root/main/trunk/greenstone2/perllib/plugins/PDFv2Plugin.pm @ 32290

Revision 32290, 32.2 KB (checked in by ak19, 13 months ago)

1. Making paged_pretty_html the default rather than pretty_html, since it's likely more users will want their converted PDF sectionalised. 2. Hopefully improved the display strings to make sense for users rather than for me.

Line 
1###########################################################################
2#
3# PDFv2Plugin.pm -- pdf plugin that uses xpdftools and pdfbox to process PDFs.
4# It only works out of the box for GS3 since it assumes the pdfbox extension
5# is installed.
6# A component of the Greenstone digital library software
7# from the New Zealand Digital Library Project at the
8# University of Waikato, New Zealand.
9#
10# Copyright (C) 1999-2001 New Zealand Digital Library Project
11#
12# This program is free software; you can redistribute it and/or modify
13# it under the terms of the GNU General Public License as published by
14# the Free Software Foundation; either version 2 of the License, or
15# (at your option) any later version.
16#
17# This program is distributed in the hope that it will be useful,
18# but WITHOUT ANY WARRANTY; without even the implied warranty of
19# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
20# GNU General Public License for more details.
21#
22# You should have received a copy of the GNU General Public License
23# along with this program; if not, write to the Free Software
24# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
25#
26###########################################################################
27package PDFv2Plugin;
28
29use strict;
30no strict 'refs'; # so we can use a var for filehandles (e.g. STDERR)
31no strict 'subs'; # allow filehandles to be variables and viceversa
32
33use ReadTextFile;
34use unicode;
35use Mojo::DOM; # for HTML parsing
36
37use PDFBoxConverter;
38use ConvertBinaryFile;
39
40@PDFv2Plugin::ISA = ('ConvertBinaryFile', 'PDFBoxConverter', 'ReadTextFile');
41
42
43my $convert_to_list =
44    [ { 'name' => "auto", # pretty_html using xpdftools' pdftohtml
45    'desc' => "{ConvertBinaryFile.convert_to.auto}" },
46      { 'name' => "text", # xpdftools' pdftotext
47    'desc' => "{ConvertBinaryFile.convert_to.text}" },
48      { 'name' => "paged_text", # pdfbox
49    'desc' => "{ConvertBinaryFile.convert_to.paged_text}" },
50     
51      { 'name' => "html", # pdfbox ## TODO: rename this to html_without_imgs?
52    'desc' => "{PDFPlugin.convert_to.html}" },
53      { 'name' => "pretty_html", # xpdftools
54    'desc' => "{PDFPlugin.convert_to.pretty_html}" },
55      { 'name' => "paged_pretty_html", # xpdftools
56    'desc' => "{PDFPlugin.convert_to.paged_pretty_html}"},
57
58      # pdfbox for all pagedimg(txt) output formats:
59      { 'name' => "pagedimg_jpg",
60    'desc' => "{ConvertBinaryFile.convert_to.pagedimg_jpg}"},
61      { 'name' => "pagedimg_png",
62    'desc' => "{ConvertBinaryFile.convert_to.pagedimg_png}"},
63     
64      { 'name' => "pagedimgtxt_jpg",
65    'desc' => "{ConvertBinaryFile.convert_to.pagedimgtxt_jpg}"},
66      { 'name' => "pagedimgtxt_png",
67    'desc' => "{ConvertBinaryFile.convert_to.pagedimgtxt_png}"},
68      ];
69
70
71my $arguments =
72    [
73     { 'name' => "convert_to",
74       'desc' => "{ConvertBinaryFile.convert_to}",
75       'type' => "enum",
76       'reqd' => "yes",
77       'list' => $convert_to_list,
78       'deft' => "paged_pretty_html" }, 
79     { 'name' => "process_exp",
80       'desc' => "{BaseImporter.process_exp}",
81       'type' => "regexp",
82       'deft' => &get_default_process_exp(),
83       'reqd' => "no" },
84     { 'name' => "block_exp",
85       'desc' => "{CommonUtil.block_exp}",
86       'type' => "regexp",
87       'deft' => &get_default_block_exp() },
88#     { 'name' => "metadata_fields",
89#       'desc' => "{HTMLPlugin.metadata_fields}",
90#       'type' => "string",
91#       'deft' => "Title,Author,Subject,Keywords" },
92#     { 'name' => "metadata_field_separator",
93#   'desc' => "{HTMLPlugin.metadata_field_separator}",
