source: main/trunk/greenstone2/perllib/plugins/PDFPlugin.pm@ 30742

Last change on this file since 30742 was 30742, checked in by kjdon, 5 years ago

paged docs without images look weird in gs3. need to make a new type, eg pagedimg so that we can differentiate those with and without images. For those without images we should do a different toc widget. for now, set thistype to Hierarchy for PDF -use_sections docs

  • Property svn:keywords set to Author Date Id Revision
File size: 16.6 KB
Line 
1###########################################################################
2#
3# PDFPlugin.pm -- reasonably with-it pdf plugin
4# A component of the Greenstone digital library software
5# from the New Zealand Digital Library Project at the
6# University of Waikato, New Zealand.
7#
8# Copyright (C) 1999-2001 New Zealand Digital Library Project
9#
10# This program is free software; you can redistribute it and/or modify
11# it under the terms of the GNU General Public License as published by
12# the Free Software Foundation; either version 2 of the License, or
13# (at your option) any later version.
14#
15# This program is distributed in the hope that it will be useful,
16# but WITHOUT ANY WARRANTY; without even the implied warranty of
17# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18# GNU General Public License for more details.
19#
20# You should have received a copy of the GNU General Public License
21# along with this program; if not, write to the Free Software
22# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
23#
24###########################################################################
25package PDFPlugin;
26
27use strict;
28no strict 'refs'; # so we can use a var for filehandles (e.g. STDERR)
29
30use ReadTextFile;
31use unicode;
32
33use AutoLoadConverters;
34use ConvertBinaryFile;
35
36@PDFPlugin::ISA = ('ConvertBinaryFile', 'AutoLoadConverters', 'ReadTextFile');
37
38
39my $convert_to_list =
40 [ { 'name' => "auto",
41 'desc' => "{ConvertBinaryFile.convert_to.auto}" },
42 { 'name' => "html",
43 'desc' => "{ConvertBinaryFile.convert_to.html}" },
44 { 'name' => "text",
45 'desc' => "{ConvertBinaryFile.convert_to.text}" },
46 { 'name' => "pagedimg_jpg",
47 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_jpg}"},
48 { 'name' => "pagedimg_gif",
49 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_gif}"},
50 { 'name' => "pagedimg_png",
51 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_png}"},
52 ];
53
54
55my $arguments =
56 [
57 { 'name' => "convert_to",
58 'desc' => "{ConvertBinaryFile.convert_to}",
59 'type' => "enum",
60 'reqd' => "yes",
61 'list' => $convert_to_list,
62 'deft' => "html" },
63 { 'name' => "process_exp",
64 'desc' => "{BasePlugin.process_exp}",
65 'type' => "regexp",
66 'deft' => &get_default_process_exp(),
67 'reqd' => "no" },
68 { 'name' => "block_exp",
69 'desc' => "{BasePlugin.block_exp}",
70 'type' => "regexp",
71 'deft' => &get_default_block_exp() },
72 { 'name' => "metadata_fields",
73 'desc' => "{HTMLPlugin.metadata_fields}",
74 'type' => "string",
75 'deft' => "Title,Author,Subject,Keywords" },
76 { 'name' => "metadata_field_separator",
77 'desc' => "{HTMLPlugin.metadata_field_separator}",
78 'type' => "string",
79 'deft' => "" },
80 { 'name' => "noimages",
81 'desc' => "{PDFPlugin.noimages}",
82 'type' => "flag" },
83 { 'name' => "allowimagesonly",
84 'desc' => "{PDFPlugin.allowimagesonly}",
85 'type' => "flag" },
86 { 'name' => "complex",
87 'desc' => "{PDFPlugin.complex}",
88 'type' => "flag" },
89 { 'name' => "nohidden",
90 'desc' => "{PDFPlugin.nohidden}",
91 'type' => "flag" },
92 { 'name' => "zoom",
93 'desc' => "{PDFPlugin.zoom}",
94 'deft' => "2",
95 'range' => "1,3", # actually the range is 0.5-3
96 'type' => "int" },
97 { 'name' => "use_sections",
98 'desc' => "{PDFPlugin.use_sections}",
99 'type' => "flag" },
100 { 'name' => "description_tags",
101 'desc' => "{HTMLPlugin.description_tags}",
102 'type' => "flag" },
103 { 'name' => "use_realistic_book",
104 'desc' => "{PDFPlugin.use_realistic_book}",
105 'type' => "flag"}
106 ];
107
108my $options = { 'name' => "PDFPlugin",
109 'desc' => "{PDFPlugin.desc}",
110 'abstract' => "no",
111 'inherits' => "yes",
112 'srcreplaceable' => "yes", # Source docs in PDF can be replaced with GS-generated html
113 'args' => $arguments };
114
115sub new {
116 my ($class) = shift (@_);
117 my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
