root/main/trunk/greenstone2/perllib/plugins/HTMLPlugin.pm @ 30022

Revision 30022, 66.8 KB (checked in by ak19, 5 years ago)

Finally committing Dr Bainbridge's suggested fix (tested) to handle the updated Tudor collection when downloaded from the web. The HTML files link to themselves with the <link rel> tag, in which case the HTML files were blocked from being processed and 0 docs were processed. Now HTMLPlugin tests whether any of the linked associated files found are to the document itself, and in such cases, they are not blocked.

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
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 Encode;
39use Unicode::Normalize 'normalize';
40
41use ReadTextFile;
42use HBPlugin;
43use ghtml;
44use unicode;
45use util;
46use FileUtils;
47use XMLParser;
48
49use File::Copy;
50
51sub BEGIN {
52    @HTMLPlugin::ISA = ('ReadTextFile', 'HBPlugin');
53    die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
54    unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan"); # for Image/Size.pm
55}
56
57use Image::Size;
58
59use strict; # every perl program should have this!
60no strict 'refs'; # make an exception so we can use variables as filehandles
61
62my $arguments =
63    [ { 'name' => "process_exp",
64    'desc' => "{BasePlugin.process_exp}",
65    'type' => "regexp",
66    'deft' =>  &get_default_process_exp() },
67      { 'name' => "block_exp",
68    'desc' => "{BasePlugin.block_exp}",
69    'type' => 'regexp',
70    'deft' =>  &get_default_block_exp() },
71      { 'name' => "nolinks",
72    'desc' => "{HTMLPlugin.nolinks}",
73    'type' => "flag" },
74      { 'name' => "keep_head",
75    'desc' => "{HTMLPlugin.keep_head}",
76    'type' => "flag" },
77      { 'name' => "no_metadata",
78    'desc' => "{HTMLPlugin.no_metadata}",
79    'type' => "flag" },
80      { 'name' => "metadata_fields",
81    'desc' => "{HTMLPlugin.metadata_fields}",
82    'type' => "string",
83    'deft' => "Title" },
84      { 'name' => "metadata_field_separator",
85    'desc' => "{HTMLPlugin.metadata_field_separator}",
86    'type' => "string",
87    'deft' => "" },
88      { 'name' => "hunt_creator_metadata",
89    'desc' => "{HTMLPlugin.hunt_creator_metadata}",
90    'type' => "flag" },
91      { 'name' => "file_is_url",
92    'desc' => "{HTMLPlugin.file_is_url}",
93    'type' => "flag" },
94      { 'name' => "assoc_files",
95    'desc' => "{HTMLPlugin.assoc_files}",
96    'type' => "regexp",
97    'deft' => &get_default_block_exp() },
98      { 'name' => "rename_assoc_files",
99    'desc' => "{HTMLPlugin.rename_assoc_files}",
100    'type' => "flag" },
101      { 'name' => "title_sub",
102    'desc' => "{HTMLPlugin.title_sub}",
103    'type' => "string",
104    'deft' => "" },
105      { 'name' => "description_tags",
106    'desc' => "{HTMLPlugin.description_tags}",
107    'type' => "flag" },
108      # retain this for backward compatibility (w3mir option was replaced by
109      # file_is_url)
110      { 'name' => "w3mir",
111#   'desc' => "{HTMLPlugin.w3mir}",
112    'type' => "flag",
113    'hiddengli' => "yes"},
114      { 'name' => "no_strip_metadata_html",
115    'desc' => "{HTMLPlugin.no_strip_metadata_html}",
116    'type' => "string",
117    'deft' => "",
118    'reqd' => "no"},
119      { 'name' => "sectionalise_using_h_tags",
120    'desc' => "{HTMLPlugin.sectionalise_using_h_tags}",
121    'type' => "flag" },
122      { 'name' => "use_realistic_book",
123        'desc' => "{HTMLPlugin.tidy_html}",
124    'type' => "flag"},
125      { 'name' => "old_style_HDL",
126        'desc' => "{HTMLPlugin.old_style_HDL}",
127    'type' => "flag"},
128      {'name' => "processing_tmp_files",
129       'desc' => "{BasePlugin.processing_tmp_files}",
130       'type' => "flag",
131       'hiddengli' => "yes"}
132      ];
133
134my $options = { 'name'     => "HTMLPlugin",
135        'desc'     => "{HTMLPlugin.desc}",
136        'abstract' => "no",
137        'inherits' => "yes",
138        'args'     => $arguments };
139
140
141sub new {
142    my ($class) = shift (@_);
143    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
144    push(@$pluginlist, $class);
145   
146    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
147    push(@{$hashArgOptLists->{"OptList"}},$options);
148   
149
150    my $self = new ReadTextFile($pluginlist,$inputargs,$hashArgOptLists);
151   
152    if ($self->{'w3mir'}) {
153    $self->{'file_is_url'} = 1;
154    }
155    $self->{'aux_files'} = {};
156    $self->{'dir_num'} = 0;
157    $self->{'file_num'} = 0;
158   
159    return bless $self, $class;
160}
161
162# may want to use (?i)\.(gif|jpe?g|jpe|png|css|js(?:@.*)?)$
163# if have eg <script language="javascript" src="img/lib.js@123">
164# blocking is now done by reading through the file and recording all the
165# images and other files
166sub get_default_block_exp {
167    my $self = shift (@_);
168   
169    #return q^(?i)\.(gif|jpe?g|jpe|jpg|png|css)$^;
170    return "";
171}
172
173sub get_default_process_exp {
174    my $self = shift (@_);
175   
176    # the last option is an attempt to encode the concept of an html query ...
177    return q^(?i)(\.html?|\.shtml|\.shm|\.asp|\.php\d?|\.cgi|.+\?.+=.*)$^;
178}
179
180sub store_block_files
181{
182    my $self =shift (@_);
183    my ($filename_full_path, $block_hash) = @_;
184
185    my $html_fname = $filename_full_path;
186   
187    my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
188    $self->{'store_content_encoding'}->{$filename_full_path} = $content_encoding;
189
190    # read in file ($text will be in utf8)
191    my $raw_text = "";
192    $self->read_file_no_decoding($filename_full_path, \$raw_text);
193
194    my $textref = \$raw_text;
195    my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
196    my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
197    $$textref =~ s/$opencom(.*?)$closecom//gs;
198
199    # Convert entities to their UTF8 equivalents
200    $$textref =~ s/&(lt|gt|amp|quot|nbsp);/&z$1;/go;
201    $$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1,0)/gseo; # on this occassion, want it left as utf8
202    $$textref =~ s/&z(lt|gt|amp|quot|nbsp);/&$1;/go;
203
204    my $attval = "\\\"[^\\\"]+\\\"|[^\\s>]+";
205    my @img_matches = ($$textref =~ m/<img[^>]*?src\s*=\s*($attval)[^>]*>/igs);
206    my @usemap_matches = ($$textref =~ m/<img[^>]*?usemap\s*=\s*($attval)[^>]*>/igs);
207    my @link_matches = ($$textref =~ m/<link[^>]*?href\s*=\s*($attval)[^>]*>/igs);
208    my @embed_matches = ($$textref =~ m/<embed[^>]*?src\s*=\s*($attval)[^>]*>/igs);
209    my @tabbg_matches = ($$textref =~ m/<(?:body|table|tr|td)[^>]*?background\s*=\s*($attval)[^>]*>/igs);
210    my @script_matches = ($$textref =~ m/<script[^>]*?src\s*=\s*($attval)[^>]*>/igs);
211
212    if(!defined $self->{'unicode_to_original_filename'}) {
213    # maps from utf8 converted link name -> original filename referrred to by (possibly URL-encoded) src url
214    $self->{'unicode_to_original_filename'} = {};
215    }
216
217    foreach my $raw_link (@img_matches, @usemap_matches, @link_matches, @embed_matches, @tabbg_matches, @script_matches) {
218
219    # remove quotes from link at start and end if necessary
220    if ($raw_link =~ m/^\"/) {
221        $raw_link =~ s/^\"//;
222        $raw_link =~ s/\"$//;
223    }
224
225    # remove any anchor names, e.g. foo.html#name becomes foo.html
226    # but watch out for any #'s that are part of entities, such as &#x3B1;
227    $raw_link =~ s/([^&])\#.*$/$1/s;
228
229    # some links may just be anchor names
230    next unless ($raw_link =~ /\S+/);
231
232    if ($raw_link !~ m@^/@ && $raw_link !~ m/^([A-Z]:?)\\/i) {
233        # Turn relative file path into full path
234        my $dirname = &File::Basename::dirname($filename_full_path);
235        $raw_link = &FileUtils::filenameConcatenate($dirname, $raw_link);
236    }
237    $raw_link = $self->eval_dir_dots($raw_link);
238
239    # this is the actual filename on the filesystem (that the link refers to)
240    my $url_original_filename = $self->opt_url_decode($raw_link);
241
242    my ($uses_bytecodes,$exceeds_bytecodes) = &unicode::analyze_raw_string($url_original_filename);
243
244    if ($exceeds_bytecodes) {
245        # We have a link to a file name that is more complicated than a raw byte filename
246        # What we do next depends on the operating system we are on
247
248        if ($ENV{'GSDLOS'} =~ /^(linux|solaris)$/i) {
249        # Assume we're dealing with a UTF-8 encoded filename
250        $url_original_filename = encode("utf8", $url_original_filename);
251        }
252        elsif ($ENV{'GSDLOS'} =~ /^darwin$/i) {
253        # HFS+ is UTF8 with decompostion
254        $url_original_filename = encode("utf8", $url_original_filename);
255        $url_original_filename = normalize('D', $url_original_filename); # Normalization Form D (decomposition)     
256        }
257        elsif ($ENV{'GSDLOS'} =~ /^windows$/i) {
258        # Don't need to do anything as later code maps Windows
259        # unicode filenames to DOS short filenames when needed     
260        }
261        else {
262        my $outhandle = $self->{'outhandle'};
263        print $outhandle "Warning: Unrecognized operating system ", $ENV{'GSDLOS'}, "\n";
264        print $outhandle "         in raw file system encoding of: $raw_link\n";
265        print $outhandle "         Assuming filesystem is UTF-8 based.\n";
266        $url_original_filename = encode("utf8", $url_original_filename);
267        }
268    }
269
270    # Convert the (currently raw) link into its Unicode version.