94#   'type' => "string",
95#   'deft' => "" },
96     { 'name' => "dpi",
97       'desc' => "{PDFv2Plugin.dpi}",
98       'deft' => "96",
99       'type' => "int" }, # 72DPI is xpdf's pdftohtml's default. pdfbox' default is 96DPI in headless mode else detected from the screen resolution, see https://pdfbox.apache.org/2.0/commandline.html#pdftoimage
100#     { 'name' => "use_sections",
101#       'desc' => "{PDFPlugin.use_sections}",
102#       'type' => "flag" },
103#     { 'name' => "description_tags",
104#       'desc' => "{HTMLPlugin.description_tags}",
105#       'type' => "flag" },
106      { 'name' => "use_realistic_book",
107        'desc' => "{PDFPlugin.use_realistic_book}",
108    'type' => "flag" }
109     ];
110
111my $options = { 'name'     => "PDFv2Plugin",
112        'desc'     => "{PDFPlugin.desc}",
113        'abstract' => "no",
114        'inherits' => "yes",
115        'srcreplaceable' => "yes", # Source docs in PDF can be replaced with GS-generated html     
116        'args'     => $arguments };
117
118sub new {
119    my ($class) = shift (@_);
120    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
121    push(@$pluginlist, $class);
122
123    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
124    push(@{$hashArgOptLists->{"OptList"}},$options);
125
126    my $pdfbox_converter_self = new PDFBoxConverter($pluginlist, $inputargs, $hashArgOptLists);
127    my $cbf_self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
128    my $self = BaseImporter::merge_inheritance($pdfbox_converter_self, $cbf_self); # this param order seems necessary to preserve the default/user-selected value for the convert_to option
129   
130    if ($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    # convert_options are passed through to gsConvert.pl by ConvertBinaryFile.pm
139   
140    # the most important option is the tool that's used to do the conversion
141    $self->{'convert_options'} = "-pdf_tool xpdftools"; # default for PDFv2Plugin. If pdfbox_conversion is on, the pdfbpox GS extension sets pdf_tool to pdfbox
142
143    # Setting dpi has meaning for xpdftools pdftohtml (so paged_pretty_html and pretty_html)
144    # and for when pdfbox outputs an image for each page (pagedimg, pagedimgtxt).
145    # dpi has no effect on (paged_)text and html output modes.
146    my $dpi = $self->{"dpi"};
147    $self->{'convert_options'} .= " -pdf_dpi $dpi";
148
149    # The old pdftohtml tool used by PDFPlugin didn't do PDF to txt conversion on Windows
150    # But PDFv2Plugin now supports PDF to txt conversion on Windows too using XPDFTools' pdftotext
151
152    if ($self->{'convert_to'} eq "auto") {
153    # defaulting to paged_pretty_html, as it's the best default option when using xpdftools
154    $self->{'convert_to'} = "paged_pretty_html";
155    &gsprintf::gsprintf(STDERR, "PDFv2Plugin: {PDFv2Plugin.auto_output_default}\n", $self->{'convert_to'});
156    }
157    if ($self->{'use_realistic_book'}) {
158    if ($self->{'convert_to'} ne "html") {
159        &gsprintf::gsprintf(STDERR, "PDFv2Plugin: {PDFPlugin.html_for_realistic_book}\n");
160        $self->{'convert_to'} = "html";
161    }
162    }
163   
164    # set convert_to_plugin and convert_to_ext
165    $self->set_standard_convert_settings();
166
167    my $secondary_plugin_name = $self->{'convert_to_plugin'};
168    my $secondary_plugin_options = $self->{'secondary_plugin_options'};
169
170    if (!defined $secondary_plugin_options->{$secondary_plugin_name}) {
171    $secondary_plugin_options->{$secondary_plugin_name} = [];
172    }
173    my $specific_options = $secondary_plugin_options->{$secondary_plugin_name};
174
175    my $associate_tail_re = $self->{'associate_tail_re'};
176    if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
177    push(@$specific_options, "-associate_tail_re", $associate_tail_re);
178    }
179    push(@$specific_options, "-file_rename_method", "none");
180   
181    if ($secondary_plugin_name eq "HTMLPlugin") {
182    # pdftohtml always produces utf8 - What about pdfbox???
