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 | ###########################################################################
|
---|
27 | package PDFv2Plugin;
|
---|
28 |
|
---|
29 | use strict;
|
---|
30 | no strict 'refs'; # so we can use a var for filehandles (e.g. STDERR)
|
---|
31 | no strict 'subs'; # allow filehandles to be variables and viceversa
|
---|
32 |
|
---|
33 | use ConvertBinaryFile;
|
---|
34 | use PDFBoxConverter;
|
---|
35 | use ReadTextFile;
|
---|
36 |
|
---|
37 | use unicode;
|
---|
38 | use Mojo::DOM; # for HTML parsing
|
---|
39 |
|
---|
40 |
|
---|
41 | @PDFv2Plugin::ISA = ('ConvertBinaryFile', 'PDFBoxConverter', 'ReadTextFile');
|
---|
42 |
|
---|
43 |
|
---|
44 | my $convert_to_list =
|
---|
45 | [ { 'name' => "auto", # pretty_html using xpdftools' pdftohtml
|
---|
46 | 'desc' => "{ConvertBinaryFile.convert_to.auto}" },
|
---|
47 | { 'name' => "text", # xpdftools' pdftotext
|
---|
48 | 'desc' => "{ConvertBinaryFile.convert_to.text}" },
|
---|
49 | { 'name' => "paged_text", # pdfbox
|
---|
50 | 'desc' => "{ConvertBinaryFile.convert_to.paged_text}" },
|
---|
51 |
|
---|
52 | { 'name' => "html", # pdfbox ## TODO: rename this to html_without_imgs?
|
---|
53 | 'desc' => "{PDFPlugin.convert_to.html}" },
|
---|
54 | { 'name' => "pretty_html", # xpdftools
|
---|
55 | 'desc' => "{PDFPlugin.convert_to.pretty_html}" },
|
---|
56 | { 'name' => "paged_pretty_html", # xpdftools
|
---|
57 | 'desc' => "{PDFPlugin.convert_to.paged_pretty_html}"},
|
---|
58 |
|
---|
59 | # pdfbox for all pagedimg(txt) output formats:
|
---|
60 | { 'name' => "pagedimg_jpg",
|
---|
61 | 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_jpg}"},
|
---|
62 | { 'name' => "pagedimg_png",
|
---|
63 | 'desc' => "{ConvertBinaryFile.convert_to.pagedimg_png}"},
|
---|
64 |
|
---|
65 | { 'name' => "pagedimgtxt_jpg",
|
---|
66 | 'desc' => "{ConvertBinaryFile.convert_to.pagedimgtxt_jpg}"},
|
---|
67 | { 'name' => "pagedimgtxt_png",
|
---|
68 | 'desc' => "{ConvertBinaryFile.convert_to.pagedimgtxt_png}"},
|
---|
69 | ];
|
---|
70 |
|
---|
71 |
|
---|
72 | my $arguments =
|
---|
73 | [
|
---|
74 | { 'name' => "convert_to",
|
---|
75 | 'desc' => "{ConvertBinaryFile.convert_to}",
|
---|
76 | 'type' => "enum",
|
---|
77 | 'reqd' => "yes",
|
---|
78 | 'list' => $convert_to_list,
|
---|
79 | 'deft' => "paged_pretty_html" },
|
---|
80 | { 'name' => "process_exp",
|
---|
81 | 'desc' => "{BaseImporter.process_exp}",
|
---|
82 | 'type' => "regexp",
|
---|
83 | 'deft' => &get_default_process_exp(),
|
---|
84 | 'reqd' => "no" },
|
---|
85 | { 'name' => "block_exp",
|
---|
86 | 'desc' => "{CommonUtil.block_exp}",
|
---|
87 | 'type' => "regexp",
|
---|
88 | 'deft' => &get_default_block_exp() },
|
---|
89 | # { 'name' => "metadata_fields",
|
---|
90 | # 'desc' => "{HTMLPlugin.metadata_fields}",
|
---|
91 | # 'type' => "string",
|
---|
92 | # 'deft' => "Title,Author,Subject,Keywords" },
|
---|
93 | # { 'name' => "metadata_field_separator",
|
---|
94 | # 'desc' => "{HTMLPlugin.metadata_field_separator}",
|
---|
95 | # 'type' => "string",
|
---|
96 | # 'deft' => "" },
|
---|
97 | { 'name' => "dpi",
|
---|
98 | 'desc' => "{PDFv2Plugin.dpi}",
|
---|
99 | 'deft' => "96",
|
---|
100 | '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
|
---|
101 | # { 'name' => "use_sections",
|
---|
102 | # 'desc' => "{PDFPlugin.use_sections}",
|
---|
103 | # 'type' => "flag" },
|
---|
104 | # { 'name' => "description_tags",
|
---|
105 | # 'desc' => "{HTMLPlugin.description_tags}",
|
---|
106 | # 'type' => "flag" },
|
---|
107 | { 'name' => "use_realistic_book",
|
---|
108 | 'desc' => "{PDFPlugin.use_realistic_book}",
|
---|
109 | 'type' => "flag" }
|
---|
110 | ];
|
---|
111 |
|
---|
112 | my $options = { 'name' => "PDFv2Plugin",
|
---|
113 | 'desc' => "{PDFv2Plugin.desc}",
|
---|
114 | 'abstract' => "no",
|
---|
115 | 'inherits' => "yes",
|
---|
116 | 'srcreplaceable' => "yes", # Source docs in PDF can be replaced with GS-generated html
|
---|
117 | 'args' => $arguments };
|
---|
118 |
|
---|
119 | sub new {
|
---|
120 | my ($class) = shift (@_);
|
---|
121 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
122 | push(@$pluginlist, $class);
|
---|
123 |
|
---|
124 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
125 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
126 |
|
---|
127 | # the 1 at the end of the first constructor call is to allow extra arguments to remain after
|
---|
128 | # parsing. There should be no extra args after the final constructor call, because all args
|
---|
129 | # should have been parsed by then, so don't pass 1 to the last constructor.
