Context Navigation

source: main/trunk/greenstone2/perllib/plugins/HTMLPlugin.pm@ 23371

Last change on this file since 23371 was 23371, checked in by davidb, 13 years ago
Further refinement of code to support HTML linking between documents when using non-ascii names on Windows
Property svn:executable set to ``* Property svn:keywords set to `Author Date Id Revision`
File size: 59.2 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
36	package HTMLPlugin;
37
38	use Encode;
39
40	use ReadTextFile;
41	use HBPlugin;
42	use ghtml;
43	use unicode;
44	use util;
45	use XMLParser;
46
47	use Image::Size;
48	use File::Copy;
49
50	sub BEGIN {
51	@HTMLPlugin::ISA = ('ReadTextFile', 'HBPlugin');
52	}
53
54	use strict; # every perl program should have this!
55	no strict 'refs'; # make an exception so we can use variables as filehandles
56
57	my $arguments =
58	[ { 'name' => "process_exp",
59	'desc' => "{BasePlugin.process_exp}",
60	'type' => "regexp",
61	'deft' => &get_default_process_exp() },
62	{ 'name' => "block_exp",
63	'desc' => "{BasePlugin.block_exp}",
64	'type' => 'regexp',
65	'deft' => &get_default_block_exp() },
66	{ 'name' => "nolinks",
67	'desc' => "{HTMLPlugin.nolinks}",
68	'type' => "flag" },
69	{ 'name' => "keep_head",
70	'desc' => "{HTMLPlugin.keep_head}",
71	'type' => "flag" },
72	{ 'name' => "no_metadata",
73	'desc' => "{HTMLPlugin.no_metadata}",
74	'type' => "flag" },
75	{ 'name' => "metadata_fields",
76	'desc' => "{HTMLPlugin.metadata_fields}",
77	'type' => "string",
78	'deft' => "Title" },
79	{ 'name' => "metadata_field_separator",
80	'desc' => "{HTMLPlugin.metadata_field_separator}",
81	'type' => "string",
82	'deft' => "" },
83	{ 'name' => "hunt_creator_metadata",
84	'desc' => "{HTMLPlugin.hunt_creator_metadata}",
85	'type' => "flag" },
86	{ 'name' => "file_is_url",
87	'desc' => "{HTMLPlugin.file_is_url}",
88	'type' => "flag" },
89	{ 'name' => "assoc_files",
90	'desc' => "{HTMLPlugin.assoc_files}",
91	'type' => "regexp",
92	'deft' => &get_default_block_exp() },
93	{ 'name' => "rename_assoc_files",
94	'desc' => "{HTMLPlugin.rename_assoc_files}",
95	'type' => "flag" },
96	{ 'name' => "title_sub",
97	'desc' => "{HTMLPlugin.title_sub}",
98	'type' => "string",
99	'deft' => "" },
100	{ 'name' => "description_tags",
101	'desc' => "{HTMLPlugin.description_tags}",
102	'type' => "flag" },
103	# retain this for backward compatibility (w3mir option was replaced by
104	# file_is_url)
105	{ 'name' => "w3mir",
106	# 'desc' => "{HTMLPlugin.w3mir}",
107	'type' => "flag",
108	'hiddengli' => "yes"},
109	{ 'name' => "no_strip_metadata_html",
110	'desc' => "{HTMLPlugin.no_strip_metadata_html}",
111	'type' => "string",
112	'deft' => "",
113	'reqd' => "no"},
114	{ 'name' => "sectionalise_using_h_tags",
115	'desc' => "{HTMLPlugin.sectionalise_using_h_tags}",
116	'type' => "flag" },
117	{ 'name' => "use_realistic_book",
118	'desc' => "{HTMLPlugin.tidy_html}",
119	'type' => "flag"},
120	{ 'name' => "old_style_HDL",
121	'desc' => "{HTMLPlugin.old_style_HDL}",
122	'type' => "flag"},
123	{'name' => "processing_tmp_files",
124	'desc' => "{BasePlugin.processing_tmp_files}",
125	'type' => "flag",
126	'hiddengli' => "yes"}
127	];
128
129	my $options = { 'name' => "HTMLPlugin",
130	'desc' => "{HTMLPlugin.desc}",
131	'abstract' => "no",
132	'inherits' => "yes",
133	'args' => $arguments };
134
135
136	sub new {
137	my ($class) = shift (@_);
138	my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
139	push(@$pluginlist, $class);
140
141	push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
142	push(@{$hashArgOptLists->{"OptList"}},$options);
143
144
145	my $self = new ReadTextFile($pluginlist,$inputargs,$hashArgOptLists);
146
147	if ($self->{'w3mir'}) {
148	$self->{'file_is_url'} = 1;
149	}
150	$self->{'aux_files'} = {};
151	$self->{'dir_num'} = 0;
152	$self->{'file_num'} = 0;
153
154	return bless $self, $class;
155	}
156
157	# may want to use (?i)\.(gif\|jpe?g\|jpe\|png\|css\|js(?:@.*)?)$
158	# if have eg <script language="javascript" src="img/lib.js@123">
159	# blocking is now done by reading through the file and recording all the
160	# images and other files
161	sub get_default_block_exp {
162	my $self = shift (@_);
163
164	#return q^(?i)\.(gif\|jpe?g\|jpe\|jpg\|png\|css)$^;
165	return "";
166	}
167
168	sub get_default_process_exp {
169	my $self = shift (@_);
170
171	# the last option is an attempt to encode the concept of an html query ...
172	return q^(?i)(\.html?\|\.shtml\|\.shm\|\.asp\|\.php\d?\|\.cgi\|.+\?.+=.*)$^;
173	}
174
175	sub store_block_files
176	{
177	my $self =shift (@_);
178	my ($filename_full_path, $block_hash) = @_;
179
180	my $html_fname = $filename_full_path;
181	my @file_blocks;
182
183	my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
184	$self->{'store_content_encoding'}->{$filename_full_path} = $content_encoding;
185
186	# read in file ($text will be in utf8)
187	my $raw_text = "";
188	$self->read_file_no_decoding($filename_full_path, \$raw_text);
189
190	my $textref = \$raw_text;
191	my $opencom = '(?:<!--\|<!(?:—\|\|--))';
192	my $closecom = '(?:-->\|(?:—\|\|--)>)';
193	$$textref =~ s/$opencom(.*?)$closecom//gs;
194
195	# Convert entities to their UTF8 equivalents
196	$$textref =~ s/&(lt\|gt\|amp\|quot\|nbsp);/&z$1;/go;
197	$$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1,0)/gseo; # on this occassion, want it left as utf8
198	$$textref =~ s/&z(lt\|gt\|amp\|quot\|nbsp);/&$1;/go;
199
200	my $attval = "\\\"[^\\\"]+\\\"\|[^\\s>]+";
201	my @img_matches = ($$textref =~ m/<img[^>]?src\s=\s($attval)[^>]>/igs);
202	my @usemap_matches = ($$textref =~ m/<img[^>]?usemap\s=\s($attval)[^>]>/igs);
203	my @link_matches = ($$textref =~ m/<link[^>]?href\s=\s($attval)[^>]>/igs);
204	my @embed_matches = ($$textref =~ m/<embed[^>]?src\s=\s($attval)[^>]>/igs);
205	my @tabbg_matches = ($$textref =~ m/<(?:body\|table\|tr\|td)[^>]?background\s=\s($attval)[^>]>/igs);
206	my @script_matches = ($$textref =~ m/<script[^>]?src\s=\s($attval)[^>]>/igs);
207
208	if(!defined $self->{'utf8_to_original_filename'}) {
209	# maps from utf8 converted link name -> original filename referrred to by (possibly URL-encoded) src url
210	$self->{'utf8_to_original_filename'} = {};
211	}
212
213	foreach my $link (@img_matches, @usemap_matches, @link_matches, @embed_matches, @tabbg_matches, @script_matches) {
214
215	# remove quotes from link at start and end if necessary
216	if ($link =~ m/^\"/) {
217	$link =~ s/^\"//;
218	$link =~ s/\"$//;
219	}
220
221	# remove any anchor names, e.g. foo.html#name becomes foo.html
222	# but watch out for any #'s that are part of entities, such as α
223	$link =~ s/([^&])\#.*$/$1/s;
224
225	# some links may just be anchor names
226	next unless ($link =~ /\S+/);
227
228	if ($link !~ m@^/@ && $link !~ m/^([A-Z]:?)\\/) {
229	# Turn relative file path into full path
230	my $dirname = &File::Basename::dirname($filename_full_path);
231	$link = &util::filename_cat($dirname, $link);
232	}
233	$link = $self->eval_dir_dots($link);
234
235	# this is the actual filename on the filesystem (that the link refers to)
236	my $url_original_filename = $self->opt_url_decode($link);
237
238	# Convert the url_original_filename into its utf8 version. Store the utf8 link along with the url_original_filename
239	my $utf8_link = "";
240	$self->decode_text($link,$content_encoding,$language,\$utf8_link);
241
242	$self->{'utf8_to_original_filename'}->{$utf8_link} = $url_original_filename;
243	# print STDERR "**** utf8_encoded_link to original src filename:\n\t$utf8_link\n\t".$self->{'utf8_to_original_filename'}->{$utf8_link}."\n";
244
245	if ($url_original_filename ne $utf8_link) {
246	my $outhandle = $self->{'outhandle'};
247
248	print $outhandle "URL Encoding $url_original_filename\n";
249	print $outhandle " ->$utf8_link\n";
250	}
251
252
253	my $unicode_url_original_filename = decode("utf8",$url_original_filename);
254
255	## print STDERR "*****!!! Blocking url original filename = $unicode_url_original_filename\n";
256
257	# Allow for possibility of raw byte version (UTF8) and Unicode versions of file
258	$block_hash->{'file_blocks'}->{$url_original_filename} = 1;
259	$block_hash->{'file_blocks'}->{$unicode_url_original_filename} = 1;
260	}
261	}
262
263	# Given a filename in any encoding, will URL decode it to get back the original filename
264	# in the original encoding. Because this method is intended to work out the original
265	# filename, it does not URL decode any filename if a file by the name of the URL-encoded*
266	# string already exists in the local folder.