183    # push(@$specific_options, "-input_encoding", "utf8");
184    push(@$specific_options, "-extract_language") if $self->{'extract_language'};
185    push(@$specific_options, "-processing_tmp_files");
186    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
187    # to extract these metadata fields from the HEAD META fields
188    if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
189        push(@$specific_options,"-metadata_fields",$self->{'metadata_fields'});
190    } else {
191        push(@$specific_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
192    }
193    if (defined $self->{'metadata_field_separator'} && $self->{'metadata_field_separator'} =~ /\S/) {
194        push(@$specific_options,"-metadata_field_separator",$self->{'metadata_field_separator'});
195    }
196    if ($self->{'use_sections'} || $self->{'description_tags'}) {
197        $self->{'description_tags'} = 1;
198        push(@$specific_options, "-description_tags");
199    }
200    if ($self->{'use_realistic_book'}) {
201        push(@$specific_options, "-use_realistic_book");
202    }
203    if($self->{'convert_to'} eq "paged_pretty_html") { # for paged pretty html, the default should be to sectionalise
204        # the single superpage, the one containing divs representing individual pages as sections, on headings
205        push(@$specific_options, "sectionalise_using_h_tags");
206    }
207    }
208    elsif ($secondary_plugin_name eq "PagedImagePlugin") {
209    push(@$specific_options, "-screenviewsize", "1000");
210    push(@$specific_options, "-enable_cache");
211    push(@$specific_options, "-processing_tmp_files");
212    }
213
214    $self = bless $self, $class;
215    $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
216    return $self;
217}
218
219sub get_default_process_exp {
220    my $self = shift (@_);
221
222    return q^(?i)\.pdf$^;
223}
224
225# so we don't inherit HTMLPlug's block exp...
226sub get_default_block_exp {
227    return "";
228}
229
230sub init {
231    my $self = shift (@_);
232
233    # ConvertBinaryFile init
234    $self->SUPER::init(@_);
235    $self->PDFBoxConverter::init(@_);
236
237}
238
239sub begin {
240    my $self = shift (@_);
241
242    $self->PDFBoxConverter::begin(@_);
243    $self->SUPER::begin(@_);
244
245}
246
247sub deinit {
248    my $self = shift (@_);
249
250    $self->PDFBoxConverter::deinit(@_);
251    $self->SUPER::deinit(@_);
252
253}
254 
255# By setting hashing to be on ga xml this ensures that two
256# PDF files that are identical except for the metadata
257# to hash to different values. Without this, when each PDF
258# file is converted to HTML there is a chance that they
259# will both be *identical* if the conversion utility does
260# not embed the metadata in the generated HTML. This is
261# certainly the case when PDFBOX is being used.
262
263# This change makes this convert to based plugin more
264# consistent with the original vision that the same document
265# with different metadata should
266# be seen as different.
267
268sub get_oid_hash_type {
269    my $self = shift (@_);
270    return "hash_on_ga_xml";
271}
272 
273
274sub tmp_area_convert_file {
275
276    my $self = shift (@_);
277    my ($output_ext, $input_filename, $textref) = @_;
278   
279    if($self->{'convert_to'} eq "text" || $self->{'convert_to'} =~ m/pretty_html$/) { # use xpdftools
280    return $self->ConvertBinaryFile::tmp_area_convert_file(@_);
281    }
282   
283    # for all other output formats, use pdfbox:
284   
285    # Here, we now do directly what AutoLoadConverters::tmp_area_convert_file(@_)
286    # does with PDFBoxConverter:
287    my ($result, $result_str, $new_filename) = $self->PDFBoxConverter::convert($input_filename, $output_ext);
288    if (defined $result && $result != 0) {
289    return $new_filename;
290    }
291    my $outhandle=$self->{'outhandle'};
292    print $outhandle "PDFBoxConverter had a conversion error\n";
293    print $outhandle "$@\n";
294    if (defined $result_str) {
295    print $outhandle "$result_str\n";
296    }
297    return "";
298}
299
300# Overriding to do some extra handling for pretty_html/paged_pretty_html output mode
301sub run_conversion_command {
302    my $self = shift (@_);
303    my ($tmp_dirname, $tmp_inputPDFname, $utf8_tailname, $lc_suffix, $tailname, $suffix) = @_;   
304
305    if($self->{'convert_to'} !~ m/pretty_html$/) {
306    return $self->ConvertBinaryFile::run_conversion_command(@_);
307    }
308
309    # else, paged_pretty_html or pretty_html
310
311    # if output mode is (paged_)pretty_html, we use Xpdf tools' pdftohtml and tell it
312    # to create a subdir called "pages" in the tmp area to puts its products
313    # in there. (Xpdf's pdftohtml needs to be passed a *non-existent* directory
314    # parameter, the "pages" subdir). If Xpdf's pdftohtml has successfully run,
315    # the intermediary output file tmp/<random-num>/pages/index.html should
316    # exist (besides other output products there)
317
318    # We let ConvertBinaryFile proceed normally, but the return value should reflect
319    # that on success it should expect the intermediary product tmpdir/pages/index.html
320    # (which is the product of xpdftohtml conversion).