|
---|
130 | my $pdfbox_converter_self = new PDFBoxConverter($pluginlist, $inputargs, $hashArgOptLists, 1);
|
---|
131 | my $cbf_self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
132 | my $self = BaseImporter::merge_inheritance($pdfbox_converter_self, $cbf_self); # this param order seems necessary to preserve the default/user-selected value for the convert_to option
|
---|
133 |
|
---|
134 | if ($self->{'info_only'}) {
|
---|
135 | # don't worry about any options etc
|
---|
136 | return bless $self, $class;
|
---|
137 | }
|
---|
138 |
|
---|
139 | $self = bless $self, $class;
|
---|
140 | $self->{'file_type'} = "PDF";
|
---|
141 |
|
---|
142 | # convert_options are passed through to gsConvert.pl by ConvertBinaryFile.pm
|
---|
143 |
|
---|
144 | # the most important option is the tool that's used to do the conversion
|
---|
145 | $self->{'convert_options'} = "-pdf_tool xpdftools"; # default for PDFv2Plugin. If pdfbox_conversion is on, the pdfbpox GS extension sets pdf_tool to pdfbox
|
---|
146 |
|
---|
147 | # Setting dpi has meaning for xpdftools pdftohtml (so paged_pretty_html and pretty_html)
|
---|
148 | # and for when pdfbox outputs an image for each page (pagedimg, pagedimgtxt).
|
---|
149 | # dpi has no effect on (paged_)text and html output modes.
|
---|
150 | my $dpi = $self->{"dpi"};
|
---|
151 | $self->{'convert_options'} .= " -pdf_dpi $dpi";
|
---|
152 |
|
---|
153 | # The old pdftohtml tool used by PDFPlugin didn't do PDF to txt conversion on Windows
|
---|
154 | # But PDFv2Plugin now supports PDF to txt conversion on Windows too using XPDFTools' pdftotext
|
---|
155 |
|
---|
156 | if ($self->{'convert_to'} eq "auto") {
|
---|
157 | # defaulting to paged_pretty_html, as it's the best default option when using xpdftools
|
---|
158 | $self->{'convert_to'} = "paged_pretty_html";
|
---|
159 | &gsprintf::gsprintf(STDERR, "PDFv2Plugin: {PDFv2Plugin.auto_output_default}\n", $self->{'convert_to'});
|
---|
160 | }
|
---|
161 | if ($self->{'use_realistic_book'}) {
|
---|
162 | if ($self->{'convert_to'} ne "html") {
|
---|
163 | &gsprintf::gsprintf(STDERR, "PDFv2Plugin: {PDFPlugin.html_for_realistic_book}\n");
|
---|
164 | $self->{'convert_to'} = "html";
|
---|
165 | }
|
---|
166 | }
|
---|
167 |
|
---|
168 | # set convert_to_plugin and convert_to_ext
|
---|
169 | $self->set_standard_convert_settings();
|
---|
170 |
|
---|
171 | my $secondary_plugin_name = $self->{'convert_to_plugin'};
|
---|
172 | my $secondary_plugin_options = $self->{'secondary_plugin_options'};
|
---|
173 |
|
---|
174 | if (!defined $secondary_plugin_options->{$secondary_plugin_name}) {
|
---|
175 | $secondary_plugin_options->{$secondary_plugin_name} = [];
|
---|
176 | }
|
---|
177 | my $specific_options = $secondary_plugin_options->{$secondary_plugin_name};
|
---|
178 |
|
---|
179 | my $associate_tail_re = $self->{'associate_tail_re'};
|
---|
180 | if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
|
---|
181 | push(@$specific_options, "-associate_tail_re", $associate_tail_re);
|
---|
182 | }
|
---|
183 | push(@$specific_options, "-file_rename_method", "none");
|
---|
184 |
|
---|
185 | if ($secondary_plugin_name eq "HTMLPlugin") {
|
---|
186 | # pdftohtml always produces utf8 - What about pdfbox???