267	#
268	# Is the following still true??
269	# Return the original filename corresponding to the parameter URL-encoded filename, and
270	# a decoded flag that is set to true iff URL-decoding had to be applied.
271	sub opt_url_decode {
272	my $self = shift (@_);
273	my ($link) = @_;
274
275	# Replace %XX's in URL with decoded value if required.
276	# Note that the filename may include the %XX in some situations
277	if ($link =~ m/\%[A-F0-9]{2}/i) {
278	if (!-e $link) {
279	$link = &unicode::url_decode($link);
280	}
281	}
282
283	return $link;
284	}
285
286	sub read_into_doc_obj
287	{
288	my $self = shift (@_);
289	my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
290
291	my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
292
293	# Lookup content_encoding worked out in file_block pass for this file
294	# Store it under the local name 'content_encoding' so its nice and
295	# easy to access
296	$self->{'content_encoding'} = $self->{'store_content_encoding'}->{$filename_full_path};
297
298	# get the input file
299	my $input_filename = $file;
300	my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
301	$suffix = lc($suffix);
302	my $tidy_filename;
303	if (($self->{'use_realistic_book'}) \|\| ($self->{'old_style_HDL'}))
304	{
305	# because the document has to be sectionalized set the description tags
306	$self->{'description_tags'} = 1;
307
308	# set the file to be tidied
309	$input_filename = &util::filename_cat($base_dir,$file) if $base_dir =~ m/\w/;
310
311	# get the tidied file
312	#my $tidy_filename = $self->tmp_tidy_file($input_filename);
313	$tidy_filename = $self->convert_tidy_or_oldHDL_file($input_filename);
314
315	# derive tmp filename from input filename
316	my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($tidy_filename, "\\.[^\\.]+\$");
317
318	# set the new input file and base_dir to be from the tidied file
319	$file = "$tailname$suffix";
320	$base_dir = $dirname;
321	}
322
323	# call the parent read_into_doc_obj
324	my ($process_status,$doc_obj) = $self->SUPER::read_into_doc_obj($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli);
325	if (($self->{'use_realistic_book'}) \|\| ($self->{'old_style_HDL'}))
326	{
327	# now we need to reset the filenames in the doc obj so that the converted filenames are not used
328	my $collect_file = &util::filename_within_collection($filename_full_path);
329	$doc_obj->set_source_filename ($collect_file, $self->{'file_rename_method'});
330	## set_source_filename does not set the doc_obj source_path which is used in archives dbs for incremental
331	# build. So set it manually.
332	$doc_obj->set_source_path($filename_full_path);
333	my $collect_conv_file = &util::filename_within_collection($tidy_filename);
334	$doc_obj->set_converted_filename($collect_conv_file);
335
336	my $plugin_filename_encoding = $self->{'filename_encoding'};
337	my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
338	$self->set_Source_metadata($doc_obj, $filename_full_path, $filename_encoding);
339	}
340
341	delete $self->{'store_content_encoding'}->{$filename_full_path};
342	$self->{'content_encoding'} = undef;
343
344	return ($process_status,$doc_obj);
345	}
346
347	# do plugin specific processing of doc_obj
348	sub process {
349	my $self = shift (@_);
350	my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
351	my $outhandle = $self->{'outhandle'};
352
353	if ($ENV{'GSDLOS'} =~ m/^windows/i) {
354	# this makes life so much easier... perl can cope with unix-style '/'s.
355	$base_dir =~ s@(\\)+@/@g;
356	$file =~ s@(\\)+@/@g;
357	}
358
359	my $filename = &util::filename_cat($base_dir,$file);
360	my $upgraded_base_dir = &util::upgrade_if_dos_filename($base_dir);
361	my $upgraded_filename = &util::upgrade_if_dos_filename($filename);
362
363	if ($ENV{'GSDLOS'} =~ m/^windows/i) {
364	# And again
365	$upgraded_base_dir =~ s@(\\)+@/@g;
366	$upgraded_filename =~ s@(\\)+@/@g;
367
368	# Need to make sure there is a '/' on the end of upgraded_base_dir
369	if ($upgraded_base_dir !~ m/\/$/) {
370	$upgraded_base_dir .= "/";
371	}
372	}
373	my $upgraded_file = &util::filename_within_directory($upgraded_filename,$upgraded_base_dir);
374
375	# reset per-doc stuff...
376	$self->{'aux_files'} = {};
377	$self->{'dir_num'} = 0;
378	$self->{'file_num'} = 0;
379
380	# process an HTML file where sections are divided by headings tags (H1, H2 ...)
381	# you can also include metadata in the format (X can be any number)
382	# <hX>Title<!--gsdl-metadata
383	# <Metadata name="name1">value1</Metadata>
384	# ...
385	# <Metadata name="nameN">valueN</Metadata>
386	#--></hX>
387	if ($self->{'sectionalise_using_h_tags'}) {
388	# description_tags should allways be activated because we convert headings to description tags
389	$self->{'description_tags'} = 1;
390
391	my $arrSections = [];
392	$$textref =~ s/<h([0-9]+)[^>]>(.?)<\/h[0-9]+>/$self->process_heading($1, $2, $arrSections, $upgraded_file)/isge;
393
394	if (scalar(@$arrSections)) {
395	my $strMetadata = $self->update_section_data($arrSections, -1);
396	if (length($strMetadata)) {
397	$strMetadata = '<!--' . $strMetadata . "\n-->\n</body>";
398	$$textref =~ s/<\/body>/$strMetadata/ig;
399	}
400	}
401	}
402
403	my $cursection = $doc_obj->get_top_section();
404
405	$self->extract_metadata ($textref, $metadata, $doc_obj, $cursection)
406	unless $self->{'no_metadata'} \|\| $self->{'description_tags'};
407
408	# Store URL for page as metadata - this can be used for an
409	# altavista style search interface. The URL won't be valid
410	# unless the file structure contains the domain name (i.e.
411	# like when w3mir is used to download a website).
