\s+$//; # remove trailing whitespace
1289
1290 # support tag<tagname>
1291 if ($field =~ m/^(.*?)<(.*?)>$/) {
1292 # "$2" is the user's preferred gs metadata name
1293 $find_fields{lc($1)}=$2; # lc = lowercase
1294 } else { # no <tagname> for mapping
1295 # "$field" is the user's preferred gs metadata name
1296 $find_fields{lc($field)}=$field; # lc = lowercase
1297 }
1298 }
1299
1300 if (defined $self->{'hunt_creator_metadata'} &&
1301 $self->{'hunt_creator_metadata'} == 1 ) {
1302 my @extra_fields =
1303 (
1304 'author',
1305 'author.email',
1306 'creator',
1307 'dc.creator',
1308 'dc.creator.corporatename',
1309 );
1310
1311 # add the creator_metadata fields to search for
1312 foreach my $field (@extra_fields) {
1313 $creator_fields{$field}=0; # add to lookup hash
1314 }
1315 }
1316
1317
1318 # find the header in the html file, which has the meta tags
1319 $$textref =~ m@<head>(.*?)</head>@si;
1320
1321 my $html_header=$1;
1322
1323 # go through every <meta... tag defined in the html and see if it is
1324 # one of the tags we want to match.
1325
1326 # special case for title - we want to remember if its been found
1327 my $found_title = 0;
1328 # this assumes that ">" won't appear. (I don't think it's allowed to...)
1329 $html_header =~ m/^/; # match the start of the string, for \G assertion
1330
1331 while ($html_header =~ m/\G.*?<meta(.*?)>/sig) {
1332 my $metatag=$1;
1333 my ($tag, $value);
1334
1335 # find the tag name
1336 $metatag =~ m/(?:name|http-equiv)\s*=\s*([\"\'])?(.*?)\1/is;
1337 $tag=$2;
1338 # in case they're not using " or ', but they should...
1339 if (! $tag) {
1340 $metatag =~ m/(?:name|http-equiv)\s*=\s*([^\s\>]+)/is;
1341 $tag=$1;
1342 }
1343
1344 if (!defined $tag) {
1345 print $outhandle "HTMLPlugin: can't find NAME in \"$metatag\"\n";
1346 next;
1347 }
1348
1349 # don't need to assign this field if it was passed in from a previous
1350 # (recursive) plugin
1351 if (defined $metadata->{$tag}) {next}
1352
1353 # find the tag content
1354 $metatag =~ m/content\s*=\s*([\"\'])?(.*?)\1/is;
1355 $value=$2;
1356
1357 if (! $value) {
1358 $metatag =~ m/(?:name|http-equiv)\s*=\s*([^\s\>]+)/is;
1359 $value=$1;
1360 }
1361 if (!defined $value) {
1362 print $outhandle "HTMLPlugin: can't find VALUE in \"$metatag\"\n";
1363 next;
1364 }
1365
1366 # clean up and add
1367 $value =~ s/\s+/ /gs;
1368 chomp($value); # remove trailing \n, if any
1369 if (exists $creator_fields{lc($tag)}) {
1370 # map this value onto greenstone's "Creator" metadata
1371 $tag='Creator';
1372 } elsif (!exists $find_fields{lc($tag)}) {
1373 next; # don't want this tag
1374 } else {
1375 # get the user's preferred capitalisation
1376 $tag = $find_fields{lc($tag)};
1377 }
1378 if (lc($tag) eq "title") {
1379 $found_title = 1;
1380 }
1381 print $outhandle " extracted \"$tag\" metadata \"$value\"\n"
1382 if ($self->{'verbosity'} > 2);
1383 if ($tag =~ m/date.*/i){
1384 $tag = lc($tag);
1385 }
1386 $doc_obj->add_utf8_metadata($section, $tag, $value);
1387
1388 }
1389
1390 # TITLE: extract the document title
1391 if (exists $find_fields{'title'} && !$found_title) {
1392 # we want a title, and didn't find one in the meta tags
1393 # see if there's a <title> tag
1394 my $title;
1395 my $from = ""; # for debugging output only
1396 if ($html_header =~ m/<title[^>]*>([^<]+)<\/title[^>]*>/is) {
1397 $title = $1;
1398 $from = "<title> tags";
1399 }
1400
1401 if (!defined $title) {
1402 $from = "first 100 chars";
1403 # if no title use first 100 or so characters
1404 $title = $$textref;
1405 $title =~ s/^\xFE\xFF//; # Remove unicode byte order mark
1406 $title =~ s/^.*?<body>//si;
1407 # ignore javascript!
1408 $title =~ s@<script.*?</script>@ @sig;
1409 $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
1410 $title =~ s/<[^>]*>/ /g; # remove all HTML tags
1411 $title = substr ($title, 0, 100);
1412 $title =~ s/\s\S*$/.../;
1413 }
1414 $title =~ s/<[^>]*>/ /g; # remove html tags
1415 $title =~ s/&nbsp;/ /g;
1416 $title =~ s/(?:&nbsp;|\xc2\xa0)/ /g; # utf-8 for nbsp...
