line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Poppler 1.01; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
=encoding utf8 |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
=head1 NAME |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
Poppler - Bindings to the poppler PDF rendering library |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
=head1 SYNOPSIS |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
use Poppler; |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
# initialize using filename |
14
|
|
|
|
|
|
|
my $pdf = Poppler::Document->new_from_file( 'file.pdf' ); |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
# or, initialize using scalar data |
17
|
|
|
|
|
|
|
open my $fh, '<:raw', 'file.pdf'; |
18
|
|
|
|
|
|
|
read ($fh, my $data, -s 'file.pdf') |
19
|
|
|
|
|
|
|
close $fh; |
20
|
|
|
|
|
|
|
my $pdf = Poppler::Document->new_from_data( $data ); |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
# get some general info |
23
|
|
|
|
|
|
|
my $n_pages = $pdf->get_n_pages; |
24
|
|
|
|
|
|
|
my $title = $pdf->get_title; |
25
|
|
|
|
|
|
|
# etc ... |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
# get the first page |
28
|
|
|
|
|
|
|
my $page = $pdf->get_page( 0 ); |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
# get page size |
31
|
|
|
|
|
|
|
my ($w, $h) = $page->get_size; |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
# or, for backward compatibility |
34
|
|
|
|
|
|
|
my $dims = $page->get_size; # a Poppler::Dimension object |
35
|
|
|
|
|
|
|
my $w = $dims->get_width; |
36
|
|
|
|
|
|
|
my $h = $dims->get_height; |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
# do other fancy things (get page links, annotations, movies, etc) |
39
|
|
|
|
|
|
|
# (see poppler-glib documentation for details) |
40
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
# render to a Cairo surface |
42
|
|
|
|
|
|
|
use Cairo; |
43
|
|
|
|
|
|
|
my $surface = Cairo::ImageSurface->create( 'argb32', 100, 100 ); |
44
|
|
|
|
|
|
|
my $cr = Cairo::Context->create( $surface ); |
45
|
|
|
|
|
|
|
$page->render_to_cairo( $cr ); |
46
|
|
|
|
|
|
|
$cr->show_page; |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
=head1 ABSTRACT |
49
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
Bindings to the poppler PDF library via the Glib interface. Allows |
51
|
|
|
|
|
|
|
querying of a PDF file structure and rendering to various output targets. |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
=head1 DESCRIPTION |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
The C module provides complete bindings to the poppler PDF library |
56
|
|
|
|
|
|
|
through the Glib interface. Find out more about poppler at |
57
|
|
|
|
|
|
|
L. |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
As of version 1.01, no XS is used directly but bindings are provided using |
60
|
|
|
|
|
|
|
GObject introspection and the L module. See the |
61
|
|
|
|
|
|
|
L for a brief example of how the module can be used. For |
62
|
|
|
|
|
|
|
detailed documentation on the available classes and methods, see the poppler |
63
|
|
|
|
|
|
|
glib documentation for the C libraries and the L |
64
|
|
|
|
|
|
|
documentation for a description of how methods are mapped between the C |
65
|
|
|
|
|
|
|
libraries and the Perl namespace. |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
=head1 CONSTRUCTORS |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
=over |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
=item new_from_file ($filename) |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
Takes a system path or URI to a PDF file as an argument and returns a |
74
|
|
|
|
|
|
|
Poppler::Document object. The C library itself requires a full |
75
|
|
|
|
|
|
|
URI (e.g. "file:///home/user/file.pdf") but this module attempts to convert |
76
|
|
|
|
|
|
|
regular system paths if provided via the L module. |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
=item new_from_data ($data) |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
Takes a PDF data chunk as an argument and returns a Poppler::Document object. |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
=back |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
=head1 METHODS |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
For details on the classes and methods available beyond the constructors |
87
|
|
|
|
|
|
|
listed above, please refer to the canonical documentation for the C library |
88
|
|
|
|
|
|
|
listed under L. A general discussion of how these classes |
89
|
|
|
|
|
|
|
and methods map to the Perl equivalents can be found in the |
90
|
|
|
|
|
|
|
L documentation. Generally speaking, a C function |
91
|
|
|
|
|
|
|