271    # Store the Unicode link along with the url_original_filename
272    my $unicode_url_original_filename = "";
273    $self->decode_text($raw_link,$content_encoding,$language,\$unicode_url_original_filename);
274
275
276    $self->{'unicode_to_original_filename'}->{$unicode_url_original_filename} = $url_original_filename;
277
278
279    if ($url_original_filename ne $unicode_url_original_filename) {
280        my $outhandle = $self->{'outhandle'};
281       
282        print $outhandle "URL Encoding $url_original_filename\n";
283        print $outhandle " ->$unicode_url_original_filename\n";
284
285        # make sure not to block the file itself, as happens when an html file links to itself
286        # e.g. if the current file is mary-boleyn/index.html and contains <link rel="canonical" href="index.html" />
287        my $unicode_html_fname = "";
288        $self->decode_text($html_fname,$content_encoding,$language,\$unicode_html_fname);       
289        if($unicode_url_original_filename ne $unicode_html_fname) {
290            # Allow for possibility of raw byte version and Unicode versions of file
291            &util::block_filename($block_hash,$unicode_url_original_filename);
292        }
293    }
294
295    # $url_original_filename = &util::upgrade_if_dos_filename($url_original_filename);
296    &util::block_filename($block_hash,$url_original_filename) if $url_original_filename ne $html_fname;
297
298            # but only add the linked file to the blocklist if the current html file does not link to itself
299       
300    }
301}
302
303# Given a filename in any encoding, will URL decode it to get back the original filename
304# in the original encoding. Because this method is intended to work out the *original*
305# filename*, it does not URL decode any filename if a file by the name of the *URL-encoded*
306# string already exists in the local folder.
307#
308sub opt_url_decode {
309    my $self = shift (@_);
310    my ($raw_link) = @_;
311
312
313    # Replace %XX's in URL with decoded value if required.
314    # Note that the filename may include the %XX in some situations
315
316##    if ($raw_link =~ m/\%[A-F0-9]{2}/i) {
317
318    if (($raw_link =~ m/\%[A-F0-9]{2}/i) || ($raw_link =~ m/\&\#x[0-9A-F]+;/i) || ($raw_link =~ m/\&\#[0-9]+;/i)) {
319    if (!-e $raw_link) {
320        $raw_link = &unicode::url_decode($raw_link,1);
321    }
322    }
323   
324    return $raw_link;
325}
326
327sub read_into_doc_obj
328{
329    my $self = shift (@_); 
330    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
331       
332    my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
333
334    # Lookup content_encoding worked out in file_block pass for this file
335    # Store it under the local name 'content_encoding' so its nice and
336    # easy to access
337    $self->{'content_encoding'} = $self->{'store_content_encoding'}->{$filename_full_path};
338
339    # get the input file
340    my $input_filename = $file;
341    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
342    $suffix = lc($suffix);   
343    my $tidy_filename;
344    if (($self->{'use_realistic_book'}) || ($self->{'old_style_HDL'}))
345    {
346    # because the document has to be sectionalized set the description tags
347    $self->{'description_tags'} = 1;
348   
349    # set the file to be tidied
350    $input_filename = &FileUtils::filenameConcatenate($base_dir,$file) if $base_dir =~ m/\w/;
351   
352    # get the tidied file
353    #my $tidy_filename = $self->tmp_tidy_file($input_filename);
354    $tidy_filename = $self->convert_tidy_or_oldHDL_file($input_filename);
355   
356    # derive tmp filename from input filename
357    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($tidy_filename, "\\.[^\\.]+\$");
358   
359    # set the new input file and base_dir to be from the tidied file
360    $file = "$tailname$suffix";
361    $base_dir = $dirname;
362    }
363   
364    # call the parent read_into_doc_obj
365    my ($process_status,$doc_obj) = $self->SUPER::read_into_doc_obj($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli);
366    if (($self->{'use_realistic_book'}) || ($self->{'old_style_HDL'}))
367    {
368    # now we need to reset the filenames in the doc obj so that the converted filenames are not used
369    my $collect_file = &util::filename_within_collection($filename_full_path);
370    $doc_obj->set_source_filename ($collect_file, $self->{'file_rename_method'});
371    ## set_source_filename does not set the doc_obj source_path which is used in archives dbs for incremental
372    # build. So set it manually.
373    $doc_obj->set_source_path($filename_full_path);
374    my $collect_conv_file = &util::filename_within_collection($tidy_filename);
375    $doc_obj->set_converted_filename($collect_conv_file);
376
377    my $plugin_filename_encoding = $self->{'filename_encoding'};
378    my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
379    $self->set_Source_metadata($doc_obj, $filename_full_path, $filename_encoding);
380    }
381
382    delete $self->{'store_content_encoding'}->{$filename_full_path};
383    $self->{'content_encoding'} = undef;
384
385    return ($process_status,$doc_obj);
386}
387
388# do plugin specific processing of doc_obj
389sub process {
390    my $self = shift (@_);
391    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
392    my $outhandle = $self->{'outhandle'};
393
394    if ($ENV{'GSDLOS'} =~ m/^windows/i) {
395    # this makes life so much easier... perl can cope with unix-style '/'s.
396    $base_dir =~ s@(\\)+@/@g;
397    $file =~ s@(\\)+@/@g;     
398    }
399
400    my $filename = &FileUtils::filenameConcatenate($base_dir,$file);
401    my $upgraded_base_dir = &util::upgrade_if_dos_filename($base_dir);
402    my $upgraded_filename = &util::upgrade_if_dos_filename($filename);
403
404    if ($ENV{'GSDLOS'} =~ m/^windows/i) {
405    # And again
406    $upgraded_base_dir =~ s@(\\)+@/@g; 
407    $upgraded_filename =~ s@(\\)+@/@g;
408   
409    # Need to make sure there is a '/' on the end of upgraded_base_dir
410    if (($upgraded_base_dir ne "") && ($upgraded_base_dir !~ m/\/$/)) {
411        $upgraded_base_dir .= "/";
412    }
413    }
414    my $upgraded_file = &util::filename_within_directory($upgraded_filename,$upgraded_base_dir);
415   
416    # reset per-doc stuff...
417    $self->{'aux_files'} = {};
418    $self->{'dir_num'} = 0;
419    $self->{'file_num'} = 0;
420
421    # process an HTML file where sections are divided by headings tags (H1, H2 ...)
422    # you can also include metadata in the format (X can be any number)
423    # <hX>Title<!--gsdl-metadata
424    #   <Metadata name="name1">value1</Metadata>
425    #   ...
426    #   <Metadata name="nameN">valueN</Metadata>
427    #--></hX>
428    if ($self->{'sectionalise_using_h_tags'}) {
429    # description_tags should allways be activated because we convert headings to description tags
430    $self->{'description_tags'} = 1;
431
432    my $arrSections = [];
433    $$textref =~ s/<h([0-9]+)[^>]*>(.*?)<\/h[0-9]+>/$self->process_heading($1, $2, $arrSections, $upgraded_file)/isge;
434
435    if (scalar(@$arrSections)) {
436        my $strMetadata = $self->update_section_data($arrSections, -1);
437        if (length($strMetadata)) {
438        $strMetadata = '<!--' . $strMetadata . "\n-->\n</body>";
439        $$textref =~ s/<\/body>/$strMetadata/ig;
440        }
441    }
442    }
443
444    my $cursection = $doc_obj->get_top_section();
445
446    $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection)
447    unless $self->{'no_metadata'} || $self->{'description_tags'};
448
449    # Store URL for page as metadata - this can be used for an
450    # altavista style search interface. The URL won't be valid
451    # unless the file structure contains the domain name (i.e.
452    # like when w3mir is used to download a website).
