root/gsdl/trunk/perllib/plugins/HTMLPlugin.pm @ 18409

Revision 18409, 51.4 KB (checked in by ak19, 11 years ago)

Need to replace character ref for underscore with underscore again for the doc.xml's URL metadata, otherwise links in HTML files referring to other HTML files whose filenames contain underscores won't work.

  • 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 ReadTextFile;
39use HBPlugin;
40use ghtml;
41use unicode;
42use util;
43use XMLParser;
44
45use Image::Size;
46use File::Copy;
47
48sub BEGIN {
49    @HTMLPlugin::ISA = ('ReadTextFile', 'HBPlugin');
50}
51
52use strict; # every perl program should have this!
53no strict 'refs'; # make an exception so we can use variables as filehandles
54
55my $arguments =
56    [ { 'name' => "process_exp",
57    'desc' => "{BasePlugin.process_exp}",
58    'type' => "regexp",
59    'deft' =>  &get_default_process_exp() },
60      { 'name' => "block_exp",
61    'desc' => "{BasePlugin.block_exp}",
62    'type' => 'regexp',
63    'deft' =>  &get_default_block_exp() },
64      { 'name' => "nolinks",
65    'desc' => "{HTMLPlugin.nolinks}",
66    'type' => "flag" },
67      { 'name' => "keep_head",
68    'desc' => "{HTMLPlugin.keep_head}",
69    'type' => "flag" },
70      { 'name' => "no_metadata",
71    'desc' => "{HTMLPlugin.no_metadata}",
72    'type' => "flag" },
73      { 'name' => "metadata_fields",
74    'desc' => "{HTMLPlugin.metadata_fields}",
75    'type' => "string",
76    'deft' => "Title" },
77      { 'name' => "hunt_creator_metadata",
78    'desc' => "{HTMLPlugin.hunt_creator_metadata}",
79    'type' => "flag" },
80      { 'name' => "file_is_url",
81    'desc' => "{HTMLPlugin.file_is_url}",
82    'type' => "flag" },
83      { 'name' => "assoc_files",
84    'desc' => "{HTMLPlugin.assoc_files}",
85    'type' => "regexp",
86    'deft' => &get_default_block_exp() },
87      { 'name' => "rename_assoc_files",
88    'desc' => "{HTMLPlugin.rename_assoc_files}",
89    'type' => "flag" },
90      { 'name' => "title_sub",
91    'desc' => "{HTMLPlugin.title_sub}",
92    'type' => "string",
93    'deft' => "" },
94      { 'name' => "description_tags",
95    'desc' => "{HTMLPlugin.description_tags}",
96    'type' => "flag" },
97      # retain this for backward compatibility (w3mir option was replaced by
98      # file_is_url)
99      { 'name' => "w3mir",
100#   'desc' => "{HTMLPlugin.w3mir}",
101    'type' => "flag",
102    'hiddengli' => "yes"},
103      { 'name' => "no_strip_metadata_html",
104    'desc' => "{HTMLPlugin.no_strip_metadata_html}",
105    'type' => "string",
106    'deft' => "",
107    'reqd' => "no"},
108      { 'name' => "sectionalise_using_h_tags",
109    'desc' => "{HTMLPlugin.sectionalise_using_h_tags}",
110    'type' => "flag" },
111      { 'name' => "use_realistic_book",
112        'desc' => "{HTMLPlugin.tidy_html}",
113    'type' => "flag"},
114      { 'name' => "old_style_HDL",
115        'desc' => "{HTMLPlugin.old_style_HDL}",
116    'type' => "flag"}
117      ];
118
119my $options = { 'name'     => "HTMLPlugin",
120        'desc'     => "{HTMLPlugin.desc}",
121        'abstract' => "no",
122        'inherits' => "yes",
123        'args'     => $arguments };
124
125
126sub HB_read_html_file {
127    my $self = shift (@_);
128    my ($htmlfile, $text) = @_;
129   
130    # load in the file
131    if (!open (FILE, $htmlfile)) {
132    print STDERR "ERROR - could not open $htmlfile\n";
133    return;
134    }
135
136    my $foundbody = 0;
137    $self->HB_gettext (\$foundbody, $text, "FILE");
138    close FILE;
139   
140    # just in case there was no <body> tag
141    if (!$foundbody) {
142    $foundbody = 1;
143    open (FILE, $htmlfile) || return;
144    $self->HB_gettext (\$foundbody, $text, "FILE");
145    close FILE;
146    }
147    # text is in utf8
148}       
149
150# converts the text to utf8, as ghtml does that for &eacute; etc.
151sub HB_gettext {
152    my $self = shift (@_);
153    my ($foundbody, $text, $handle) = @_;
154   
155    my $line = "";
156    while (defined ($line = <$handle>)) {
157    # look for body tag
158    if (!$$foundbody) {
159        if ($line =~ s/^.*<body[^>]*>//i) {
160        $$foundbody = 1;
161        } else {
162        next;
163        }
164    }
165   
166    # check for symbol fonts
167    if ($line =~ m/<font [^>]*?face\s*=\s*\"?(\w+)\"?/i) {
168        my $font = $1;
169        print STDERR "HBPlug::HB_gettext - warning removed font $font\n"
170        if ($font !~ m/^arial$/i);
171    }
172
173    $$text .= $line;
174    }
175
176    if ($self->{'input_encoding'} eq "iso_8859_1") {
177    # convert to utf-8
178    $$text=&unicode::unicode2utf8(&unicode::convert2unicode("iso_8859_1", $text));
179    }
180    # convert any alphanumeric character entities to their utf-8
181    # equivalent for indexing purposes
182    #&ghtml::convertcharentities ($$text);
183
184    $$text =~ s/\s+/ /g; # remove \n's
185}
186
187sub HB_clean_section {
188    my $self = shift (@_);
189    my ($section) = @_;
190
191    # remove tags without a starting tag from the section
192    my ($tag, $tagstart);
193    while ($section =~ m/<\/([^>]{1,10})>/) {
194    $tag = $1;
195    $tagstart = index($section, "<$tag");
196    last if (($tagstart >= 0) && ($tagstart < index($section, "<\/$tag")));
197    $section =~ s/<\/$tag>//;
198    }
199   
200    # remove extra paragraph tags
201    while ($section =~ s/<p\b[^>]*>\s*<p\b/<p/ig) {}
202   
203    # remove extra stuff at the end of the section
204    while ($section =~ s/(<u>|<i>|<b>|<p\b[^>]*>|&nbsp;|\s)$//i) {}
205   
206    # add a newline at the beginning of each paragraph
207    $section =~ s/(.)\s*<p\b/$1\n\n<p/gi;
208   
209    # add a newline every 80 characters at a word boundary
210    # Note: this regular expression puts a line feed before
211    # the last word in each section, even when it is not
212    # needed.