412
413	# URL metadata (even invalid ones) are used to support internal
414	# links, so even if 'file_is_url' is off, still need to store info
415
416	my ($tailname,$dirname) = &File::Basename::fileparse($upgraded_file);
417
418	# my $utf8_file = $self->filename_to_utf8_metadata($file);
419	# $utf8_file =~ s/&\#095;/_/g;
420	my $utf8_file = &unicode::raw_filename_to_url_encoded($tailname);
421
422
423	my $web_url = "http://";
424	if(defined $dirname) { # local directory
425	# Check for "ftp" in the domain name of the directory
426	# structure to determine if this URL should be a ftp:// URL
427	# This check is not infallible, but better than omitting the
428	# check, which would cause all files downloaded from ftp sites
429	# via mirroring with wget to have potentially erroneous http:// URLs
430	# assigned in their metadata
431	if ($dirname =~ /^[^\/]*ftp/i)
432	{
433	$web_url = "ftp://";
434	}
435	$dirname = $self->eval_dir_dots($dirname);
436	$dirname .= &util::get_dirsep() if $dirname ne ""; # if there's a directory, it should end on "/"
437	$web_url = $web_url.$dirname.$utf8_file;
438	} else {
439	$web_url = $web_url.$utf8_file;
440	}
441	$web_url =~ s/\\/\//g;
442
443	if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
444	print STDERR "******* upgraded_file: $upgraded_file\n";
445	print STDERR "******* adding URL metadata: $utf8_file\n";
446	}
447
448
449	$doc_obj->add_utf8_metadata($cursection, "URL", $web_url);
450
451	if ($self->{'file_is_url'}) {
452	$doc_obj->add_metadata($cursection, "weblink", "<a href=\"$web_url\">");
453	$doc_obj->add_metadata($cursection, "webicon", "_iconworld_");
454	$doc_obj->add_metadata($cursection, "/weblink", "</a>");
455	}
456
457	if ($self->{'description_tags'}) {
458	# remove the html header - note that doing this here means any
459	# sections defined within the header will be lost (so all <Section>
460	# tags must appear within the body of the HTML)
461	my ($head_keep) = ($$textref =~ m/^(.?)<body[^>]>/is);
462
463	$$textref =~ s/^.?<body[^>]>//is;
464	$$textref =~ s/(<\/body[^>]>\|<\/html[^>]>)//isg;
465
466	my $opencom = '(?:<!--\|<!(?:—\|\|--))';
467	my $closecom = '(?:-->\|(?:—\|\|--)>)';
468
469	my $lt = '(?:<\|<)';
470	my $gt = '(?:>\|>)';
471	my $quot = '(?:"\|"\|”\|“)';
472
473	my $dont_strip = '';
474	if ($self->{'no_strip_metadata_html'}) {
475	($dont_strip = $self->{'no_strip_metadata_html'}) =~ s{,}{\|}g;
476	}
477
478	my $found_something = 0; my $top = 1;
479	while ($$textref =~ s/^(.?)$opencom(.?)$closecom//s) {
480	my $text = $1;
481	my $comment = $2;
482	if (defined $text) {
483	# text before a comment - note that getting to here
484	# doesn't necessarily mean there are Section tags in
485	# the document
486	$self->process_section(\$text, $upgraded_base_dir, $upgraded_file, $doc_obj, $cursection);
487	}
488	while ($comment =~ s/$lt(.*?)$gt//s) {
489	my $tag = $1;
490	if ($tag eq "Section") {
491	$found_something = 1;
492	$cursection = $doc_obj->insert_section($doc_obj->get_end_child($cursection)) unless $top;
493	$top = 0;
494	} elsif ($tag eq "/Section") {
495	$found_something = 1;
496	$cursection = $doc_obj->get_parent_section ($cursection);
497	} elsif ($tag =~ m/^Metadata name=$quot(.*?)$quot/s) {
498	my $metaname = $1;
499	my $accumulate = $tag =~ m/mode=${quot}accumulate${quot}/ ? 1 : 0;
500	$comment =~ s/^(.*?)$lt\/Metadata$gt//s;
501	my $metavalue = $1;
502	$metavalue =~ s/^\s+//;
503	$metavalue =~ s/\s+$//;
504	# assume that no metadata value intentionally includes
505	# carriage returns or HTML tags (if they're there they
506	# were probably introduced when converting to HTML from
507	# some other format).
508	# actually some people want to have html tags in their
509	# metadata.
510	$metavalue =~ s/[\cJ\cM]/ /sg;
511	$metavalue =~ s/<[^>]+>//sg
512	unless $dont_strip && ($dont_strip eq 'all' \|\| $metaname =~ m/^($dont_strip)$/);
513	$metavalue =~ s/\s+/ /sg;
514	if ($metaname =~ /\./) { # has a namespace
515	$metaname = "ex.$metaname";
516	}
517	if ($accumulate) {
518	$doc_obj->add_utf8_metadata($cursection, $metaname, $metavalue);
519	} else {
520	$doc_obj->set_utf8_metadata_element($cursection, $metaname, $metavalue);
521	}
522	} elsif ($tag eq "Description" \|\| $tag eq "/Description") {
523	# do nothing with containing Description tags
524	} else {
525	# simple HTML tag (probably created by the conversion
526	# to HTML from some other format) - we'll ignore it and
527	# hope for the best ;-)
528	}
529	}
530	}
531	if ($cursection ne "") {
532	print $outhandle "HTMLPlugin: WARNING: $upgraded_file contains unmatched <Section></Section> tags\n";
533	}
534
535	$$textref =~ s/^.?<body[^>]>//is;
536	$$textref =~ s/(<\/body[^>]>\|<\/html[^>]>)//isg;
537	if ($$textref =~ m/\S/) {
538	if (!$found_something) {
539	if ($self->{'verbosity'} > 2) {
540	print $outhandle "HTMLPlugin: WARNING: $upgraded_file appears to contain no Section tags so\n";
541	print $outhandle " will be processed as a single section document\n";
542	}
543
544	# go ahead and process single-section document
545	$self->process_section($textref, $upgraded_base_dir, $upgraded_file, $doc_obj, $cursection);
546
547	# if document contains no Section tags we'll go ahead
548	# and extract metadata (this won't have been done
549	# above as the -description_tags option prevents it)
550	my $complete_text = $head_keep.$doc_obj->get_text($cursection);
551	$self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
552	unless $self->{'no_metadata'};
553
554	} else {
555	print $outhandle "HTMLPlugin: WARNING: $upgraded_file contains the following text outside\n";
556	print $outhandle " of the final closing </Section> tag. This text will\n";
557	print $outhandle " be ignored.";
558
559	my ($text);
560	if (length($$textref) > 30) {
561	$text = substr($$textref, 0, 30) . "...";
562	} else {
563	$text = $$textref;
564	}
565	$text =~ s/\n/ /isg;
566	print $outhandle " ($text)\n";
567	}
568	} elsif (!$found_something) {
569
570	if ($self->{'verbosity'} > 2) {
571	# may get to here if document contained no valid Section
572	# tags but did contain some comments. The text will have
573	# been processed already but we should print the warning
574	# as above and extract metadata
575	print $outhandle "HTMLPlugin: WARNING: $upgraded_file appears to contain no Section tags and\n";
576	print $outhandle " is blank or empty. Metadata will be assigned if present.\n";
577	}
578
579	my $complete_text = $head_keep.$doc_obj->get_text($cursection);
580	$self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
581	unless $self->{'no_metadata'};
582	}
583
584	} else {
585
586	# remove header and footer
587	if (!$self->{'keep_head'} \|\| $self->{'description_tags'}) {
588	$$textref =~ s/^.?<body[^>]>//is;
589	$$textref =~ s/(<\/body[^>]>\|<\/html[^>]>)//isg;
590	}
591
592	# single section document
593	$self->process_section($textref, $upgraded_base_dir, $upgraded_file, $doc_obj, $cursection);
594	}
595
596	return 1;
597	}
598
599
600	sub process_heading
601	{
602	my ($self, $nHeadNo, $strHeadingText, $arrSections, $file) = @_;
603	$strHeadingText = '' if (!defined($strHeadingText));
604
605	my $strMetadata = $self->update_section_data($arrSections, int($nHeadNo));
606
607	my $strSecMetadata = '';
608	while ($strHeadingText =~ s/<!--gsdl-metadata(.*?)-->//is)
609	{
610	$strSecMetadata .= $1;
611	}
612
613	$strHeadingText =~ s/^\s+//g;
614	$strHeadingText =~ s/\s+$//g;
615	$strSecMetadata =~ s/^\s+//g;
616	$strSecMetadata =~ s/\s+$//g;
617
618	$strMetadata .= "\n<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">" . $strHeadingText . "</Metadata>\n";
619
620	if (length($strSecMetadata)) {
621	$strMetadata .= "\t\t" . $strSecMetadata . "\n";
622	}
623
624	$strMetadata .= "\t</Description>\n";
625
626	return "<!--" . $strMetadata . "-->";
627	}
628
629
630	sub update_section_data
631	{
632	my ($self, $arrSections, $nCurTocNo) = @_;
633	my ($strBuffer, $nLast, $nSections) = ('', 0, scalar(@$arrSections));
634
635	if ($nSections == 0) {
636	push @$arrSections, $nCurTocNo;
637	return $strBuffer;
638	}
639	$nLast = $arrSections->[$nSections - 1];
640	if ($nCurTocNo > $nLast) {
641	push @$arrSections, $nCurTocNo;
642	return $strBuffer;
643	}
644	for(my $i = $nSections - 1; $i >= 0; $i--) {
645	if ($nCurTocNo <= $arrSections->[$i]) {
646	$strBuffer .= "\n</Section>";
647	pop @$arrSections;
648	}
649	}
650	push @$arrSections, $nCurTocNo;
651	return $strBuffer;
652	}
653
654
655	# note that process_section may be called multiple times for a single
656	# section (relying on the fact that add_utf8_text appends the text to any
657	# that may exist already).