|
---|
187 | # push(@$specific_options, "-input_encoding", "utf8");
|
---|
188 | push(@$specific_options, "-extract_language") if $self->{'extract_language'};
|
---|
189 | push(@$specific_options, "-processing_tmp_files");
|
---|
190 | # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
|
---|
191 | # to extract these metadata fields from the HEAD META fields
|
---|
192 | if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
|
---|
193 | push(@$specific_options,"-metadata_fields",$self->{'metadata_fields'});
|
---|
194 | } else {
|
---|
195 | push(@$specific_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
|
---|
196 | }
|
---|
197 | if (defined $self->{'metadata_field_separator'} && $self->{'metadata_field_separator'} =~ /\S/) {
|
---|
198 | push(@$specific_options,"-metadata_field_separator",$self->{'metadata_field_separator'});
|
---|
199 | }
|
---|
200 | if ($self->{'use_sections'} || $self->{'description_tags'}) {
|
---|
201 | $self->{'description_tags'} = 1;
|
---|
202 | push(@$specific_options, "-description_tags");
|
---|
203 | }
|
---|
204 | if ($self->{'use_realistic_book'}) {
|
---|
205 | push(@$specific_options, "-use_realistic_book");
|
---|
206 | }
|
---|
207 | if($self->{'convert_to'} eq "paged_pretty_html") { # for paged pretty html, the default should be to sectionalise
|
---|
208 | # the single superpage, the one containing divs representing individual pages as sections, on headings
|
---|
209 | push(@$specific_options, "sectionalise_using_h_tags");
|
---|
210 | }
|
---|
211 | }
|
---|
212 | elsif ($secondary_plugin_name eq "PagedImagePlugin") {
|
---|
213 | push(@$specific_options, "-screenviewsize", "1000");
|
---|
214 | push(@$specific_options, "-enable_cache");
|
---|
215 | push(@$specific_options, "-processing_tmp_files");
|
---|
216 | }
|
---|
217 |
|
---|
218 | $self = bless $self, $class;
|
---|
219 | $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
|
---|
220 | return $self;
|
---|
221 | }
|
---|
222 |
|
---|
223 | sub get_default_process_exp {
|
---|
224 | my $self = shift (@_);
|
---|
225 |
|
---|
226 | return q^(?i)\.pdf$^;
|
---|
227 | }
|
---|
228 |
|
---|
229 | # so we don't inherit HTMLPlug's block exp...
|
---|
230 | sub get_default_block_exp {
|
---|
231 | return "";
|
---|
232 | }
|
---|
233 |
|
---|
234 | sub init {
|
---|
235 | my $self = shift (@_);
|
---|
236 |
|
---|
237 | # ConvertBinaryFile init
|
---|
238 | $self->SUPER::init(@_);
|
---|
239 | $self->PDFBoxConverter::init(@_);
|
---|
240 |
|
---|
241 | }
|
---|
242 |
|
---|
243 | sub begin {
|
---|
244 | my $self = shift (@_);
|
---|
245 |
|
---|
246 | $self->PDFBoxConverter::begin(@_);
|
---|
247 | $self->SUPER::begin(@_);
|
---|
248 |
|
---|
249 | }
|
---|
250 |
|
---|
251 | sub deinit {
|
---|
252 | my $self = shift (@_);
|
---|
253 |
|
---|
254 | $self->PDFBoxConverter::deinit(@_);
|
---|
255 | $self->SUPER::deinit(@_);
|
---|
256 |
|
---|
257 | }
|
---|
258 |
|
---|
259 | # By setting hashing to be on ga xml this ensures that two
|
---|
260 | # PDF files that are identical except for the metadata
|
---|
261 | # to hash to different values. Without this, when each PDF
|
---|
262 | # file is converted to HTML there is a chance that they
|
---|
263 | # will both be *identical* if the conversion utility does
|
---|
264 | # not embed the metadata in the generated HTML. This is
|
---|
265 | # certainly the case when PDFBOX is being used.
|
---|
266 |
|
---|
267 | # This change makes this convert to based plugin more
|
---|
268 | # consistent with the original vision that the same document
|
---|
269 | # with different metadata should
|
---|
270 | # be seen as different.
|
---|
271 |
|
---|
272 | sub get_oid_hash_type {
|
---|
273 | my $self = shift (@_);
|
---|
274 | return "hash_on_ga_xml";
|
---|
275 | }
|
---|
276 |
|
---|
277 |
|
---|
278 | sub tmp_area_convert_file {
|
---|
279 |
|
---|
280 | my $self = shift (@_);
|
---|
281 | my ($output_ext, $input_filename, $textref) = @_;
|
---|
282 |
|
---|
283 | if($self->{'convert_to'} eq "text" || $self->{'convert_to'} =~ m/pretty_html$/) { # use xpdftools
|
---|
284 | return $self->ConvertBinaryFile::tmp_area_convert_file(@_);
|
---|
285 | }
|
---|
286 |
|
---|
287 | # for all other output formats, use pdfbox:
|
---|
288 |
|
---|
289 | # Here, we now do directly what AutoLoadConverters::tmp_area_convert_file(@_)
|
---|
290 | # does with PDFBoxConverter:
|
---|
291 | my ($result, $result_str, $new_filename) = $self->PDFBoxConverter::convert($input_filename, $output_ext);
|
---|
292 | if (defined $result && $result != 0) {
|
---|
293 | return $new_filename;
|
---|
294 | }
|
---|
295 | my $outhandle=$self->{'outhandle'};
|
---|
296 | print $outhandle "PDFBoxConverter had a conversion error\n";
|
---|
297 | print $outhandle "$@\n";
|
---|
298 | if (defined $result_str) {
|
---|
299 | print $outhandle "$result_str\n";
|
---|
300 | }
|
---|
301 | return "";
|
---|
302 | }
|
---|
303 |
|
---|
304 | # Overriding to do some extra handling for pretty_html/paged_pretty_html output mode
|
---|
305 | sub run_conversion_command {
|
---|
306 | my $self = shift (@_);
|
---|
307 | my ($tmp_dirname, $tmp_inputPDFname, $utf8_tailname, $lc_suffix, $tailname, $suffix) = @_;
|
---|
308 |
|
---|
309 | if($self->{'convert_to'} !~ m/pretty_html$/) {
|
---|
310 | return $self->ConvertBinaryFile::run_conversion_command(@_);
|
---|
311 | }
|
---|
312 |
|
---|
313 | # else, paged_pretty_html or pretty_html
|
---|
314 |
|
---|
315 | # if output mode is (paged_)pretty_html, we use Xpdf tools' pdftohtml and tell it
|
---|
316 | # to create a subdir called "pages" in the tmp area to puts its products
|
---|
317 | # in there. (Xpdf's pdftohtml needs to be passed a *non-existent* directory
|
---|
318 | # parameter, the "pages" subdir). If Xpdf's pdftohtml has successfully run,
|
---|
319 | # the intermediary output file tmp/<random-num>/pages/index.html should
|
---|
320 | # exist (besides other output products there)
|
---|
321 |
|
---|
322 | # We let ConvertBinaryFile proceed normally, but the return value should reflect
|
---|
323 | # that on success it should expect the intermediary product tmpdir/pages/index.html
|
---|
324 | # (which is the product of xpdftohtml conversion).