1417 $title =~ s/\s+/ /gs; # collapse multiple spaces
1418 $title =~ s/^\s*//; # remove leading spaces
1419 $title =~ s/\s*$//; # remove trailing spaces
1420
1421 $title =~ s/^$self->{'title_sub'}// if ($self->{'title_sub'});
1422 $title =~ s/^\s+//s; # in case title_sub introduced any...
1423 $doc_obj->add_utf8_metadata ($section, 'Title', $title);
1424 print $outhandle " extracted Title metadata \"$title\" from $from\n"
1425 if ($self->{'verbosity'} > 2);
1426 }
1427
1428 # add FileFormat metadata
1429 $doc_obj->add_metadata($section,"FileFormat", "HTML");
1430
1431 # Special, for metadata names such as tagH1 - extracts
1432 # the text between the first <H1> and </H1> tags into "H1" metadata.
1433
1434 foreach my $field (keys %find_fields) {
1435 if ($field !~ m/^tag([a-z0-9]+)$/i) {next}
1436 my $tag = $1;
1437 if ($$textref =~ m@<$tag[^>]*>(.*?)</$tag[^>]*>@g) {
1438 my $content = $1;
1439 $content =~ s/&nbsp;/ /g;
1440 $content =~ s/<[^>]*>/ /g;
1441 $content =~ s/^\s+//;
1442 $content =~ s/\s+$//;
1443 $content =~ s/\s+/ /gs;
1444 if ($content) {
1445 $tag=$find_fields{"tag$tag"}; # get the user's capitalisation
1446 $tag =~ s/^tag//i;
1447 $doc_obj->add_utf8_metadata ($section, $tag, $content);
1448 print $outhandle " extracted \"$tag\" metadata \"$content\"\n"
1449 if ($self->{'verbosity'} > 2);
1450 }
1451 }
1452 }
1453}
1454
1455
1456# evaluate any "../" to next directory up
1457# evaluate any "./" as here
1458sub eval_dir_dots {
1459 my $self = shift (@_);
1460 my ($filename) = @_;
1461 my $dirsep_os = &util::get_os_dirsep();
1462 my @dirsep = split(/$dirsep_os/,$filename);
1463
1464 my @eval_dirs = ();
1465 foreach my $d (@dirsep) {
1466 if ($d eq "..") {
1467 pop(@eval_dirs);
1468
1469 } elsif ($d eq ".") {
1470 # do nothing!
1471
1472 } else {
1473 push(@eval_dirs,$d);
1474 }
1475 }
1476
1477 # Need to fiddle with number of elements in @eval_dirs if the
1478 # first one is the empty string. This is because of a
1479 # modification to util::filename_cat that supresses the addition
1480 # of a leading '/' character (or \ if windows) (intended to help
1481 # filename cat with relative paths) if the first entry in the
1482 # array is the empty string. Making the array start with *two*
1483 # empty strings is a way to defeat this "smart" option.
1484 #
1485 if (scalar(@eval_dirs) > 0) {
1486 if ($eval_dirs[0] eq ""){
1487 unshift(@eval_dirs,"");
1488 }
1489 }
1490
1491 my $evaluated_filename = (scalar @eval_dirs > 0) ? &util::filename_cat(@eval_dirs) : "";
1492 return $evaluated_filename;
1493}
1494
1495sub replace_usemap_links {
1496 my $self = shift (@_);
1497 my ($front, $link, $back) = @_;
1498
1499 # remove quotes from link at start and end if necessary
1500 if ($link=~/^[\"\']/) {
1501 $link=~s/^[\"\']//;
1502 $link=~s/[\"\']$//;
1503 $front.='"';
1504 $back="\"$back";
1505 }
1506
1507 $link =~ s/^\.\///;
1508 return $front . $link . $back;
1509}
1510
1511sub inc_filecount {
1512 my $self = shift (@_);
1513
1514 if ($self->{'file_num'} == 1000) {
1515 $self->{'dir_num'} ++;
1516 $self->{'file_num'} = 0;
1517 } else {
1518 $self->{'file_num'} ++;
1519 }
1520}
1521
1522
1523# Extend read_file so that strings like &eacute; are
1524# converted to UTF8 internally.
1525#
1526# We don't convert &lt; or &gt; or &amp; or &quot; in case
1527# they interfere with the GML files
1528
1529sub read_file {
1530 my $self = shift(@_);
1531 my ($filename, $encoding, $language, $textref) = @_;
1532
1533 $self->SUPER::read_file($filename, $encoding, $language, $textref);
1534
1535 # Convert entities to their UTF8 equivalents
1536 $$textref =~ s/&(lt|gt|amp|quot|nbsp);/&z$1;/go;
1537 $$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1)/gseo;
1538 $$textref =~ s/&z(lt|gt|amp|quot|nbsp);/&$1;/go;
1539}
1540
15411;
Note: See TracBrowser for help on using the repository browser.