source: main/trunk/greenstone2/perllib/plugins/PDFv2Plugin.pm@ 32280

Last change on this file since 32280 was 32280, checked in by ak19, 6 years ago

Implementing PDFv2paged_text (with pdfbox)

File size: 32.4 KB
Line 
1###########################################################################
2#
3# PDFv2Plugin.pm -- pdf plugin that uses xpdftools or, if switched on,
4# pdfbox, to process PDFs.
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) 1999-2001 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###########################################################################
26package PDFv2Plugin;
27
28use strict;
29no strict 'refs'; # so we can use a var for filehandles (e.g. STDERR)
30no strict 'subs'; # allow filehandles to be variables and viceversa
31
32use ReadTextFile;
33use unicode;
34use Mojo::DOM; # for HTML parsing
35
36use AutoLoadConverters;
37use ConvertBinaryFile;
38
39@PDFv2Plugin::ISA = ('ConvertBinaryFile', 'AutoLoadConverters', 'ReadTextFile');
40
41
42my $convert_to_list =
43 [ { 'name' => "auto",
44 'desc' => "{ConvertBinaryFile.convert_to.auto}" },
45 { 'name' => "text", # xpdftools
46 'desc' => "{ConvertBinaryFile.convert_to.text}" },
47 { 'name' => "paged_text", # xpdftools
48 'desc' => "{ConvertBinaryFile.convert_to.paged_text}" },
49
50 { 'name' => "html", # pdfbox ## TODO: rename this to html_without_imgs
51 'desc' => "{PDFPlugin.convert_to.html}" },
52 { 'name' => "pretty_html", # xpdftools
53 'desc' => "{PDFPlugin.convert_to.pretty_html}" },
54 { 'name' => "paged_pretty_html", # xpdftools
55 'desc' => "{PDFPlugin.convert_to.paged_pretty_html}"},
56
57 #pdfbox
58 { 'name' => "pagedimg_jpg",
59 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_jpg}"},
60 { 'name' => "pagedimg_png",
61 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_png}"},
62
63 { 'name' => "pagedimgtxt_jpg",
64 'desc' => "{ConvertBinaryFile.convert_to.pagedimgtxt_jpg}"},
65 { 'name' => "pagedimgtxt_png",
66 'desc' => "{ConvertBinaryFile.convert_to.pagedimgtxt_png}"},
67 ];
68
69
70my $arguments =
71 [
72 { 'name' => "convert_to",
73 'desc' => "{ConvertBinaryFile.convert_to}",
74 'type' => "enum",
75 'reqd' => "yes",
76 'list' => $convert_to_list,
77 'deft' => "pretty_html" },
78 { 'name' => "process_exp",
79 'desc' => "{BaseImporter.process_exp}",
80 'type' => "regexp",
81 'deft' => &get_default_process_exp(),
82 'reqd' => "no" },
83 { 'name' => "block_exp",
84 'desc' => "{CommonUtil.block_exp}",
85 'type' => "regexp",
86 'deft' => &get_default_block_exp() },
87 { 'name' => "metadata_fields",
88 'desc' => "{HTMLPlugin.metadata_fields}",
89 'type' => "string",
90 'deft' => "Title,Author,Subject,Keywords" },
91 { 'name' => "metadata_field_separator",
92 'desc' => "{HTMLPlugin.metadata_field_separator}",
93 'type' => "string",
94 'deft' => "" },
95# { 'name' => "noimages",
96# 'desc' => "{PDFPlugin.noimages}",
97# 'type' => "flag" },
98# { 'name' => "allowimagesonly",
99# 'desc' => "{PDFPlugin.allowimagesonly}",
100# 'type' => "flag" },
101# { 'name' => "complex",
102# 'desc' => "{PDFPlugin.complex}",
103# 'type' => "flag" },
104# { 'name' => "nohidden",
105# 'desc' => "{PDFPlugin.nohidden}",
106# 'type' => "flag" },
107 { 'name' => "zoom",
108 'desc' => "{PDFv2Plugin.zoom}",
109 'deft' => "1",
110 'type' => "string" }, # xpdftools' zoom takes fractions
111 { 'name' => "use_sections",
112 'desc' => "{PDFPlugin.use_sections}",
113 'type' => "flag" },
114 { 'name' => "description_tags",
115 'desc' => "{HTMLPlugin.description_tags}",
116 'type' => "flag" },
117 { 'name' => "use_realistic_book",
118 'desc' => "{PDFPlugin.use_realistic_book}",
119 'type' => "flag"}
120 ];
121
122my $options = { 'name' => "PDFv2Plugin",
123 'desc' => "{PDFPlugin.desc}",
124 'abstract' => "no",