453
454    # URL metadata (even invalid ones) are used to support internal
455    # links, so even if 'file_is_url' is off, still need to store info
456
457    my ($tailname,$dirname) = &File::Basename::fileparse($upgraded_file);
458
459#    my $utf8_file = $self->filename_to_utf8_metadata($file);
460#    $utf8_file =~ s/&\#095;/_/g;
461#    variable below used to be utf8_file
462
463    my $url_encoded_file = &unicode::raw_filename_to_url_encoded($tailname);
464    my $utf8_url_encoded_file = &unicode::raw_filename_to_utf8_url_encoded($tailname);
465
466    my $web_url = "http://";
467    my $utf8_web_url = "http://";
468    if(defined $dirname) { # local directory
469        # Check for "ftp" in the domain name of the directory
470        #  structure to determine if this URL should be a ftp:// URL
471        # This check is not infallible, but better than omitting the
472        #  check, which would cause all files downloaded from ftp sites
473        #  via mirroring with wget to have potentially erroneous http:// URLs
474        #  assigned in their metadata
475        if ($dirname =~ /^[^\/]*ftp/i)
476    {
477      $web_url = "ftp://";
478      $utf8_web_url = "ftp://";
479    }
480    $dirname = $self->eval_dir_dots($dirname);
481    $dirname .= &util::get_dirsep() if $dirname ne ""; # if there's a directory, it should end on "/"
482
483    $web_url = $web_url.$dirname.$url_encoded_file;
484    $utf8_web_url = $utf8_web_url.$dirname.$utf8_url_encoded_file;
485    } else {
486    $web_url = $web_url.$url_encoded_file;
487    $utf8_web_url = $utf8_web_url.$utf8_url_encoded_file;
488    }
489    $web_url =~ s/\\/\//g;
490    $utf8_web_url =~ s/\\/\//g;
491
492    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
493    print STDERR "*******DEBUG: upgraded_file:       $upgraded_file\n";
494    print STDERR "*******DEBUG: adding URL metadata: $utf8_url_encoded_file\n";
495    }
496
497
498    $doc_obj->add_utf8_metadata($cursection, "URL", $web_url);
499    $doc_obj->add_utf8_metadata($cursection, "UTF8URL", $utf8_web_url);
500
501    if ($self->{'file_is_url'}) {
502    $doc_obj->add_metadata($cursection, "weblink", "<a href=\"$web_url\">");
503    $doc_obj->add_metadata($cursection, "webicon", "_iconworld_");
504    $doc_obj->add_metadata($cursection, "/weblink", "</a>");
505    }
506
507    if ($self->{'description_tags'}) {
508    # remove the html header - note that doing this here means any
509    # sections defined within the header will be lost (so all <Section>
510    # tags must appear within the body of the HTML)
511    my ($head_keep) = ($$textref =~ m/^(.*?)<body[^>]*>/is);
512
513    $$textref =~ s/^.*?<body[^>]*>//is;
514    $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
515
516    my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
517    my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
518
519    my $lt = '(?:<|&lt;)';
520    my $gt = '(?:>|&gt;)';
521    my $quot = '(?:"|&quot;|&rdquo;|&ldquo;)';
522
523    my $dont_strip = '';
524    if ($self->{'no_strip_metadata_html'}) {
525        ($dont_strip = $self->{'no_strip_metadata_html'}) =~ s{,}{|}g;
526    }
527
528    my $found_something = 0; my $top = 1;
529    while ($$textref =~ s/^(.*?)$opencom(.*?)$closecom//s) {
530        my $text = $1;
531        my $comment = $2;
532        if (defined $text) {
533        # text before a comment - note that getting to here
534        # doesn't necessarily mean there are Section tags in
535        # the document
536        $self->process_section(\$text, $upgraded_base_dir, $upgraded_file, $doc_obj, $cursection);
537        }
538        while ($comment =~ s/$lt(.*?)$gt//s) {
539        my $tag = $1;
540        if ($tag eq "Section") {
541            $found_something = 1;
542            $cursection = $doc_obj->insert_section($doc_obj->get_end_child($cursection)) unless $top;
543            $top = 0;
544        } elsif ($tag eq "/Section") {
545            $found_something = 1;
546            $cursection = $doc_obj->get_parent_section ($cursection);
547        } elsif ($tag =~ m/^Metadata name=$quot(.*?)$quot/s) {
548            my $metaname = $1;
549            my $accumulate = $tag =~ m/mode=${quot}accumulate${quot}/ ? 1 : 0;
550            $comment =~ s/^(.*?)$lt\/Metadata$gt//s;
551            my $metavalue = $1;
552            $metavalue =~ s/^\s+//;
553            $metavalue =~ s/\s+$//;
554                    # assume that no metadata value intentionally includes
555                    # carriage returns or HTML tags (if they're there they
556                    # were probably introduced when converting to HTML from
557                    # some other format).
558            # actually some people want to have html tags in their
559            # metadata.
560            $metavalue =~ s/[\cJ\cM]/ /sg;
561            $metavalue =~ s/<[^>]+>//sg
562            unless $dont_strip && ($dont_strip eq 'all' || $metaname =~ m/^($dont_strip)$/);
563            $metavalue =~ s/\s+/ /sg;
564            if ($metaname =~ /\./) { # has a namespace
565            $metaname = "ex.$metaname";
566            }
567            if ($accumulate) {
568            $doc_obj->add_utf8_metadata($cursection, $metaname, $metavalue);
569            } else {
570            $doc_obj->set_utf8_metadata_element($cursection, $metaname, $metavalue);   
571            }
572        } elsif ($tag eq "Description" || $tag eq "/Description") {
573            # do nothing with containing Description tags
574        } else {
575            # simple HTML tag (probably created by the conversion
576            # to HTML from some other format) - we'll ignore it and
577            # hope for the best ;-)
578        }
579        }
580    }
581    if ($cursection ne "") {
582        print $outhandle "HTMLPlugin: WARNING: $upgraded_file contains unmatched <Section></Section> tags\n";
583    }
584
585    $$textref =~ s/^.*?<body[^>]*>//is;
586    $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
587    if ($$textref =~ m/\S/) {
588        if (!$found_something) {
589        if ($self->{'verbosity'} > 2) {
590            print $outhandle "HTMLPlugin: WARNING: $upgraded_file appears to contain no Section tags so\n";
591            print $outhandle "          will be processed as a single section document\n";
592        }
593
594        # go ahead and process single-section document
595        $self->process_section($textref, $upgraded_base_dir, $upgraded_file, $doc_obj, $cursection);
596
597        # if document contains no Section tags we'll go ahead
598        # and extract metadata (this won't have been done
599        # above as the -description_tags option prevents it)
600        my $complete_text = $head_keep.$doc_obj->get_text($cursection);
601        $self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
602            unless $self->{'no_metadata'};
603
604        } else {
605        print $outhandle "HTMLPlugin: WARNING: $upgraded_file contains the following text outside\n";
606        print $outhandle "          of the final closing </Section> tag. This text will\n";
607        print $outhandle "          be ignored.";
608
609        my ($text);
610        if (length($$textref) > 30) {
611            $text = substr($$textref, 0, 30) . "...";
612        } else {
613            $text = $$textref;
614        }
615        $text =~ s/\n/ /isg;
616        print $outhandle " ($text)\n";
617        }
618    } elsif (!$found_something) {
619
620        if ($self->{'verbosity'} > 2) {
621        # may get to here if document contained no valid Section
622        # tags but did contain some comments. The text will have
623        # been processed already but we should print the warning
624        # as above and extract metadata
625        print $outhandle "HTMLPlugin: WARNING: $upgraded_file appears to contain no Section tags and\n";
626        print $outhandle "          is blank or empty.  Metadata will be assigned if present.\n";
627        }
628
629        my $complete_text = $head_keep.$doc_obj->get_text($cursection);
630        $self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
631        unless $self->{'no_metadata'};
632    }
633   
634    } else {
635
636    # remove header and footer
637    if (!$self->{'keep_head'} || $self->{'description_tags'}) {
638        $$textref =~ s/^.*?<body[^>]*>//is;
639        $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
640    }
641
642    $self->{'css_assoc_files'} = {};
643   
644    # single section document
645    $self->process_section($textref, $upgraded_base_dir, $upgraded_file, $doc_obj, $cursection);
646   
647    #my $upgraded_filename_dirname = &File::Basename::dirname($upgraded_filename);
648       
649    $self->acquire_css_associated_files($doc_obj, $cursection);
650   
651    $self->{'css_assoc_files'} = {};
652    }
653
654    return 1;
655}
656
657
658sub process_heading
659{
660    my ($self, $nHeadNo, $strHeadingText, $arrSections, $file) = @_;
661    $strHeadingText = '' if (!defined($strHeadingText));
662
663    my $strMetadata = $self->update_section_data($arrSections, int($nHeadNo));
664
665    my $strSecMetadata = '';
666    while ($strHeadingText =~ s/<!--gsdl-metadata(.*?)-->//is)
667    {
668    $strSecMetadata .= $1;
669    }
670
671    $strHeadingText =~ s/^\s+//g;
672    $strHeadingText =~ s/\s+$//g;
673    $strSecMetadata =~ s/^\s+//g;
674    $strSecMetadata =~ s/\s+$//g;
675
676    $strMetadata .= "\n<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">" . $strHeadingText . "</Metadata>\n";
677
678    if (length($strSecMetadata)) {
679    $strMetadata .= "\t\t" . $strSecMetadata . "\n";
680    }
681
682    $strMetadata .= "\t</Description>\n";
683
684    return "<!--" . $strMetadata . "-->";
685}
686
687
688sub update_section_data
689{
690    my ($self, $arrSections, $nCurTocNo) = @_;
691    my ($strBuffer, $nLast, $nSections) = ('', 0, scalar(@$arrSections));
692
693    if ($nSections == 0) {
694    push @$arrSections, $nCurTocNo;
695    return $strBuffer;
696    }
697    $nLast = $arrSections->[$nSections - 1];
698    if ($nCurTocNo > $nLast) {
699    push @$arrSections, $nCurTocNo;
700    return $strBuffer;
701    }
702    for(my $i = $nSections - 1; $i >= 0; $i--) {
703    if ($nCurTocNo <= $arrSections->[$i]) {
704        $strBuffer .= "\n</Section>";
705        pop @$arrSections;
706    }
707    }
708    push @$arrSections, $nCurTocNo;
709    return $strBuffer;
710}
711
712
713# note that process_section may be called multiple times for a single
714# section (relying on the fact that add_utf8_text appends the text to any
715# that may exist already).