321    my $output_filename = $self->ConvertBinaryFile::run_conversion_command(@_);
322    $output_filename = &FileUtils::filenameConcatenate($tmp_dirname, "pages", "index.html");
323
324    # However, when convert_post_process() is done, it should have output the final
325    # product of the (paged_)pretty_html conversion: an html file of the same name and in the
326    # same tmp location as the input PDF file.
327
328    my ($name_prefix, $output_dir, $ext)
329    = &File::Basename::fileparse($tmp_inputPDFname, "\\.[^\\.]+\$");
330    $self->{'conv_filename_after_post_process'} = &FileUtils::filenameConcatenate($output_dir, $name_prefix.".html");
331#    print STDERR "@@@@@ final paged html file will be: " . $self->{'conv_filename_after_post_process'} . "\n";
332
333    return $output_filename;
334}
335
336sub convert_post_process
337{
338    my $self = shift (@_);
339    my ($conv_filename) = @_;
340
341    my $outhandle=$self->{'outhandle'};
342
343    if($self->{'convert_to'} =~ /pretty_html/) { # (paged_)pretty_html
344    # special post-processing for (paged_)pretty_html mode, as HTML pages generated
345    # by xpdf's pdftohtml need to be massaged into the form we want
346    $self->xpdftohtml_convert_post_process($conv_filename);
347    }
348    else { # use original PDFPlugin's usual post processing
349    $self->default_convert_post_process($conv_filename);
350    }
351}
352
353# Called after gsConvert.pl has been run to convert a PDF to (paged_)pretty_html
354# using Xpdftools' pdftohtml
355# This method will do some cleanup of the HTML files produced after XPDF has produced
356# an HTML doc for each PDF page: it first gets rid of the default index.html.
357# Instead, it constructs a single html page containing each original HTML page
358# <body> nested as divs instead, with simple section information inserted at the top
359# of each 'page' <div> and some further styling customisation. This HTML manipulation
360# is to be done with the Mojo::DOM perl package.
361# Note that since xpdf's pdftohtml would have failed if the output dir already
362# existed and for simpler naming, the output files are created in a new "pages"
363# subdirectory of the tmp location parent of $conv_filename instead
364sub xpdftohtml_convert_post_process
365{
366    my $self = shift (@_);
367    my ($pages_index_html) = @_; # = tmp/<rand>/pages/index.html for (paged_)pretty_html output mode
368    my $output_filename = $self->{'conv_filename_after_post_process'};
369
370    # Read in all the html files in tmp's "pages" subdir, except for index.html.
371    # and use it to create a new html file called $self->{'conv_filename_after_post_process'}
372    # which will consist of a slightly modified version of
373    # each of the other html files concatenated together.
374
375    my $outhandle=$self->{'outhandle'};
376
377    my ($tailname, $pages_subdir, $suffix)
378    = &File::Basename::fileparse($pages_index_html, "\\.[^\\.]+\$");
379
380    # Code from util::create_itemfile()
381    # Read in all the files
382    opendir(DIR, $pages_subdir) || die "can't opendir $pages_subdir: $!";
383    my @page_files = grep {-f "$pages_subdir/$_"} readdir(DIR);
384    closedir DIR;
385    # Sort files in the directory by page_num
386    # files are named index.html, page1.html, page2.html, ..., pagen.html
387    sub page_number {
388    my ($dir) = @_;
389    my ($pagenum) =($dir =~ m/^page(\d+)\.html?$/i);
390    $pagenum = 0 unless defined $pagenum; # index.html will be given pagenum=0
391    return $pagenum;
392    }
393    # sort the files in the directory in the order of page_num rather than lexically.
394    @page_files = sort { page_number($a) <=> page_number($b) } @page_files;
395
396    #my $num_html_pages = (scalar(@page_files) - 1)/2; # skip index file.
397              # For every html file there's an img file, so halve the total num.
398              # What about other file types that may potentially be there too???
399    my $num_html_pages = 0;
400    foreach my $pagefile (@page_files) {
401    $num_html_pages++ if $pagefile =~ m/\.html?$/ && $pagefile !~ /^index\.html?/i;
402    }
403
404    # Prepare to create our new html page that will contain all the individual
405    # htmls generated by xpdf's pdftohtml in sequence.