125 'inherits' => "yes",
126 'srcreplaceable' => "yes", # Source docs in PDF can be replaced with GS-generated html
127 'args' => $arguments };
128
129sub new {
130 my ($class) = shift (@_);
131 my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
132 push(@$pluginlist, $class);
133
134 push(@$inputargs,"-title_sub");
135 push(@$inputargs,'^(Page\s+\d+)?(\s*1\s+)?');
136
137 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
138 push(@{$hashArgOptLists->{"OptList"}},$options);
139
140 my $auto_converter_self = new AutoLoadConverters($pluginlist,$inputargs,$hashArgOptLists,["PDFBoxConverter"],1);
141 my $cbf_self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
142 my $self = BaseImporter::merge_inheritance($auto_converter_self, $cbf_self);
143
144 if ($self->{'info_only'}) {
145 # don't worry about any options etc
146 return bless $self, $class;
147 }
148
149 $self = bless $self, $class;
150 $self->{'file_type'} = "PDF";
151
152 # convert_options are passed through to gsConvert.pl by ConvertBinaryFile.pm
153
154 # the most important option is the tool that's used to do the conversion
155 $self->{'convert_options'} = "-pdf_tool xpdftools"; # default for PDFv2Plugin. If pdfbox_conversion is on, the pdfbpox GS extension sets pdf_tool to pdfbox
156
157 # pdf_zoom is supported by xpdftools' pdftohtml. So for pretty_html and paged_pretty_html
158 my $zoom = $self->{"zoom"};
159 $self->{'convert_options'} .= " -pdf_zoom $zoom";
160
161 # check convert_to
162
163 # Not all available conversion output options are possible with xpdftools, as some are
164 # only handled by pdfbox. If a format is unavailable with xpdftools, default to pretty_html
165 if ($self->{'convert_to'} =~ /^html$/) {
166 &gsprintf::gsprintf(STDERR, "{PDFv2Plugin.conversion_needs_pdfbox}\n", ($self->{'convert_to'}, "pretty_html"));
167# $self->{'convert_to'} = "pretty_html";
168 }
169 elsif ($self->{'convert_to'} =~ /^pagedimg/) {
170 &gsprintf::gsprintf(STDERR, "{PDFv2Plugin.conversion_needs_pdfbox}\n", ($self->{'convert_to'}, "paged_pretty_html"));
171# $self->{'convert_to'} = "paged_pretty_html";
172 }
173 elsif ($self->{'convert_to'} =~ /^paged_text$/) {
174 # TODO
175 print STDERR "@@@ Conversion to " . $self->{'convert_to'} , " with Xpdf Tools is not yet implemented.\n";
176 #print STDERR "@@@ Converting to text instead.\n";
177 #$self->{'convert_to'} = "text";
178 }
179
180 # PDFv2Plugin now supports PDF to txt conversion on Windows too:
181 # using XPDF Tools (incl pdftotext) on Windows/Linux/Mac
182 elsif ($self->{'convert_to'} eq "text" && $ENV{'GSDLOS'} =~ /^windows$/i) {
183 &gsprintf::gsprintf(STDERR, "{PDFv2Plugin.win_pdftotext_info}\n");
184 }
185 elsif ($self->{'convert_to'} eq "auto") {
186 # choose pretty_html is the best default option when using xpdftools
187 $self->{'convert_to'} = "pretty_html";
188 }
189 if ($self->{'use_realistic_book'}) {
190 if ($self->{'convert_to'} ne "html") {
191 print STDERR "PDFs will be converted to HTML for realistic book functionality\n";
192 $self->{'convert_to'} = "html";
193 }
194 }
195 # set convert_to_plugin and convert_to_ext
196 $self->set_standard_convert_settings();
197
198 my $secondary_plugin_name = $self->{'convert_to_plugin'};
199 my $secondary_plugin_options = $self->{'secondary_plugin_options'};
200
201 if (!defined $secondary_plugin_options->{$secondary_plugin_name}) {
202 $secondary_plugin_options->{$secondary_plugin_name} = [];
203 }
204 my $specific_options = $secondary_plugin_options->{$secondary_plugin_name};
205
206 # following title_sub removes "Page 1" added by pdftohtml, and a leading
207 # "1", which is often the page number at the top of the page. Bad Luck
208 # if your document title actually starts with "1 " - is there a better way?