213    $section =~ s/(.{1,80})\s/$1\n/g;
214   
215    # fix up the image links
216    $section =~ s/<img[^>]*?src=\"?([^\">]+)\"?[^>]*>/
217    <center><img src=\"$1\" \/><\/center><br\/>/ig;
218    $section =~ s/&lt;&lt;I&gt;&gt;\s*([^\.]+\.(png|jpg|gif))/
219    <center><img src=\"$1\" \/><\/center><br\/>/ig;
220
221    return $section;
222}
223
224# Will convert the oldHDL format to the new HDL format (using the Section tag) 
225sub convert_to_newHDLformat
226{
227    my $self = shift (@_);
228    my ($file,$cnfile) = @_;
229    my $input_filename = $file;
230    my $tmp_filename = $cnfile;
231   
232    # write HTML tmp file with new HDL format
233    open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
234   
235    # read in the file and do basic html cleaning (removing header etc)
236    my $html = "";
237    $self->HB_read_html_file ($input_filename, \$html);
238   
239    # process the file one section at a time
240    my $curtoclevel = 1;
241    my $firstsection = 1;
242    my $toclevel = 0;
243    while (length ($html) > 0) {
244    if ($html =~ s/^.*?(?:<p\b[^>]*>)?((<b>|<i>|<u>|\s)*)&lt;&lt;TOC(\d+)&gt;&gt;\s*(.*?)<p\b/<p/i) {
245        $toclevel = $3;
246        my $title = $4;
247        my $sectiontext = "";
248        if ($html =~ s/^(.*?)((?:<p\b[^>]*>)?((<b>|<i>|<u>|\s)*)&lt;&lt;TOC\d+&gt;&gt;)/$2/i) {
249        $sectiontext = $1;
250        } else {
251        $sectiontext = $html;
252        $html = "";
253        }
254
255        # remove tags and extra spaces from the title
256        $title =~ s/<\/?[^>]+>//g;
257        $title =~ s/^\s+|\s+$//g;
258
259        # close any sections below the current level and
260        # create a new section (special case for the firstsection)
261        print PROD "<!--\n";
262        while (($curtoclevel > $toclevel) ||
263           (!$firstsection && $curtoclevel == $toclevel)) {
264        $curtoclevel--;
265        print PROD "</Section>\n";
266        }
267        if ($curtoclevel+1 < $toclevel) {
268        print STDERR "WARNING - jump in toc levels in $input_filename " .
269            "from $curtoclevel to $toclevel\n";
270        }
271        while ($curtoclevel < $toclevel) {
272        $curtoclevel++;
273        }
274
275        if ($curtoclevel == 1) {
276            # add the header tag
277        print PROD "-->\n";
278            print PROD "<HTML>\n<HEAD>\n<TITLE>$title</TITLE>\n</HEAD>\n<BODY>\n";
279        print PROD "<!--\n";
280        }
281       
282        print PROD "<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">$title</Metadata>\n\t</Description>\n";
283       
284        print PROD "-->\n";
285       
286        # clean up the section html
287        $sectiontext = $self->HB_clean_section($sectiontext);
288
289        print PROD "$sectiontext\n";     
290
291    } else {
292        print STDERR "WARNING - leftover text\n" , $self->shorten($html),
293        "\nin $input_filename\n";
294        last;
295    }
296    $firstsection = 0;
297    }
298   
299    print PROD "<!--\n";
300    while ($curtoclevel > 0) {
301    $curtoclevel--;
302    print PROD "</Section>\n";
303    }
304    print PROD "-->\n";
305   
306    close (PROD) || die("Error Closing File: $tmp_filename $!");
307   
308    return $tmp_filename;
309}       
310
311sub shorten {
312    my $self = shift (@_);
313    my ($text) = @_;
314
315    return "\"$text\"" if (length($text) < 100);
316
317    return "\"" . substr ($text, 0, 50) . "\" ... \"" .
318    substr ($text, length($text)-50) . "\"";
319}
320
321sub convert_tidy_or_oldHDL_file
322{
323    my $self = shift (@_);
324    my ($file) = @_;
325    my $input_filename = $file;
326   
327    if (-d $input_filename)
328    {
329        return $input_filename;
330    }
331   
332    # get the input filename
333    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
334    my $base_dirname = $dirname;
335    $suffix = lc($suffix);   
336   
337    # derive tmp filename from input filename
338    # Remove any white space from filename -- no risk of name collision, and
339    # makes later conversion by utils simpler. Leave spaces in path...
340    # tidy up the filename with space, dot, hyphen between
341    $tailname =~ s/\s+//g;
342    $tailname =~ s/\.+//g;
343    $tailname =~ s/\-+//g;
344    # convert to utf-8 otherwise we have problems with the doc.xml file
345    # later on
346    &unicode::ensure_utf8(\$tailname);
347   
348    # softlink to collection tmp dir
349    my $tmp_dirname = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "tidytmp");
350    &util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
351   
352    my $test_dirname = "";
353    my $f_separator = &util::get_os_dirsep();
354   
355    if ($dirname =~ m/import$f_separator/)
356    {
357        $test_dirname = $'; #'
358   
359    #print STDERR "init $'\n";
360   
361    while ($test_dirname =~ m/[$f_separator]/)
362    {
363        my $folderdirname = $`;
364        $tmp_dirname = &util::filename_cat($tmp_dirname,$folderdirname);
365        &util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);
366        $test_dirname = $'; #'
367    }
368    }
369   
370    my $tmp_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
371   
372    # tidy or convert the input file if it is a HTML-like file or it is accepted by the process_exp
373    if (($suffix eq ".htm") || ($suffix eq ".html") || ($suffix eq ".shtml"))
374    {   
375        #convert the input file to a new style HDL
376        my $hdl_output_filename = $input_filename;
377        if ($self->{'old_style_HDL'})
378        {
379        $hdl_output_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
380        $hdl_output_filename = $self->convert_to_newHDLformat($input_filename,$hdl_output_filename);
381        }
382   
383    #just for checking copy all other file from the base dir to tmp dir if it is not exists
384    opendir(DIR,$base_dirname) or die "Can't open base directory : $base_dirname!";
385    my @files = grep {!/^\.+$/} readdir(DIR);
386    close(DIR);
387
388    foreach my $file (@files)
389    {
390        my $src_file = &util::filename_cat($base_dirname,$file);
391        my $dest_file = &util::filename_cat($tmp_dirname,$file);
392        if ((!-e $dest_file) && (!-d $src_file))
393        {
394        # just copy the original file back to the tmp directory
395        copy($src_file,$dest_file) or die "Can't copy file $src_file to $dest_file $!";
396        }
397    }
398   
399    # tidy the input file
400    my $tidy_output_filename = $hdl_output_filename;
401    if ($self->{'use_realistic_book'})
402    {
403        $tidy_output_filename = &util::filename_cat($tmp_dirname, "$tailname$suffix");
404        $tidy_output_filename = $self->tmp_tidy_file($hdl_output_filename,$tidy_output_filename);
405    }
406    $tmp_filename = $tidy_output_filename;
407    }
408    else
409    {
410        if (!-e $tmp_filename)
411    {
412        # just copy the original file back to the tmp directory
413        copy($input_filename,$tmp_filename) or die "Can't copy file $input_filename to $tmp_filename $!";
414    }
415    }
416   
417    return $tmp_filename;
418}
419
420
421# Will make the html input file as a proper XML file with removed font tag and
422# image size added to the img tag.
423# The tidying process takes place in a collection specific 'tmp' directory so
424# that we don't accidentally damage the input.