658	sub process_section {
659	my $self = shift (@_);
660	my ($textref, $base_dir, $file, $doc_obj, $cursection) = @_;
661	# trap links
662	if (!$self->{'nolinks'}) {
663	# usemap="./#index" not handled correctly => change to "#index"
664	## $$textref =~ s/(<img[^>]?usemap\s=\s[\"\']?)([^\"\'>\s]+)([\"\']?[^>]>)/
665
666	$$textref =~ s/(<img[^>]?usemap\s=\s)((?:[\"][^\"]+[\"])\|(?:[\'][^\']+[\'])\|(?:[^\s\/>]+))([^>]>)/
667	$self->replace_usemap_links($1, $2, $3)/isge;
668
669	## $$textref =~ s/(<(?:a\|area\|frame\|link\|script)\s+[^>]?\s(?:href\|src)\s=\s[\"\']?)([^\"\'>\s]+)([\"\']?[^>]*>)/
670
671	$$textref =~ s/(<(?:a\|area\|frame\|link\|script)\s+[^>]?\s(?:href\|src)\s=\s)((?:[\"][^\"]+[\"])\|(?:[\'][^\']+[\'])\|(?:[^\s\/>]+))([^>]*>)/
672	$self->replace_href_links ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
673	}
674
675	# trap images
676
677	# Previously, by default, HTMLPlugin would embed <img> tags inside anchor tags
678	# i.e. <a href="image><img src="image"></a> in order to overcome a problem that
679	# turned regular text succeeding images into links. That is, by embedding <imgs>
680	# inside <a href=""></a>, the text following images were no longer misbehaving.
681	# However, there would be many occasions whereby images were not meant to link
682	# to their source images but where the images would link to another web page.
683	# To allow this, the no_image_links option was introduced: it would prevent
684	# the behaviour of embedding images into links that referenced the source images.
685
686	# Somewhere along the line, the problem of normal text turning into links when
687	# such text followed images which were not embedded in <a href=""></a> ceased
688	# to occur. This is why the following lines have been commented out (as well as
689	# two lines in replace_images). They appear to no longer apply.
690
691	# If at any time, there is a need for having images embedded in <a> anchor tags,
692	# then it might be better to turn that into an HTMLPlugin option rather than make
693	# it the default behaviour. Also, eventually, no_image_links needs to become
694	# a deprecated option for HTMLPlugin as it has now become the default behaviour.
695
696	#if(!$self->{'no_image_links'}){
697	$$textref =~ s/(<(?:img\|embed\|table\|tr\|td)[^>]?(?:src\|background)\s=\s)((?:[\"][^\"]+[\"])\|(?:[\'][^\']+[\'])\|(?:[^\s\/>]+))([^>]>)/
698	$self->replace_images ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
699	#}
700
701	# add text to document object
702	# turn \ into \\ so that the rest of greenstone doesn't think there
703	# is an escape code following. (Macro parsing loses them...)
704	$$textref =~ s/\\/\\\\/go;
705
706	$doc_obj->add_utf8_text($cursection, $$textref);
707	}
708
709	sub replace_images {
710	my $self = shift (@_);
711	my ($front, $link, $back, $base_dir,
712	$file, $doc_obj, $section) = @_;
713
714	# remove quotes from link at start and end if necessary
715	if ($link=~/^[\"\']/) {
716	$link=~s/^[\"\']//;
717	$link=~s/[\"\']$//;
718	$front.='"';
719	$back="\"$back";
720	}
721
722	$link =~ s/\n/ /g;
723
724	# Hack to overcome Windows wv 0.7.1 bug that causes embedded images to be broken
725	# If the Word file path has spaces in it, wv messes up and you end up with
726	# absolute paths for the images, and without the "file://" prefix
727	# So check for this special case and massage the data to be correct
728	if ($ENV{'GSDLOS'} =~ m/^windows/i && $self->{'plugin_type'} eq "WordPlug" && $link =~ m/^[A-Za-z]\:\\/) {
729	$link =~ s/^.*\\([^\\]+)$/$1/;
730	}
731
732	my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
733
734	my $img_file = $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section);
735
736	# print STDERR "** link = $link\n href = $href\n** img_file = $img_file, rl = $rl\n";
737
738	my $anchor_name = $img_file;
739	#$anchor_name =~ s/^.*\///;
740	#$anchor_name = "<a name=\"$anchor_name\" ></a>";
741
742	my $image_link = $front . $img_file .$back;
743	return $image_link;
744
745	# The reasons for why the following two lines are no longer necessary can be
746	# found in subroutine process_section
747	#my $anchor_link = "<a href=\"$img_file\" >".$image_link."</a>";
748	#return $anchor_link;
749
750	#return $front . $img_file . $back . $anchor_name;
751	}
752
753	sub replace_href_links {
754	my $self = shift (@_);
755	my ($front, $link, $back, $base_dir, $file, $doc_obj, $section) = @_;
756
757	# remove quotes from link at start and end if necessary
758	if ($link=~/^[\"\']/) {
759	$link=~s/^[\"\']//;
760	$link=~s/[\"\']$//;
761	$front.='"';
762	$back="\"$back";
763	}
764
765	# attempt to sort out targets - frames are not handled
766	# well in this plugin and some cases will screw things
767	# up - e.g. the _parent target (so we'll just remove
768	# them all ;-)
769	$front =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
770	$back =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
771	$front =~ s/target=\"?_parent\"?//is;
772	$back =~ s/target=\"?_parent\"?//is;
773
774	return $front . $link . $back if $link =~ m/^\#/s;
775	$link =~ s/\n/ /g;
776
777	# Find file referred to by $link on file system
778	# This is more complicated than it sounds when char encodings
779	# is taken in to account
780	my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
781
782	# href may use '\'s where '/'s should be on Windows
783	$href =~ s/\\/\//g;
784	my ($filename) = $href =~ m/^(?:.?):(?:\/\/)?(.)/;
785
786
787	##### leave all these links alone (they won't be picked up by intermediate
788	##### pages). I think that's safest when dealing with frames, targets etc.
789	##### (at least until I think of a better way to do it). Problems occur with
790	##### mailto links from within small frames, the intermediate page is displayed
791	##### within that frame and can't be seen. There is still potential for this to
792	##### happen even with html pages - the solution seems to be to somehow tell
793	##### the browser from the server side to display the page being sent (i.e.
794	##### the intermediate page) in the top level window - I'm not sure if that's
795	##### possible - the following line should probably be deleted if that can be done
796	return $front . $link . $back if $href =~ m/^(mailto\|news\|gopher\|nntp\|telnet\|javascript):/is;
797
798	if (($rl == 0) \|\| ($filename =~ m/$self->{'process_exp'}/) \|\|
799	($href =~ m/\/$/) \|\| ($href =~ m/^(mailto\|news\|gopher\|nntp\|telnet\|javascript):/i)) {
800
801	if ($ENV{'GSDLOS'} =~ m/^windows$/) {
802
803	# Don't do any encoding for now, as not clear what
804	# the right thing to do is to support filename
805	# encoding on Windows when they are not UTF16
806	#
807	}
808	else {
809	# => Unix-based system
810
811	# If web page didn't give encoding, then default to utf8
812	my $content_encoding= $self->{'content_encoding'} \|\| "utf8";
813
814	if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
815	print STDERR "**** Encoding with '$content_encoding', href: $href\n";
816	}
817
818	$href = encode($content_encoding,$href);
819	}
820
821	$href = &unicode::raw_filename_to_url_encoded($href);
822	$href = &unicode::filename_to_url($href);
823
824	&ghtml::urlsafe ($href);
825
826	if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
827	print STDERR "****** href=$href\n";
828	}
829
830
831	return $front . "_httpextlink_&rl=" . $rl . "&href=" . $href . $hash_part . $back;
832	} else {
833	# link is to some other type of file (e.g., an image) so we'll
834	# need to associate that file
835	return $front . $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) . $back;
836	}
837	}
838
839	sub add_file {
840	my $self = shift (@_);
841	my ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) = @_;
842	my ($newname);
843
844	my $filename = $href;
845	if ($base_dir eq "") {
846	# remove http:/ thereby leaving one slash at the start
847	$filename =~ s/^[^:]*:\///;
848	}
849	else {
850	# remove http://
851	$filename =~ s/^[^:]*:\/\///;
852	}
853
854	$filename = &util::filename_cat($base_dir, $filename);
855
856	if (($self->{'use_realistic_book'}) \|\| ($self->{'old_style_HDL'})) {
857	# we are processing a tidytmp file - want paths to be in import
858	$filename =~ s/([\\\/])tidytmp([\\\/])/$1import$2/;
859	}
860
861	# Replace %XX's in URL with decoded value if required. Note that the
862	# filename may include the %XX in some situations. If the original
863	# file's name was in URL encoding, the following method will not decode
864	# it.