209 push(@$specific_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
210 my $associate_tail_re = $self->{'associate_tail_re'};
211 if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
212 push(@$specific_options, "-associate_tail_re", $associate_tail_re);
213 }
214 push(@$specific_options, "-file_rename_method", "none");
215
216 if ($secondary_plugin_name eq "HTMLPlugin") {
217 # pdftohtml always produces utf8 - What about pdfbox???
218 # push(@$specific_options, "-input_encoding", "utf8");
219 push(@$specific_options, "-extract_language") if $self->{'extract_language'};
220 push(@$specific_options, "-processing_tmp_files");
221 # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
222 # to extract these metadata fields from the HEAD META fields
223 if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
224 push(@$specific_options,"-metadata_fields",$self->{'metadata_fields'});
225 } else {
226 push(@$specific_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
227 }
228 if (defined $self->{'metadata_field_separator'} && $self->{'metadata_field_separator'} =~ /\S/) {
229 push(@$specific_options,"-metadata_field_separator",$self->{'metadata_field_separator'});
230 }
231 if ($self->{'use_sections'} || $self->{'description_tags'}) {
232 $self->{'description_tags'} = 1;
233 push(@$specific_options, "-description_tags");
234 }
235 if ($self->{'use_realistic_book'}) {
236 push(@$specific_options, "-use_realistic_book");
237 }
238 if($self->{'convert_to'} eq "paged_pretty_html") { # for paged pretty html, the default should be to sectionalise
239 # the single superpage, the one containing divs representing individual pages as sections, on headings
240 push(@$specific_options, "sectionalise_using_h_tags");
241 }
242 }
243 elsif ($secondary_plugin_name eq "PagedImagePlugin") {
244 push(@$specific_options, "-screenviewsize", "1000");
245 push(@$specific_options, "-enable_cache");
246 push(@$specific_options, "-processing_tmp_files");
247 }
248
249 $self = bless $self, $class;
250 $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
251 return $self;
252}
253
254sub get_default_process_exp {
255 my $self = shift (@_);
256
257 return q^(?i)\.pdf$^;
258}
259
260# so we don't inherit HTMLPlug's block exp...
261sub get_default_block_exp {
262 return "";
263}
264
265sub init {
266 my $self = shift (@_);
267
268 # ConvertBinaryFile init
269 $self->SUPER::init(@_);
270 $self->AutoLoadConverters::init(@_);
271
272}
273
274sub begin {
275 my $self = shift (@_);
276
277 $self->AutoLoadConverters::begin(@_);
278 $self->SUPER::begin(@_);
279
280}
281
282sub deinit {
283 my $self = shift (@_);
284
285 $self->AutoLoadConverters::deinit(@_);
286 $self->SUPER::deinit(@_);
287
288}
289
290# By setting hashing to be on ga xml this ensures that two
291# PDF files that are identical except for the metadata
292# to hash to different values. Without this, when each PDF
293# file is converted to HTML there is a chance that they
294# will both be *identical* if the conversion utility does
295# not embed the metadata in the generated HTML. This is
296# certainly the case when PDFBOX is being used.
297
298# This change makes this convert to based plugin more
299# consistent with the original vision that the same document
300# with different metadata should
301# be seen as different.
302
303sub get_oid_hash_type {
304 my $self = shift (@_);
305 return "hash_on_ga_xml";
306}
307
308
309sub tmp_area_convert_file {
310
311 my $self = shift (@_);
312 return $self->AutoLoadConverters::tmp_area_convert_file(@_);
313
314}
315
316# Overriding to do some extra handling for pretty_html/paged_pretty_html output mode
317sub run_conversion_command {
318 my $self = shift (@_);
319 my ($tmp_dirname, $tmp_inputPDFname, $utf8_tailname, $lc_suffix, $tailname, $suffix) = @_;
320
321 if($self->{'convert_to'} !~ /pretty_html/) {
322 return $self->ConvertBinaryFile::run_conversion_command(@_);
323 }
324
325 # else, paged_pretty_html or pretty_html
326
327 # if output mode is (paged_)pretty_html, we use Xpdf tools' pdftohtml and tell it
328 # to create a subdir called "pages" in the tmp area to puts its products
329 # in there. (Xpdf's pdftohtml needs to be passed a *non-existent* directory
330 # parameter, the "pages" subdir). If Xpdf's pdftohtml has successfully run,
331 # the intermediary output file tmp/<random-num>/pages/index.html should
332 # exist (besides other output products there)
333
334 # We let ConvertBinaryFile proceed normally, but the return value should reflect
335 # that on success it should expect the intermediary product tmpdir/pages/index.html
336 # (which is the product of xpdftohtml conversion).