716sub process_section {
717    my $self = shift (@_);
718    my ($textref, $base_dir, $file, $doc_obj, $cursection) = @_;
719
720    my @styleTagsText = ($$textref =~ m/<style[^>]*>([^<]*)<\/style>/sg);
721    if(scalar(@styleTagsText) > 0)
722    {
723        my $css_filename_dirname = &File::Basename::dirname(&FileUtils::filenameConcatenate($base_dir, $file));
724        foreach my $styleText (@styleTagsText)
725        {
726            $self->acquire_css_associated_files_from_text_block($styleText, $css_filename_dirname);
727        }
728    }
729
730    # trap links
731    if (!$self->{'nolinks'}) {
732    # usemap="./#index" not handled correctly => change to "#index"
733##  $$textref =~ s/(<img[^>]*?usemap\s*=\s*[\"\']?)([^\"\'>\s]+)([\"\']?[^>]*>)/
734
735##  my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
736##  my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
737
738    $$textref =~ s/(<img[^>]*?usemap\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
739        $self->replace_usemap_links($1, $2, $3)/isge;
740
741    $$textref =~ s/(<(?:a|area|frame|link|script)\s+[^>]*?\s*(?:href|src)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
742        $self->replace_href_links ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
743   
744##  $$textref =~ s/($opencom.*?)?+(<(?:a|area|frame|link|script)\s+[^>]*?\s*(?:href|src)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)(.*?$closecom)?+/
745#       $self->replace_href_links ($1, $2, $3, $4, $5, $base_dir, $file, $doc_obj, $cursection)/isge;
746    }
747
748    # trap images
749
750    # Previously, by default, HTMLPlugin would embed <img> tags inside anchor tags
751    # i.e. <a href="image><img src="image"></a> in order to overcome a problem that
752    # turned regular text succeeding images into links. That is, by embedding <imgs>
753    # inside <a href=""></a>, the text following images were no longer misbehaving.
754    # However, there would be many occasions whereby images were not meant to link
755    # to their source images but where the images would link to another web page.
756    # To allow this, the no_image_links option was introduced: it would prevent
757    # the behaviour of embedding images into links that referenced the source images.
758
759    # Somewhere along the line, the problem of normal text turning into links when
760    # such text followed images which were not embedded in <a href=""></a> ceased
761    # to occur. This is why the following lines have been commented out (as well as
762    # two lines in replace_images). They appear to no longer apply.
763
764    # If at any time, there is a need for having images embedded in <a> anchor tags,
765    # then it might be better to turn that into an HTMLPlugin option rather than make
766    # it the default behaviour. Also, eventually, no_image_links needs to become
767    # a deprecated option for HTMLPlugin as it has now become the default behaviour.
768
769    #if(!$self->{'no_image_links'}){
770    $$textref =~ s/(<(?:img|embed|table|tr|td)[^>]*?(?:src|background)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
771    $self->replace_images ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
772    #}
773
774    # add text to document object
775    # turn \ into \\ so that the rest of greenstone doesn't think there
776    # is an escape code following. (Macro parsing loses them...)
777    $$textref =~ s/\\/\\\\/go;
778   
779    $doc_obj->add_utf8_text($cursection, $$textref);
780}
781
782sub replace_images {
783    my $self = shift (@_);
784    my ($front, $link, $back, $base_dir,
785    $file, $doc_obj, $section) = @_;
786
787    # remove quotes from link at start and end if necessary
788    if ($link=~/^[\"\']/) {
789    $link=~s/^[\"\']//;
790    $link=~s/[\"\']$//;
791    $front.='"';
792    $back="\"$back";
793    }
794
795    $link =~ s/\n/ /g;
796
797    # Hack to overcome Windows wv 0.7.1 bug that causes embedded images to be broken
798    # If the Word file path has spaces in it, wv messes up and you end up with
799    #   absolute paths for the images, and without the "file://" prefix
800    # So check for this special case and massage the data to be correct
801    if ($ENV{'GSDLOS'} =~ m/^windows/i && $self->{'plugin_type'} eq "WordPlug" && $link =~ m/^[A-Za-z]\:\\/) {
802    $link =~ s/^.*\\([^\\]+)$/$1/;
803    }
804   
805    my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
806
807    my $img_file =  $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section);
808
809#    print STDERR "**** link = $link\n**** href = $href\n**** img_file = $img_file, rl = $rl\n\n";
810
811    my $anchor_name = $img_file;
812    #$anchor_name =~ s/^.*\///;
813    #$anchor_name = "<a name=\"$anchor_name\" ></a>";
814
815    my $image_link = $front . $img_file .$back;
816    return $image_link;
817
818    # The reasons for why the following two lines are no longer necessary can be
819    # found in subroutine process_section
820    #my $anchor_link = "<a href=\"$img_file\" >".$image_link."</a>"; 
821    #return $anchor_link;   
822   
823    #return $front . $img_file . $back . $anchor_name;
824}
825
826sub replace_href_links {
827    my $self = shift (@_);
828    my ($front, $link, $back, $base_dir, $file, $doc_obj, $section) = @_;
829   
830    if($front =~ m/^<link / && $link =~ m/\.css"$/)
831    {
832        my $actual_link = $link;
833        $actual_link =~ s/^"(.*)"$/$1/;
834       
835        my $directory = &File::Basename::dirname($file);
836       
837        my $css_filename = &FileUtils::filenameConcatenate($base_dir, $directory, $actual_link);   
838        $self->retrieve_css_associated_files($css_filename);
839    }
840   
841    # remove quotes from link at start and end if necessary
842    if ($link=~/^[\"\']/) {
843    $link=~s/^[\"\']//;
844    $link=~s/[\"\']$//;
845    $front.='"';
846    $back="\"$back";
847    }
848
849    # attempt to sort out targets - frames are not handled
850    # well in this plugin and some cases will screw things
851    # up - e.g. the _parent target (so we'll just remove
852    # them all ;-)
853    $front =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
854    $back =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
855    $front =~ s/target=\"?_parent\"?//is;
856    $back =~ s/target=\"?_parent\"?//is;
857
858    if($link =~ m/^\#/s)
859    {
860        return $front . "_httpsamepagelink_" . $link . $back;   
861    }
862   
863    $link =~ s/\n/ /g;
864
865    # Find file referred to by $link on file system
866    # This is more complicated than it sounds when char encodings
867    # is taken in to account
868    my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
869
870    # href may use '\'s where '/'s should be on Windows
871    $href =~ s/\\/\//g;
872    my ($filename) = $href =~ m/^(?:.*?):(?:\/\/)?(.*)/;
873
874    ##### leave all these links alone (they won't be picked up by intermediate
875    ##### pages). I think that's safest when dealing with frames, targets etc.
876    ##### (at least until I think of a better way to do it). Problems occur with
877    ##### mailto links from within small frames, the intermediate page is displayed
878    ##### within that frame and can't be seen. There is still potential for this to
879    ##### happen even with html pages - the solution seems to be to somehow tell
880    ##### the browser from the server side to display the page being sent (i.e.
881    ##### the intermediate page) in the top level window - I'm not sure if that's
882    ##### possible - the following line should probably be deleted if that can be done
883    return $front . $link . $back if $href =~ m/^(mailto|news|gopher|nntp|telnet|javascript):/is;
884
885    if (($rl == 0) || ($filename =~ m/$self->{'process_exp'}/) ||
886    ($href =~ m/\/$/) || ($href =~ m/^(mailto|news|gopher|nntp|telnet|javascript):/i)) {
887
888    if ($ENV{'GSDLOS'} =~ m/^windows$/) {
889
890        # Don't do any encoding for now, as not clear what
891        # the right thing to do is to support filename
892        # encoding on Windows when they are not UTF16
893        #
894    }
895    else {
896        # => Unix-based system
897
898        # If web page didn't give encoding, then default to utf8
899        my $content_encoding= $self->{'content_encoding'} || "utf8";
900
901        if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
902        print STDERR "**** Encoding with '$content_encoding', href: $href\n";
903        }
904
905        # on Darwin, the unicode filenames are stored on the file
906        # system in decomposed form, so any href link (including when
907        # URL-encoded) should refer to the decomposed name of the file
908        if ($ENV{'GSDLOS'} =~ /^darwin$/i) {
909        $href = normalize('D', $href); # Normalization Form D (decomposition)
910        }
911
912        $href = encode($content_encoding,$href);
913    }
914
915    $href = &unicode::raw_filename_to_utf8_url_encoded($href); 
916    $href = &unicode::filename_to_url($href);
917
918    &ghtml::urlsafe ($href);
919
920    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
921        print STDERR "******DEBUG: href=$href\n";   
922    }
923
924
925    return $front . "_httpextlink_&amp;rl=" . $rl . "&amp;href=" . $href . $hash_part . $back;
926    } else {
927    # link is to some other type of file (e.g., an image) so we'll
928    # need to associate that file
929    return $front . $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) . $back;
930    }
931}
932
933sub retrieve_css_associated_files {
934    my $self = shift (@_);
935    my ($css_filename) = @_;
936   
937    my $css_filename_dirname = &File::Basename::dirname($css_filename);
938   
939    open (CSSFILE, $css_filename) || return;
940    sysread (CSSFILE, my $file_string, -s CSSFILE);
941   
942    $self->acquire_css_associated_files_from_text_block($file_string, $css_filename_dirname) unless !defined $file_string;
943   
944    close CSSFILE;
945}
946
947sub acquire_css_associated_files_from_text_block {
948    my $self = shift (@_);
949    my ($text, $css_filename_dirname) = @_;
950   
951    my @image_urls = ($text =~ m/background-image:\s*url[^;]*;/sg);
952    foreach my $img_url (@image_urls)
953    {
954        $img_url =~ s/^.*url.*\((.*)\).*$/$1/;
955        $img_url =~ s/^\s*"?([^"]*)"?\s*$/$1/;
956
957        $self->{'css_assoc_files'}->{&FileUtils::filenameConcatenate($css_filename_dirname, $img_url)} = $img_url;
958    }
959}
960
961sub acquire_css_associated_files {
962    my $self = shift(@_);
963   
964    my ($doc_obj, $section) = @_;
965   
966    foreach my $image_filename (keys %{$self->{'css_assoc_files'}})
967    {
968        $doc_obj->associate_file($image_filename, $self->{'css_assoc_files'}->{$image_filename}, undef, $section); 
969    }
970}
971
972sub add_file {
973    my $self = shift (@_);
974    my ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) = @_;
975    my ($newname);
976
977    my $filename = $href;
978    if ($base_dir eq "") {
979    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
980        # remove http://
981        $filename =~ s/^[^:]*:\/\///;
982    }
983    else {
984        # remove http:/ thereby leaving one slash at the start as
985        # part of full pathname
986        $filename =~ s/^[^:]*:\///;
987    }
988    }
989    else {
990    # remove http://
991    $filename =~ s/^[^:]*:\/\///;
992    }
993   
994    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
995        $filename =~ s@\/@\\@g;
996    }
997   
998    $filename = &FileUtils::filenameConcatenate($base_dir, $filename);
999
1000    if (($self->{'use_realistic_book'}) || ($self->{'old_style_HDL'})) {
1001    # we are processing a tidytmp file - want paths to be in import
1002    $filename =~ s/([\\\/])tidytmp([\\\/])/$1import$2/;
1003    }
1004
1005    # Replace %XX's in URL with decoded value if required. Note that the
1006    # filename may include the %XX in some situations. If the *original*
1007    # file's name was in URL encoding, the following method will not decode
1008    # it.