425sub tmp_tidy_file
426{
427    my $self = shift (@_);
428    my ($file,$cnfile) = @_;
429    my $input_filename = $file;
430    my $tmp_filename = $cnfile;
431   
432    # get the input filename
433    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
434
435    require HTML::TokeParser::Simple;
436   
437    # create HTML parser to decode the input file
438    my $parser = HTML::TokeParser::Simple->new($input_filename);
439
440    # write HTML tmp file without the font tag and image size are added to the img tag
441    open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
442    while (my $token = $parser->get_token())
443    {
444    # is it an img tag
445    if ($token->is_start_tag('img'))
446    {
447        # get the attributes
448        my $attr = $token->return_attr;
449
450        # get the full path to the image
451        my $img_file = &util::filename_cat($dirname,$attr->{src});
452       
453        # set the width and height attribute
454        ($attr->{width}, $attr->{height}) = imgsize($img_file);
455
456        # recreate the tag
457        print PROD "<img";
458        print PROD map { qq { $_="$attr->{$_}"} } keys %$attr;
459        print PROD ">";
460    }
461    # is it a font tag
462    else
463    {
464        if (($token->is_start_tag('font')) || ($token->is_end_tag('font')))
465        {
466        # remove font tag
467        print PROD "";
468        }
469        else
470        {
471        # print without changes
472        print PROD $token->as_is;
473        }
474    }
475    }
476    close (PROD) || die("Error Closing File: $tmp_filename $!");
477
478    # run html-tidy on the tmp file to make it a proper XML file
479    my $tidyfile = `tidy -utf8 -wrap 0 -asxml "$tmp_filename"`;
480
481    # write result back to the tmp file
482    open (PROD, ">$tmp_filename") || die("Error Writing to File: $tmp_filename $!");
483    print PROD $tidyfile;
484    close (PROD) || die("Error Closing File: $tmp_filename $!");
485   
486    # return the output filename
487    return $tmp_filename;
488}
489
490sub read_into_doc_obj
491{
492    my $self = shift (@_); 
493    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
494       
495    # get the input file
496    my $input_filename = $file;
497    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
498    $suffix = lc($suffix);   
499   
500    if (($self->{'use_realistic_book'}) || ($self->{'old_style_HDL'}))
501    {
502    # because the document has to be sectionalized set the description tags
503    $self->{'description_tags'} = 1;
504   
505    # set the file to be tidied
506    $input_filename = &util::filename_cat($base_dir,$file) if $base_dir =~ m/\w/;
507   
508    # get the tidied file
509    #my $tidy_filename = $self->tmp_tidy_file($input_filename);
510    my $tidy_filename = $self->convert_tidy_or_oldHDL_file($input_filename);
511   
512    # derive tmp filename from input filename
513    my ($tailname, $dirname, $suffix) = &File::Basename::fileparse($tidy_filename, "\\.[^\\.]+\$");
514   
515    # set the new input file and base_dir to be from the tidied file
516    $file = "$tailname$suffix";
517    $base_dir = $dirname;
518    }
519   
520    # call the parent read_into_doc_obj
521    my ($process_status,$doc_obj) = $self->SUPER::read_into_doc_obj($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli);
522   
523    return ($process_status,$doc_obj);
524}
525
526sub new {
527    my ($class) = shift (@_);
528    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
529    push(@$pluginlist, $class);
530   
531    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
532    push(@{$hashArgOptLists->{"OptList"}},$options);
533   
534
535    my $self = new ReadTextFile($pluginlist,$inputargs,$hashArgOptLists);
536   
537    if ($self->{'w3mir'}) {
538    $self->{'file_is_url'} = 1;
539    }
540    $self->{'aux_files'} = {};
541    $self->{'dir_num'} = 0;
542    $self->{'file_num'} = 0;
543   
544    return bless $self, $class;
545}
546
547# may want to use (?i)\.(gif|jpe?g|jpe|png|css|js(?:@.*)?)$
548# if have eg <script language="javascript" src="img/lib.js@123">
549sub get_default_block_exp {
550    my $self = shift (@_);
551   
552    #return q^(?i)\.(gif|jpe?g|jpe|jpg|png|css)$^;
553    return "";
554}
555
556sub get_default_process_exp {
557    my $self = shift (@_);
558   
559    # the last option is an attempt to encode the concept of an html query ...
560    return q^(?i)(\.html?|\.shtml|\.shm|\.asp|\.php\d?|\.cgi|.+\?.+=.*)$^;
561}
562
563sub store_block_files
564{
565    my $self =shift (@_);
566    my ($filename_full_path, $block_hash) = @_;
567
568    my $html_fname = $filename_full_path;
569    my @file_blocks;
570   
571    my ($language, $encoding) = $self->textcat_get_language_encoding ($filename_full_path);
572
573    # read in file ($text will be in utf8)
574    my $raw_text = "";
575    $self->read_file_no_decoding ($filename_full_path, \$raw_text);
576
577    my $textref = \$raw_text;
578    my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
579    my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
580    $$textref =~ s/$opencom(.*?)$closecom//gs;
581
582    my $attval = "\\\"[^\\\"]+\\\"|[^\\s>]+";
583    my @img_matches = ($$textref =~ m/<img[^>]*?src\s*=\s*($attval)[^>]*>/igs);
584    my @usemap_matches = ($$textref =~ m/<img[^>]*?usemap\s*=\s*($attval)[^>]*>/igs);
585    my @link_matches = ($$textref =~ m/<link[^>]*?href\s*=\s*($attval)[^>]*>/igs);
586    my @embed_matches = ($$textref =~ m/<embed[^>]*?src\s*=\s*($attval)[^>]*>/igs);
587    my @tabbg_matches = ($$textref =~ m/<(?:body|table|tr|td)[^>]*?background\s*=\s*($attval)[^>]*>/igs);
588    my @script_matches = ($$textref =~ m/<script[^>]*?src\s*=\s*($attval)[^>]*>/igs);
589
590    if(!defined $self->{'utf8_to_original_filename'}) {
591    # maps from utf8 converted link name -> original filename referrred to by (possibly URL-encoded) src url
592    $self->{'utf8_to_original_filename'} = {};
593    }
594
595    foreach my $link (@img_matches, @usemap_matches, @link_matches, @embed_matches, @tabbg_matches, @script_matches) {
596
597    # remove quotes from link at start and end if necessary
598    if ($link=~/^\"/) {
599        $link=~s/^\"//;
600        $link=~s/\"$//;
601    }
602
603    $link =~ s/\#.*$//s; # remove any anchor names, e.g. foo.html#name becomes foo.html
604    # some links may just be anchor names
605    next unless ($link =~ /\S+/);
606
607    if ($link !~ m@^/@ && $link !~ m/^([A-Z]:?)\\/) {
608        # Turn relative file path into full path
609        my $dirname = &File::Basename::dirname($filename_full_path);
610        $link = &util::filename_cat($dirname, $link);
611    }
612    $link = $self->eval_dir_dots($link);
613
614    # this is the actual filename on the filesystem (that the link refers to)
615    my $url_original_filename = $self->opt_url_decode($link);
616
617    # Convert the url_original_filename into its utf8 version. Store the utf8 link along with the url_original_filename
618    my $utf8_link = "";
619    $self->decode_text($link,$encoding,$language,\$utf8_link);
620
621    $self->{'utf8_to_original_filename'}->{$utf8_link} = $url_original_filename;
622#   print STDERR "**** utf8_encoded_link to original src filename:\n\t$utf8_link\n\t".$self->{'utf8_to_original_filename'}->{$utf8_link}."\n";
623
624    if ($url_original_filename ne $utf8_link) {
625        my $outhandle = $self->{'outhandle'};
626
627        print $outhandle "URL Encoding $url_original_filename\n";
628        print $outhandle " ->$utf8_link\n";
629    }
630
631    $block_hash->{'file_blocks'}->{$url_original_filename} = 1;
632    }
633}
634
635# Given a filename in any encoding, will URL decode it to get back the original filename
636# in the original encoding. Because this method is intended to work out the *original*
637# filename*, it does not URL decode any filename if a file by the name of the *URL-encoded*
638# string already exists in the local folder.
639# Return the original filename corresponding to the parameter URL-encoded filename, and
640# a decoded flag that is set to true iff URL-decoding had to be applied.