337 my $output_filename = $self->ConvertBinaryFile::run_conversion_command(@_);
338 $output_filename = &FileUtils::filenameConcatenate($tmp_dirname, "pages", "index.html");
339
340 # However, when convert_post_process() is done, it should have output the final
341 # product of the (paged_)pretty_html conversion: an html file of the same name and in the
342 # same tmp location as the input PDF file.
343
344 my ($name_prefix, $output_dir, $ext)
345 = &File::Basename::fileparse($tmp_inputPDFname, "\\.[^\\.]+\$");
346 $self->{'conv_filename_after_post_process'} = &FileUtils::filenameConcatenate($output_dir, $name_prefix.".html");
347# print STDERR "@@@@@ final paged html file will be: " . $self->{'conv_filename_after_post_process'} . "\n";
348
349 return $output_filename;
350}
351
352sub convert_post_process
353{
354 my $self = shift (@_);
355 my ($conv_filename) = @_;
356
357 my $outhandle=$self->{'outhandle'};
358
359 if($self->{'convert_to'} =~ /pretty_html/) { # (paged_)pretty_html
360 # special post-processing for (paged_)pretty_html mode, as HTML pages generated
361 # by xpdf's pdftohtml need to be massaged into the form we want
362 $self->xpdftohtml_convert_post_process($conv_filename);
363 }
364 else { # use PDFPlugin's usual post processing
365 $self->default_convert_post_process($conv_filename);
366 }
367}
368
369# Called after gsConvert.pl has been run to convert a PDF to (paged_)pretty_html
370# using Xpdftools' pdftohtml
371# This method will do some cleanup of the HTML files produced after XPDF has produced
372# an HTML doc for each PDF page: it first gets rid of the default index.html.
373# Instead, it constructs a single html page containing each original HTML page
374# <body> nested as divs instead, with simple section information inserted at the top
375# of each 'page' <div> and some further styling customisation. This HTML manipulation
376# is to be done with the Mojo::DOM perl package.
377# Note that since xpdf's pdftohtml would have failed if the output dir already
378# existed and for simpler naming, the output files are created in a new "pages"
379# subdirectory of the tmp location parent of $conv_filename instead
380sub xpdftohtml_convert_post_process
381{
382 my $self = shift (@_);
383 my ($pages_index_html) = @_; # = tmp/<rand>/pages/index.html for (paged_)pretty_html output mode
384 my $output_filename = $self->{'conv_filename_after_post_process'};
385
386 # Read in all the html files in tmp's "pages" subdir, except for index.html.
387 # and use it to create a new html file called $self->{'conv_filename_after_post_process'}
388 # which will consist of a slightly modified version of
389 # each of the other html files concatenated together.
390
391 my $outhandle=$self->{'outhandle'};
392
393 my ($tailname, $pages_subdir, $suffix)
394 = &File::Basename::fileparse($pages_index_html, "\\.[^\\.]+\$");
395
396 # Code from util::create_itemfile()
397 # Read in all the files
398 opendir(DIR, $pages_subdir) || die "can't opendir $pages_subdir: $!";
399 my @page_files = grep {-f "$pages_subdir/$_"} readdir(DIR);
400 closedir DIR;
401 # Sort files in the directory by page_num
402 # files are named index.html, page1.html, page2.html, ..., pagen.html
403 sub page_number {
404 my ($dir) = @_;
405 my ($pagenum) =($dir =~ m/^page(\d+)\.html?$/i);
406 $pagenum = 0 unless defined $pagenum; # index.html will be given pagenum=0
407 return $pagenum;
408 }
409 # sort the files in the directory in the order of page_num rather than lexically.
410 @page_files = sort { page_number($a) <=> page_number($b) } @page_files;
411
412 #my $num_html_pages = (scalar(@page_files) - 1)/2; # skip index file.
413 # For every html file there's an img file, so halve the total num.
414 # What about other file types that may potentially be there too???
415 my $num_html_pages = 0;
416 foreach my $pagefile (@page_files) {
417 $num_html_pages++ if $pagefile =~ m/\.html?$/ && $pagefile !~ /^index\.html?/i;
418 }
419
420 # Prepare to create our new html page that will contain all the individual
421 # htmls generated by xpdf's pdftohtml in sequence.