118 push(@$pluginlist, $class);
119
120 push(@$inputargs,"-title_sub");
121 push(@$inputargs,'^(Page\s+\d+)?(\s*1\s+)?');
122
123 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
124 push(@{$hashArgOptLists->{"OptList"}},$options);
125
126 my $auto_converter_self = new AutoLoadConverters($pluginlist,$inputargs,$hashArgOptLists,["PDFBoxConverter"],1);
127 my $cbf_self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
128 my $self = BasePlugin::merge_inheritance($auto_converter_self, $cbf_self);
129
130 if ($self->{'info_only'}) {
131 # don't worry about any options etc
132 return bless $self, $class;
133 }
134
135 $self = bless $self, $class;
136 $self->{'file_type'} = "PDF";
137
138 # these are passed through to gsConvert.pl by ConvertBinaryFile.pm
139 my $zoom = $self->{"zoom"};
140 $self->{'convert_options'} = "-pdf_zoom $zoom";
141 $self->{'convert_options'} .= " -pdf_complex" if $self->{"complex"};
142 $self->{'convert_options'} .= " -pdf_nohidden" if $self->{"nohidden"};
143 $self->{'convert_options'} .= " -pdf_ignore_images" if $self->{"noimages"};
144 $self->{'convert_options'} .= " -pdf_allow_images_only" if $self->{"allowimagesonly"};
145
146 # check convert_to
147 if ($self->{'convert_to'} eq "text" && $ENV{'GSDLOS'} =~ /^windows$/i) {
148 print STDERR "Windows does not support pdf to text. PDFs will be converted to HTML instead\n";
149 $self->{'convert_to'} = "html";
150 }
151 elsif ($self->{'convert_to'} eq "auto") {
152 # choose html ?? is this the best option
153 $self->{'convert_to'} = "html";
154 }
155 if ($self->{'use_realistic_book'}) {
156 if ($self->{'convert_to'} ne "html") {
157 print STDERR "PDFs will be converted to HTML for realistic book functionality\n";
158 $self->{'convert_to'} = "html";
159 }
160 }
161 # set convert_to_plugin and convert_to_ext
162 $self->set_standard_convert_settings();
163
164 my $secondary_plugin_name = $self->{'convert_to_plugin'};
165 my $secondary_plugin_options = $self->{'secondary_plugin_options'};
166
167 if (!defined $secondary_plugin_options->{$secondary_plugin_name}) {
168 $secondary_plugin_options->{$secondary_plugin_name} = [];
169 }
170 my $specific_options = $secondary_plugin_options->{$secondary_plugin_name};
171
172 # following title_sub removes "Page 1" added by pdftohtml, and a leading
173 # "1", which is often the page number at the top of the page. Bad Luck
174 # if your document title actually starts with "1 " - is there a better way?
175 push(@$specific_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
176 my $associate_tail_re = $self->{'associate_tail_re'};
177 if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
178 push(@$specific_options, "-associate_tail_re", $associate_tail_re);
179 }
180 push(@$specific_options, "-file_rename_method", "none");
181
182 if ($secondary_plugin_name eq "HTMLPlugin") {
183 # pdftohtml always produces utf8 - What about pdfbox???
184 # push(@$specific_options, "-input_encoding", "utf8");
185 push(@$specific_options, "-extract_language") if $self->{'extract_language'};
186 push(@$specific_options, "-processing_tmp_files");
187 # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
188 # to extract these metadata fields from the HEAD META fields
189 if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
190 push(@$specific_options,"-metadata_fields",$self->{'metadata_fields'});
191 } else {
192 push(@$specific_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
193 }
194 if (defined $self->{'metadata_field_separator'} && $self->{'metadata_field_separator'} =~ /\S/) {
195 push(@$specific_options,"-metadata_field_separator",$self->{'metadata_field_separator'});
196 }
197 if ($self->{'use_sections'} || $self->{'description_tags'}) {
198 $self->{'description_tags'} = 1;
199 push(@$specific_options, "-description_tags");
200 }
201 if ($self->{'use_realistic_book'}) {
202 push(@$specific_options, "-use_realistic_book");
203 }
204 }
205 elsif ($secondary_plugin_name eq "PagedImagePlugin") {
206 push(@$specific_options, "-screenviewsize", "1000");
207 push(@$specific_options, "-enable_cache");
208 push(@$specific_options, "-processing_tmp_files");
209 }
210
211 $self = bless $self, $class;
212 $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
213 return $self;
214}
215
216sub get_default_process_exp {
217 my $self = shift (@_);
218
219 return q^(?i)\.pdf$^;
220}
221
222# so we don't inherit HTMLPlug's block exp...