406    # First write the opening html tags out to the output file. These are the
407    # same tags and their contents, including <meta>, as is generated by
408    # Xpdf's pdftohtml for each of its individual html pages.
409    my $start_text = "<html>\n<head>\n";
410    my ($output_tailname, $tmp_subdir, $html_suffix)
411    = &File::Basename::fileparse($output_filename, "\\.[^\\.]+\$");
412    $start_text .= "<title>$output_tailname</title>\n";
413    $start_text .= "<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n";
414    $start_text .= "</head>\n<body>\n\n";
415   
416    if($self->{'convert_to'} =~ /paged_pretty_html/) { # then add the <h>tags for sectionalising
417    $start_text .= "<h1>$output_tailname</h1>\n\n";
418    }
419
420    #handle content encodings the same way that default_convert_post_process does
421    # $self->utf8_write_file ($start_text, $conv_filename); # will close file after write   
422    # Don't want to build a giant string in memory of all the pages concatenated
423    # and then write it out in one go. Instead, build up the final single page
424    # by writing each modified (paged_)pretty_html file out to it as this is processed.
425    # Copying file open/close code from CommonUtil::utf8_write_file()
426    if (!open (OUTFILE, ">:utf8", $output_filename)) {
427    &gsprintf::gsprintf(STDERR, "PDFv2Plugin::xpdftohtml_convert_post_process {CommonUtil.could_not_open_for_writing} ($!)\n", $output_filename);
428    die "\n";
429    }
430    print OUTFILE $start_text;
431
432    # Get the contents of each individual HTML page generated by Xpdf, after first
433    # modifying each, and write each out into our single all-encompassing html
434    foreach my $pagefile (@page_files) {
435    if ($pagefile =~ m/\.html?$/ && $pagefile !~ /^index\.html?/i) {
436        my $page_num = page_number($pagefile);   
437        # get full path to pagefile
438        $pagefile = &FileUtils::filenameConcatenate($pages_subdir, $pagefile);
439#       print STDERR "@@@ About to process html file $pagefile (num $page_num)\n";
440        my $modified_page_contents = $self->_process_pretty_html_page($pagefile, $page_num, $num_html_pages);
441        print OUTFILE "$modified_page_contents\n\n";
442    }
443    }
444
445    # we've now created a single HTML file by concatenating (a modified version)
446    # of each paged html file
447    print OUTFILE "</body>\n</html>\n"; # write out closing tags
448    close OUTFILE; # done
449
450    # Get rid of all the htm(l) files incl index.html in the associated "pages"
451    # subdir, since we've now processed them all into a single html file
452    # one folder level up and we don't want HTMLPlugin to process all of them next.
453    &FileUtils::removeFilesFiltered($pages_subdir, "\.html?\$"); #  no specific whitelist, but blacklist htm(l)
454
455    # now the tmp area should contain a single html file contain all the html pages'
456    # contents in sequence, and a "pages" subdir containing the screenshot images
457    # of each page.   
458    # HTMLPlugin will process these further in the plugin pipeline
459}
460
461# For whatever reason, most html <tags> don't get printed out in GLI
462# So when debugging, use this function to print them out as [tags] instead.
463sub _debug_print_html
464{
465    my $self = shift (@_);
466    my ($string_or_dom) = @_;
467
468    # can't seem to determine type of string with ref/reftype
469    # https://stackoverflow.com/questions/1731333/how-do-i-tell-what-type-of-value-is-in-a-perl-variable
470    # Not needed, as $dom objects seem to get correctly stringified in string contexts
471    # $dom.to_string/$dom.stringify seem to get called, no need to call them
472    # https://stackoverflow.com/questions/5214543/what-is-stringification-in-perl
473    my $escapedTxt = $string_or_dom;
474    $escapedTxt =~ s@\<@[@sg;
475    $escapedTxt =~ s@\>@]@sg;
476
477    print STDERR "#### $escapedTxt\n";
478}
479
480# Helper function for (paged_)pretty_html
481# to read in each page of pretty_html generated by Xpdf's pdftohtml
482# then modify the html suitably using the HTML parsing functions offered by
483# Mojo::DOM, then return the modified HTML content as a string.