|
---|
325 | my $output_filename = $self->ConvertBinaryFile::run_conversion_command(@_);
|
---|
326 | $output_filename = &FileUtils::filenameConcatenate($tmp_dirname, "pages", "index.html");
|
---|
327 |
|
---|
328 | # However, when convert_post_process() is done, it should have output the final
|
---|
329 | # product of the (paged_)pretty_html conversion: an html file of the same name and in the
|
---|
330 | # same tmp location as the input PDF file.
|
---|
331 |
|
---|
332 | my ($name_prefix, $output_dir, $ext)
|
---|
333 | = &File::Basename::fileparse($tmp_inputPDFname, "\\.[^\\.]+\$");
|
---|
334 | $self->{'conv_filename_after_post_process'} = &FileUtils::filenameConcatenate($output_dir, $name_prefix.".html");
|
---|
335 | # print STDERR "@@@@@ final paged html file will be: " . $self->{'conv_filename_after_post_process'} . "\n";
|
---|
336 |
|
---|
337 | return $output_filename;
|
---|
338 | }
|
---|
339 |
|
---|
340 | sub convert_post_process
|
---|
341 | {
|
---|
342 | my $self = shift (@_);
|
---|
343 | my ($conv_filename) = @_;
|
---|
344 |
|
---|
345 | my $outhandle=$self->{'outhandle'};
|
---|
346 |
|
---|
347 | if($self->{'convert_to'} =~ /pretty_html/) { # (paged_)pretty_html
|
---|
348 | # special post-processing for (paged_)pretty_html mode, as HTML pages generated
|
---|
349 | # by xpdf's pdftohtml need to be massaged into the form we want
|
---|
350 | $self->xpdftohtml_convert_post_process($conv_filename);
|
---|
351 | }
|
---|
352 | else { # use original PDFPlugin's usual post processing
|
---|
353 | $self->default_convert_post_process($conv_filename);
|
---|
354 | }
|
---|
355 | }
|
---|
356 |
|
---|
357 | # Called after gsConvert.pl has been run to convert a PDF to (paged_)pretty_html
|
---|
358 | # using Xpdftools' pdftohtml
|
---|
359 | # This method will do some cleanup of the HTML files produced after XPDF has produced
|
---|
360 | # an HTML doc for each PDF page: it first gets rid of the default index.html.
|
---|
361 | # Instead, it constructs a single html page containing each original HTML page
|
---|
362 | # <body> nested as divs instead, with simple section information inserted at the top
|
---|
363 | # of each 'page' <div> and some further styling customisation. This HTML manipulation
|
---|
364 | # is to be done with the Mojo::DOM perl package.
|
---|
365 | # Note that since xpdf's pdftohtml would have failed if the output dir already
|
---|
366 | # existed and for simpler naming, the output files are created in a new "pages"
|
---|
367 | # subdirectory of the tmp location parent of $conv_filename instead
|
---|
368 | sub xpdftohtml_convert_post_process
|
---|
369 | {
|
---|
370 | my $self = shift (@_);
|
---|
371 | my ($pages_index_html) = @_; # = tmp/<rand>/pages/index.html for (paged_)pretty_html output mode
|
---|
372 | my $output_filename = $self->{'conv_filename_after_post_process'};
|
---|
373 |
|
---|
374 | # Read in all the html files in tmp's "pages" subdir, except for index.html.
|
---|
375 | # and use it to create a new html file called $self->{'conv_filename_after_post_process'}
|
---|
376 | # which will consist of a slightly modified version of
|
---|
377 | # each of the other html files concatenated together.
|
---|
378 |
|
---|
379 | my $outhandle=$self->{'outhandle'};
|
---|
380 |
|
---|
381 | my ($tailname, $pages_subdir, $suffix)
|
---|
382 | = &File::Basename::fileparse($pages_index_html, "\\.[^\\.]+\$");
|
---|
383 |
|
---|
384 | # Code from util::create_itemfile()
|
---|
385 | # Read in all the files
|
---|
386 | opendir(DIR, $pages_subdir) || die "can't opendir $pages_subdir: $!";
|
---|
387 | my @page_files = grep {-f "$pages_subdir/$_"} readdir(DIR);
|
---|
388 | closedir DIR;
|
---|
389 | # Sort files in the directory by page_num
|
---|
390 | # files are named index.html, page1.html, page2.html, ..., pagen.html
|
---|
391 | sub page_number {
|
---|
392 | my ($dir) = @_;
|
---|
393 | my ($pagenum) =($dir =~ m/^page(\d+)\.html?$/i);
|
---|
394 | $pagenum = 0 unless defined $pagenum; # index.html will be given pagenum=0
|
---|
395 | return $pagenum;
|
---|
396 | }
|
---|
397 | # sort the files in the directory in the order of page_num rather than lexically.
|
---|
398 | @page_files = sort { page_number($a) <=> page_number($b) } @page_files;
|
---|
399 |
|
---|
400 | #my $num_html_pages = (scalar(@page_files) - 1)/2; # skip index file.