223sub get_default_block_exp {
224 return "";
225}
226
227sub init {
228 my $self = shift (@_);
229
230 # ConvertBinaryFile init
231 $self->SUPER::init(@_);
232 $self->AutoLoadConverters::init(@_);
233
234}
235
236sub begin {
237 my $self = shift (@_);
238
239 $self->AutoLoadConverters::begin(@_);
240 $self->SUPER::begin(@_);
241
242}
243
244sub deinit {
245 my $self = shift (@_);
246
247 $self->AutoLoadConverters::deinit(@_);
248 $self->SUPER::deinit(@_);
249
250}
251
252# By setting hashing to be on ga xml this ensures that two
253# PDF files that are identical except for the metadata
254# to hash to different values. Without this, when each PDF
255# file is converted to HTML there is a chance that they
256# will both be *identical* if the conversion utility does
257# not embed the metadata in the generated HTML. This is
258# certainly the case when PDFBOX is being used.
259
260# This change makes this convert to based plugin more
261# consistent with the original vision that the same document
262# with different metadata should
263# be seen as different.
264
265sub get_oid_hash_type {
266 my $self = shift (@_);
267 return "hash_on_ga_xml";
268}
269
270
271sub tmp_area_convert_file {
272
273 my $self = shift (@_);
274 return $self->AutoLoadConverters::tmp_area_convert_file(@_);
275
276}
277
278sub convert_post_process
279{
280 my $self = shift (@_);
281 my ($conv_filename) = @_;
282
283 my $outhandle=$self->{'outhandle'};
284
285 #$self->{'input_encoding'} = "utf8"; # The output is always in utf8 (is it?? it is for html, but what about other types?)
286 #my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
287
288 # read in file ($text will be in utf8)
289 my $text = "";
290 # encoding will be utf8 for html files - what about other types? will we do this step for them anyway?
291 $self->read_file ($conv_filename, "utf8", "", \$text);
292
293 # To support the use_sections option with PDFBox: Greenstone splits PDFs into pages for
294 # sections. The PDFPlugin code wants each new page to be prefixed with <a name=pagenum></a>,
295 # which it then splits on to generate page-based sections. However, that's not what PDFBox
296 # generates in its HTML output. Fortunately, PDFBox does have its own page-separator: it
297 # embeds each page in an extra div. The div opener is:
298 # <div style=\"page-break-before:always; page-break-after:always\">
299 # The PDFPlugin now looks for this and prefixes <a name=0></a> to each such div. (The
300 # pagenumber is fixed at 0 since I'm unable to work out how to increment the pagenum during
301 # a regex substitution even with regex extensions on.) Later, when we process each section
302 # to get the pagenum, PDFBox's output for this is pre-processed by having a loopcounter
303 # that increments the pagenum for each subsequent section.
304
305 #$pdfbox_pageheader="\<div style=\"page-break-before:always; page-break-after:always\">";
306 my $loopcounter = 0; # used later on!
307 $text =~ s@\<div style=\"page-break-before:always; page-break-after:always\">@<a name=$loopcounter></a><div style=\"page-break-before:always; page-break-after:always\">@g;
308
309
310 # Calculate number of pages based on <a ...> tags (we have a <a name=1> etc
311 # for each page). Metadata based on this calculation not set until process()
312 #
313 # Note: this is done even if we are not breaking the document into pages as it might
314 # be useful to give an indication of document length in browser through setting
315 # num_pages as metadata.
316 # Clean html from low and hight surrogates D800–DFFF
317 $text =~ s@[\N{U+D800}-\N{U+DFFF}]@\ @g;
318 my @pages = ($text =~ m/\<[Aa] name=\"?\w+\"?>/ig); #<div style=\"?page-break-before:always; page-break-after:always\"?>
319 my $num_pages = scalar(@pages);
320 $self->{'num_pages'} = $num_pages;
321
322 if ($self->{'use_sections'}
323 && $self->{'converted_to'} eq "HTML") {
324
325 print $outhandle "PDFPlugin: Calculating sections...\n";
326
327 # we have "<a name=1></a>" etc for each page
328 # it may be <A name=
329 my @sections = split('<[Aa] name=', $text);
330
331 my $top_section = "";
332
333 if (scalar (@sections) == 1) { #only one section - no split!
334 print $outhandle "PDFPlugin: warning - no sections found\n";
335 } else {
336 $top_section .= shift @sections; # keep HTML header etc as top_section
337 }
338
339 # handle first section specially for title? Or all use first 100...
340
341 my $title = $sections[0];
342 $title =~ s/^\"?\w+\"?>//; # specific for pdftohtml...