865	my $utf8_filename = $filename;
866	my $opt_decode_utf8_filename = $self->opt_url_decode($utf8_filename);
867
868	my $content_encoding= $self->{'content_encoding'} \|\| "utf8";
869
870	# The filenames that come through the HTML file have been decoded
871	# into Unicode aware Perl strings. Need to convert them back
872	# to their initial raw-byte encoding to match the file that
873	# exists on the file system
874	$filename = encode($content_encoding, $opt_decode_utf8_filename);
875
876	# some special processing if the intended filename was converted to utf8, but
877	# the actual file still needs to be renamed
878	if (!&util::fd_exists($filename)) {
879	# try the original filename stored in map
880	if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
881	print STDERR "******!! orig filename did not exist: $filename\n";
882	}
883
884	## print STDERR "**** trying to look up utf8_filename: $utf8_filename\n";
885
886	my $original_filename = $self->{'utf8_to_original_filename'}->{$utf8_filename};
887
888	if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
889	print STDERR "****** From lookup utf8_filename, now trying for: $original_filename\n";
890	}
891
892	if (defined $original_filename && -e $original_filename) {
893	if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
894	print STDERR "****** Found match!\n";
895	}
896	$filename = $original_filename;
897	}
898	}
899
900	my ($ext) = $filename =~ m/(\.[^\.]*)$/;
901
902	if ($rl == 0) {
903	if ((!defined $ext) \|\| ($ext !~ m/$self->{'assoc_files'}/)) {
904	return "_httpextlink_&rl=0&el=prompt&href=" . $href . $hash_part;
905	}
906	else {
907	return "_httpextlink_&rl=0&el=direct&href=" . $href . $hash_part;
908	}
909	}
910
911	if ((!defined $ext) \|\| ($ext !~ m/$self->{'assoc_files'}/)) {
912	return "_httpextlink_&rl=" . $rl . "&href=" . $href . $hash_part;
913	}
914	# add the original image file as a source file
915	if (!$self->{'processing_tmp_files'} ) {
916	$doc_obj->associate_source_file($filename);
917	}
918	if ($self->{'rename_assoc_files'}) {
919	if (defined $self->{'aux_files'}->{$href}) {
920	$newname = $self->{'aux_files'}->{$href}->{'dir_num'} . "/" .
921	$self->{'aux_files'}->{$href}->{'file_num'} . $ext;
922	} else {
923	$newname = $self->{'dir_num'} . "/" . $self->{'file_num'} . $ext;
924	$self->{'aux_files'}->{$href} = {'dir_num' => $self->{'dir_num'}, 'file_num' => $self->{'file_num'}};
925	$self->inc_filecount ();
926	}
927	$doc_obj->associate_file($filename, $newname, undef, $section);
928	return "_httpdocimg_/$newname";
929	} else {
930	if(&unicode::is_url_encoded($utf8_filename)) {
931	# use the possibly-decoded filename instead to avoid double URL encoding
932	($newname) = $filename =~ m/([^\/\\]*)$/;
933	} else {
934	($newname) = $utf8_filename =~ m/([^\/\\]*)$/;
935	}
936
937	# Make sure this name uses only ASCII characters.
938	# We use either base64 or URL encoding, as these preserve original encoding
939	$newname = &util::rename_file($newname, $self->{'file_rename_method'});
940
941	### print STDERR "***** associating $filename (raw-byte/utf8)-> $newname\n";
942	$doc_obj->associate_file($filename, $newname, undef, $section);
943
944	# Since the generated image will be URL-encoded to avoid file-system/browser mess-ups
945	# of filenames, URL-encode the additional percent signs of the URL-encoded filename
946	my $newname_url = $newname;
947	$newname_url = &unicode::filename_to_url($newname_url);
948	return "_httpdocimg_/$newname_url";
949	}
950	}
951
952
953	sub format_link {
954	my $self = shift (@_);
955	my ($link, $base_dir, $file) = @_;
956
957	# strip off hash part, e.g. #foo, but watch out for any entities, e.g. α
958	my ($before_hash, $hash_part) = $link =~ m/^(.?[^&])(\#.)?$/;
959
960	$hash_part = "" if !defined $hash_part;
961	if (!defined $before_hash \|\| $before_hash !~ m/[\w\.\/]/) {
962	my $outhandle = $self->{'outhandle'};
963	print $outhandle "HTMLPlugin: ERROR - badly formatted tag ignored ($link)\n"
964	if $self->{'verbosity'};
965	return ($link, "", 0);
966	}
967
968	if ($before_hash =~ s@^((?:http\|https\|ftp\|file\|mms)://)@@i) {
969	my $type = $1;
970
971	if ($link =~ m/^(http\|ftp):/i) {
972	# Turn url (using /) into file name (possibly using \ on windows)
973	my @http_dir_split = split('/', $before_hash);
974	$before_hash = &util::filename_cat(@http_dir_split);
975	}
976
977	$before_hash = $self->eval_dir_dots($before_hash);
978
979	my $linkfilename = &util::filename_cat ($base_dir, $before_hash);
980
981	my $rl = 0;
982	$rl = 1 if (-e $linkfilename);
983
984	# make sure there's a slash on the end if it's a directory
985	if ($before_hash !~ m/\/$/) {
986	$before_hash .= "/" if (-d $linkfilename);
987	}
988	return ($type . $before_hash, $hash_part, $rl);
989
990	} elsif ($link !~ m/^(mailto\|news\|gopher\|nntp\|telnet\|javascript):/i && $link !~ m/^\//) {
991
992	if ($before_hash =~ s@^/@@ \|\| $before_hash =~ m/\\/) {
993
994	# the first directory will be the domain name if file_is_url
995	# to generate archives, otherwise we'll assume all files are
996	# from the same site and base_dir is the root
997
998	if ($self->{'file_is_url'}) {
999	my @dirs = split /[\/\\]/, $file;
1000	my $domname = shift (@dirs);
1001	$before_hash = &util::filename_cat($domname, $before_hash);
1002	$before_hash =~ s@\\@/@g; # for windows
1003	}
1004	else
1005	{
1006	# see if link shares directory with source document
1007	# => turn into relative link if this is so!
1008
1009	if ($ENV{'GSDLOS'} =~ m/^windows/i) {
1010	# too difficult doing a pattern match with embedded '\'s...
1011	my $win_before_hash=$before_hash;
1012	$win_before_hash =~ s@(\\)+@/@g;
1013	# $base_dir is already similarly "converted" on windows.
1014	if ($win_before_hash =~ s@^$base_dir/@@o) {
1015	# if this is true, we removed a prefix
1016	$before_hash=$win_before_hash;
1017	}
1018	}
1019	else {
1020	# before_hash has lost leading slash by this point,
1021	# -> add back in prior to substitution with $base_dir
1022	$before_hash = "/$before_hash";
1023
1024	$before_hash = &util::filename_cat("",$before_hash);
1025	$before_hash =~ s@^$base_dir/@@;
1026	}
1027	}
1028	} else {
1029	# Turn relative file path into full path
1030	my $dirname = &File::Basename::dirname($file);
1031	$before_hash = &util::filename_cat($dirname, $before_hash);
1032	$before_hash = $self->eval_dir_dots($before_hash);
1033	}
1034
1035	my $linkfilename = &util::filename_cat ($base_dir, $before_hash);
1036	# make sure there's a slash on the end if it's a directory
1037	if ($before_hash !~ m/\/$/) {
1038	$before_hash .= "/" if (-d $linkfilename);
1039	}
1040	return ("http://" . $before_hash, $hash_part, 1);
1041	} else {
1042	# mailto, news, nntp, telnet, javascript or gopher link
1043	return ($before_hash, "", 0);
1044	}
1045	}
1046
1047	sub extract_first_NNNN_characters {
1048	my $self = shift (@_);
1049	my ($textref, $doc_obj, $thissection) = @_;
1050
1051	foreach my $size (split /,/, $self->{'first'}) {
1052	my $tmptext = $$textref;
1053	# skip to the body
1054	$tmptext =~ s/.<body[^>]>//i;
1055	# remove javascript
1056	$tmptext =~ s@<script.*?</script>@ @sig;
1057	$tmptext =~ s/<[^>]*>/ /g;
1058	$tmptext =~ s/ / /g;
1059	$tmptext =~ s/^\s+//;
1060	$tmptext =~ s/\s+$//;
1061	$tmptext =~ s/\s+/ /gs;
1062	$tmptext = &unicode::substr ($tmptext, 0, $size);
1063	$tmptext =~ s/\s\S*$/…/; # adds an ellipse (...)