484# For paged_pretty_html, some additional modification is done to sectionalise the final html
485# See https://mojolicious.org/perldoc/Mojo/DOM
486sub _process_pretty_html_page
487{
488    my $self = shift (@_);
489    my ($pagefile, $page_num, $num_html_pages) = @_;
490
491    my $text = "";
492
493    # handling content encoding the same way default_convert_post_process does
494    $self->read_file ($pagefile, "utf8", "", \$text);
495
496    my $dom = Mojo::DOM->new($text);
497
498#    $self->_debug_print_html($dom);
499
500    # there's a <style> element on the <html>, we need to shift it into the <div>
501    # tag that we'll be creating. We'll first slightly modify the <style> element
502    # store the first style element, which is the only one and in the <body>
503    # we'll later insert it as child of an all-encompassing div that we'll create
504    my $page_style_tag_str = $dom->at('html')->at('style')->to_string;
505    # In the style tag, convert id style references to class style references
506    my $css_class = ".p".$page_num."f";
507    $page_style_tag_str =~ s@\#f@$css_class@sg;
508    my $style_element = Mojo::DOM->new($page_style_tag_str)->at('style'); # modified   
509#$self->_debug_print_html($style_element);
510
511    # need to know the image's height to set the height of the surrounding
512    # div that's to replace this page's <body>:
513    my $img_height = $dom->find('img')->[0]{height};
514
515    # 2. Adjust the img#background src attribute to point to the pages subdir for imgs
516    # 3. Set that img tag's class=background, and change its id to background+$page_num
517    my $bg_img_tag=$dom->find('img#background')->[0];
518    my $img_src_str = $bg_img_tag->{src};
519    $img_src_str = "pages/$img_src_str";
520    $bg_img_tag->attr(src => $img_src_str); # reset
521#$self->_debug_print_html($bg_img_tag);
522    # set both class and modified id attributes in one step:
523    $bg_img_tag->attr({class => "background", id => "background".$page_num});
524#$self->_debug_print_html($bg_img_tag);
525
526    # get all the <span> nested inside <div class="txt"> elements and
527    # 1. set their class attr to be "p + page_num + id-of-the-span",
528    # 2. then delete the id, because the span ids have been reused when element
529    # ids ought to be unique. Which is why we set the modified ids to be the
530    # value of the class attribute instead
531    $dom->find('div.txt span')->each(sub {
532    $_->attr(class => "p". $page_num. $_->{id});
533    delete $_->{id};
534                     }); # both changes done in one find() operation
535#$self->_debug_print_html($dom->find('div.txt span')->last);
536
537    # Finally can create our new dom, starting with a div tag for the current page
538    # Must be: <div id="$page_num" style="position:relative; height:$img_height;"/>
539#    my $new_dom = Mojo::DOM->new_tag('div', id => "page".$page_num, style => "position: relative; height: ".$img_height."px;" )
540    my $new_dom = Mojo::DOM->new_tag('div', style => "position: relative; height: ".$img_height."px;" );
541#$self->_debug_print_html($new_dom);
542    $new_dom->at('div')->append_content($style_element)->root;
543
544
545#$self->_debug_print_html($new_dom);
546    # Copy across all the old html's body tag's child nodes into the new dom's new div tag
547    $dom->at('body')->child_nodes->each(sub { $new_dom->at('div')->append_content($_)}); #$_->to_string
548#$self->_debug_print_html($new_dom);
549
550    # build up the outer div
551    my $inner_div_str = $new_dom->to_string;
552    my $page_div = "<div id=\"page".$page_num."\">\n";
553
554    # If paged_pretty_html, then add <h>tags for sectionalising
555    if($self->{'convert_to'} =~ /paged_pretty_html/) {
556   
557    # Append a page range bucket heading if applicable: if we have more than 10 pages
558    # to display in the current bucket AND we're on the first page of each bucket of 10 pages.
559    # Dr Bainbridge thinks for now we need only consider PDFs where the
560    # total number of pages < 1000 and create buckets of size 10 (e.g. 1-10, ... 51-60, ...)
561    # If number of remaining pages >= 10, then create new bucket heading
562    # e.g. "Pages 30-40"
563    if(($page_num % 10) == 1 && ($num_html_pages - $page_num) > 10) {
564        # Double-digit page numbers that start with 2
565        # i.e. 21 to 29 (and 30) should be in 21 to 30 range
566        my $start_range = $page_num - ($page_num % 10) + 1;
567        my $end_range = $page_num + 10 - ($page_num % 10);
568        $page_div .= "<h2 style=\"font-size:1em;font-weight:normal;\">Pages ".$start_range . "-" . $end_range."</h2>\n";
569    }
570
571    # Every page is now a section too, not just buckets.