|
---|
401 | # For every html file there's an img file, so halve the total num.
|
---|
402 | # What about other file types that may potentially be there too???
|
---|
403 | my $num_html_pages = 0;
|
---|
404 | foreach my $pagefile (@page_files) {
|
---|
405 | $num_html_pages++ if $pagefile =~ m/\.html?$/ && $pagefile !~ /^index\.html?/i;
|
---|
406 | }
|
---|
407 |
|
---|
408 | # Prepare to create our new html page that will contain all the individual
|
---|
409 | # htmls generated by xpdf's pdftohtml in sequence.
|
---|
410 | # First write the opening html tags out to the output file. These are the
|
---|
411 | # same tags and their contents, including <meta>, as is generated by
|
---|
412 | # Xpdf's pdftohtml for each of its individual html pages.
|
---|
413 | my $start_text = "<html>\n<head>\n";
|
---|
414 | my ($output_tailname, $tmp_subdir, $html_suffix)
|
---|
415 | = &File::Basename::fileparse($output_filename, "\\.[^\\.]+\$");
|
---|
416 | $start_text .= "<title>$output_tailname</title>\n";
|
---|
417 | $start_text .= "<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">\n";
|
---|
418 | $start_text .= "</head>\n<body>\n\n";
|
---|
419 |
|
---|
420 | if($self->{'convert_to'} =~ /paged_pretty_html/) { # then add the <h>tags for sectionalising
|
---|
421 | $start_text .= "<h1>$output_tailname</h1>\n\n";
|
---|
422 | }
|
---|
423 |
|
---|
424 | #handle content encodings the same way that default_convert_post_process does
|
---|
425 | # $self->utf8_write_file ($start_text, $conv_filename); # will close file after write
|
---|
426 | # Don't want to build a giant string in memory of all the pages concatenated
|
---|
427 | # and then write it out in one go. Instead, build up the final single page
|
---|
428 | # by writing each modified (paged_)pretty_html file out to it as this is processed.
|
---|
429 | # Copying file open/close code from CommonUtil::utf8_write_file()
|
---|
430 | if (!open (OUTFILE, ">:utf8", $output_filename)) {
|
---|
431 | &gsprintf::gsprintf(STDERR, "PDFv2Plugin::xpdftohtml_convert_post_process {CommonUtil.could_not_open_for_writing} ($!)\n", $output_filename);
|
---|
432 | die "\n";
|
---|
433 | }
|
---|
434 | print OUTFILE $start_text;
|
---|
435 |
|
---|
436 | # Get the contents of each individual HTML page generated by Xpdf, after first
|
---|
437 | # modifying each, and write each out into our single all-encompassing html
|
---|
438 | foreach my $pagefile (@page_files) {
|
---|
439 | if ($pagefile =~ m/\.html?$/ && $pagefile !~ /^index\.html?/i) {
|
---|
440 | my $page_num = page_number($pagefile);
|
---|
441 | # get full path to pagefile
|
---|
442 | $pagefile = &FileUtils::filenameConcatenate($pages_subdir, $pagefile);
|
---|
443 | # print STDERR "@@@ About to process html file $pagefile (num $page_num)\n";
|
---|
444 | my $modified_page_contents = $self->_process_pretty_html_page($pagefile, $page_num, $num_html_pages);
|
---|
445 | # check for surrogates which are illegal in utf8 - need converting to proper codepoint
|
---|
446 | if ($modified_page_contents =~ /[\x{D800}-\x{DFFF}]/) {
|
---|
447 | $modified_page_contents =~ s/([\x{D800}-\x{DBFF}])([\x{DC00}-\x{DFFF}])/desurrogate($1, $2)/ge;
|
---|
448 | }
|
---|
449 | print OUTFILE "$modified_page_contents\n\n";
|
---|
450 | }
|
---|
451 | }
|
---|
452 |
|
---|
453 | # we've now created a single HTML file by concatenating (a modified version)
|
---|
454 | # of each paged html file
|
---|
455 | print OUTFILE "</body>\n</html>\n"; # write out closing tags
|
---|
456 | close OUTFILE; # done
|
---|
457 |
|
---|
458 | # Get rid of all the htm(l) files incl index.html in the associated "pages"
|
---|
459 | # subdir, since we've now processed them all into a single html file
|
---|
460 | # one folder level up and we don't want HTMLPlugin to process all of them next.
|
---|
461 | &FileUtils::removeFilesFiltered($pages_subdir, "\.html?\$"); # no specific whitelist, but blacklist htm(l)
|
---|
462 |
|
---|
463 | # now the tmp area should contain a single html file contain all the html pages'
|
---|
464 | # contents in sequence, and a "pages" subdir containing the screenshot images
|
---|
465 | # of each page.
|
---|
466 | # HTMLPlugin will process these further in the plugin pipeline
|
---|
467 | }
|
---|
468 |
|
---|
469 | # Turn a surrogate pair into a character
|
---|
470 | sub desurrogate
|
---|
471 | {
|
---|
472 | my ($hi, $lo) = @_;
|
---|
473 | my $codepoint = 0x10000 + (ord($hi) - 0xD800) * 0x400 + (ord($lo) - 0xDC00);
|
---|
474 | return chr($codepoint);
|
---|
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.
|
---|
479 | sub _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
|
---|
502 | sub _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 | # Also need to know the img width to set the surrounding div width to be no
|
---|
531 | # more than the img/page's screenshot width, so that the links in the GS doc
|
---|
532 | # view's table of contents are still at the top (z-index) and responsive.