641sub opt_url_decode {
642    my $self = shift (@_);
643    my ($link) = @_;
644
645    # Replace %XX's in URL with decoded value if required.
646    # Note that the filename may include the %XX in some situations
647    if ($link =~ m/\%[A-F0-9]{2}/i) {
648    if (!-e $link) {
649        $link = &unicode::url_decode($link);
650    }
651    }
652
653    return $link;
654}
655
656
657# do plugin specific processing of doc_obj
658sub process {
659    my $self = shift (@_);
660    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
661    my $outhandle = $self->{'outhandle'};
662
663    if ($ENV{'GSDLOS'} =~ m/^windows/i) {
664    # this makes life so much easier... perl can cope with unix-style '/'s.
665    $base_dir =~ s@(\\)+@/@g;
666    $file =~ s@(\\)+@/@g;
667    }
668   
669    # reset per-doc stuff...
670    $self->{'aux_files'} = {};
671    $self->{'dir_num'} = 0;
672    $self->{'file_num'} = 0;
673
674    # process an HTML file where sections are divided by headings tags (H1, H2 ...)
675    # you can also include metadata in the format (X can be any number)
676    # <hX>Title<!--gsdl-metadata
677    #   <Metadata name="name1">value1</Metadata>
678    #   ...
679    #   <Metadata name="nameN">valueN</Metadata>
680    #--></hX>
681    if ($self->{'sectionalise_using_h_tags'}) {
682    # description_tags should allways be activated because we convert headings to description tags
683    $self->{'description_tags'} = 1;
684
685    my $arrSections = [];
686    $$textref =~ s/<h([0-9]+)[^>]*>(.*?)<\/h[0-9]+>/$self->process_heading($1, $2, $arrSections, $file)/isge;
687
688    if (scalar(@$arrSections)) {
689        my $strMetadata = $self->update_section_data($arrSections, -1);
690        if (length($strMetadata)) {
691        $strMetadata = '<!--' . $strMetadata . "\n-->\n</body>";
692        $$textref =~ s/<\/body>/$strMetadata/ig;
693        }
694    }
695    }
696
697    my $cursection = $doc_obj->get_top_section();
698
699    $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection)
700    unless $self->{'no_metadata'} || $self->{'description_tags'};
701
702    # Store URL for page as metadata - this can be used for an
703    # altavista style search interface. The URL won't be valid
704    # unless the file structure contains the domain name (i.e.
705    # like when w3mir is used to download a website).
706
707    # URL metadata (even invalid ones) are used to support internal
708    # links, so even if 'file_is_url' is off, still need to store info
709
710    my ($tailname,$dirname,$suffix) = &File::Basename::fileparse($file, "\\.[^\\.]+\$");
711    my $utf8_file = $self->filename_to_utf8_metadata($file);
712    $utf8_file =~ s/&\#095;/_/g;
713    my $web_url = "http://";
714    if(defined $dirname) { # local directory
715    $dirname = $self->eval_dir_dots($dirname);
716    $dirname .= "/" if $dirname ne ""; # if there's a directory, it should end on "/"
717    $web_url = $web_url.$dirname.$utf8_file;
718    } else {
719    $web_url = $web_url.$utf8_file;
720    }
721
722    $doc_obj->add_utf8_metadata($cursection, "URL", $web_url);
723
724    if ($self->{'file_is_url'}) {
725    $doc_obj->add_metadata($cursection, "weblink", "<a href=\"$web_url\">");
726    $doc_obj->add_metadata($cursection, "webicon", "_iconworld_");
727    $doc_obj->add_metadata($cursection, "/weblink", "</a>");
728    }
729
730    if ($self->{'description_tags'}) {
731    # remove the html header - note that doing this here means any
732    # sections defined within the header will be lost (so all <Section>
733    # tags must appear within the body of the HTML)
734    my ($head_keep) = ($$textref =~ m/^(.*?)<body[^>]*>/is);
735
736    $$textref =~ s/^.*?<body[^>]*>//is;
737    $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
738
739    my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
740    my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
741
742    my $lt = '(?:<|&lt;)';
743    my $gt = '(?:>|&gt;)';
744    my $quot = '(?:"|&quot;|&rdquo;|&ldquo;)';
745
746    my $dont_strip = '';
747    if ($self->{'no_strip_metadata_html'}) {
748        ($dont_strip = $self->{'no_strip_metadata_html'}) =~ s{,}{|}g;
749    }
750
751    my $found_something = 0; my $top = 1;
752    while ($$textref =~ s/^(.*?)$opencom(.*?)$closecom//s) {
753        my $text = $1;
754        my $comment = $2;
755        if (defined $text) {
756        # text before a comment - note that getting to here
757        # doesn't necessarily mean there are Section tags in
758        # the document
759        $self->process_section(\$text, $base_dir, $file, $doc_obj, $cursection);
760        }
761        while ($comment =~ s/$lt(.*?)$gt//s) {
762        my $tag = $1;
763        if ($tag eq "Section") {
764            $found_something = 1;
765            $cursection = $doc_obj->insert_section($doc_obj->get_end_child($cursection)) unless $top;
766            $top = 0;
767        } elsif ($tag eq "/Section") {
768            $found_something = 1;
769            $cursection = $doc_obj->get_parent_section ($cursection);
770        } elsif ($tag =~ m/^Metadata name=$quot(.*?)$quot/s) {
771            my $metaname = $1;
772            my $accumulate = $tag =~ m/mode=${quot}accumulate${quot}/ ? 1 : 0;
773            $comment =~ s/^(.*?)$lt\/Metadata$gt//s;
774            my $metavalue = $1;
775            $metavalue =~ s/^\s+//;
776            $metavalue =~ s/\s+$//;
777                    # assume that no metadata value intentionally includes
778                    # carriage returns or HTML tags (if they're there they
779                    # were probably introduced when converting to HTML from
780                    # some other format).
781            # actually some people want to have html tags in their
782            # metadata.