such as 'poppler_document_get_title' would map to |
92
|
|
|
|
|
|
|
'Poppler::Document->get_title'. Methods that return a G |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
=cut |
95
|
|
|
|
|
|
|
|
96
|
1
|
|
|
1
|
|
25441
|
use strict; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
40
|
|
97
|
1
|
|
|
1
|
|
6
|
use warnings; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
40
|
|
98
|
1
|
|
|
1
|
|
6
|
use Carp qw/croak/; |
|
1
|
|
|
|
|
6
|
|
|
1
|
|
|
|
|
68
|
|
99
|
1
|
|
|
1
|
|
7
|
use Cwd qw/abs_path/; |
|
1
|
|
|
|
|
7
|
|
|
1
|
|
|
|
|
50
|
|
100
|
1
|
|
|
1
|
|
6
|
use Exporter; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
46
|
|
101
|
1
|
|
|
1
|
|
1289
|
use File::ShareDir; |
|
1
|
|
|
|
|
35038
|
|
|
1
|
|
|
|
|
59
|
|
102
|
1
|
|
|
1
|
|
1541
|
use Glib::Object::Introspection; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
use URI::file; |
104
|
|
|
|
|
|
|
use FindBin; |
105
|
|
|
|
|
|
|
use Poppler::Page::Dimension; |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
our @ISA = qw(Exporter); |
108
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
my $_POPPLER_BASENAME = 'Poppler'; |
110
|
|
|
|
|
|
|
my $_POPPLER_VERSION = '0.18'; |
111
|
|
|
|
|
|
|
my $_POPPLER_PACKAGE = 'Poppler'; |
112
|
|
|
|
|
|
|
|
113
|
|
|
|
|
|
|
=head2 Customizations and overrides |
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
In order to make things more Perlish, C customizes the API generated |
116
|
|
|
|
|
|
|
by L in a few spots: |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
=over |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
=cut |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
# - Customizations ---------------------------------------------------------- # |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
=item * The array ref normally returned by the following functions is flattened |
125
|
|
|
|
|
|
|
into a list: |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
=over |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
=item Poppler::Document::get_attachments |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
=item Poppler::Page::get_link_mapping |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
=item Poppler::Page::find_text |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
=item Poppler::Page::find_text_with_options |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
=item Poppler::Page::get_annot_mapping |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
=item Poppler::Page::get_form_field_mapping |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
=item Poppler::Page::get_image_mapping |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
=item Poppler::Page::get_link_mapping |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
=item Poppler::Page::get_selection_region |
146
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
=item Poppler::Page::get_text_attributes |
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
=item Poppler::Page::get_text_attributes_for_area |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
=back |
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
=cut |
154
|
|
|
|
|
|
|
|
155
|
|
|
|
|
|
|
my @_POPPLER_FLATTEN_ARRAY_REF_RETURN_FOR = qw/ |
156
|
|
|
|
|
|
|
Poppler::Document::get_attachments |
157
|
|
|
|
|
|
|
Poppler::Page::get_link_mapping |
158
|
|
|
|
|
|
|
Poppler::Page::find_text |
159
|
|
|
|
|
|
|
Poppler::Page::find_text_with_options |
160
|
|
|
|
|
|
|
Poppler::Page::get_annot_mapping |
161
|
|
|
|
|
|
|
Poppler::Page::get_form_field_mapping |
162
|
|
|
|
|
|
|
Poppler::Page::get_image_mapping |
163
|
|
|
|
|
|
|
Poppler::Page::get_link_mapping |
164
|
|
|
|
|
|
|
Poppler::Page::get_selection_region |
165
|
|
|
|
|
|
|
Poppler::Page::get_text_attributes |
166
|
|
|
|
|
|
|
Poppler::Page::get_text_attributes_for_area |
167
|
|
|
|
|
|
|
/; |
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
=item * The following functions normally return a boolean and additional out |
170
|
|
|
|
|
|
|
arguments, where the boolean indicates whether the out arguments are valid. |
171
|
|
|
|
|
|
|
They are altered such that when the boolean is true, only the additional out |
172
|
|
|
|
|
|
|
arguments are returned, and when the boolean is false, an empty list is |
173
|
|
|
|
|
|
|
returned. |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
=over |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
=item Poppler::Document::get_id |
178
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
=item Poppler::Page::get_text_layout |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
=item Poppler::Page::get_text_layout_for_area |
182
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
=item Poppler::Page::get_thumbnail_size |