|
---|
533 | my $img_width = $dom->find('img')->[0]{width};
|
---|
534 |
|
---|
535 | # 2. Adjust the img#background src attribute to point to the pages subdir for imgs
|
---|
536 | # 3. Set that img tag's class=background, and change its id to background+$page_num
|
---|
537 | my $bg_img_tag=$dom->find('img#background')->[0];
|
---|
538 | my $img_src_str = $bg_img_tag->{src};
|
---|
539 | $img_src_str = "pages/$img_src_str";
|
---|
540 | $bg_img_tag->attr(src => $img_src_str); # reset
|
---|
541 | #$self->_debug_print_html($bg_img_tag);
|
---|
542 | # set both class and modified id attributes in one step:
|
---|
543 | $bg_img_tag->attr({class => "background", id => "background".$page_num});
|
---|
544 | #$self->_debug_print_html($bg_img_tag);
|
---|
545 |
|
---|
546 | # get all the <span> nested inside <div class="txt"> elements and
|
---|
547 | # 1. set their class attr to be "p + page_num + id-of-the-span",
|
---|
548 | # 2. then delete the id, because the span ids have been reused when element
|
---|
549 | # ids ought to be unique. Which is why we set the modified ids to be the
|
---|
550 | # value of the class attribute instead
|
---|
551 | $dom->find('div.txt span')->each(sub {
|
---|
552 | $_->attr(class => "p". $page_num. $_->{id});
|
---|
553 | delete $_->{id};
|
---|
554 | }); # both changes done in one find() operation
|
---|
555 | #$self->_debug_print_html($dom->find('div.txt span')->last);
|
---|
556 |
|
---|
557 | # Finally can create our new dom, starting with a div tag for the current page
|
---|
558 | # Must be: <div id="$page_num" style="position:relative; height:$img_height;"/>
|
---|
559 | # my $new_dom = Mojo::DOM->new_tag('div', id => "page".$page_num, style => "position: relative; height: ".$img_height."px;" )
|
---|
560 | my $new_dom = Mojo::DOM->new_tag('div', style => "position: relative; height: ".$img_height."px; width: ".$img_width."px;");
|
---|
561 | #$self->_debug_print_html($new_dom);
|
---|
562 | $new_dom->at('div')->append_content($style_element)->root;
|
---|
563 |
|
---|
564 |
|
---|
565 | #$self->_debug_print_html($new_dom);
|
---|
566 | # Copy across all the old html's body tag's child nodes into the new dom's new div tag
|
---|
567 | $dom->at('body')->child_nodes->each(sub { $new_dom->at('div')->append_content($_)}); #$_->to_string
|
---|
568 | #$self->_debug_print_html($new_dom);
|
---|
569 |
|
---|
570 | # build up the outer div
|
---|
571 | my $inner_div_str = $new_dom->to_string;
|
---|
572 | my $page_div = "<div id=\"page".$page_num."\">\n";
|
---|
573 |
|
---|
574 | # If paged_pretty_html, then add <h>tags for sectionalising
|
---|
575 | if($self->{'convert_to'} =~ /paged_pretty_html/) {
|
---|
576 |
|
---|
577 | # Append a page range bucket heading if applicable: if we have more than 10 pages
|
---|
578 | # to display in the current bucket AND we're on the first page of each bucket of 10 pages.
|
---|
579 | # Dr Bainbridge thinks for now we need only consider PDFs where the
|
---|
580 | # total number of pages < 1000 and create buckets of size 10 (e.g. 1-10, ... 51-60, ...)
|
---|
581 | # If number of remaining pages >= 10, then create new bucket heading
|
---|
582 | # e.g. "Pages 30-40"
|
---|
583 | my $header_text = ""; #we want the <h2/3> tags to come outside the <div page=1> tag, otherwise we end up with unclosed divs in
|
---|
584 | # some sections
|
---|
585 | if(($page_num % 10) == 1 && ($num_html_pages > 10)) {
|
---|
586 | # Double-digit page numbers that start with 2
|
---|
587 | # i.e. 21 to 29 (and 30) should be in 21 to 30 range
|
---|
588 | my $start_range = $page_num - ($page_num % 10) + 1;
|
---|
589 | my $end_range = $page_num + 10 - ($page_num % 10);
|
---|
590 | #$page_div .= "<h2 style=\"font-size:1em;font-weight:normal;\">Pages ".$start_range . "-" . $end_range."</h2>\n";
|
---|
591 | $header_text = "<h2 style=\"font-size:1em;font-weight:normal;\">Pages ".$start_range . "-" . $end_range."</h2>\n";
|
---|
592 | }
|
---|
593 |
|
---|
594 | # Every page is now a section too, not just buckets.
|
---|
595 | # Whether we're starting a new bucket or not, add a simpler heading: just the pagenumber, "Page #"
|
---|
596 | # However, this should be <H3> when there are buckets and <H2> when there aren't any.
|
---|
597 | my $hx="h2";
|
---|
598 | if($num_html_pages > 10) {
|
---|
599 | $hx = "h3";
|
---|
600 | }
|
---|
601 | #$page_div .= "<$hx style=\"font-size:1em;font-weight:normal;\">Page ".$page_num."</$hx>\n";
|
---|
602 | $header_text .= "<$hx style=\"font-size:1em;font-weight:normal;\">Page ".$page_num."</$hx>\n";
|
---|
603 | $page_div = $header_text . $page_div;
|
---|
604 | }
|
---|
605 |
|
---|
606 | $page_div .= $inner_div_str;
|
---|
607 | $page_div .= "\n</div>";
|
---|
608 |
|
---|
609 | # Finished processing a single html page of the (paged_)pretty_html output generated by
|
---|
610 | # Xpdf's pdftohtml: finished massaging that single html page into the right form
|
---|
611 | return $page_div;
|
---|
612 | }
|
---|
613 |
|
---|
614 | # This subroutine is called to do the PDFv2Plugin post-processing for all cases
|
---|
615 | # except the "pretty_html" or "paged_pretty_html" conversion modes.