422 # First write the opening html tags out to the output file. These are the
423 # same tags and their contents, including <meta>, as is generated by
424 # Xpdf's pdftohtml for each of its individual html pages.
425 my $start_text = "<html>\n<head>\n";
426 my ($output_tailname, $tmp_subdir, $html_suffix)
427 = &File::Basename::fileparse($output_filename, "\\.[^\\.]+\$");
428 $start_text .= "<title>$output_tailname</title>\n";
429 $start_text .= "<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n";
430 $start_text .= "</head>\n<body>\n\n";
431
432 if($self->{'convert_to'} =~ /paged_pretty_html/) { # then add the <h>tags for sectionalising
433 $start_text .= "<h1>$output_tailname</h1>\n\n";
434 }
435
436 #handle content encodings the same way that default_convert_post_process does
437 # $self->utf8_write_file ($start_text, $conv_filename); # will close file after write
438 # Don't want to build a giant string in memory of all the pages concatenated
439 # and then write it out in one go. Instead, build up the final single page
440 # by writing each modified (paged_)pretty_html file out to it as this is processed.
441 # Copying file open/close code from CommonUtil::utf8_write_file()
442 if (!open (OUTFILE, ">:utf8", $output_filename)) {
443 gsprintf(STDERR, "PDFv2Plugin::xpdftohtml_convert_post_process {CommonUtil.could_not_open_for_writing} ($!)\n", $output_filename);
444 die "\n";
445 }
446 print OUTFILE $start_text;
447
448 # Get the contents of each individual HTML page generated by Xpdf, after first
449 # modifying each, and write each out into our single all-encompassing html
450 foreach my $pagefile (@page_files) {
451 if ($pagefile =~ m/\.html?$/ && $pagefile !~ /^index\.html?/i) {
452 my $page_num = page_number($pagefile);
453 # get full path to pagefile
454 $pagefile = &FileUtils::filenameConcatenate($pages_subdir, $pagefile);
455# print STDERR "@@@ About to process html file $pagefile (num $page_num)\n";
456 my $modified_page_contents = $self->_process_pretty_html_page($pagefile, $page_num, $num_html_pages);
457 print OUTFILE "$modified_page_contents\n\n";
458 }
459 }
460
461 # we've now created a single HTML file by concatenating (a modified version)
462 # of each paged html file
463 print OUTFILE "</body>\n</html>\n"; # write out closing tags
464 close OUTFILE; # done
465
466 # Get rid of all the htm(l) files incl index.html in the associated "pages"
467 # subdir, since we've now processed them all into a single html file
468 # one folder level up and we don't want HTMLPlugin to process all of them next.
469 &FileUtils::removeFilesFiltered($pages_subdir, "\.html?\$"); # no specific whitelist, but blacklist htm(l)
470
471 # now the tmp area should contain a single html file contain all the html pages'
472 # contents in sequence, and a "pages" subdir containing the screenshot images
473 # of each page.
474 # HTMLPlugin will process these further in the plugin pipeline
475}
476
477# For whatever reason, most html <tags> don't get printed out in GLI
478# So when debugging, use this function to print them out as [tags] instead.
479sub _debug_print_html
480{
481 my $self = shift (@_);
482 my ($string_or_dom) = @_;
483
484 # can't seem to determine type of string with ref/reftype
485 # https://stackoverflow.com/questions/1731333/how-do-i-tell-what-type-of-value-is-in-a-perl-variable
486 # Not needed, as $dom objects seem to get correctly stringified in string contexts
487 # $dom.to_string/$dom.stringify seem to get called, no need to call them
488 # https://stackoverflow.com/questions/5214543/what-is-stringification-in-perl
489 my $escapedTxt = $string_or_dom;
490 $escapedTxt =~ s@\<@[@sg;
491 $escapedTxt =~ s@\>@]@sg;
492
493 print STDERR "#### $escapedTxt\n";
494}
495
496# Helper function for (paged_)pretty_html
497# to read in each page of pretty_html generated by Xpdf's pdftohtml
498# then modify the html suitably using the HTML parsing functions offered by
499# Mojo::DOM, then return the modified HTML content as a string.