1009    my $unicode_filename = $filename;
1010    my $opt_decode_unicode_filename = $self->opt_url_decode($unicode_filename);
1011
1012    # wvWare can generate <img src="StrangeNoGraphicData"> tags, but with no
1013    # (it seems) accompanying file
1014    if ($opt_decode_unicode_filename =~ m/StrangeNoGraphicData$/) { return ""; }
1015
1016    my $content_encoding= $self->{'content_encoding'} || "utf8";
1017
1018    if ($ENV{'GSDLOS'} =~ /^(linux|solaris)$/i) {
1019    # The filenames that come through the HTML file have been decoded
1020    # into Unicode aware Perl strings.  Need to convert them back
1021    # to their initial raw-byte encoding to match the file that
1022    # exists on the file system
1023    $filename = encode($content_encoding, $opt_decode_unicode_filename);
1024    }
1025    elsif ($ENV{'GSDLOS'} =~ /^darwin$/i) {
1026    # HFS+ is UTF8 with decompostion
1027    $filename = encode($content_encoding, $opt_decode_unicode_filename);
1028    $filename = normalize('D', $filename); # Normalization Form D (decomposition)
1029
1030    }
1031    elsif ($ENV{'GSDLOS'} =~ /^windows$/i) {
1032    my $long_filename = Win32::GetLongPathName($opt_decode_unicode_filename);
1033
1034    if (defined $long_filename) {
1035        my $short_filename = Win32::GetLongPathName($long_filename);
1036        $filename = $short_filename;
1037    }
1038#   else {
1039#       print STDERR "***** failed to map href to real file:\n";
1040#       print STDERR "****** $href -> $opt_decode_unicode_filename\n";
1041#   }
1042    }
1043    else {
1044    my $outhandle = $self->{'outhandle'};
1045    print $outhandle "Warning: Unrecognized operating system ", $ENV{'GSDLOS'}, "\n";
1046    print $outhandle "         in file system encoding of href: $href\n";
1047    print $outhandle "         No character encoding done.\n";
1048    }
1049
1050
1051    # some special processing if the intended filename was converted to utf8, but
1052    # the actual file still needs to be renamed
1053    #if (!&util::fd_exists($filename)) {
1054    if (!&FileUtils::fileExists($filename)) {
1055    # try the original filename stored in map
1056    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
1057        print STDERR "******!! orig filename did not exist: $filename\n";
1058    }
1059
1060##  print STDERR "**** trying to look up unicode_filename: $unicode_filename\n";
1061
1062    my $original_filename = $self->{'unicode_to_original_filename'}->{$unicode_filename};
1063
1064    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
1065        print STDERR "******   From lookup unicode_filename, now trying for: $original_filename\n";
1066    }
1067
1068    if (defined $original_filename && -e $original_filename) {
1069        if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
1070            print STDERR "******   Found match!\n";
1071        }
1072        $filename = $original_filename;
1073    }
1074    }
1075   
1076    my ($ext) = $filename =~ m/(\.[^\.]*)$/;
1077
1078    if ($rl == 0) {
1079    if ((!defined $ext) || ($ext !~ m/$self->{'assoc_files'}/)) {
1080        return "_httpextlink_&amp;rl=0&amp;el=prompt&amp;href=" . $href . $hash_part;
1081    }
1082    else {
1083        return "_httpextlink_&amp;rl=0&amp;el=direct&amp;href=" . $href . $hash_part;
1084    }
1085    }
1086
1087    if ((!defined $ext) || ($ext !~ m/$self->{'assoc_files'}/)) {
1088    return "_httpextlink_&amp;rl=" . $rl . "&amp;href=" . $href . $hash_part;
1089    }
1090    # add the original image file as a source file
1091    if (!$self->{'processing_tmp_files'} ) {
1092    $doc_obj->associate_source_file($filename);
1093    }
1094    if ($self->{'rename_assoc_files'}) {
1095    if (defined $self->{'aux_files'}->{$href}) {
1096        $newname = $self->{'aux_files'}->{$href}->{'dir_num'} . "/" .
1097        $self->{'aux_files'}->{$href}->{'file_num'} . $ext;
1098    } else {
1099        $newname = $self->{'dir_num'} . "/" . $self->{'file_num'} . $ext;
1100        $self->{'aux_files'}->{$href} = {'dir_num' => $self->{'dir_num'}, 'file_num' => $self->{'file_num'}};
1101        $self->inc_filecount ();
1102    }
1103    $doc_obj->associate_file($filename, $newname, undef, $section);
1104    return "_httpdocimg_/$newname";
1105    } else {
1106    if(&unicode::is_url_encoded($unicode_filename)) {
1107        # use the possibly-decoded filename instead to avoid double URL encoding
1108        ($newname) = $filename =~ m/([^\/\\]*)$/;
1109    } else {
1110        ($newname) = $unicode_filename =~ m/([^\/\\]*)$/;
1111    }
1112
1113    # Make sure this name uses only ASCII characters.
1114    # We use either base64 or URL encoding, as these preserve original encoding
1115    $newname = &util::rename_file($newname, $self->{'file_rename_method'});
1116
1117### print STDERR "***** associating $filename (raw-byte/utf8)-> $newname\n";
1118    $doc_obj->associate_file($filename, $newname, undef, $section);
1119
1120    # Since the generated image will be URL-encoded to avoid file-system/browser mess-ups
1121    # of filenames, URL-encode the additional percent signs of the URL-encoded filename
1122    my $newname_url = $newname;
1123    $newname_url = &unicode::filename_to_url($newname_url);
1124    return "_httpdocimg_/$newname_url";
1125    }
1126}
1127
1128
1129sub format_link {
1130    my $self = shift (@_);
1131    my ($link, $base_dir, $file) = @_;
1132
1133    # strip off hash part, e.g. #foo, but watch out for any entities, e.g. &#x3B1;
1134    my ($before_hash, $hash_part) = $link =~ m/^(.*?[^&])(\#.*)?$/;
1135   
1136    $hash_part = "" if !defined $hash_part;
1137    if (!defined $before_hash || $before_hash !~ m/[\w\.\/]/) {
1138        my $outhandle = $self->{'outhandle'};
1139        print $outhandle "HTMLPlugin: ERROR - badly formatted tag ignored ($link)\n"
1140            if $self->{'verbosity'};
1141        return ($link, "", 0);
1142    }
1143   
1144    if ($before_hash =~ s@^((?:http|https|ftp|file|mms)://)@@i) {
1145        my $type = $1;
1146        my $before_hash_file = $before_hash;
1147       
1148    if ($link =~ m/^(http|ftp):/i) {
1149       
1150        # Turn url (using /) into file name (possibly using \ on windows)
1151        my @http_dir_split = split('/', $before_hash_file);
1152        $before_hash_file = &FileUtils::filenameConcatenate(@http_dir_split);       
1153    }
1154   
1155    # want to maintain two version of "before_hash": one representing the URL, the other using filesystem specific directory separator
1156    $before_hash_file = $self->eval_dir_dots($before_hash_file);
1157    my $before_hash_url = $before_hash_file;
1158    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
1159        $before_hash_url =~ s@\\@\/@g;
1160    }
1161   
1162    my $linkfilename = &FileUtils::filenameConcatenate($base_dir, $before_hash_file);
1163   
1164    my $rl = 0;
1165    $rl = 1 if (-e $linkfilename);
1166
1167    # make sure there's a slash on the end if it's a directory
1168    if ($before_hash_url !~ m/\/$/) {
1169        $before_hash_url .= "/" if (-d $linkfilename);
1170    }
1171    return ($type . $before_hash_url, $hash_part, $rl);
1172   
1173    } elsif ($link !~ m/^(mailto|news|gopher|nntp|telnet|javascript):/i && $link !~ m/^\//) {
1174
1175    if ($before_hash =~ s@^/@@ || $before_hash =~ m/\\/) {
1176
1177        # the first directory will be the domain name if file_is_url
1178        # to generate archives, otherwise we'll assume all files are
1179        # from the same site and base_dir is the root
1180
1181        if ($self->{'file_is_url'}) {
1182        my @dirs = split /[\/\\]/, $file;
1183        my $domname = shift (@dirs);
1184        $before_hash = &FileUtils::filenameConcatenate($domname, $before_hash);
1185        $before_hash =~ s@\\@/@g; # for windows
1186        }
1187        else
1188        {
1189        # see if link shares directory with source document
1190        # => turn into relative link if this is so!
1191       
1192        if ($ENV{'GSDLOS'} =~ m/^windows/i) {
1193            # too difficult doing a pattern match with embedded '\'s...