|
---|
616 | # This is what PDFPlugin always used to do:
|
---|
617 | sub default_convert_post_process
|
---|
618 | {
|
---|
619 | my $self = shift (@_);
|
---|
620 | my ($conv_filename) = @_;
|
---|
621 | my $outhandle=$self->{'outhandle'};
|
---|
622 |
|
---|
623 | #$self->{'input_encoding'} = "utf8"; # TODO: The output is always in utf8 (is it?? it is for html, but what about other types?)
|
---|
624 | #my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
|
---|
625 |
|
---|
626 | # read in file ($text will be in utf8)
|
---|
627 | my $text = "";
|
---|
628 | # encoding will be utf8 for html files - what about other types? will we do this step for them anyway?
|
---|
629 | $self->read_file ($conv_filename, "utf8", "", \$text);
|
---|
630 |
|
---|
631 | # To support the use_sections option with PDFBox: Greenstone splits PDFs into pages for
|
---|
632 | # sections. The PDFPlugin code wants each new page to be prefixed with <a name=pagenum></a>,
|
---|
633 | # which it then splits on to generate page-based sections. However, that's not what PDFBox
|
---|
634 | # generates in its HTML output. Fortunately, PDFBox does have its own page-separator: it
|
---|
635 | # embeds each page in an extra div. The div opener is:
|
---|
636 | # <div style=\"page-break-before:always; page-break-after:always\">
|
---|
637 | # The PDFPlugin now looks for this and prefixes <a name=0></a> to each such div. (The
|
---|
638 | # pagenumber is fixed at 0 since I'm unable to work out how to increment the pagenum during
|
---|
639 | # a regex substitution even with regex extensions on.) Later, when we process each section
|
---|
640 | # to get the pagenum, PDFBox's output for this is pre-processed by having a loopcounter
|
---|
641 | # that increments the pagenum for each subsequent section.
|
---|
642 |
|
---|
643 | #$pdfbox_pageheader="\<div style=\"page-break-before:always; page-break-after:always\">";
|
---|
644 | my $loopcounter = 0; # used later on!
|
---|
645 | $text =~ s@\<div style=\"page-break-before:always; page-break-after:always\">@<a name=$loopcounter></a><div style=\"page-break-before:always; page-break-after:always\">@g;
|
---|
646 |
|
---|
647 |
|
---|
648 | # Calculate number of pages based on <a ...> tags (we have a <a name=1> etc
|
---|
649 | # for each page). Metadata based on this calculation not set until process()
|
---|
650 | #
|
---|
651 | # Note: this is done even if we are not breaking the document into pages as it might
|
---|
652 | # be useful to give an indication of document length in browser through setting
|
---|
653 | # num_pages as metadata.
|
---|
654 | # Clean html from low and hight surrogates D800âDFFF
|
---|
655 | $text =~ s@[\N{U+D800}-\N{U+DFFF}]@\ @g;
|
---|
656 | my @pages = ($text =~ m/\<[Aa] name=\"?\w+\"?>/ig); #<div style=\"?page-break-before:always; page-break-after:always\"?>
|
---|
657 | my $num_pages = scalar(@pages);
|
---|
658 | $self->{'num_pages'} = $num_pages;
|
---|
659 |
|
---|
660 | if ($self->{'use_sections'}
|
---|
661 | && $self->{'converted_to'} eq "HTML") {
|
---|
662 |
|
---|
663 | print $outhandle "PDFv2Plugin: Calculating sections...\n";
|
---|
664 |
|
---|
665 | # we have "<a name=1></a>" etc for each page
|
---|
666 | # it may be <A name=
|
---|
667 | my @sections = split('<[Aa] name=', $text);
|
---|
668 |
|
---|
669 | my $top_section = "";
|
---|
670 |
|
---|
671 | if (scalar (@sections) == 1) { #only one section - no split!
|
---|
672 | print $outhandle "PDFv2Plugin: warning - no sections found\n";
|
---|
673 | } else {
|
---|
674 | $top_section .= shift @sections; # keep HTML header etc as top_section
|
---|
675 | }
|
---|
676 |
|
---|
677 | # handle first section specially for title? Or all use first 100...
|
---|
678 |
|
---|
679 | my $title = $sections[0];
|
---|
680 | $title =~ s/^\"?\w+\"?>//; # specific for old pdftohtml...
|
---|
681 | $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
|
---|
682 | $title =~ s/<[^>]*>/ /g;
|
---|
683 | $title =~ s/(?: |\xc2\xa0)/ /g; # utf-8 for nbsp...
|
---|
684 | $title =~ s/^\s+//s;
|
---|
685 | $title =~ s/\s+$//;
|
---|
686 | $title =~ s/\s+/ /gs;
|
---|
687 | $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?
|
---|
688 | $title = substr ($title, 0, 100);
|
---|
689 | $title =~ s/\s\S*$/.../;
|
---|
690 |
|
---|
691 |
|
---|
692 | if (scalar (@sections) == 1) { # no sections found
|
---|
693 | $top_section .= $sections[0];
|
---|
694 | @sections=();
|
---|
695 | } else {
|
---|
696 | $top_section .= "<!--<Section>\n<Metadata name=\"Title\">$title</Metadata>\n-->\n <!--</Section>-->\n";
|
---|
697 | }
|
---|
698 |
|
---|
699 | # add metadata per section...