343 $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
344 $title =~ s/<[^>]*>/ /g;
345 $title =~ s/(?:&nbsp;|\xc2\xa0)/ /g; # utf-8 for nbsp...
346 $title =~ s/^\s+//s;
347 $title =~ s/\s+$//;
348 $title =~ s/\s+/ /gs;
349 $title =~ s/^$self->{'title_sub'}// if ($self->{'title_sub'});
350 $title =~ s/^\s+//s; # in case title_sub introduced any...
351 $title = substr ($title, 0, 100);
352 $title =~ s/\s\S*$/.../;
353
354
355 if (scalar (@sections) == 1) { # no sections found
356 $top_section .= $sections[0];
357 @sections=();
358 } else {
359 $top_section .= "<!--<Section>\n<Metadata name=\"Title\">$title</Metadata>\n-->\n <!--</Section>-->\n";
360 }
361
362 # add metadata per section...
363 foreach my $section (@sections) {
364 # section names are not always just digits, may be like "outline"
365 $section =~ s@^\"?(\w+)\"?></a>@@; # leftover from split expression...
366
367 $title = $1; # Greenstone does magic if sections are titled digits
368
369 # A title of pagenum=0 means use_sections is being applied on output from PDFBox,
370 # which didn't originally have a <a name=incremented pagenumber></a> to split each page.
371 # Our Perl code then prefixed <a name=0></a> to it. Now need to increment the pagenum here:
372 if($loopcounter > 0 || ($title eq 0 && $loopcounter == 0)) { # implies use_sections with PDFBox
373 $title = ++$loopcounter;
374 }
375
376 if (! defined($title) ) {
377 print STDERR "no title: $section\n";
378 $title = " "; # get rid of the undefined warning in next line
379 }
380 my $newsection = "<!-- from PDFPlugin -->\n<!-- <Section>\n";
381 $newsection .= "<Metadata name=\"Title\">" . $title
382 . "</Metadata>\n--><br />\n";
383 $newsection .= $section;
384 $newsection .= "<!--</Section>-->\n";
385 $section = $newsection;
386 }
387
388 $text=join('', ($top_section, @sections));
389 }
390
391 if ($self->{'use_sections'}
392 && $self->{'converted_to'} eq "text") {
393 print STDERR "**** When converting PDF to text, cannot apply use_sections\n";
394 }
395
396
397 # The following should no longer be needed, now that strings
398 # read in are Unicode aware (in the Perl sense) rather than
399 # raw binary strings that just happen to be UTF-8 compliant
400
401 # turn any high bytes that aren't valid utf-8 into utf-8.
402## unicode::ensure_utf8(\$text);
403
404 # Write it out again!
405 $self->utf8_write_file (\$text, $conv_filename);
406}
407
408
409# do plugin specific processing of doc_obj for HTML type
410sub process {
411 my $self = shift (@_);
412 my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
413
414 my $result = $self->process_type($base_dir,$file,$doc_obj);
415
416 # fix up the extracted date metadata to be in Greenstone date format,
417 # and fix the capitalisation of 'date'
418 my $cursection = $doc_obj->get_top_section();
419 foreach my $datemeta (@{$doc_obj->get_metadata($cursection, "date")}) {
420 $doc_obj->delete_metadata($cursection, "date", $datemeta);
421
422 # We're just interested in the date bit, not the time
423 # some pdf creators (eg "Acrobat 5.0 Scan Plug-in for Windows")
424 # set a /CreationDate, and set /ModDate to 000000000. pdftohtml
425 # extracts the ModDate, so it is 0...
426 $datemeta =~ /(\d+)-(\d+)-(\d+)/;
427 my ($year, $month, $day) = ($1,$2,$3);
428 if (defined($year) && defined($month) && defined($day)) {
429 if ($year == 0) {next}
430 if ($year < 100) {$year += 1900} # just to be safe
431 if ($month =~ /^\d$/) {$month="0$month"} # single digit
432 if ($day =~ /^\d$/) {$day="0$day"} # single digit
433 my $date="$year$month$day";
434 $doc_obj->add_utf8_metadata($cursection, "Date", $date);
435 }
436 }
437
438 $doc_obj->add_utf8_metadata($cursection, "NumPages", $self->{'num_pages'}) if defined $self->{'num_pages'};
439
440 if ($self->{'use_sections'} && $self->{'converted_to'} eq "HTML") {
441 # For gs2 we explicitly make it a paged document, cos greenstone won't get it
442 # right if any section has an empty title, or one with letters in it
443 if (&util::is_gs3()) {
444 # but for gs3, paged docs currently use image slider which is ugly if there are no images
445 $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Hierarchy");
446 } else {
447 $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Paged");
448 }
449 }
450
451 return $result;
452}
453
4541;
Note: See TracBrowser for help on using the repository browser.