783            $metavalue =~ s/[\cJ\cM]/ /sg;
784            $metavalue =~ s/<[^>]+>//sg
785            unless $dont_strip && ($dont_strip eq 'all' || $metaname =~ m/^($dont_strip)$/);
786            $metavalue =~ s/\s+/ /sg;
787            if ($accumulate) {
788            $doc_obj->add_utf8_metadata($cursection, $metaname, $metavalue);
789            } else {
790            $doc_obj->set_utf8_metadata_element($cursection, $metaname, $metavalue);   
791            }
792        } elsif ($tag eq "Description" || $tag eq "/Description") {
793            # do nothing with containing Description tags
794        } else {
795            # simple HTML tag (probably created by the conversion
796            # to HTML from some other format) - we'll ignore it and
797            # hope for the best ;-)
798        }
799        }
800    }
801    if ($cursection ne "") {
802        print $outhandle "HTMLPlugin: WARNING: $file contains unmatched <Section></Section> tags\n";
803    }
804
805    $$textref =~ s/^.*?<body[^>]*>//is;
806    $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
807    if ($$textref =~ m/\S/) {
808        if (!$found_something) {
809        if ($self->{'verbosity'} > 2) {
810            print $outhandle "HTMLPlugin: WARNING: $file appears to contain no Section tags so\n";
811            print $outhandle "          will be processed as a single section document\n";
812        }
813
814        # go ahead and process single-section document
815        $self->process_section($textref, $base_dir, $file, $doc_obj, $cursection);
816
817        # if document contains no Section tags we'll go ahead
818        # and extract metadata (this won't have been done
819        # above as the -description_tags option prevents it)
820        my $complete_text = $head_keep.$doc_obj->get_text($cursection);
821        $self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
822            unless $self->{'no_metadata'};
823
824        } else {
825        print $outhandle "HTMLPlugin: WARNING: $file contains the following text outside\n";
826        print $outhandle "          of the final closing </Section> tag. This text will\n";
827        print $outhandle "          be ignored.";
828
829        my ($text);
830        if (length($$textref) > 30) {
831            $text = substr($$textref, 0, 30) . "...";
832        } else {
833            $text = $$textref;
834        }
835        $text =~ s/\n/ /isg;
836        print $outhandle " ($text)\n";
837        }
838    } elsif (!$found_something) {
839
840        if ($self->{'verbosity'} > 2) {
841        # may get to here if document contained no valid Section
842        # tags but did contain some comments. The text will have
843        # been processed already but we should print the warning
844        # as above and extract metadata
845        print $outhandle "HTMLPlugin: WARNING: $file appears to contain no Section tags and\n";
846        print $outhandle "          is blank or empty.  Metadata will be assigned if present.\n";
847        }
848
849        my $complete_text = $head_keep.$doc_obj->get_text($cursection);
850        $self->extract_metadata (\$complete_text, $metadata, $doc_obj, $cursection)
851        unless $self->{'no_metadata'};
852    }
853   
854    } else {
855
856    # remove header and footer
857    if (!$self->{'keep_head'} || $self->{'description_tags'}) {
858        $$textref =~ s/^.*?<body[^>]*>//is;
859        $$textref =~ s/(<\/body[^>]*>|<\/html[^>]*>)//isg;
860    }
861
862    # single section document
863    $self->process_section($textref, $base_dir, $file, $doc_obj, $cursection);
864    }
865    return 1;
866}
867
868
869sub process_heading
870{
871    my ($self, $nHeadNo, $strHeadingText, $arrSections, $file) = @_;
872    $strHeadingText = '' if (!defined($strHeadingText));
873
874    my $strMetadata = $self->update_section_data($arrSections, int($nHeadNo));
875
876    my $strSecMetadata = '';
877    while ($strHeadingText =~ s/<!--gsdl-metadata(.*?)-->//is)
878    {
879    $strSecMetadata .= $1;
880    }
881
882    $strHeadingText =~ s/^\s+//g;
883    $strHeadingText =~ s/\s+$//g;
884    $strSecMetadata =~ s/^\s+//g;
885    $strSecMetadata =~ s/\s+$//g;
886
887    $strMetadata .= "\n<Section>\n\t<Description>\n\t\t<Metadata name=\"Title\">" . $strHeadingText . "</Metadata>\n";
888
889    if (length($strSecMetadata)) {
890    $strMetadata .= "\t\t" . $strSecMetadata . "\n";
891    }
892
893    $strMetadata .= "\t</Description>\n";
894
895    return "<!--" . $strMetadata . "-->";
896}
897
898
899sub update_section_data
900{
901    my ($self, $arrSections, $nCurTocNo) = @_;
902    my ($strBuffer, $nLast, $nSections) = ('', 0, scalar(@$arrSections));
903
904    if ($nSections == 0) {
905    push @$arrSections, $nCurTocNo;
906    return $strBuffer;
907    }
908    $nLast = $arrSections->[$nSections - 1];
909    if ($nCurTocNo > $nLast) {
910    push @$arrSections, $nCurTocNo;
911    return $strBuffer;
912    }
913    for(my $i = $nSections - 1; $i >= 0; $i--) {
914    if ($nCurTocNo <= $arrSections->[$i]) {
915        $strBuffer .= "\n</Section>";
916        pop @$arrSections;
917    }
918    }
919    push @$arrSections, $nCurTocNo;
920    return $strBuffer;
921}
922
923
924# note that process_section may be called multiple times for a single
925# section (relying on the fact that add_utf8_text appends the text to any
926# that may exist already).
927sub process_section {
928    my $self = shift (@_);
929    my ($textref, $base_dir, $file, $doc_obj, $cursection) = @_;
930    # trap links
931    if (!$self->{'nolinks'}) {
932    # usemap="./#index" not handled correctly => change to "#index"
933##  $$textref =~ s/(<img[^>]*?usemap\s*=\s*[\"\']?)([^\"\'>\s]+)([\"\']?[^>]*>)/
934
935    $$textref =~ s/(<img[^>]*?usemap\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
936        $self->replace_usemap_links($1, $2, $3)/isge;
937
938##  $$textref =~ s/(<(?:a|area|frame|link|script)\s+[^>]*?\s*(?:href|src)\s*=\s*[\"\']?)([^\"\'>\s]+)([\"\']?[^>]*>)/
939
940    $$textref =~ s/(<(?:a|area|frame|link|script)\s+[^>]*?\s*(?:href|src)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
941        $self->replace_href_links ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
942    }
943
944    # trap images
945
946    # Previously, by default, HTMLPlugin would embed <img> tags inside anchor tags
947    # i.e. <a href="image><img src="image"></a> in order to overcome a problem that
948    # turned regular text succeeding images into links. That is, by embedding <imgs>
949    # inside <a href=""></a>, the text following images were no longer misbehaving.
950    # However, there would be many occasions whereby images were not meant to link
951    # to their source images but where the images would link to another web page.
952    # To allow this, the no_image_links option was introduced: it would prevent
953    # the behaviour of embedding images into links that referenced the source images.
954
955    # Somewhere along the line, the problem of normal text turning into links when
956    # such text followed images which were not embedded in <a href=""></a> ceased
957    # to occur. This is why the following lines have been commented out (as well as
958    # two lines in replace_images). They appear to no longer apply.
959
960    # If at any time, there is a need for having images embedded in <a> anchor tags,
961    # then it might be better to turn that into an HTMLPlugin option rather than make
962    # it the default behaviour. Also, eventually, no_image_links needs to become
963    # a deprecated option for HTMLPlugin as it has now become the default behaviour.
964
965    #if(!$self->{'no_image_links'}){
966    $$textref =~ s/(<(?:img|embed|table|tr|td)[^>]*?(?:src|background)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/
967    $self->replace_images ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
968    #}
969
970    # add text to document object
971    # turn \ into \\ so that the rest of greenstone doesn't think there
972    # is an escape code following. (Macro parsing loses them...)