|
---|
700 | foreach my $section (@sections) {
|
---|
701 | # section names are not always just digits, may be like "outline"
|
---|
702 | $section =~ s@^\"?(\w+)\"?></a>@@; # leftover from split expression...
|
---|
703 |
|
---|
704 | $title = $1; # Greenstone does magic if sections are titled digits
|
---|
705 |
|
---|
706 | # A title of pagenum=0 means use_sections is being applied on output from PDFBox,
|
---|
707 | # which didn't originally have a <a name=incremented pagenumber></a> to split each page.
|
---|
708 | # Our Perl code then prefixed <a name=0></a> to it. Now need to increment the pagenum here:
|
---|
709 | if($loopcounter > 0 || ($title eq 0 && $loopcounter == 0)) { # implies use_sections with PDFBox
|
---|
710 | $title = ++$loopcounter;
|
---|
711 | }
|
---|
712 |
|
---|
713 | if (! defined($title) ) {
|
---|
714 | print STDERR "no title: $section\n";
|
---|
715 | $title = " "; # get rid of the undefined warning in next line
|
---|
716 | }
|
---|
717 | my $newsection = "<!-- from PDFv2Plugin -->\n<!-- <Section>\n";
|
---|
718 | $newsection .= "<Metadata name=\"Title\">" . $title
|
---|
719 | . "</Metadata>\n--><br />\n";
|
---|
720 | $newsection .= $section;
|
---|
721 | $newsection .= "<!--</Section>-->\n";
|
---|
722 | $section = $newsection;
|
---|
723 | }
|
---|
724 |
|
---|
725 | $text=join('', ($top_section, @sections));
|
---|
726 | }
|
---|
727 |
|
---|
728 | if ($self->{'use_sections'}
|
---|
729 | && $self->{'converted_to'} eq "text") {
|
---|
730 | print STDERR "**** When converting PDF to text, cannot apply use_sections\n";
|
---|
731 | }
|
---|
732 |
|
---|
733 |
|
---|
734 | # The following should no longer be needed, now that strings
|
---|
735 | # read in are Unicode aware (in the Perl sense) rather than
|
---|
736 | # raw binary strings that just happen to be UTF-8 compliant
|
---|
737 |
|
---|
738 | # turn any high bytes that aren't valid utf-8 into utf-8.
|
---|
739 | ## unicode::ensure_utf8(\$text);
|
---|
740 |
|
---|
741 | # Write it out again!
|
---|
742 | $self->utf8_write_file (\$text, $conv_filename);
|
---|
743 | }
|
---|
744 |
|
---|
745 |
|
---|
746 | # do plugin specific processing of doc_obj for HTML type
|
---|
747 | sub process {
|
---|
748 | my $self = shift (@_);
|
---|
749 | my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
|
---|
750 |
|
---|
751 | my $result = $self->process_type($base_dir,$file,$doc_obj);
|
---|
752 |
|
---|
753 | # fix up the extracted date metadata to be in Greenstone date format,
|
---|
754 | # and fix the capitalisation of 'date'
|
---|
755 | my $cursection = $doc_obj->get_top_section();
|
---|
756 | foreach my $datemeta (@{$doc_obj->get_metadata($cursection, "date")}) {
|
---|
757 | $doc_obj->delete_metadata($cursection, "date", $datemeta);
|
---|
758 |
|
---|
759 | # We're just interested in the date bit, not the time
|
---|
760 | # some pdf creators (eg "Acrobat 5.0 Scan Plug-in for Windows")
|
---|
761 | # set a /CreationDate, and set /ModDate to 000000000. pdftohtml
|
---|
762 | # extracts the ModDate, so it is 0...
|
---|
763 | $datemeta =~ /(\d+)-(\d+)-(\d+)/;
|
---|
764 | my ($year, $month, $day) = ($1,$2,$3);
|
---|
765 | if (defined($year) && defined($month) && defined($day)) {
|
---|
766 | if ($year == 0) {next}
|
---|
767 | if ($year < 100) {$year += 1900} # just to be safe
|
---|
768 | if ($month =~ /^\d$/) {$month="0$month"} # single digit
|
---|
769 | if ($day =~ /^\d$/) {$day="0$day"} # single digit
|
---|
770 | my $date="$year$month$day";
|
---|
771 | $doc_obj->add_utf8_metadata($cursection, "Date", $date);
|
---|
772 | }
|
---|
773 | }
|
---|
774 |
|
---|
775 | $doc_obj->add_utf8_metadata($cursection, "NumPages", $self->{'num_pages'}) if defined $self->{'num_pages'};
|
---|
776 |
|
---|
777 | if ($self->{'use_sections'} && $self->{'converted_to'} eq "HTML") {
|
---|
778 | # For gs2 we explicitly make it a paged document, cos greenstone won't get it
|
---|
779 | # right if any section has an empty title, or one with letters in it
|
---|
780 | if (&util::is_gs3()) {
|
---|
781 | # but for gs3, paged docs currently use image slider which is ugly if there are no images
|
---|
782 | $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Hierarchy");
|
---|
783 | } else {
|
---|
784 | $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Paged");
|
---|
785 | }
|
---|
786 | }
|
---|
787 |
|
---|
788 | return $result;
|
---|
789 | }
|
---|
790 |
|
---|
791 | 1;
|
---|