184
|
|
|
|
|
|
|
|
185
|
|
|
|
|
|
|
=back |
186
|
|
|
|
|
|
|
|
187
|
|
|
|
|
|
|
=cut |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
my @_POPPLER_HANDLE_SENTINEL_BOOLEAN_FOR = qw/ |
190
|
|
|
|
|
|
|
Poppler::Document::get_id |
191
|
|
|
|
|
|
|
Poppler::Page::get_text_layout |
192
|
|
|
|
|
|
|
Poppler::Page::get_text_layout_for_area |
193
|
|
|
|
|
|
|
Poppler::Page::get_thumbnail_size |
194
|
|
|
|
|
|
|
/; |
195
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
|
197
|
|
|
|
|
|
|
# - Wiring ------------------------------------------------------------------ # |
198
|
|
|
|
|
|
|
|
199
|
|
|
|
|
|
|
sub import { |
200
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
Glib::Object::Introspection->setup ( |
202
|
|
|
|
|
|
|
search_path => File::ShareDir::dist_dir('Poppler'), |
203
|
|
|
|
|
|
|
basename => $_POPPLER_BASENAME, |
204
|
|
|
|
|
|
|
version => $_POPPLER_VERSION, |
205
|
|
|
|
|
|
|
package => $_POPPLER_PACKAGE, |
206
|
|
|
|
|
|
|
flatten_array_ref_return_for => \@_POPPLER_FLATTEN_ARRAY_REF_RETURN_FOR, |
207
|
|
|
|
|
|
|
handle_sentinel_boolean_for => \@_POPPLER_HANDLE_SENTINEL_BOOLEAN_FOR, |
208
|
|
|
|
|
|
|
); |
209
|
|
|
|
|
|
|
|
210
|
|
|
|
|
|
|
# call into Exporter for the unrecognized arguments; handles exporting and |
211
|
|
|
|
|
|
|
# version checking |
212
|
|
|
|
|
|
|
Poppler->export_to_level (1, @_); |
213
|
|
|
|
|
|
|
|
214
|
|
|
|
|
|
|
} |
215
|
|
|
|
|
|
|
|
216
|
|
|
|
|
|
|
# - Overrides --------------------------------------------------------------- # |
217
|
|
|
|
|
|
|
|
218
|
|
|
|
|
|
|
=item * Perl reimplementations of C, |
219
|
|
|
|
|
|
|
C, and C |
220
|
|
|
|
|
|
|
are provided which remove the need to provide filenames as URIs (e.g. |
221
|
|
|
|
|
|
|
"file:///absolute/path"). The module accepts either real URIs or regular |
222
|
|
|
|
|
|
|
system paths and will convert as necessary using the C module. Any of |
223
|
|
|
|
|
|
|
these formats should work: |
224
|
|
|
|
|
|
|
|
225
|
|
|
|
|
|
|
$p = Poppler::Document->new_from_file( 'file:///home/user/file.pdf' ); |
226
|
|
|
|
|
|
|
$p = Poppler::Document->new_from_file( '/home/user/file.pdf' ); |
227
|
|
|
|
|
|
|
$p = Poppler::Document->new_from_file( 'file.pdf' ); |
228
|
|
|
|
|
|
|
|
229
|
|
|
|
|
|
|
# likewise for save() |
230
|
|
|
|
|
|
|
# likewise for save_a_copy() |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
=cut |
233
|
|
|
|
|
|
|
|
234
|
|
|
|
|
|
|
sub Poppler::Document::new_from_file { |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
my ($class, $fn, $pwd) = @_; |
237
|
|
|
|
|
|
|
|
238
|
|
|
|
|
|
|
$fn = URI::file->new_abs($fn) |
239
|
|
|
|
|
|
|
if (! URI->new($fn)->has_recognized_scheme); |
240
|
|
|
|
|
|
|
my $doc = Glib::Object::Introspection->invoke( |
241
|
|
|
|
|
|
|
'Poppler', 'Document', 'new_from_file', $class, $fn, $pwd); |
242
|
|
|
|
|
|
|
|
243
|
|
|
|
|
|
|
return $doc; |
244
|
|
|
|
|
|
|
|
245
|
|
|
|
|
|
|
} |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
sub Poppler::Document::save { |
248
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
my ($class, $fn) = @_; |
250
|
|
|
|
|
|
|
|
251
|
|
|
|
|
|
|
$fn = URI::file->new_abs($fn) |
252
|
|
|
|
|
|
|
if (! URI->new($fn)->has_recognized_scheme); |
253
|
|
|
|
|
|
|
my $bool = Glib::Object::Introspection->invoke( |
254
|
|
|
|
|
|
|
'Poppler', 'Document', 'save', $class, $fn); |
255
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
return $bool; |
257
|
|
|
|
|
|
|
|
258
|
|
|
|
|
|
|
} |
259
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
sub Poppler::Document::save_a_copy { |
261
|
|
|
|
|
|
|
|
262
|
|
|
|
|
|
|
my ($class, $fn) = @_; |
263
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
$fn = URI::file->new_abs($fn) |
265
|
|
|
|
|
|
|
if (! URI->new($fn)->has_recognized_scheme); |
266
|
|
|
|
|
|
|
my $bool = Glib::Object::Introspection->invoke( |
267
|
|
|
|
|
|
|
'Poppler', 'Document', 'save_a_copy', $class, $fn); |
268
|
|
|
|
|
|
|
|
269
|
|
|
|
|
|
|
return $bool; |
270
|
|
|
|
|
|
|
|
271
|
|
|
|
|
|
|
} |
272
|
|
|
|
|
|
|
|
273
|
|
|
|
|
|
|
sub Poppler::Page::get_size { |
274
|
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
my ($class) = @_; |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
my ($w,$h) = Glib::Object::Introspection->invoke( |
278
|
|
|
|
|
|
|
'Poppler', 'Page', 'get_size', $class); |
279
|
|
|
|
|
|
|
return Poppler::Page::Dimension->new($w, $h) |
280
|
|
|
|
|
|
|
if (! wantarray); |
281
|
|
|
|
|
|
|
return($w, $h); |
282
|
|
|
|
|
|
|
|
283
|
|
|
|
|
|
|
} |
284
|
|
|
|
|
|
|
|
285
|
|
|
|
|
|
|
=back |
286
|
|
|
|
|
|
|
|
287
|
|
|
|
|
|
|
=cut |
288
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
1; |
290
|
|
|
|
|
|
|
|
291
|
|
|
|
|
|
|
__END__ |