1194            my $win_before_hash=$before_hash;
1195            $win_before_hash =~ s@(\\)+@/@g;
1196            # $base_dir is already similarly "converted" on windows.
1197            if ($win_before_hash =~ s@^$base_dir/@@o) {
1198            # if this is true, we removed a prefix
1199            $before_hash=$win_before_hash;
1200            }
1201        }
1202        else {
1203            # before_hash has lost leading slash by this point,
1204            # -> add back in prior to substitution with $base_dir
1205            $before_hash = "/$before_hash";
1206
1207            $before_hash = &FileUtils::filenameConcatenate("",$before_hash);
1208            $before_hash =~ s@^$base_dir/@@;
1209        }
1210        }
1211    } else {
1212        # Turn relative file path into full path
1213        my $dirname = &File::Basename::dirname($file);
1214        $before_hash = &FileUtils::filenameConcatenate($dirname, $before_hash);
1215        $before_hash = $self->eval_dir_dots($before_hash);   
1216        $before_hash =~ s@\\@/@g; # for windows         
1217    }
1218
1219    my $linkfilename = &FileUtils::filenameConcatenate($base_dir, $before_hash);
1220
1221
1222#   print STDERR "**** linkfilename = $linkfilename\n";
1223#   if (!&util::fd_exists($linkfilename)) {
1224#       print STDERR "***** Warning: Could not find $linkfilename\n";
1225#   }
1226
1227
1228    # make sure there's a slash on the end if it's a directory
1229    if ($before_hash !~ m/\/$/) {
1230        $before_hash .= "/" if (-d $linkfilename);
1231    }
1232
1233#   print STDERR "*** returning: $before_hash\n";
1234
1235    return ("http://" . $before_hash, $hash_part, 1);
1236    } else {
1237    # mailto, news, nntp, telnet, javascript or gopher link
1238    return ($before_hash, "", 0);
1239    }
1240}
1241
1242sub extract_first_NNNN_characters {
1243    my $self = shift (@_);
1244    my ($textref, $doc_obj, $thissection) = @_;
1245   
1246    foreach my $size (split /,/, $self->{'first'}) {
1247    my $tmptext =  $$textref;
1248    # skip to the body
1249    $tmptext =~ s/.*<body[^>]*>//i;
1250    # remove javascript
1251    $tmptext =~ s@<script.*?</script>@ @sig;
1252    $tmptext =~ s/<[^>]*>/ /g;
1253    $tmptext =~ s/&nbsp;/ /g;
1254    $tmptext =~ s/^\s+//;
1255    $tmptext =~ s/\s+$//;
1256    $tmptext =~ s/\s+/ /gs;
1257    $tmptext = &unicode::substr ($tmptext, 0, $size);
1258    $tmptext =~ s/\s\S*$/&#8230;/; # adds an ellipse (...)
1259    $doc_obj->add_utf8_metadata ($thissection, "First$size", $tmptext);
1260    }
1261}
1262
1263
1264sub extract_metadata {
1265    my $self = shift (@_);
1266    my ($textref, $metadata, $doc_obj, $section) = @_;
1267    my $outhandle = $self->{'outhandle'};
1268    # if we don't want metadata, we may as well not be here ...
1269    return if (!defined $self->{'metadata_fields'});
1270   
1271    my $separator = $self->{'metadata_field_separator'};
1272    if ($separator eq "") {
1273    undef $separator;
1274    }
1275
1276    # metadata fields to extract/save. 'key' is the (lowercase) name of the
1277    # html meta, 'value' is the metadata name for greenstone to use
1278    my %find_fields = ();
1279
1280    my %creator_fields = (); # short-cut for lookups
1281
1282
1283    foreach my $field (split /,/, $self->{'metadata_fields'}) {
1284        $field =~ s/^\s+//; # remove leading whitespace
1285        $field =~ s/\s+$//; # remove trailing whitespace
1286   
1287    # support tag<tagname>
1288    if ($field =~ m/^(.*?)\s*<(.*?)>$/) {
1289        # "$2" is the user's preferred gs metadata name
1290        $find_fields{lc($1)}=$2; # lc = lowercase
1291    } else { # no <tagname> for mapping
1292        # "$field" is the user's preferred gs metadata name
1293        $find_fields{lc($field)}=$field; # lc = lowercase
1294    }
1295    }
1296
1297    if (defined $self->{'hunt_creator_metadata'} &&
1298    $self->{'hunt_creator_metadata'} == 1 ) {
1299    my @extra_fields =
1300        (
1301         'author',
1302         'author.email',
1303         'creator',
1304         'dc.creator',
1305         'dc.creator.corporatename',
1306         );
1307
1308    # add the creator_metadata fields to search for
1309    foreach my $field (@extra_fields) {
1310        $creator_fields{$field}=0; # add to lookup hash
1311    }
1312    }
1313
1314
1315    # find the header in the html file, which has the meta tags
1316    $$textref =~ m@<head>(.*?)</head>@si;
1317
1318    my $html_header=$1;
1319
1320    # go through every <meta... tag defined in the html and see if it is
1321    # one of the tags we want to match.
1322   
1323    # special case for title - we want to remember if its been found
1324    my $found_title = 0;
1325    # this assumes that ">" won't appear. (I don't think it's allowed to...)
1326    $html_header =~ m/^/; # match the start of the string, for \G assertion
1327   
1328    while ($html_header =~ m/\G.*?<meta(.*?)>/sig) {
1329    my $metatag=$1;
1330    my ($tag, $value);
1331
1332    # find the tag name
1333    $metatag =~ m/(?:name|http-equiv)\s*=\s*([\"\'])?(.*?)\1/is;
1334    $tag=$2;
1335    # in case they're not using " or ', but they should...
1336    if (! $tag) {
1337        $metatag =~ m/(?:name|http-equiv)\s*=\s*([^\s\>]+)/is;
1338        $tag=$1;
1339    }
1340
1341    if (!defined $tag) {
1342        print $outhandle "HTMLPlugin: can't find NAME in \"$metatag\"\n";
1343        next;
1344    }
1345   
1346    # don't need to assign this field if it was passed in from a previous
1347    # (recursive) plugin
1348    if (defined $metadata->{$tag}) {next}
1349
1350    # find the tag content
1351    $metatag =~ m/content\s*=\s*([\"\'])?(.*?)\1/is;
1352    $value=$2;
1353
1354    # The following code assigns the metaname to value if value is
1355    # empty. Why would we do this?
1356    #if (! $value) {
1357    #    $metatag =~ m/(?:name|http-equiv)\s*=\s*([^\s\>]+)/is;
1358    #    $value=$1;
1359    #}
1360    if (!defined $value || $value eq "") {
1361        print $outhandle "HTMLPlugin: can't find VALUE in <meta $metatag >\n" if ($self->{'verbosity'} > 2);
1362        next;
1363    }
1364   
1365    # clean up and add
1366    $value =~ s/\s+/ /gs;
1367    chomp($value); # remove trailing \n, if any
1368    if (exists $creator_fields{lc($tag)}) {
1369        # map this value onto greenstone's "Creator" metadata
1370        $tag='Creator';
1371    } elsif (!exists $find_fields{lc($tag)}) {
1372        next; # don't want this tag
1373    } else {
1374        # get the user's preferred capitalisation
1375        $tag = $find_fields{lc($tag)};
1376    }
1377    if (lc($tag) eq "title") {
1378        $found_title = 1;
1379    }
1380
1381    if ($self->{'verbosity'} > 2) {
1382        print $outhandle " extracted \"$tag\" metadata \"$value\"\n";
1383    }
1384
1385    if ($tag =~ /\./) {
1386        # there is a . so has a namespace, add ex.
1387        $tag = "ex.$tag";
1388    }
1389    if (defined $separator) {
1390        my @values = split($separator, $value);
1391        foreach my $v (@values) {
1392        $doc_obj->add_utf8_metadata($section, $tag, $v) if $v =~ /\S/;
1393        }
1394    }
1395    else {
1396        $doc_obj->add_utf8_metadata($section, $tag, $value);
1397    }
1398    }
1399   
1400    # TITLE: extract the document title
1401    if (exists $find_fields{'title'} && !$found_title) {
1402    # we want a title, and didn't find one in the meta tags
1403    # see if there's a <title> tag
1404    my $title;
1405    my $from = ""; # for debugging output only
1406    if ($html_header =~ m/<title[^>]*>([^<]+)<\/title[^>]*>/is) {
1407        $title = $1;
1408        $from = "<title> tags";
1409    }
1410
1411    if (!defined $title) {
1412        $from = "first 100 chars";
1413        # if no title use first 100 or so characters
1414        $title = $$textref;
1415        $title =~ s/^\xFE\xFF//; # Remove unicode byte order mark       
1416        $title =~ s/^.*?<body>//si;
1417        # ignore javascript!
1418        $title =~ s@<script.*?</script>@ @sig;
1419        $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
1420        $title =~ s/<[^>]*>/ /g; # remove all HTML tags
1421        $title =~ s@\r@@g; # remove Windows carriage returns to ensure that titles of pdftohtml docs are consistent (the same 100 chars) across windows and linux
1422        $title = substr ($title, 0, 100);
1423        $title =~ s/\s\S*$/.../;
1424    }
1425    $title =~ s/<[^>]*>/ /g; # remove html tags
1426    $title =~ s/&nbsp;/ /g;
1427    $title =~ s/(?:&nbsp;|\xc2\xa0)/ /g; # utf-8 for nbsp...
1428    $title =~ s/\s+/ /gs; # collapse multiple spaces
1429    $title =~ s/^\s*//;   # remove leading spaces
1430    $title =~ s/\s*$//;   # remove trailing spaces
1431
1432    $title =~ s/^$self->{'title_sub'}// if ($self->{'title_sub'});
1433    $title =~ s/^\s+//s; # in case title_sub introduced any...