572    # Whether we're starting a new bucket or not, add a simpler heading: just the pagenumber, "Page #"
573    # However, this should be <H3> when there are buckets and <H2> when there aren't any.
574    if($num_html_pages > 10) {
575        $page_div .= "<h3 style=\"font-size:1em;font-weight:normal;\">Page ".$page_num."</h3>\n";       
576    } else { # PDF has less than 10 pages in total
577        $page_div .= "<h2 style=\"font-size:1em;font-weight:normal;\">Page ".$page_num."</h2>\n";       
578    }
579    }
580
581    $page_div .= $inner_div_str;
582    $page_div .= "\n</div>";
583
584    # Finished processing a single html page of the (paged_)pretty_html output generated by
585    # Xpdf's pdftohtml: finished massaging that single html page into the right form
586    return $page_div;
587}
588
589# This subroutine is called to do the PDFv2Plugin post-processing for all cases
590# except the "pretty_html" or "paged_pretty_html" conversion modes.
591# This is what PDFPlugin always used to do:
592sub default_convert_post_process
593{
594    my $self = shift (@_);
595    my ($conv_filename) = @_;
596    my $outhandle=$self->{'outhandle'};
597
598    #$self->{'input_encoding'} = "utf8"; # TODO: The output is always in utf8 (is it?? it is for html, but what about other types?)
599    #my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
600
601    # read in file ($text will be in utf8)
602    my $text = "";
603    # encoding will be utf8 for html files - what about other types? will we do this step for them anyway?
604    $self->read_file ($conv_filename, "utf8", "", \$text);
605
606    # To support the use_sections option with PDFBox: Greenstone splits PDFs into pages for
607    # sections. The PDFPlugin code wants each new page to be prefixed with <a name=pagenum></a>,
608    # which it then splits on to generate page-based sections. However, that's not what PDFBox
609    # generates in its HTML output. Fortunately, PDFBox does have its own page-separator: it
610    # embeds each page in an extra div. The div opener is:
611    # <div style=\"page-break-before:always; page-break-after:always\">
612    # The PDFPlugin now looks for this and prefixes <a name=0></a> to each such div. (The
613    # pagenumber is fixed at 0 since I'm unable to work out how to increment the pagenum during
614    # a regex substitution even with regex extensions on.) Later, when we process each section
615    # to get the pagenum, PDFBox's output for this is pre-processed by having a loopcounter
616    # that increments the pagenum for each subsequent section.
617
618    #$pdfbox_pageheader="\<div style=\"page-break-before:always; page-break-after:always\">";
619    my $loopcounter = 0; # used later on!
620    $text =~ s@\<div style=\"page-break-before:always; page-break-after:always\">@<a name=$loopcounter></a><div style=\"page-break-before:always; page-break-after:always\">@g;
621
622
623    # Calculate number of pages based on <a ...> tags (we have a <a name=1> etc
624    # for each page).  Metadata based on this calculation not set until process()
625    #
626    # Note: this is done even if we are not breaking the document into pages as it might
627    # be useful to give an indication of document length in browser through setting
628    # num_pages as metadata.
629    # Clean html from low and hight surrogates D800–DFFF
630    $text =~ s@[\N{U+D800}-\N{U+DFFF}]@\ @g;
631    my @pages = ($text =~ m/\<[Aa] name=\"?\w+\"?>/ig); #<div style=\"?page-break-before:always; page-break-after:always\"?>
632    my $num_pages = scalar(@pages);
633    $self->{'num_pages'} = $num_pages;
634
635    if ($self->{'use_sections'}
636    && $self->{'converted_to'} eq "HTML") {
637
638    print $outhandle "PDFv2Plugin: Calculating sections...\n";
639
640    # we have "<a name=1></a>" etc for each page
641    # it may be <A name=
642    my @sections = split('<[Aa] name=', $text);
643
644    my $top_section = "";
645
646    if (scalar (@sections) == 1) { #only one section - no split!
647        print $outhandle "PDFv2Plugin: warning - no sections found\n";
648    } else {
649        $top_section .= shift @sections; # keep HTML header etc as top_section
650    }
651
652    # handle first section specially for title? Or all use first 100...
653   
654    my $title = $sections[0];
655    $title =~ s/^\"?\w+\"?>//; # specific for old pdftohtml...
656    $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
657    $title =~ s/<[^>]*>/ /g;
658    $title =~ s/(?:&nbsp;|\xc2\xa0)/ /g; # utf-8 for nbsp...