500# For paged_pretty_html, some additional modification is done to sectionalise the final html
501# See https://mojolicious.org/perldoc/Mojo/DOM
502sub _process_pretty_html_page
503{
504 my $self = shift (@_);
505 my ($pagefile, $page_num, $num_html_pages) = @_;
506
507 my $text = "";
508
509 # handling content encoding the same way default_convert_post_process does
510 $self->read_file ($pagefile, "utf8", "", \$text);
511
512 my $dom = Mojo::DOM->new($text);
513
514# $self->_debug_print_html($dom);
515
516 # there's a <style> element on the <html>, we need to shift it into the <div>
517 # tag that we'll be creating. We'll first slightly modify the <style> element
518 # store the first style element, which is the only one and in the <body>
519 # we'll later insert it as child of an all-encompassing div that we'll create
520 my $page_style_tag_str = $dom->at('html')->at('style')->to_string;
521 # In the style tag, convert id style references to class style references
522 my $css_class = ".p".$page_num."f";
523 $page_style_tag_str =~ s@\#f@$css_class@sg;
524 my $style_element = Mojo::DOM->new($page_style_tag_str)->at('style'); # modified
525#$self->_debug_print_html($style_element);
526
527 # need to know the image's height to set the height of the surrounding
528 # div that's to replace this page's <body>:
529 my $img_height = $dom->find('img')->[0]{height};
530
531 # 2. Adjust the img#background src attribute to point to the pages subdir for imgs
532 # 3. Set that img tag's class=background, and change its id to background+$page_num
533 my $bg_img_tag=$dom->find('img#background')->[0];
534 my $img_src_str = $bg_img_tag->{src};
535 $img_src_str = "pages/$img_src_str";
536 $bg_img_tag->attr(src => $img_src_str); # reset
537#$self->_debug_print_html($bg_img_tag);
538 # set both class and modified id attributes in one step:
539 $bg_img_tag->attr({class => "background", id => "background".$page_num});
540#$self->_debug_print_html($bg_img_tag);
541
542 # get all the <span> nested inside <div class="txt"> elements and
543 # 1. set their class attr to be "p + page_num + id-of-the-span",
544 # 2. then delete the id, because the span ids have been reused when element
545 # ids ought to be unique. Which is why we set the modified ids to be the
546 # value of the class attribute instead
547 $dom->find('div.txt span')->each(sub {
548 $_->attr(class => "p". $page_num. $_->{id});
549 delete $_->{id};
550 }); # both changes done in one find() operation
551#$self->_debug_print_html($dom->find('div.txt span')->last);
552
553 # Finally can create our new dom, starting with a div tag for the current page
554 # Must be: <div id="$page_num" style="position:relative; height:$img_height;"/>
555# my $new_dom = Mojo::DOM->new_tag('div', id => "page".$page_num, style => "position: relative; height: ".$img_height."px;" )
556 my $new_dom = Mojo::DOM->new_tag('div', style => "position: relative; height: ".$img_height."px;" );
557#$self->_debug_print_html($new_dom);
558 $new_dom->at('div')->append_content($style_element)->root;
559
560
561#$self->_debug_print_html($new_dom);
562 # Copy across all the old html's body tag's child nodes into the new dom's new div tag
563 $dom->at('body')->child_nodes->each(sub { $new_dom->at('div')->append_content($_)}); #$_->to_string
564#$self->_debug_print_html($new_dom);
565
566 # build up the outer div
567 my $inner_div_str = $new_dom->to_string;
568 my $page_div = "<div id=\"page".$page_num."\">\n";
569
570 if($self->{'convert_to'} =~ /paged_pretty_html/) { # then add the <h>tags for sectionalising
571
572 # Append a page range bucket heading if applicable: if we have more than 10 pages
573 # to display in the current bucket AND we're on the first page of each bucket of 10 pages.
574 # Dr Bainbridge thinks for now we need only consider PDFs where the
575 # total number of pages < 1000 and create buckets of size 10 (e.g. 1-10, ... 51-60, ...)