973    $$textref =~ s/\\/\\\\/go;
974   
975    $doc_obj->add_utf8_text($cursection, $$textref);
976}
977
978sub replace_images {
979    my $self = shift (@_);
980    my ($front, $link, $back, $base_dir,
981    $file, $doc_obj, $section) = @_;
982
983    # remove quotes from link at start and end if necessary
984    if ($link=~/^[\"\']/) {
985    $link=~s/^[\"\']//;
986    $link=~s/[\"\']$//;
987    $front.='"';
988    $back="\"$back";
989    }
990
991    $link =~ s/\n/ /g;
992
993    # Hack to overcome Windows wv 0.7.1 bug that causes embedded images to be broken
994    # If the Word file path has spaces in it, wv messes up and you end up with
995    #   absolute paths for the images, and without the "file://" prefix
996    # So check for this special case and massage the data to be correct
997    if ($ENV{'GSDLOS'} =~ m/^windows/i && $self->{'plugin_type'} eq "WordPlug" && $link =~ m/^[A-Za-z]\:\\/) {
998    $link =~ s/^.*\\([^\\]+)$/$1/;
999    }
1000   
1001    my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
1002
1003    my $img_file =  $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section);
1004
1005#    print STDERR "**** link = $link\n**** href = $href\n**** img_file = $img_file, rl = $rl\n";
1006
1007    my $anchor_name = $img_file;
1008    #$anchor_name =~ s/^.*\///;
1009    #$anchor_name = "<a name=\"$anchor_name\" ></a>";
1010
1011    my $image_link = $front . $img_file .$back;
1012    return $image_link;
1013
1014    # The reasons for why the following two lines are no longer necessary can be
1015    # found in subroutine process_section
1016    #my $anchor_link = "<a href=\"$img_file\" >".$image_link."</a>"; 
1017    #return $anchor_link;   
1018   
1019    #return $front . $img_file . $back . $anchor_name;
1020}
1021
1022sub replace_href_links {
1023    my $self = shift (@_);
1024    my ($front, $link, $back, $base_dir, $file, $doc_obj, $section) = @_;
1025
1026    # remove quotes from link at start and end if necessary
1027    if ($link=~/^[\"\']/) {
1028    $link=~s/^[\"\']//;
1029    $link=~s/[\"\']$//;
1030    $front.='"';
1031    $back="\"$back";
1032    }
1033
1034    # attempt to sort out targets - frames are not handled
1035    # well in this plugin and some cases will screw things
1036    # up - e.g. the _parent target (so we'll just remove
1037    # them all ;-)
1038    $front =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
1039    $back =~ s/(target=\"?)_top(\"?)/$1_gsdltop_$2/is;
1040    $front =~ s/target=\"?_parent\"?//is;
1041    $back =~ s/target=\"?_parent\"?//is;
1042
1043    return $front . $link . $back if $link =~ m/^\#/s;
1044    $link =~ s/\n/ /g;
1045
1046    # Find file referred to by $link on file system
1047    # This is more complicated than it sounds when char encodings
1048    # is taken in to account
1049    my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
1050 
1051    # href may use '\'s where '/'s should be on Windows
1052    $href =~ s/\\/\//g;
1053    my ($filename) = $href =~ m/^(?:.*?):(?:\/\/)?(.*)/;
1054   
1055
1056    ##### leave all these links alone (they won't be picked up by intermediate
1057    ##### pages). I think that's safest when dealing with frames, targets etc.
1058    ##### (at least until I think of a better way to do it). Problems occur with
1059    ##### mailto links from within small frames, the intermediate page is displayed
1060    ##### within that frame and can't be seen. There is still potential for this to
1061    ##### happen even with html pages - the solution seems to be to somehow tell
1062    ##### the browser from the server side to display the page being sent (i.e.
1063    ##### the intermediate page) in the top level window - I'm not sure if that's
1064    ##### possible - the following line should probably be deleted if that can be done
1065    return $front . $link . $back if $href =~ m/^(mailto|news|gopher|nntp|telnet|javascript):/is;
1066
1067
1068    if (($rl == 0) || ($filename =~ m/$self->{'process_exp'}/) ||
1069    ($href =~ m/\/$/) || ($href =~ m/^(mailto|news|gopher|nntp|telnet|javascript):/i)) {
1070    &ghtml::urlsafe ($href);
1071    return $front . "_httpextlink_&rl=" . $rl . "&href=" . $href . $hash_part . $back;
1072    } else {
1073    # link is to some other type of file (eg image) so we'll
1074    # need to associate that file
1075    return $front . $self->add_file ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) . $back;
1076    }
1077}
1078
1079sub add_file {
1080    my $self = shift (@_);
1081    my ($href, $rl, $hash_part, $base_dir, $doc_obj, $section) = @_;
1082    my ($newname);
1083
1084    my $filename = $href;
1085    if ($base_dir eq "") {
1086    # remove http:/ thereby leaving one slash at the start
1087    $filename =~ s/^[^:]*:\///;
1088    }
1089    else {
1090    # remove http://
1091    $filename =~ s/^[^:]*:\/\///;
1092    }
1093
1094    $filename = &util::filename_cat($base_dir, $filename);
1095    # Replace %XX's in URL with decoded value if required. Note that the filename may include the %XX in some
1096    # situations. If the *original* file's name was in URL encoding, the following method will not decode it.
1097    my $utf8_filename = $filename;
1098    $filename = $self->opt_url_decode($utf8_filename);
1099
1100    # some special processing if the intended filename was converted to utf8, but
1101    # the actual file still needs to be renamed
1102    if (!-e $filename) {
1103    # try the original filename stored in map
1104    my $original_filename = $self->{'utf8_to_original_filename'}->{$filename};
1105    if (defined $original_filename && -e $original_filename) {
1106        $filename = $original_filename;
1107    }
1108    }
1109   
1110    my ($ext) = $filename =~ m/(\.[^\.]*)$/;
1111
1112    if ($rl == 0) {
1113    if ((!defined $ext) || ($ext !~ m/$self->{'assoc_files'}/)) {
1114        return "_httpextlink_&rl=0&el=prompt&href=" . $href . $hash_part;
1115    }
1116    else {
1117        return "_httpextlink_&rl=0&el=direct&href=" . $href . $hash_part;
1118    }
1119    }
1120
1121    if ((!defined $ext) || ($ext !~ m/$self->{'assoc_files'}/)) {
1122    return "_httpextlink_&rl=" . $rl . "&href=" . $href . $hash_part;
1123    }
1124    if ($self->{'rename_assoc_files'}) {
1125    if (defined $self->{'aux_files'}->{$href}) {
1126        $newname = $self->{'aux_files'}->{$href}->{'dir_num'} . "/" .
1127        $self->{'aux_files'}->{$href}->{'file_num'} . $ext;
1128    } else {
1129        $newname = $self->{'dir_num'} . "/" . $self->{'file_num'} . $ext;
1130        $self->{'aux_files'}->{$href} = {'dir_num' => $self->{'dir_num'}, 'file_num' => $self->{'file_num'}};
1131        $self->inc_filecount ();
1132    }
1133    $doc_obj->associate_file($filename, $newname, undef, $section);
1134    return "_httpdocimg_/$newname";
1135    } else {
1136    if(&unicode::is_url_encoded($utf8_filename)) {
1137        # use the possibly-decoded filename instead to avoid double URL encoding
1138        ($newname) = $filename =~ m/([^\/\\]*)$/;
1139    } else {
1140        ($newname) = $utf8_filename =~ m/([^\/\\]*)$/;
1141    }
1142
1143    # Make sure this name uses only ASCII characters.