659    $title =~ s/^\s+//s;
660    $title =~ s/\s+$//;
661    $title =~ s/\s+/ /gs;
662    $title =~ s/^\s+//s; # in case title_sub (of old PDFPlugin's old pdftohtml) introduced any... Generally still useful to remove spaces at the start?
663    $title = substr ($title, 0, 100);
664    $title =~ s/\s\S*$/.../;
665
666
667    if (scalar (@sections) == 1) { # no sections found
668        $top_section .= $sections[0];
669        @sections=();
670    } else {
671        $top_section .= "<!--<Section>\n<Metadata name=\"Title\">$title</Metadata>\n-->\n <!--</Section>-->\n";
672    }
673
674    # add metadata per section...
675    foreach my $section (@sections) {
676        # section names are not always just digits, may be like "outline"
677        $section =~ s@^\"?(\w+)\"?></a>@@; # leftover from split expression...
678
679        $title = $1; # Greenstone does magic if sections are titled digits
680
681        # A title of pagenum=0 means use_sections is being applied on output from PDFBox,
682        # which didn't originally have a <a name=incremented pagenumber></a> to split each page.
683        # Our Perl code then prefixed <a name=0></a> to it. Now need to increment the pagenum here:
684        if($loopcounter > 0 || ($title eq 0 && $loopcounter == 0)) { # implies use_sections with PDFBox
685        $title = ++$loopcounter;
686        }
687
688        if (! defined($title) ) {
689        print STDERR "no title: $section\n";
690        $title = " "; # get rid of the undefined warning in next line
691        }
692        my $newsection = "<!-- from PDFv2Plugin -->\n<!-- <Section>\n";
693        $newsection .= "<Metadata name=\"Title\">" . $title
694        . "</Metadata>\n--><br />\n";
695        $newsection .= $section;
696        $newsection .= "<!--</Section>-->\n";
697        $section = $newsection;
698    }
699
700    $text=join('', ($top_section, @sections));
701    }
702
703    if ($self->{'use_sections'}
704    && $self->{'converted_to'} eq "text") {
705    print STDERR "**** When converting PDF to text, cannot apply use_sections\n";
706    }
707
708
709    # The following should no longer be needed, now that strings
710    # read in are Unicode aware (in the Perl sense) rather than
711    # raw binary strings that just happen to be UTF-8 compliant
712
713    # turn any high bytes that aren't valid utf-8 into utf-8.
714##    unicode::ensure_utf8(\$text);
715
716    # Write it out again!
717    $self->utf8_write_file (\$text, $conv_filename);
718}
719
720
721# do plugin specific processing of doc_obj for HTML type
722sub process {
723    my $self = shift (@_);
724    my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
725
726    my $result = $self->process_type($base_dir,$file,$doc_obj);
727
728    # fix up the extracted date metadata to be in Greenstone date format,
729    # and fix the capitalisation of 'date'
730    my $cursection = $doc_obj->get_top_section();
731    foreach my $datemeta (@{$doc_obj->get_metadata($cursection, "date")}) {
732    $doc_obj->delete_metadata($cursection, "date", $datemeta);
733
734    # We're just interested in the date bit, not the time
735    # some pdf creators (eg "Acrobat 5.0 Scan Plug-in for Windows")
736    # set a /CreationDate, and set /ModDate to 000000000. pdftohtml
737    # extracts the ModDate, so it is 0...
738    $datemeta =~ /(\d+)-(\d+)-(\d+)/;
739    my ($year, $month, $day) = ($1,$2,$3);
740    if (defined($year) && defined($month) && defined($day)) {
741        if ($year == 0) {next}
742        if ($year < 100) {$year += 1900} # just to be safe
743        if ($month =~ /^\d$/) {$month="0$month"} # single digit
744        if ($day =~ /^\d$/) {$day="0$day"} # single digit
745        my $date="$year$month$day";
746        $doc_obj->add_utf8_metadata($cursection, "Date", $date);
747    }
748    }
749
750    $doc_obj->add_utf8_metadata($cursection, "NumPages", $self->{'num_pages'}) if defined $self->{'num_pages'};
751   
752    if ($self->{'use_sections'} && $self->{'converted_to'} eq "HTML") {
753    # For gs2 we explicitly make it a paged document, cos greenstone won't get it
754    # right if any section has an empty title, or one with letters in it
755    if (&util::is_gs3()) {
756        # but for gs3, paged docs currently use image slider which is ugly if there are no images
757        $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Hierarchy");
758    } else {
759        $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Paged");
760    }
761    }
762
763    return $result;
764}
765
7661;
Note: See TracBrowser for help on using the browser.