576 # If number of remaining pages >= 10, then create new bucket heading
577 # e.g. "Pages 30-40"
578 if(($page_num % 10) == 1 && ($num_html_pages - $page_num) > 10) {
579 # Double-digit page numbers that start with 2
580 # i.e. 21 to 29 (and 30) should be in 21 to 30 range
581 my $start_range = $page_num - ($page_num % 10) + 1;
582 my $end_range = $page_num + 10 - ($page_num % 10);
583 $page_div .= "<h2 style=\"font-size:1em;font-weight:normal;\">Pages ".$start_range . "-" . $end_range."</h2>\n";
584 }
585
586 # No sectionalising for 10 pages or under. Otherwise, every page is a section too, not just buckets
587 if($num_html_pages > 10) {
588 # Whether we're starting a new bucket or not, add a simpler heading: just the pagenumber, "Page #"
589 $page_div .= "<h3 style=\"font-size:1em;font-weight:normal;\">Page ".$page_num."</h3>\n";
590 }
591 }
592
593 $page_div .= $inner_div_str;
594 $page_div .= "\n</div>";
595
596 # Finished processing a single html page of the (paged_)pretty_html output generated by
597 # Xpdf's pdftohtml: finished massaging that single html page into the right form
598 return $page_div;
599}
600
601# This subroutine is called to do the PDFv2Plugin post-processing for all cases
602# except the "pretty_html" or "paged_pretty_html" conversion modes.
603# This is what PDFPlugin always used to do:
604sub default_convert_post_process
605{
606 my $self = shift (@_);
607 my ($conv_filename) = @_;
608 my $outhandle=$self->{'outhandle'};
609
610 #$self->{'input_encoding'} = "utf8"; # TODO: The output is always in utf8 (is it?? it is for html, but what about other types?)
611 #my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
612
613 # read in file ($text will be in utf8)
614 my $text = "";
615 # encoding will be utf8 for html files - what about other types? will we do this step for them anyway?
616 $self->read_file ($conv_filename, "utf8", "", \$text);
617
618 # To support the use_sections option with PDFBox: Greenstone splits PDFs into pages for
619 # sections. The PDFPlugin code wants each new page to be prefixed with <a name=pagenum></a>,
620 # which it then splits on to generate page-based sections. However, that's not what PDFBox
621 # generates in its HTML output. Fortunately, PDFBox does have its own page-separator: it
622 # embeds each page in an extra div. The div opener is:
623 # <div style=\"page-break-before:always; page-break-after:always\">
624 # The PDFPlugin now looks for this and prefixes <a name=0></a> to each such div. (The
625 # pagenumber is fixed at 0 since I'm unable to work out how to increment the pagenum during
626 # a regex substitution even with regex extensions on.) Later, when we process each section
627 # to get the pagenum, PDFBox's output for this is pre-processed by having a loopcounter
628 # that increments the pagenum for each subsequent section.
629
630 #$pdfbox_pageheader="\<div style=\"page-break-before:always; page-break-after:always\">";
631 my $loopcounter = 0; # used later on!
632 $text =~ s@\<div style=\"page-break-before:always; page-break-after:always\">@<a name=$loopcounter></a><div style=\"page-break-before:always; page-break-after:always\">@g;
633
634
635 # Calculate number of pages based on <a ...> tags (we have a <a name=1> etc
636 # for each page). Metadata based on this calculation not set until process()
637 #
638 # Note: this is done even if we are not breaking the document into pages as it might
639 # be useful to give an indication of document length in browser through setting
640 # num_pages as metadata.
641 # Clean html from low and hight surrogates D800–DFFF
642 $text =~ s@[\N{U+D800}-\N{U+DFFF}]@\ @g;
643 my @pages = ($text =~ m/\<[Aa] name=\"?\w+\"?>/ig); #<div style=\"?page-break-before:always; page-break-after:always\"?>
644 my $num_pages = scalar(@pages);
645 $self->{'num_pages'} = $num_pages;
646
647 if ($self->{'use_sections'}
648 && $self->{'converted_to'} eq "HTML") {
649
650 print $outhandle "PDFv2Plugin: Calculating sections...\n";
651
652 # we have "<a name=1></a>" etc for each page
653 # it may be <A name=
654 my @sections = split('<[Aa] name=', $text);
655
656 my $top_section = "";
657
658 if (scalar (@sections) == 1) { #only one section - no split!
659 print $outhandle "PDFv2Plugin: warning - no sections found\n";
660 } else {
661 $top_section .= shift @sections; # keep HTML header etc as top_section
662 }
663
664 # handle first section specially for title? Or all use first 100...
665
666 my $title = $sections[0];
667 $title =~ s/^\"?\w+\"?>//; # specific for pdftohtml...
668 $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
669 $title =~ s/<[^>]*>/ /g;
670 $title =~ s/(?:&nbsp;|\xc2\xa0)/ /g; # utf-8 for nbsp...
671 $title =~ s/^\s+//s;
672 $title =~ s/\s+$//;
673 $title =~ s/\s+/ /gs;
674 $title =~ s/^$self->{'title_sub'}// if ($self->{'title_sub'});
675 $title =~ s/^\s+//s; # in case title_sub introduced any...