1144    # We use either base64 or URL encoding, as these preserve original encoding
1145    $newname = &util::rename_file($newname, $self->{'file_rename_method'});
1146
1147    $doc_obj->associate_file($filename, $newname, undef, $section);
1148
1149    # Since the generated image will be URL-encoded to avoid file-system/browser mess-ups
1150    # of filenames, URL-encode the additional percent signs of the URL-encoded filename
1151    my $newname_url = $newname;
1152    $newname_url = &unicode::filename_to_url($newname_url);
1153    return "_httpdocimg_/$newname_url";
1154    }
1155}
1156
1157
1158sub format_link {
1159    my $self = shift (@_);
1160    my ($link, $base_dir, $file) = @_;
1161
1162    my ($before_hash, $hash_part) = $link =~ m/^([^\#]*)(\#?.*)$/;
1163   
1164    $hash_part = "" if !defined $hash_part;
1165    if (!defined $before_hash || $before_hash !~ m/[\w\.\/]/) {
1166    my $outhandle = $self->{'outhandle'};
1167    print $outhandle "HTMLPlugin: ERROR - badly formatted tag ignored ($link)\n"
1168        if $self->{'verbosity'};
1169    return ($link, "", 0);
1170    }
1171
1172    if ($before_hash =~ s@^((?:http|ftp|file)://)@@i) {
1173    my $type = $1;
1174
1175    if ($link =~ m/^(http|ftp):/i) {
1176        # Turn url (using /) into file name (possibly using \ on windows)
1177        my @http_dir_split = split('/', $before_hash);
1178        $before_hash = &util::filename_cat(@http_dir_split);
1179    }
1180
1181    $before_hash = $self->eval_dir_dots($before_hash);
1182   
1183    my $linkfilename = &util::filename_cat ($base_dir, $before_hash);
1184   
1185    my $rl = 0;
1186    $rl = 1 if (-e $linkfilename);
1187
1188    # make sure there's a slash on the end if it's a directory
1189    if ($before_hash !~ m/\/$/) {
1190        $before_hash .= "/" if (-d $linkfilename);
1191    }
1192    return ($type . $before_hash, $hash_part, $rl);
1193   
1194    } elsif ($link !~ m/^(mailto|news|gopher|nntp|telnet|javascript):/i && $link !~ m/^\//) {
1195
1196    if ($before_hash =~ s@^/@@ || $before_hash =~ m/\\/) {
1197
1198        # the first directory will be the domain name if file_is_url
1199        # to generate archives, otherwise we'll assume all files are
1200        # from the same site and base_dir is the root
1201
1202        if ($self->{'file_is_url'}) {
1203        my @dirs = split /[\/\\]/, $file;
1204        my $domname = shift (@dirs);
1205        $before_hash = &util::filename_cat($domname, $before_hash);
1206        $before_hash =~ s@\\@/@g; # for windows
1207        }
1208        else
1209        {
1210        # see if link shares directory with source document
1211        # => turn into relative link if this is so!
1212       
1213        if ($ENV{'GSDLOS'} =~ m/^windows/i) {
1214            # too difficult doing a pattern match with embedded '\'s...
1215            my $win_before_hash=$before_hash;
1216            $win_before_hash =~ s@(\\)+@/@g;
1217            # $base_dir is already similarly "converted" on windows.
1218            if ($win_before_hash =~ s@^$base_dir/@@o) {
1219            # if this is true, we removed a prefix
1220            $before_hash=$win_before_hash;
1221            }
1222        }
1223        else {
1224            # before_hash has lost leading slash by this point,
1225            # -> add back in prior to substitution with $base_dir
1226            $before_hash = "/$before_hash";
1227
1228            $before_hash = &util::filename_cat("",$before_hash);
1229            $before_hash =~ s@^$base_dir/@@;
1230        }
1231        }
1232    } else {
1233        # Turn relative file path into full path
1234        my $dirname = &File::Basename::dirname($file);
1235        $before_hash = &util::filename_cat($dirname, $before_hash);
1236        $before_hash = $self->eval_dir_dots($before_hash);             
1237    }
1238
1239    my $linkfilename = &util::filename_cat ($base_dir, $before_hash);
1240    # make sure there's a slash on the end if it's a directory
1241    if ($before_hash !~ m/\/$/) {
1242        $before_hash .= "/" if (-d $linkfilename);
1243    }
1244    return ("http://" . $before_hash, $hash_part, 1);
1245    } else {
1246    # mailto, news, nntp, telnet, javascript or gopher link
1247    return ($before_hash, "", 0);
1248    }
1249}
1250
1251sub extract_first_NNNN_characters {
1252    my $self = shift (@_);
1253    my ($textref, $doc_obj, $thissection) = @_;
1254   
1255    foreach my $size (split /,/, $self->{'first'}) {
1256    my $tmptext =  $$textref;
1257    # skip to the body
1258    $tmptext =~ s/.*<body[^>]*>//i;
1259    # remove javascript
1260    $tmptext =~ s@<script.*?</script>@ @sig;
1261    $tmptext =~ s/<[^>]*>/ /g;
1262    $tmptext =~ s/&nbsp;/ /g;
1263    $tmptext =~ s/^\s+//;
1264    $tmptext =~ s/\s+$//;
1265    $tmptext =~ s/\s+/ /gs;
1266    $tmptext = &unicode::substr ($tmptext, 0, $size);
1267    $tmptext =~ s/\s\S*$/&#8230;/; # adds an ellipse (...)
1268    $doc_obj->add_utf8_metadata ($thissection, "First$size", $tmptext);
1269    }
1270}
1271
1272
1273sub extract_metadata {
1274    my $self = shift (@_);
1275    my ($textref, $metadata, $doc_obj, $section) = @_;
1276    my $outhandle = $self->{'outhandle'};
1277    # if we don't want metadata, we may as well not be here ...
1278    return if (!defined $self->{'metadata_fields'});
1279
1280    # metadata fields to extract/save. 'key' is the (lowercase) name of the
1281    # html meta, 'value' is the metadata name for greenstone to use
1282    my %find_fields = ();
1283
1284    my %creator_fields = (); # short-cut for lookups
1285
1286
1287    foreach my $field (split /,/, $self->{'metadata_fields'}) {
1288        $field =~ s/^\s+//; # remove leading whitespace
1289        $field =~ s/\s+$//; # remove trailing whitespace
1290
1291    # support tag<tagname>
1292    if ($field =~ m/^(.*?)<(.*?)>$/) {
1293        # "$2" is the user's preferred gs metadata name
1294        $find_fields{lc($1)}=$2; # lc = lowercase
1295    } else { # no <tagname> for mapping
1296        # "$field" is the user's preferred gs metadata name
1297        $find_fields{lc($field)}=$field; # lc = lowercase
1298    }
1299    }
1300
1301    if (defined $self->{'hunt_creator_metadata'} &&
1302    $self->{'hunt_creator_metadata'} == 1 ) {
1303    my @extra_fields =
1304        (
1305         'author',
1306         'author.email',
1307         'creator',
1308         'dc.creator',
1309         'dc.creator.corporatename',
1310         );
1311
1312    # add the creator_metadata fields to search for
1313    foreach my $field (@extra_fields) {
1314        $creator_fields{$field}=0; # add to lookup hash
1315    }
1316    }
1317
1318
1319    # find the header in the html file, which has the meta tags
1320    $$textref =~ m@<head>(.*?)</head>@si;
1321
1322    my $html_header=$1;
1323
1324    # go through every <meta... tag defined in the html and see if it is
1325    # one of the tags we want to match.
1326   
1327    # special case for title - we want to remember if its been found
1328    my $found_title = 0;
1329    # this assumes that ">" won't appear. (I don't think it's allowed to...)
1330    $html_header =~ m/^/; # match the start of the string, for \G assertion
1331   
1332    while ($html_header =~ m/\G.*?<meta(.*?)>/sig) {
1333    my $metatag=$1;
1334    my ($tag, $value);
1335
1336    # find the tag name
1337    $metatag =~ m/(?:name|http-equiv)\s*=\s*([\"\'])?(.*?)\1/is;
1338    $tag=$2;
1339    # in case they're not using " or ', but they should...