1434    $doc_obj->add_utf8_metadata ($section, "Title", $title);
1435    print $outhandle " extracted Title metadata \"$title\" from $from\n"
1436        if ($self->{'verbosity'} > 2);
1437    }
1438   
1439    # add FileFormat metadata
1440    $doc_obj->add_metadata($section,"FileFormat", "HTML");
1441
1442    # Special, for metadata names such as tagH1 - extracts
1443    # the text between the first <H1> and </H1> tags into "H1" metadata.
1444
1445    foreach my $field (keys %find_fields) {
1446    if ($field !~ m/^tag([a-z0-9]+)$/i) {next}
1447    my $tag = $1;
1448    if ($$textref =~ m@<$tag[^>]*>(.*?)</$tag[^>]*>@g) {
1449        my $content = $1;
1450        $content =~ s/&nbsp;/ /g;
1451        $content =~ s/<[^>]*>/ /g;
1452        $content =~ s/^\s+//;
1453        $content =~ s/\s+$//;
1454        $content =~ s/\s+/ /gs;
1455        if ($content) {
1456        $tag=$find_fields{"tag$tag"}; # get the user's capitalisation
1457        $tag =~ s/^tag//i;
1458        $doc_obj->add_utf8_metadata ($section, $tag, $content);
1459        print $outhandle " extracted \"$tag\" metadata \"$content\"\n"
1460            if ($self->{'verbosity'} > 2);
1461        }
1462    }
1463    }   
1464}
1465
1466
1467# evaluate any "../" to next directory up
1468# evaluate any "./" as here
1469sub eval_dir_dots {
1470    my $self = shift (@_);
1471    my ($filename) = @_;
1472    my $dirsep_os = &util::get_os_dirsep();
1473    my @dirsep = split(/$dirsep_os/,$filename);
1474
1475    my @eval_dirs = ();
1476    foreach my $d (@dirsep) {
1477    if ($d eq "..") {
1478        pop(@eval_dirs);
1479       
1480    } elsif ($d eq ".") {
1481        # do nothing!
1482
1483    } else {
1484        push(@eval_dirs,$d);
1485    }
1486    }
1487
1488    # Need to fiddle with number of elements in @eval_dirs if the
1489    # first one is the empty string.  This is because of a
1490    # modification to FileUtils::filenameConcatenate that supresses the addition
1491    # of a leading '/' character (or \ if windows) (intended to help
1492    # filename cat with relative paths) if the first entry in the
1493    # array is the empty string.  Making the array start with *two*
1494    # empty strings is a way to defeat this "smart" option.
1495    #
1496    if (scalar(@eval_dirs) > 0) {
1497    if ($eval_dirs[0] eq ""){
1498        unshift(@eval_dirs,"");
1499    }
1500    }
1501   
1502    my $evaluated_filename = (scalar @eval_dirs > 0) ? &FileUtils::filenameConcatenate(@eval_dirs) : "";
1503    return $evaluated_filename;
1504}
1505
1506sub replace_usemap_links {
1507    my $self = shift (@_);
1508    my ($front, $link, $back) = @_;
1509
1510    # remove quotes from link at start and end if necessary
1511    if ($link=~/^[\"\']/) {
1512    $link=~s/^[\"\']//;
1513    $link=~s/[\"\']$//;
1514    $front.='"';
1515    $back="\"$back";
1516    }
1517
1518    $link =~ s/^\.\///;
1519    return $front . $link . $back;
1520}
1521
1522sub inc_filecount {
1523    my $self = shift (@_);
1524
1525    if ($self->{'file_num'} == 1000) {
1526    $self->{'dir_num'} ++;
1527    $self->{'file_num'} = 0;
1528    } else {
1529    $self->{'file_num'} ++;
1530    }
1531}
1532
1533
1534# Extend read_file so that strings like &eacute; are
1535# converted to UTF8 internally. 
1536#
1537# We don't convert &lt; or &gt; or &amp; or &quot; in case
1538# they interfere with the GML files
1539
1540sub read_file {
1541    my $self = shift(@_);
1542    my ($filename, $encoding, $language, $textref) = @_;
1543
1544    $self->SUPER::read_file($filename, $encoding, $language, $textref);
1545
1546    # Convert entities to their Unicode code-point equivalents
1547    $$textref =~ s/&(lt|gt|amp|quot|nbsp);/&z$1;/go;
1548    $$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1,1)/gseo;
1549    $$textref =~ s/&z(lt|gt|amp|quot|nbsp);/&$1;/go;
1550
1551}
1552
1553sub HB_read_html_file {
1554    my $self = shift (@_);
1555    my ($htmlfile, $text) = @_;
1556   
1557    # load in the file
1558    if (!open (FILE, $htmlfile)) {
1559    print STDERR "ERROR - could not open $htmlfile\n";
1560    return;
1561    }
1562
1563    my $foundbody = 0;
1564    $self->HB_gettext (\$foundbody, $text, "FILE");
1565    close FILE;
1566   
1567    # just in case there was no <body> tag
1568    if (!$foundbody) {
1569    $foundbody = 1;
1570    open (FILE, $htmlfile) || return;
1571    $self->HB_gettext (\$foundbody, $text, "FILE");
1572    close FILE;
1573    }
1574    # text is in utf8
1575}       
1576
1577# converts the text to utf8, as ghtml does that for &eacute; etc.
1578sub HB_gettext {
1579    my $self = shift (@_);
1580    my ($foundbody, $text, $handle) = @_;
1581   
1582    my $line = "";
1583    while (defined ($line = <$handle>)) {
1584    # look for body tag
1585    if (!$$foundbody) {
1586        if ($line =~ s/^.*<body[^>]*>//i) {
1587        $$foundbody = 1;
1588        } else {
1589        next;
1590        }
1591    }
1592   
1593    # check for symbol fonts
1594    if ($line =~ m/<font [^>]*?face\s*=\s*\"?(\w+)\"?/i) {
1595        my $font = $1;
1596        print STDERR "HBPlug::HB_gettext - warning removed font $font\n"
1597        if ($font !~ m/^arial$/i);
1598    }
1599
1600    $$text .= $line;
1601    }
1602
1603    if ($self->{'input_encoding'} eq "iso_8859_1") {
1604    # convert to utf-8
1605    $$text=&unicode::unicode2utf8(&unicode::convert2unicode("iso_8859_1", $text));
1606    }
1607    # convert any alphanumeric character entities to their utf-8
1608    # equivalent for indexing purposes
1609    #&ghtml::convertcharentities ($$text);
1610
1611    $$text =~ s/\s+/ /g; # remove \n's
1612
1613    # At this point $$text is a binary byte string
1614    # => turn it into a Unicode aware string, so full
1615    # Unicode aware pattern matching can be used.
1616    # For instance: 's/\x{0101}//g' or '[[:upper:]]'
1617    #
1618
1619    $$text = decode("utf8",$$text);
1620}
1621
1622sub HB_clean_section {
1623    my $self = shift (@_);
1624    my ($section) = @_;
1625
1626    # remove tags without a starting tag from the section
1627    my ($tag, $tagstart);
1628    while ($section =~ m/<\/([^>]{1,10})>/) {
1629    $tag = $1;
1630    $tagstart = index($section, "<$tag");
1631    last if (($tagstart >= 0) && ($tagstart < index($section, "<\/$tag")));
1632    $section =~ s/<\/$tag>//;
1633    }
1634   
1635    # remove extra paragraph tags
1636    while ($section =~ s/<p\b[^>]*>\s*<p\b/<p/ig) {}
1637   
1638    # remove extra stuff at the end of the section
1639    while ($section =~ s/(<u>|<i>|<b>|<p\b[^>]*>|&nbsp;|\s)$//i) {}
1640   
1641    # add a newline at the beginning of each paragraph
1642    $section =~ s/(.)\s*<p\b/$1\n\n<p/gi;
1643   
1644    # add a newline every 80 characters at a word boundary
1645    # Note: this regular expression puts a line feed before
1646    # the last word in each section, even when it is not
1647    # needed.
1648    $section =~ s/(.{1,80})\s/$1\n/g;
1649   
1650    # fix up the image links
1651    $section =~ s/<img[^>]*?src=\"?([^\">]+)\"?[^>]*>/
1652    <center><img src=\"$1\" \/><\/center><br\/>/ig;
1653    $section =~ s/&lt;&lt;I&gt;&gt;\s*([^\.]+\.(png|jpg|gif))/
1654    <center><img src=\"$1\" \/><\/center><br\/>/ig;
1655
1656    return $section;
1657}
1658
1659# Will convert the oldHDL format to the new HDL format (using the Section tag) 
1660sub convert_to_newHDLformat
1661{
1662    my $self = shift (@_);
1663    my ($file,$cnfile) = @_;
1664    my $input_filename = $file;
1665    my $tmp_filename = $cnfile;
1666   
1667    # write HTML tmp file with new HDL format
1668    open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
1669   
1670    # read in the file and do basic html cleaning (removing header etc)
1671    my $html = "";
1672    $self->HB_read_html_file ($input_filename, \$html);
1673   
1674    # process the file one section at a time
1675    my $curtoclevel = 1;
1676    my $firstsection = 1;
1677    my $toclevel = 0;
1678    while (length ($html) > 0) {
1679    if ($html =~ s/^.*?(?:<p\b[^>]*>)?((<b>|<i>|<u>|\s)*)&lt;&lt;TOC(\d+)&gt;&gt;\s*(.*?)<p\b/<p/i) {
1680        $toclevel = $3;
1681        my $title = $4;
1682        my $sectiontext = "";
1683        if ($html =~ s/^(.*?)((?:<p\b[^>]*>)?((<b>|<i>|<u>|\s)*)&lt;&lt;TOC\d+&gt;&gt;)/$2/i) {
1684        $sectiontext = $1;
1685        } else {
1686        $sectiontext = $html;
1687        $html = "";
1688        }
1689
1690        # remove tags and extra spaces from the title
1691        $title =~ s/<\/?[^>]+>//g;
1692        $title =~ s/^\s+|\s+$//g;
1693
1694        # close any sections below the current level and
1695        # create a new section (special case for the firstsection)
1696        print PROD "<!--\n";
1697        while (($curtoclevel > $toclevel) ||
1698           (!$firstsection && $curtoclevel == $toclevel)) {
1699        $curtoclevel--;
1700        print PROD "</Section>\n";
1701        }
1702        if ($curtoclevel+1 < $toclevel) {
1703        print STDERR "WARNING - jump in toc levels in $input_filename " .