1064	$doc_obj->add_utf8_metadata ($thissection, "First$size", $tmptext);
1065	}
1066	}
1067
1068
1069	sub extract_metadata {
1070	my $self = shift (@_);
1071	my ($textref, $metadata, $doc_obj, $section) = @_;
1072	my $outhandle = $self->{'outhandle'};
1073	# if we don't want metadata, we may as well not be here ...
1074	return if (!defined $self->{'metadata_fields'});
1075
1076	my $separator = $self->{'metadata_field_separator'};
1077	if ($separator eq "") {
1078	undef $separator;
1079	}
1080
1081	# metadata fields to extract/save. 'key' is the (lowercase) name of the
1082	# html meta, 'value' is the metadata name for greenstone to use
1083	my %find_fields = ();
1084
1085	my %creator_fields = (); # short-cut for lookups
1086
1087
1088	foreach my $field (split /,/, $self->{'metadata_fields'}) {
1089	$field =~ s/^\s+//; # remove leading whitespace
1090	$field =~ s/\s+$//; # remove trailing whitespace
1091
1092	# support tag<tagname>
1093	if ($field =~ m/^(.?)\s<(.*?)>$/) {
1094	# "$2" is the user's preferred gs metadata name
1095	$find_fields{lc($1)}=$2; # lc = lowercase
1096	} else { # no <tagname> for mapping
1097	# "$field" is the user's preferred gs metadata name
1098	$find_fields{lc($field)}=$field; # lc = lowercase
1099	}
1100	}
1101
1102	if (defined $self->{'hunt_creator_metadata'} &&
1103	$self->{'hunt_creator_metadata'} == 1 ) {
1104	my @extra_fields =
1105	(
1106	'author',
1107	'author.email',
1108	'creator',
1109	'dc.creator',
1110	'dc.creator.corporatename',
1111	);
1112
1113	# add the creator_metadata fields to search for
1114	foreach my $field (@extra_fields) {
1115	$creator_fields{$field}=0; # add to lookup hash
1116	}
1117	}
1118
1119
1120	# find the header in the html file, which has the meta tags
1121	$$textref =~ m@<head>(.*?)</head>@si;
1122
1123	my $html_header=$1;
1124
1125	# go through every <meta... tag defined in the html and see if it is
1126	# one of the tags we want to match.
1127
1128	# special case for title - we want to remember if its been found
1129	my $found_title = 0;
1130	# this assumes that ">" won't appear. (I don't think it's allowed to...)
1131	$html_header =~ m/^/; # match the start of the string, for \G assertion
1132
1133	while ($html_header =~ m/\G.?<meta(.?)>/sig) {
1134	my $metatag=$1;
1135	my ($tag, $value);
1136
1137	# find the tag name
1138	$metatag =~ m/(?:name\|http-equiv)\s=\s([\"\'])?(.*?)\1/is;
1139	$tag=$2;
1140	# in case they're not using " or ', but they should...
1141	if (! $tag) {
1142	$metatag =~ m/(?:name\|http-equiv)\s=\s([^\s\>]+)/is;
1143	$tag=$1;
1144	}
1145
1146	if (!defined $tag) {
1147	print $outhandle "HTMLPlugin: can't find NAME in \"$metatag\"\n";
1148	next;
1149	}
1150
1151	# don't need to assign this field if it was passed in from a previous
1152	# (recursive) plugin
1153	if (defined $metadata->{$tag}) {next}
1154
1155	# find the tag content
1156	$metatag =~ m/content\s=\s([\"\'])?(.*?)\1/is;
1157	$value=$2;
1158
1159	if (! $value) {
1160	$metatag =~ m/(?:name\|http-equiv)\s=\s([^\s\>]+)/is;
1161	$value=$1;
1162	}
1163	if (!defined $value) {
1164	print $outhandle "HTMLPlugin: can't find VALUE in \"$metatag\"\n";
1165	next;
1166	}
1167
1168	# clean up and add
1169	$value =~ s/\s+/ /gs;
1170	chomp($value); # remove trailing \n, if any
1171	if (exists $creator_fields{lc($tag)}) {
1172	# map this value onto greenstone's "Creator" metadata
1173	$tag='Creator';
1174	} elsif (!exists $find_fields{lc($tag)}) {
1175	next; # don't want this tag
1176	} else {
1177	# get the user's preferred capitalisation
1178	$tag = $find_fields{lc($tag)};
1179	}
1180	if (lc($tag) eq "title") {
1181	$found_title = 1;
1182	}
1183
1184	if ($self->{'verbosity'} > 2) {
1185	print $outhandle " extracted \"$tag\" metadata \"$value\"\n";
1186	}
1187
1188	if ($tag =~ /\./) {
1189	# there is a . so has a namespace, add ex.
1190	$tag = "ex.$tag";
1191	}
1192	if (defined $separator) {
1193	my @values = split($separator, $value);
1194	foreach my $v (@values) {
1195	$doc_obj->add_utf8_metadata($section, $tag, $v) if $v =~ /\S/;
1196	}
1197	}
1198	else {
1199	$doc_obj->add_utf8_metadata($section, $tag, $value);
1200	}
1201	}
1202
1203	# TITLE: extract the document title
1204	if (exists $find_fields{'title'} && !$found_title) {
1205	# we want a title, and didn't find one in the meta tags
1206	# see if there's a <title> tag
1207	my $title;
1208	my $from = ""; # for debugging output only
1209	if ($html_header =~ m/<title[^>]>([^<]+)<\/title[^>]>/is) {
1210	$title = $1;
1211	$from = "<title> tags";
1212	}
1213
1214	if (!defined $title) {
1215	$from = "first 100 chars";
1216	# if no title use first 100 or so characters
1217	$title = $$textref;
1218	$title =~ s/^\xFE\xFF//; # Remove unicode byte order mark
1219	$title =~ s/^.*?<body>//si;
1220	# ignore javascript!
1221	$title =~ s@<script.*?</script>@ @sig;
1222	$title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
1223	$title =~ s/<[^>]*>/ /g; # remove all HTML tags
1224	$title = substr ($title, 0, 100);
1225	$title =~ s/\s\S*$/.../;
1226	}
1227	$title =~ s/<[^>]*>/ /g; # remove html tags
1228	$title =~ s/ / /g;
1229	$title =~ s/(?: \|\xc2\xa0)/ /g; # utf-8 for nbsp...
1230	$title =~ s/\s+/ /gs; # collapse multiple spaces
1231	$title =~ s/^\s*//; # remove leading spaces
1232	$title =~ s/\s*$//; # remove trailing spaces
1233
1234	$title =~ s/^$self->{'title_sub'}// if ($self->{'title_sub'});
1235	$title =~ s/^\s+//s; # in case title_sub introduced any...
1236	$doc_obj->add_utf8_metadata ($section, "Title", $title);
1237	print $outhandle " extracted Title metadata \"$title\" from $from\n"
1238	if ($self->{'verbosity'} > 2);
1239	}
1240
1241	# add FileFormat metadata
1242	$doc_obj->add_metadata($section,"FileFormat", "HTML");
1243
1244	# Special, for metadata names such as tagH1 - extracts
1245	# the text between the first <H1> and </H1> tags into "H1" metadata.
1246
1247	foreach my $field (keys %find_fields) {
1248	if ($field !~ m/^tag([a-z0-9]+)$/i) {next}
1249	my $tag = $1;
1250	if ($$textref =~ m@<$tag[^>]>(.?)</$tag[^>]*>@g) {
1251	my $content = $1;
1252	$content =~ s/ / /g;
1253	$content =~ s/<[^>]*>/ /g;
1254	$content =~ s/^\s+//;
1255	$content =~ s/\s+$//;
1256	$content =~ s/\s+/ /gs;
1257	if ($content) {
1258	$tag=$find_fields{"tag$tag"}; # get the user's capitalisation
1259	$tag =~ s/^tag//i;
1260	$doc_obj->add_utf8_metadata ($section, $tag, $content);
1261	print $outhandle " extracted \"$tag\" metadata \"$content\"\n"
1262	if ($self->{'verbosity'} > 2);
1263	}
1264	}
1265	}
1266	}
1267
1268
1269	# evaluate any "../" to next directory up
1270	# evaluate any "./" as here
1271	sub eval_dir_dots {
1272	my $self = shift (@_);
1273	my ($filename) = @_;
1274	my $dirsep_os = &util::get_os_dirsep();
1275	my @dirsep = split(/$dirsep_os/,$filename);
1276
1277	my @eval_dirs = ();
1278	foreach my $d (@dirsep) {
1279	if ($d eq "..") {
1280	pop(@eval_dirs);
1281
1282	} elsif ($d eq ".") {
1283	# do nothing!