1340    if (! $tag) {
1341        $metatag =~ m/(?:name|http-equiv)\s*=\s*([^\s\>]+)/is;
1342        $tag=$1;
1343    }
1344
1345    if (!defined $tag) {
1346        print $outhandle "HTMLPlugin: can't find NAME in \"$metatag\"\n";
1347        next;
1348    }
1349   
1350    # don't need to assign this field if it was passed in from a previous
1351    # (recursive) plugin
1352    if (defined $metadata->{$tag}) {next}
1353
1354    # find the tag content
1355    $metatag =~ m/content\s*=\s*([\"\'])?(.*?)\1/is;
1356    $value=$2;
1357
1358    if (! $value) {
1359        $metatag =~ m/(?:name|http-equiv)\s*=\s*([^\s\>]+)/is;
1360        $value=$1;
1361    }
1362    if (!defined $value) {
1363        print $outhandle "HTMLPlugin: can't find VALUE in \"$metatag\"\n";
1364        next;
1365    }
1366
1367    # clean up and add
1368    $value =~ s/\s+/ /gs;
1369    chomp($value); # remove trailing \n, if any
1370    if (exists $creator_fields{lc($tag)}) {
1371        # map this value onto greenstone's "Creator" metadata
1372        $tag='Creator';
1373    } elsif (!exists $find_fields{lc($tag)}) {
1374        next; # don't want this tag
1375    } else {
1376        # get the user's preferred capitalisation
1377        $tag = $find_fields{lc($tag)};
1378    }
1379    if (lc($tag) eq "title") {
1380        $found_title = 1;
1381    }
1382    print $outhandle " extracted \"$tag\" metadata \"$value\"\n"
1383        if ($self->{'verbosity'} > 2);
1384    if ($tag =~ m/date.*/i){
1385        $tag = lc($tag);
1386    }
1387    $doc_obj->add_utf8_metadata($section, $tag, $value);
1388
1389    }
1390   
1391    # TITLE: extract the document title
1392    if (exists $find_fields{'title'} && !$found_title) {
1393    # we want a title, and didn't find one in the meta tags
1394    # see if there's a <title> tag
1395    my $title;
1396    my $from = ""; # for debugging output only
1397    if ($html_header =~ m/<title[^>]*>([^<]+)<\/title[^>]*>/is) {
1398        $title = $1;
1399        $from = "<title> tags";
1400    }
1401
1402    if (!defined $title) {
1403        $from = "first 100 chars";
1404        # if no title use first 100 or so characters
1405        $title = $$textref;
1406        $title =~ s/^\xFE\xFF//; # Remove unicode byte order mark       
1407        $title =~ s/^.*?<body>//si;
1408        # ignore javascript!
1409        $title =~ s@<script.*?</script>@ @sig;
1410        $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
1411        $title =~ s/<[^>]*>/ /g; # remove all HTML tags
1412        $title = substr ($title, 0, 100);
1413        $title =~ s/\s\S*$/.../;
1414    }
1415    $title =~ s/<[^>]*>/ /g; # remove html tags
1416    $title =~ s/&nbsp;/ /g;
1417    $title =~ s/(?:&nbsp;|\xc2\xa0)/ /g; # utf-8 for nbsp...
1418    $title =~ s/\s+/ /gs; # collapse multiple spaces
1419    $title =~ s/^\s*//;   # remove leading spaces
1420    $title =~ s/\s*$//;   # remove trailing spaces
1421
1422    $title =~ s/^$self->{'title_sub'}// if ($self->{'title_sub'});
1423    $title =~ s/^\s+//s; # in case title_sub introduced any...
1424    $doc_obj->add_utf8_metadata ($section, 'Title', $title);
1425    print $outhandle " extracted Title metadata \"$title\" from $from\n"
1426        if ($self->{'verbosity'} > 2);
1427    }
1428   
1429    # add FileFormat metadata
1430    $doc_obj->add_metadata($section,"FileFormat", "HTML");
1431
1432    # Special, for metadata names such as tagH1 - extracts
1433    # the text between the first <H1> and </H1> tags into "H1" metadata.
1434
1435    foreach my $field (keys %find_fields) {
1436    if ($field !~ m/^tag([a-z0-9]+)$/i) {next}
1437    my $tag = $1;
1438    if ($$textref =~ m@<$tag[^>]*>(.*?)</$tag[^>]*>@g) {
1439        my $content = $1;
1440        $content =~ s/&nbsp;/ /g;
1441        $content =~ s/<[^>]*>/ /g;
1442        $content =~ s/^\s+//;
1443        $content =~ s/\s+$//;
1444        $content =~ s/\s+/ /gs;
1445        if ($content) {
1446        $tag=$find_fields{"tag$tag"}; # get the user's capitalisation
1447        $tag =~ s/^tag//i;
1448        $doc_obj->add_utf8_metadata ($section, $tag, $content);
1449        print $outhandle " extracted \"$tag\" metadata \"$content\"\n"
1450            if ($self->{'verbosity'} > 2);
1451        }
1452    }
1453    }   
1454}
1455
1456
1457# evaluate any "../" to next directory up
1458# evaluate any "./" as here
1459sub eval_dir_dots {
1460    my $self = shift (@_);
1461    my ($filename) = @_;
1462    my $dirsep_os = &util::get_os_dirsep();
1463    my @dirsep = split(/$dirsep_os/,$filename);
1464
1465    my @eval_dirs = ();
1466    foreach my $d (@dirsep) {
1467    if ($d eq "..") {
1468        pop(@eval_dirs);
1469       
1470    } elsif ($d eq ".") {
1471        # do nothing!
1472
1473    } else {
1474        push(@eval_dirs,$d);
1475    }
1476    }
1477
1478    # Need to fiddle with number of elements in @eval_dirs if the
1479    # first one is the empty string.  This is because of a
1480    # modification to util::filename_cat that supresses the addition
1481    # of a leading '/' character (or \ if windows) (intended to help
1482    # filename cat with relative paths) if the first entry in the
1483    # array is the empty string.  Making the array start with *two*
1484    # empty strings is a way to defeat this "smart" option.
1485    #
1486    if (scalar(@eval_dirs) > 0) {
1487    if ($eval_dirs[0] eq ""){
1488        unshift(@eval_dirs,"");
1489    }
1490    }
1491   
1492    my $evaluated_filename = (scalar @eval_dirs > 0) ? &util::filename_cat(@eval_dirs) : "";
1493    return $evaluated_filename;
1494}
1495
1496sub replace_usemap_links {
1497    my $self = shift (@_);
1498    my ($front, $link, $back) = @_;
1499
1500    # remove quotes from link at start and end if necessary
1501    if ($link=~/^[\"\']/) {
1502    $link=~s/^[\"\']//;
1503    $link=~s/[\"\']$//;
1504    $front.='"';
1505    $back="\"$back";
1506    }
1507
1508    $link =~ s/^\.\///;
1509    return $front . $link . $back;
1510}
1511
1512sub inc_filecount {
1513    my $self = shift (@_);
1514
1515    if ($self->{'file_num'} == 1000) {
1516    $self->{'dir_num'} ++;
1517    $self->{'file_num'} = 0;
1518    } else {
1519    $self->{'file_num'} ++;
1520    }
1521}
1522
1523
1524# Extend read_file so that strings like &eacute; are
1525# converted to UTF8 internally. 
1526#
1527# We don't convert &lt; or &gt; or &amp; or &quot; in case
1528# they interfere with the GML files
1529
1530sub read_file {
1531    my $self = shift(@_);
1532    my ($filename, $encoding, $language, $textref) = @_;
1533
1534    $self->SUPER::read_file($filename, $encoding, $language, $textref);
1535
1536    # Convert entities to their UTF8 equivalents
1537    $$textref =~ s/&(lt|gt|amp|quot|nbsp);/&z$1;/go;
1538    $$textref =~ s/&([^;]+);/&ghtml::getcharequiv($1,1)/gseo;
1539    $$textref =~ s/&z(lt|gt|amp|quot|nbsp);/&$1;/go;
1540}
1541
15421;
Note: See TracBrowser for help on using the browser.