676 $title = substr ($title, 0, 100);
677 $title =~ s/\s\S*$/.../;
678
679
680 if (scalar (@sections) == 1) { # no sections found
681 $top_section .= $sections[0];
682 @sections=();
683 } else {
684 $top_section .= "<!--<Section>\n<Metadata name=\"Title\">$title</Metadata>\n-->\n <!--</Section>-->\n";
685 }
686
687 # add metadata per section...
688 foreach my $section (@sections) {
689 # section names are not always just digits, may be like "outline"
690 $section =~ s@^\"?(\w+)\"?></a>@@; # leftover from split expression...
691
692 $title = $1; # Greenstone does magic if sections are titled digits
693
694 # A title of pagenum=0 means use_sections is being applied on output from PDFBox,
695 # which didn't originally have a <a name=incremented pagenumber></a> to split each page.
696 # Our Perl code then prefixed <a name=0></a> to it. Now need to increment the pagenum here:
697 if($loopcounter > 0 || ($title eq 0 && $loopcounter == 0)) { # implies use_sections with PDFBox
698 $title = ++$loopcounter;
699 }
700
701 if (! defined($title) ) {
702 print STDERR "no title: $section\n";
703 $title = " "; # get rid of the undefined warning in next line
704 }
705 my $newsection = "<!-- from PDFv2Plugin -->\n<!-- <Section>\n";
706 $newsection .= "<Metadata name=\"Title\">" . $title
707 . "</Metadata>\n--><br />\n";
708 $newsection .= $section;
709 $newsection .= "<!--</Section>-->\n";
710 $section = $newsection;
711 }
712
713 $text=join('', ($top_section, @sections));
714 }
715
716 if ($self->{'use_sections'}
717 && $self->{'converted_to'} eq "text") {
718 print STDERR "**** When converting PDF to text, cannot apply use_sections\n";
719 }
720
721
722 # The following should no longer be needed, now that strings
723 # read in are Unicode aware (in the Perl sense) rather than
724 # raw binary strings that just happen to be UTF-8 compliant
725
726 # turn any high bytes that aren't valid utf-8 into utf-8.
727## unicode::ensure_utf8(\$text);
728
729 # Write it out again!
730 $self->utf8_write_file (\$text, $conv_filename);
731}
732
733
734# do plugin specific processing of doc_obj for HTML type
735sub process {
736 my $self = shift (@_);
737 my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
738
739 my $result = $self->process_type($base_dir,$file,$doc_obj);
740
741 # fix up the extracted date metadata to be in Greenstone date format,
742 # and fix the capitalisation of 'date'
743 my $cursection = $doc_obj->get_top_section();
744 foreach my $datemeta (@{$doc_obj->get_metadata($cursection, "date")}) {
745 $doc_obj->delete_metadata($cursection, "date", $datemeta);
746
747 # We're just interested in the date bit, not the time
748 # some pdf creators (eg "Acrobat 5.0 Scan Plug-in for Windows")
749 # set a /CreationDate, and set /ModDate to 000000000. pdftohtml
750 # extracts the ModDate, so it is 0...
751 $datemeta =~ /(\d+)-(\d+)-(\d+)/;
752 my ($year, $month, $day) = ($1,$2,$3);
753 if (defined($year) && defined($month) && defined($day)) {
754 if ($year == 0) {next}
755 if ($year < 100) {$year += 1900} # just to be safe
756 if ($month =~ /^\d$/) {$month="0$month"} # single digit
757 if ($day =~ /^\d$/) {$day="0$day"} # single digit
758 my $date="$year$month$day";
759 $doc_obj->add_utf8_metadata($cursection, "Date", $date);
760 }
761 }
762
763 $doc_obj->add_utf8_metadata($cursection, "NumPages", $self->{'num_pages'}) if defined $self->{'num_pages'};
764
765 if ($self->{'use_sections'} && $self->{'converted_to'} eq "HTML") {
766 # For gs2 we explicitly make it a paged document, cos greenstone won't get it
767 # right if any section has an empty title, or one with letters in it
768 if (&util::is_gs3()) {
769 # but for gs3, paged docs currently use image slider which is ugly if there are no images
770 $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Hierarchy");
771 } else {
772 $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Paged");
773 }
774 }
775
776 return $result;
777}
778
7791;
Note: See TracBrowser for help on using the repository browser.