1284
1285	} else {
1286	push(@eval_dirs,$d);
1287	}
1288	}
1289
1290	# Need to fiddle with number of elements in @eval_dirs if the
1291	# first one is the empty string. This is because of a
1292	# modification to util::filename_cat that supresses the addition
1293	# of a leading '/' character (or \ if windows) (intended to help
1294	# filename cat with relative paths) if the first entry in the
1295	# array is the empty string. Making the array start with two
1296	# empty strings is a way to defeat this "smart" option.
1297	#
1298	if (scalar(@eval_dirs) > 0) {
1299	if ($eval_dirs[0] eq ""){
1300	unshift(@eval_dirs,"");
1301	}
1302	}
1303
1304	my $evaluated_filename = (scalar @eval_dirs > 0) ? &util::filename_cat(@eval_dirs) : "";
1305	return $evaluated_filename;
1306	}
1307
1308	sub replace_usemap_links {
1309	my $self = shift (@_);
1310	my ($front, $link, $back) = @_;
1311
1312	# remove quotes from link at start and end if necessary
1313	if ($link=~/^[\"\']/) {
1314	$link=~s/^[\"\']//;
1315	$link=~s/[\"\']$//;
1316	$front.='"';
1317	$back="\"$back";
1318	}
1319
1320	$link =~ s/^\.\///;
1321	return $front . $link . $back;
1322	}
1323
1324	sub inc_filecount {
1325	my $self = shift (@_);
1326
1327	if ($self->{'file_num'} == 1000) {
1328	$self->{'dir_num'} ++;
1329	$self->{'file_num'} = 0;
1330	} else {
1331	$self->{'file_num'} ++;
1332	}
1333	}
1334
1335
1336	# Extend read_file so that strings like é are
1337	# converted to UTF8 internally.
1338	#
1339	# We don't convert < or > or & or " in case
1340	# they interfere with the GML files
1341
1342	sub read_file {
1343	my $self = shift(@_);
1344	my ($filename, $encoding, $language, $textref) = @_;
1345
1346	$self->SUPER::read_file($filename, $encoding, $language, $textref);
1347
1348	# Convert entities to their Unicode code-point equivalents
1349	$$textref =~ s/&(lt\|gt\|amp\|quot\|nbsp);/&z$1;/go;
1350	$$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1,1)/gseo;
1351	$$textref =~ s/&z(lt\|gt\|amp\|quot\|nbsp);/&$1;/go;
1352
1353	}
1354
1355	sub HB_read_html_file {
1356	my $self = shift (@_);
1357	my ($htmlfile, $text) = @_;
1358
1359	# load in the file
1360	if (!open (FILE, $htmlfile)) {
1361	print STDERR "ERROR - could not open $htmlfile\n";
1362	return;
1363	}
1364
1365	my $foundbody = 0;
1366	$self->HB_gettext (\$foundbody, $text, "FILE");
1367	close FILE;
1368
1369	# just in case there was no <body> tag
1370	if (!$foundbody) {
1371	$foundbody = 1;
1372	open (FILE, $htmlfile) \|\| return;
1373	$self->HB_gettext (\$foundbody, $text, "FILE");
1374	close FILE;
1375	}
1376	# text is in utf8
1377	}
1378
1379	# converts the text to utf8, as ghtml does that for é etc.
1380	sub HB_gettext {
1381	my $self = shift (@_);
1382	my ($foundbody, $text, $handle) = @_;
1383
1384	my $line = "";
1385	while (defined ($line = <$handle>)) {
1386	# look for body tag
1387	if (!$$foundbody) {
1388	if ($line =~ s/^.<body[^>]>//i) {
1389	$$foundbody = 1;
1390	} else {
1391	next;
1392	}
1393	}
1394
1395	# check for symbol fonts
1396	if ($line =~ m/<font [^>]?face\s=\s*\"?(\w+)\"?/i) {
1397	my $font = $1;
1398	print STDERR "HBPlug::HB_gettext - warning removed font $font\n"
1399	if ($font !~ m/^arial$/i);
1400	}
1401
1402	$$text .= $line;
1403	}
1404
1405	if ($self->{'input_encoding'} eq "iso_8859_1") {
1406	# convert to utf-8
1407	$$text=&unicode::unicode2utf8(&unicode::convert2unicode("iso_8859_1", $text));
1408	}
1409	# convert any alphanumeric character entities to their utf-8
1410	# equivalent for indexing purposes
1411	#&ghtml::convertcharentities ($$text);
1412
1413	$$text =~ s/\s+/ /g; # remove \n's
1414
1415	# At this point $$text is a binary byte string
1416	# => turn it into a Unicode aware string, so full
1417	# Unicode aware pattern matching can be used.
1418	# For instance: 's/\x{0101}//g' or '[[:upper:]]'
1419	#
1420
1421	$$text = decode("utf8",$$text);
1422	}
1423
1424	sub HB_clean_section {
1425	my $self = shift (@_);
1426	my ($section) = @_;
1427
1428	# remove tags without a starting tag from the section
1429	my ($tag, $tagstart);
1430	while ($section =~ m/<\/([^>]{1,10})>/) {
1431	$tag = $1;
1432	$tagstart = index($section, "<$tag");
1433	last if (($tagstart >= 0) && ($tagstart < index($section, "<\/$tag")));
1434	$section =~ s/<\/$tag>//;
1435	}
1436
1437	# remove extra paragraph tags
1438	while ($section =~ s/<p\b[^>]>\s<p\b/<p/ig) {}
1439
1440	# remove extra stuff at the end of the section
1441	while ($section =~ s/(<u>\|<i>\|<b>\|<p\b[^>]*>\| \|\s)$//i) {}
1442
1443	# add a newline at the beginning of each paragraph
1444	$section =~ s/(.)\s*<p\b/$1\n\n<p/gi;
1445
1446	# add a newline every 80 characters at a word boundary
1447	# Note: this regular expression puts a line feed before
1448	# the last word in each section, even when it is not
1449	# needed.
1450	$section =~ s/(.{1,80})\s/$1\n/g;
1451
1452	# fix up the image links
1453	$section =~ s/<img[^>]?src=\"?([^\">]+)\"?[^>]>/
1454	<center><img src=\"$1\" \/><\/center><br\/>/ig;
1455	$section =~ s/<<I>>\s*([^\.]+\.(png\|jpg\|gif))/
1456	<center><img src=\"$1\" \/><\/center><br\/>/ig;
1457
1458	return $section;
1459	}
1460
1461	# Will convert the oldHDL format to the new HDL format (using the Section tag)
1462	sub convert_to_newHDLformat
1463	{
1464	my $self = shift (@_);
1465	my ($file,$cnfile) = @_;
1466	my $input_filename = $file;
1467	my $tmp_filename = $cnfile;
1468
1469	# write HTML tmp file with new HDL format
1470	open (PROD, ">$tmp_filename") \|\| die("Error Writing to File: $tmp_filename $!");
1471
1472	# read in the file and do basic html cleaning (removing header etc)
1473	my $html = "";
1474	$self->HB_read_html_file ($input_filename, \$html);
1475
1476	# process the file one section at a time
1477	my $curtoclevel = 1;
1478	my $firstsection = 1;
1479	my $toclevel = 0;
1480	while (length ($html) > 0) {
1481	if ($html =~ s/^.?(?:<p\b[^>]>)?((<b>\|<i>\|<u>\|\s))<<TOC(\d+)>>\s(.*?)<p\b/<p/i) {
1482	$toclevel = $3;
1483	my $title = $4;
1484	my $sectiontext = "";
1485	if ($html =~ s/^(.?)((?:<p\b[^>]>)?((<b>\|<i>\|<u>\|\s)*)<<TOC\d+>>)/$2/i) {
1486	$sectiontext = $1;
1487	} else {
1488	$sectiontext = $html;
1489	$html = "";
1490	}
1491
1492	# remove tags and extra spaces from the title
1493	$title =~ s/<\/?[^>]+>//g;
1494	$title =~ s/^\s+\|\s+$//g;
1495
1496	# close any sections below the current level and
1497	# create a new section (special case for the firstsection)
1498	print PROD "<!--\n";
1499	while (($curtoclevel > $toclevel) \|\|
1500	(!$firstsection && $curtoclevel == $toclevel)) {
1501	$curtoclevel--;
1502	print PROD "</Section>\n";
1503	}
1504	if ($curtoclevel+1 < $toclevel) {
1505	print STDERR "WARNING - jump in toc levels in $input_filename " .