1704            "from $curtoclevel to $toclevel\n";
1705        }
1706        while ($curtoclevel < $toclevel) {
1707        $curtoclevel++;
1708        }
1709
1710        if ($curtoclevel == 1) {
1711            # add the header tag
1712        print PROD "-->\n";
1713            print PROD "<HTML>\n<HEAD>\n<TITLE>$title</TITLE>\n</HEAD>\n<BODY>\n";
1714        print PROD "<!--\n";
1715        }
1716       
1717        print PROD "<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">$title</Metadata>\n\t</Description>\n";
1718       
1719        print PROD "-->\n";
1720       
1721        # clean up the section html
1722        $sectiontext = $self->HB_clean_section($sectiontext);
1723
1724        print PROD "$sectiontext\n";     
1725
1726    } else {
1727        print STDERR "WARNING - leftover text\n" , $self->shorten($html),
1728        "\nin $input_filename\n";
1729        last;
1730    }
1731    $firstsection = 0;
1732    }
1733   
1734    print PROD "<!--\n";
1735    while ($curtoclevel > 0) {
1736    $curtoclevel--;
1737    print PROD "</Section>\n";
1738    }
1739    print PROD "-->\n";
1740   
1741    close (PROD) || die("Error Closing File: $tmp_filename $!");
1742   
1743    return $tmp_filename;
1744}       
1745
1746sub shorten {
1747    my $self = shift (@_);
1748    my ($text) = @_;
1749
1750    return "\"$text\"" if (length($text) < 100);
1751
1752    return "\"" . substr ($text, 0, 50) . "\" ... \"" .
1753    substr ($text, length($text)-50) . "\"";
1754}
1755
1756sub convert_tidy_or_oldHDL_file
1757{
1758    my $self = shift (@_);
1759    my ($file) = @_;
1760    my $input_filename = $file;
1761   
1762    if (-d $input_filename)
1763    {
1764        return $input_filename;
1765    }
1766   
1767    # get the input filename
1768    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
1769    my $base_dirname = $dirname;
1770    $suffix = lc($suffix);   
1771   
1772    # derive tmp filename from input filename
1773    # Remove any white space from filename -- no risk of name collision, and
1774    # makes later conversion by utils simpler. Leave spaces in path...
1775    # tidy up the filename with space, dot, hyphen between
1776    $tailname =~ s/\s+//g;
1777    $tailname =~ s/\.+//g;
1778    $tailname =~ s/\-+//g;
1779    # convert to utf-8 otherwise we have problems with the doc.xml file
1780    # later on
1781    &unicode::ensure_utf8(\$tailname);
1782   
1783    # softlink to collection tmp dir
1784    my $tmp_dirname = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tidytmp");
1785    &FileUtils::makeDirectory($tmp_dirname) if (!-e $tmp_dirname);
1786   
1787    my $test_dirname = "";
1788    my $f_separator = &util::get_os_dirsep();
1789   
1790    if ($dirname =~ m/import$f_separator/)
1791    {
1792        $test_dirname = $'; #'
1793   
1794    #print STDERR "init $'\n";
1795   
1796    while ($test_dirname =~ m/[$f_separator]/)
1797    {
1798        my $folderdirname = $`;
1799        $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname,$folderdirname);
1800        &FileUtils::makeDirectory($tmp_dirname) if (!-e $tmp_dirname);
1801        $test_dirname = $'; #'
1802    }
1803    }
1804   
1805    my $tmp_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$suffix");
1806   
1807    # tidy or convert the input file if it is a HTML-like file or it is accepted by the process_exp
1808    if (($suffix eq ".htm") || ($suffix eq ".html") || ($suffix eq ".shtml"))
1809    {   
1810        #convert the input file to a new style HDL
1811        my $hdl_output_filename = $input_filename;
1812        if ($self->{'old_style_HDL'})
1813        {
1814        $hdl_output_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$suffix");
1815        $hdl_output_filename = $self->convert_to_newHDLformat($input_filename,$hdl_output_filename);
1816        }
1817   
1818    #just for checking copy all other file from the base dir to tmp dir if it is not exists
1819    opendir(DIR,$base_dirname) or die "Can't open base directory : $base_dirname!";
1820    my @files = grep {!/^\.+$/} readdir(DIR);
1821    close(DIR);
1822
1823    foreach my $file (@files)
1824    {
1825        my $src_file = &FileUtils::filenameConcatenate($base_dirname,$file);
1826        my $dest_file = &FileUtils::filenameConcatenate($tmp_dirname,$file);
1827        if ((!-e $dest_file) && (!-d $src_file))
1828        {
1829        # just copy the original file back to the tmp directory
1830        copy($src_file,$dest_file) or die "Can't copy file $src_file to $dest_file $!";
1831        }
1832    }
1833   
1834    # tidy the input file
1835    my $tidy_output_filename = $hdl_output_filename;
1836    if ($self->{'use_realistic_book'})
1837    {
1838        $tidy_output_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$suffix");
1839        $tidy_output_filename = $self->tmp_tidy_file($hdl_output_filename,$tidy_output_filename);
1840    }
1841    $tmp_filename = $tidy_output_filename;
1842    }
1843    else
1844    {
1845        if (!-e $tmp_filename)
1846    {
1847        # just copy the original file back to the tmp directory
1848        copy($input_filename,$tmp_filename) or die "Can't copy file $input_filename to $tmp_filename $!";
1849    }
1850    }
1851   
1852    return $tmp_filename;
1853}
1854
1855
1856# Will make the html input file as a proper XML file with removed font tag and
1857# image size added to the img tag.
1858# The tidying process takes place in a collection specific 'tmp' directory so
1859# that we don't accidentally damage the input.
1860sub tmp_tidy_file
1861{
1862    my $self = shift (@_);
1863    my ($file,$cnfile) = @_;
1864    my $input_filename = $file;
1865    my $tmp_filename = $cnfile;
1866   
1867    # get the input filename
1868    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
1869
1870    require HTML::TokeParser::Simple;
1871   
1872    # create HTML parser to decode the input file
1873    my $parser = HTML::TokeParser::Simple->new($input_filename);
1874
1875    # write HTML tmp file without the font tag and image size are added to the img tag
1876    open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
1877    while (my $token = $parser->get_token())
1878    {
1879    # is it an img tag
1880    if ($token->is_start_tag('img'))
1881    {
1882        # get the attributes
1883        my $attr = $token->return_attr;
1884
1885        # get the full path to the image
1886        my $img_file = &FileUtils::filenameConcatenate($dirname,$attr->{src});
1887       
1888        # set the width and height attribute
1889        ($attr->{width}, $attr->{height}) = imgsize($img_file);
1890
1891        # recreate the tag
1892        print PROD "<img";
1893        print PROD map { qq { $_="$attr->{$_}"} } keys %$attr;
1894        print PROD ">";
1895    }
1896    # is it a font tag
1897    else
1898    {
1899        if (($token->is_start_tag('font')) || ($token->is_end_tag('font')))
1900        {
1901        # remove font tag
1902        print PROD "";
1903        }
1904        else
1905        {
1906        # print without changes
1907        print PROD $token->as_is;
1908        }
1909    }
1910    }
1911    close (PROD) || die("Error Closing File: $tmp_filename $!");
1912
1913    # run html-tidy on the tmp file to make it a proper XML file
1914
1915    my $outhandle = $self->{'outhandle'};
1916    print $outhandle "Converting HTML to be XML compliant:\n";
1917
1918    my $tidy_cmd = "tidy";
1919    $tidy_cmd .= " -q" if ($self->{'verbosity'} <= 2);
1920    $tidy_cmd .= " -raw -wrap 0 -asxml \"$tmp_filename\"";
1921    if ($self->{'verbosity'} <= 2) {
1922    if ($ENV{'GSDLOS'} =~ m/^windows/i) {
1923        $tidy_cmd .= " 2>nul";
1924    }
1925    else {
1926        $tidy_cmd .= " 2>/dev/null";
1927    }
1928    print $outhandle "  => $tidy_cmd\n";
1929    }
1930
1931    my $tidyfile = `$tidy_cmd`;
1932
1933    # write result back to the tmp file
1934    open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
1935    print PROD $tidyfile;
1936    close (PROD) || die("Error Closing File: $tmp_filename $!");
1937   
1938    # return the output filename
1939    return $tmp_filename;
1940}
1941
1942sub associate_cover_image
1943{
1944    my $self = shift(@_);
1945    my ($doc_obj, $filename) = @_;
1946    if (($self->{'use_realistic_book'}) || ($self->{'old_style_HDL'}))
1947    {
1948    # we will have cover image in tidytmp, but want it from import
1949    $filename =~ s/([\\\/])tidytmp([\\\/])/$1import$2/;
1950    }
1951    $self->SUPER::associate_cover_image($doc_obj, $filename);
1952}
1953
1954   
19551;
Note: See TracBrowser for help on using the browser.