1506	"from $curtoclevel to $toclevel\n";
1507	}
1508	while ($curtoclevel < $toclevel) {
1509	$curtoclevel++;
1510	}
1511
1512	if ($curtoclevel == 1) {
1513	# add the header tag
1514	print PROD "-->\n";
1515	print PROD "<HTML>\n<HEAD>\n<TITLE>$title</TITLE>\n</HEAD>\n<BODY>\n";
1516	print PROD "<!--\n";
1517	}
1518
1519	print PROD "<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">$title</Metadata>\n\t</Description>\n";
1520
1521	print PROD "-->\n";
1522
1523	# clean up the section html
1524	$sectiontext = $self->HB_clean_section($sectiontext);
1525
1526	print PROD "$sectiontext\n";
1527
1528	} else {
1529	print STDERR "WARNING - leftover text\n" , $self->shorten($html),
1530	"\nin $input_filename\n";
1531	last;
1532	}
1533	$firstsection = 0;
1534	}
1535
1536	print PROD "<!--\n";
1537	while ($curtoclevel > 0) {
1538	$curtoclevel--;
1539	print PROD "</Section>\n";
1540	}
1541	print PROD "-->\n";
1542
1543	close (PROD) \|\| die("Error Closing File: $tmp_filename $!");
1544
1545	return $tmp_filename;
1546	}
1547
1548	sub shorten {
1549	my $self = shift (@_);
1550	my ($text) = @_;
1551
1552	return "\"$text\"" if (length($text) < 100);
1553
1554	return "\"" . substr ($text, 0, 50) . "\" ... \"" .
1555	substr ($text, length($text)-50) . "\"";
1556	}
1557
1558	sub convert_tidy_or_oldHDL_file
1559	{
1560	my $self = shift (@_);
1561	my ($file) = @_;
1562	my $input_filename = $file;
1563
1564	if (-d $input_filename)
1565	{
1566	return $input_filename;
1567	}
1568
1569	# get the input filename
1570	my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
1571	my $base_dirname = $dirname;
1572	$suffix = lc($suffix);
1573
1574	# derive tmp filename from input filename
1575	# Remove any white space from filename -- no risk of name collision, and
1576	# makes later conversion by utils simpler. Leave spaces in path...
1577	# tidy up the filename with space, dot, hyphen between
1578	$tailname =~ s/\s+//g;
1579	$tailname =~ s/\.+//g;
1580	$tailname =~ s/\-+//g;
1581	# convert to utf-8 otherwise we have problems with the doc.xml file
1582	# later on
1583	&unicode::ensure_utf8(\$tailname);
1584
1585	# softlink to collection tmp dir
1586	my $tmp_dirname = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "tidytmp");
1587	&util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
1588
1589	my $test_dirname = "";
1590	my $f_separator = &util::get_os_dirsep();
1591
1592	if ($dirname =~ m/import$f_separator/)
1593	{
1594	$test_dirname = $'; #'
1595
1596	#print STDERR "init $'\n";
1597
1598	while ($test_dirname =~ m/[$f_separator]/)
1599	{
1600	my $folderdirname = $`;
1601	$tmp_dirname = &util::filename_cat($tmp_dirname,$folderdirname);
1602	&util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
1603	$test_dirname = $'; #'
1604	}
1605	}
1606
1607	my $tmp_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
1608
1609	# tidy or convert the input file if it is a HTML-like file or it is accepted by the process_exp
1610	if (($suffix eq ".htm") \|\| ($suffix eq ".html") \|\| ($suffix eq ".shtml"))
1611	{
1612	#convert the input file to a new style HDL
1613	my $hdl_output_filename = $input_filename;
1614	if ($self->{'old_style_HDL'})
1615	{
1616	$hdl_output_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
1617	$hdl_output_filename = $self->convert_to_newHDLformat($input_filename,$hdl_output_filename);
1618	}
1619
1620	#just for checking copy all other file from the base dir to tmp dir if it is not exists
1621	opendir(DIR,$base_dirname) or die "Can't open base directory : $base_dirname!";
1622	my @files = grep {!/^\.+$/} readdir(DIR);
1623	close(DIR);
1624
1625	foreach my $file (@files)
1626	{
1627	my $src_file = &util::filename_cat($base_dirname,$file);
1628	my $dest_file = &util::filename_cat($tmp_dirname,$file);
1629	if ((!-e $dest_file) && (!-d $src_file))
1630	{
1631	# just copy the original file back to the tmp directory
1632	copy($src_file,$dest_file) or die "Can't copy file $src_file to $dest_file $!";
1633	}
1634	}
1635
1636	# tidy the input file
1637	my $tidy_output_filename = $hdl_output_filename;
1638	if ($self->{'use_realistic_book'})
1639	{
1640	$tidy_output_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
1641	$tidy_output_filename = $self->tmp_tidy_file($hdl_output_filename,$tidy_output_filename);
1642	}
1643	$tmp_filename = $tidy_output_filename;
1644	}
1645	else
1646	{
1647	if (!-e $tmp_filename)
1648	{
1649	# just copy the original file back to the tmp directory
1650	copy($input_filename,$tmp_filename) or die "Can't copy file $input_filename to $tmp_filename $!";
1651	}
1652	}
1653
1654	return $tmp_filename;
1655	}
1656
1657
1658	# Will make the html input file as a proper XML file with removed font tag and
1659	# image size added to the img tag.
1660	# The tidying process takes place in a collection specific 'tmp' directory so
1661	# that we don't accidentally damage the input.
1662	sub tmp_tidy_file
1663	{
1664	my $self = shift (@_);
1665	my ($file,$cnfile) = @_;
1666	my $input_filename = $file;
1667	my $tmp_filename = $cnfile;
1668
1669	# get the input filename
1670	my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
1671
1672	require HTML::TokeParser::Simple;
1673
1674	# create HTML parser to decode the input file
1675	my $parser = HTML::TokeParser::Simple->new($input_filename);
1676
1677	# write HTML tmp file without the font tag and image size are added to the img tag
1678	open (PROD, ">$tmp_filename") \|\| die("Error Writing to File: $tmp_filename $!");
1679	while (my $token = $parser->get_token())
1680	{
1681	# is it an img tag
1682	if ($token->is_start_tag('img'))
1683	{
1684	# get the attributes
1685	my $attr = $token->return_attr;
1686
1687	# get the full path to the image
1688	my $img_file = &util::filename_cat($dirname,$attr->{src});
1689
1690	# set the width and height attribute
1691	($attr->{width}, $attr->{height}) = imgsize($img_file);
1692
1693	# recreate the tag
1694	print PROD "<img";
1695	print PROD map { qq { $_="$attr->{$_}"} } keys %$attr;
1696	print PROD ">";
1697	}
1698	# is it a font tag
1699	else
1700	{
1701	if (($token->is_start_tag('font')) \|\| ($token->is_end_tag('font')))
1702	{
1703	# remove font tag
1704	print PROD "";
1705	}
1706	else
1707	{
1708	# print without changes
1709	print PROD $token->as_is;
1710	}
1711	}
1712	}
1713	close (PROD) \|\| die("Error Closing File: $tmp_filename $!");
1714
1715	# run html-tidy on the tmp file to make it a proper XML file
1716
1717	my $outhandle = $self->{'outhandle'};
1718	print $outhandle "Converting HTML to be XML compliant:\n";
1719
1720	my $tidy_cmd = "tidy";
1721	$tidy_cmd .= " -q" if ($self->{'verbosity'} <= 2);
1722	$tidy_cmd .= " -raw -wrap 0 -asxml \"$tmp_filename\"";
1723	if ($self->{'verbosity'} <= 2) {
1724	if ($ENV{'GSDLOS'} =~ m/^windows/i) {
1725	$tidy_cmd .= " 2>nul";
1726	}
1727	else {
1728	$tidy_cmd .= " 2>/dev/null";
1729	}
1730	print $outhandle " => $tidy_cmd\n";
1731	}
1732
1733	my $tidyfile = `$tidy_cmd`;
1734
1735	# write result back to the tmp file
1736	open (PROD, ">$tmp_filename") \|\| die("Error Writing to File: $tmp_filename $!");
1737	print PROD $tidyfile;
1738	close (PROD) \|\| die("Error Closing File: $tmp_filename $!");
1739
1740	# return the output filename
1741	return $tmp_filename;
1742	}
1743
1744	sub associate_cover_image
1745	{
1746	my $self = shift(@_);
1747	my ($doc_obj, $filename) = @_;
1748	if (($self->{'use_realistic_book'}) \|\| ($self->{'old_style_HDL'}))
1749	{
1750	# we will have cover image in tidytmp, but want it from import
1751	$filename =~ s/([\\\/])tidytmp([\\\/])/$1import$2/;
1752	}
1753	$self->SUPER::associate_cover_image($doc_obj, $filename);
1754	}
1755
1756
1757	1;

Note: See TracBrowser for help on using the repository browser.

Download in other formats: