biblicit 2.0.3 → 2.0.4
Sign up to get free protection for your applications and to get access to all the features.
- data/README.md +0 -2
- data/biblicit.gemspec +1 -1
- data/parscit/bin/citeExtract.pl +9 -161
- data/parscit/bin/sectExtract.pl +0 -14
- data/parscit/lib/ParsCit/Controller.pm +0 -59
- data/parscit/lib/ParsCit/PreProcess.pm +0 -4
- data/parscit/lib/ParsCit/Tr2crfpp.pm +1 -7
- metadata +4 -24
- data/parscit/bin/sectLabel/processOmniXML.pl +0 -1427
- data/parscit/bin/sectLabel/processOmniXML_new.pl +0 -1025
- data/parscit/bin/sectLabel/processOmniXMLv2.pl +0 -1529
- data/parscit/bin/sectLabel/processOmniXMLv3.pl +0 -964
- data/parscit/bin/sectLabel/simplifyOmniXML.pl +0 -382
- data/parscit/bin/xml2train.pl +0 -193
- data/parscit/lib/Omni/Config.pm +0 -93
- data/parscit/lib/Omni/Omnicell.pm +0 -263
- data/parscit/lib/Omni/Omnicol.pm +0 -292
- data/parscit/lib/Omni/Omnidd.pm +0 -328
- data/parscit/lib/Omni/Omnidoc.pm +0 -153
- data/parscit/lib/Omni/Omniframe.pm +0 -223
- data/parscit/lib/Omni/Omniline.pm +0 -423
- data/parscit/lib/Omni/Omnipage.pm +0 -282
- data/parscit/lib/Omni/Omnipara.pm +0 -232
- data/parscit/lib/Omni/Omnirun.pm +0 -303
- data/parscit/lib/Omni/Omnitable.pm +0 -336
- data/parscit/lib/Omni/Omniword.pm +0 -162
- data/parscit/lib/Omni/Traversal.pm +0 -313
- data/parscit/lib/SectLabel/AAMatching.pm +0 -1949
data/parscit/lib/Omni/Omnidd.pm
DELETED
@@ -1,328 +0,0 @@
|
|
1
|
-
package Omni::Omnidd;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omnicol;
|
9
|
-
use Omni::Omnipara;
|
10
|
-
use Omni::Omniframe;
|
11
|
-
use Omni::Omnitable;
|
12
|
-
|
13
|
-
# Extern libraries
|
14
|
-
use XML::Twig;
|
15
|
-
use XML::Parser;
|
16
|
-
|
17
|
-
# Global variables
|
18
|
-
my $tag_list = $Omni::Config::tag_list;
|
19
|
-
my $att_list = $Omni::Config::att_list;
|
20
|
-
my $obj_list = $Omni::Config::obj_list;
|
21
|
-
|
22
|
-
# Temporary variables
|
23
|
-
|
24
|
-
###
|
25
|
-
# A dd object in Omnipage xml: a dd, don't know what it is, but its structure
|
26
|
-
# is quite similar to a column
|
27
|
-
#
|
28
|
-
# Do Hoang Nhat Huy, 11 Jan 2011
|
29
|
-
###
|
30
|
-
# Initialization
|
31
|
-
sub new
|
32
|
-
{
|
33
|
-
my ($class) = @_;
|
34
|
-
|
35
|
-
# dd: a dd can have many tables, or pictures, may be paras, and columns
|
36
|
-
my @objs = ();
|
37
|
-
|
38
|
-
# Class members
|
39
|
-
my $self = { '_self' => $obj_list->{ 'OMNIDD' },
|
40
|
-
'_raw' => undef,
|
41
|
-
'_content' => undef,
|
42
|
-
'_bottom' => undef,
|
43
|
-
'_top' => undef,
|
44
|
-
'_left' => undef,
|
45
|
-
'_right' => undef,
|
46
|
-
'_bottom_dist' => undef,
|
47
|
-
'_top_dist' => undef,
|
48
|
-
'_left_dist' => undef,
|
49
|
-
'_right_dist' => undef,
|
50
|
-
'_objs' => \@objs };
|
51
|
-
|
52
|
-
bless $self, $class;
|
53
|
-
return $self;
|
54
|
-
}
|
55
|
-
|
56
|
-
#
|
57
|
-
sub set_raw
|
58
|
-
{
|
59
|
-
my ($self, $raw) = @_;
|
60
|
-
|
61
|
-
# Save the raw xml <column> ... </column>
|
62
|
-
$self->{ '_raw' } = $raw;
|
63
|
-
|
64
|
-
# Parse the raw string
|
65
|
-
my $twig_roots = { $tag_list->{ 'DD' } => 1 };
|
66
|
-
my $twig_handlers = { $tag_list->{ 'DD' } => sub { parse(@_, \$self); } };
|
67
|
-
|
68
|
-
# XML::Twig
|
69
|
-
my $twig = new XML::Twig( twig_roots => $twig_roots,
|
70
|
-
twig_handlers => $twig_handlers,
|
71
|
-
pretty_print => 'indented' );
|
72
|
-
|
73
|
-
# Start the XML parsing
|
74
|
-
$twig->parse($raw, \$self);
|
75
|
-
$twig->purge;
|
76
|
-
}
|
77
|
-
|
78
|
-
sub get_raw
|
79
|
-
{
|
80
|
-
my ($self) = @_;
|
81
|
-
return $self->{ '_raw' };
|
82
|
-
}
|
83
|
-
|
84
|
-
sub parse
|
85
|
-
{
|
86
|
-
my ($twig, $node, $self) = @_;
|
87
|
-
|
88
|
-
# At first, content is blank
|
89
|
-
my $tmp_content = "";
|
90
|
-
# because there's no object
|
91
|
-
my @tmp_objs = ();
|
92
|
-
|
93
|
-
# Get <dd> node attributes
|
94
|
-
my $tmp_bottom = GetNodeAttr($node, $att_list->{ 'BOTTOM' });
|
95
|
-
my $tmp_top = GetNodeAttr($node, $att_list->{ 'TOP' });
|
96
|
-
my $tmp_left = GetNodeAttr($node, $att_list->{ 'LEFT' });
|
97
|
-
my $tmp_right = GetNodeAttr($node, $att_list->{ 'RIGHT' });
|
98
|
-
my $tmp_bottom_dist = GetNodeAttr($node, $att_list->{ 'BOTTOMDIST' });
|
99
|
-
my $tmp_top_dist = GetNodeAttr($node, $att_list->{ 'TOPDIST' });
|
100
|
-
my $tmp_left_dist = GetNodeAttr($node, $att_list->{ 'LEFTDIST' });
|
101
|
-
my $tmp_right_dist = GetNodeAttr($node, $att_list->{ 'RIGHTDIST' });
|
102
|
-
|
103
|
-
# Check if there's any paragraph, col, table, or picture
|
104
|
-
# The large number of possible children is due to the
|
105
|
-
# ambiguous structure of the Omnipage XML
|
106
|
-
my $dd_tag = $tag_list->{ 'DD' };
|
107
|
-
my $img_tag = $tag_list->{ 'PICTURE' };
|
108
|
-
my $para_tag = $tag_list->{ 'PARA' };
|
109
|
-
my $table_tag = $tag_list->{ 'TABLE' };
|
110
|
-
my $column_tag = $tag_list->{ 'COLUMN' };
|
111
|
-
my $frame_tag = $tag_list->{ 'FRAME' };
|
112
|
-
|
113
|
-
my $child = undef;
|
114
|
-
# Get the first child in the body text
|
115
|
-
$child = $node->first_child();
|
116
|
-
|
117
|
-
while (defined $child)
|
118
|
-
{
|
119
|
-
my $xpath = $child->path();
|
120
|
-
|
121
|
-
# if this child is a <para> tag
|
122
|
-
if ($xpath =~ m/\/$para_tag$/)
|
123
|
-
{
|
124
|
-
my $para = new Omni::Omnipara();
|
125
|
-
|
126
|
-
# Set raw content
|
127
|
-
$para->set_raw($child->sprint());
|
128
|
-
|
129
|
-
# Update paragraph list
|
130
|
-
push @tmp_objs, $para;
|
131
|
-
|
132
|
-
# Update content
|
133
|
-
$tmp_content = $tmp_content . $para->get_content() . "\n";
|
134
|
-
}
|
135
|
-
# if this child is a <dd> tag
|
136
|
-
elsif ($xpath =~ m/\/$dd_tag$/)
|
137
|
-
{
|
138
|
-
my $dd = new Omni::Omnidd();
|
139
|
-
|
140
|
-
# Set raw content
|
141
|
-
$dd->set_raw($child->sprint());
|
142
|
-
|
143
|
-
# Nested <dd> is not allowed so we copy the objects
|
144
|
-
my $objects = $dd->get_objs_ref();
|
145
|
-
|
146
|
-
# Update <dd> objects list
|
147
|
-
push @tmp_objs, @{ $objects };
|
148
|
-
|
149
|
-
# Update content
|
150
|
-
$tmp_content = $tmp_content . $dd->get_content() . "\n";
|
151
|
-
}
|
152
|
-
# if this child is a <table> tag
|
153
|
-
elsif ($xpath =~ m/\/$table_tag$/)
|
154
|
-
{
|
155
|
-
my $table = new Omni::Omnitable();
|
156
|
-
|
157
|
-
# Set raw content
|
158
|
-
$table->set_raw($child->sprint());
|
159
|
-
|
160
|
-
# Update paragraph list
|
161
|
-
push @tmp_objs, $table;
|
162
|
-
|
163
|
-
# Update content
|
164
|
-
$tmp_content = $tmp_content . $table->get_content() . "\n";
|
165
|
-
}
|
166
|
-
# if this child is a <picture> tag
|
167
|
-
elsif ($xpath =~ m/\/$img_tag$/)
|
168
|
-
{
|
169
|
-
#my $img = new Omni::Omniimg();
|
170
|
-
|
171
|
-
# Set raw content
|
172
|
-
#$img->set_raw($child->sprint());
|
173
|
-
|
174
|
-
# Update paragraph list
|
175
|
-
#push @tmp_objs, $img;
|
176
|
-
|
177
|
-
# Update content
|
178
|
-
#$tmp_content = $tmp_content . $img->get_content() . "\n";
|
179
|
-
}
|
180
|
-
# if this child is a <column> tag
|
181
|
-
elsif ($xpath =~ m/\/$column_tag$/)
|
182
|
-
{
|
183
|
-
#my $col = new Omni::Omnicol();
|
184
|
-
|
185
|
-
# Set raw content
|
186
|
-
#$col->set_raw($child->sprint());
|
187
|
-
|
188
|
-
# Update paragraph list
|
189
|
-
#push @tmp_objs, $col;
|
190
|
-
|
191
|
-
# Update content
|
192
|
-
#$tmp_content = $tmp_content . $col->get_content() . "\n";
|
193
|
-
}
|
194
|
-
# if this child is <frame>
|
195
|
-
elsif ($xpath =~ m/\/$frame_tag$/)
|
196
|
-
{
|
197
|
-
my $frame = new Omni::Omniframe();
|
198
|
-
|
199
|
-
# Set raw content
|
200
|
-
$frame->set_raw($child->sprint());
|
201
|
-
|
202
|
-
# Update column list
|
203
|
-
push @tmp_objs, $frame;
|
204
|
-
|
205
|
-
# Update content
|
206
|
-
$tmp_content = $tmp_content . $frame->get_content() . "\n";
|
207
|
-
}
|
208
|
-
|
209
|
-
# Little brother
|
210
|
-
if ($child->is_last_child)
|
211
|
-
{
|
212
|
-
last;
|
213
|
-
}
|
214
|
-
else
|
215
|
-
{
|
216
|
-
$child = $child->next_sibling();
|
217
|
-
}
|
218
|
-
}
|
219
|
-
|
220
|
-
# Copy information from temporary variables to class members
|
221
|
-
$$self->{ '_bottom' } = $tmp_bottom;
|
222
|
-
$$self->{ '_top' } = $tmp_top;
|
223
|
-
$$self->{ '_left' } = $tmp_left;
|
224
|
-
$$self->{ '_right' } = $tmp_right;
|
225
|
-
$$self->{ '_bottom_dist' } = $tmp_bottom_dist;
|
226
|
-
$$self->{ '_top_dist' } = $tmp_top_dist;
|
227
|
-
$$self->{ '_left_dist' } = $tmp_left_dist;
|
228
|
-
$$self->{ '_right_dist' } = $tmp_right_dist;
|
229
|
-
|
230
|
-
# Copy content
|
231
|
-
$$self->{ '_content' } = $tmp_content;
|
232
|
-
|
233
|
-
# Copy all objects
|
234
|
-
@{$$self->{ '_objs' } } = @tmp_objs;
|
235
|
-
}
|
236
|
-
|
237
|
-
sub get_name
|
238
|
-
{
|
239
|
-
my ($self) = @_;
|
240
|
-
return $self->{ '_self' };
|
241
|
-
}
|
242
|
-
|
243
|
-
sub get_objs_ref
|
244
|
-
{
|
245
|
-
my ($self) = @_;
|
246
|
-
return $self->{ '_objs' };
|
247
|
-
}
|
248
|
-
|
249
|
-
sub get_content
|
250
|
-
{
|
251
|
-
my ($self) = @_;
|
252
|
-
return $self->{ '_content' };
|
253
|
-
}
|
254
|
-
|
255
|
-
sub get_bottom_pos
|
256
|
-
{
|
257
|
-
my ($self) = @_;
|
258
|
-
return $self->{ '_bottom' };
|
259
|
-
}
|
260
|
-
|
261
|
-
sub get_top_pos
|
262
|
-
{
|
263
|
-
my ($self) = @_;
|
264
|
-
return $self->{ '_top' };
|
265
|
-
}
|
266
|
-
|
267
|
-
sub get_left_pos
|
268
|
-
{
|
269
|
-
my ($self) = @_;
|
270
|
-
return $self->{ '_left' };
|
271
|
-
}
|
272
|
-
|
273
|
-
sub get_right_pos
|
274
|
-
{
|
275
|
-
my ($self) = @_;
|
276
|
-
return $self->{ '_right' };
|
277
|
-
}
|
278
|
-
|
279
|
-
sub get_bottom_distance
|
280
|
-
{
|
281
|
-
my ($self) = @_;
|
282
|
-
return $self->{ '_bottom_dist' };
|
283
|
-
}
|
284
|
-
|
285
|
-
sub get_top_distance
|
286
|
-
{
|
287
|
-
my ($self) = @_;
|
288
|
-
return $self->{ '_top_dist' };
|
289
|
-
}
|
290
|
-
|
291
|
-
sub get_left_distance
|
292
|
-
{
|
293
|
-
my ($self) = @_;
|
294
|
-
return $self->{ '_left_dist' };
|
295
|
-
}
|
296
|
-
|
297
|
-
sub get_right_distance
|
298
|
-
{
|
299
|
-
my ($self) = @_;
|
300
|
-
return $self->{ '_right_dist' };
|
301
|
-
}
|
302
|
-
|
303
|
-
# Support functions
|
304
|
-
sub GetNodeAttr
|
305
|
-
{
|
306
|
-
my ($node, $attr) = @_;
|
307
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
308
|
-
}
|
309
|
-
|
310
|
-
sub SetNodeAttr
|
311
|
-
{
|
312
|
-
my ($node, $attr, $value) = @_;
|
313
|
-
$node->set_att($attr, $value);
|
314
|
-
}
|
315
|
-
|
316
|
-
sub GetNodeText
|
317
|
-
{
|
318
|
-
my ($node) = @_;
|
319
|
-
return $node->text;
|
320
|
-
}
|
321
|
-
|
322
|
-
sub SetNodeText
|
323
|
-
{
|
324
|
-
my ($node, $value) = @_;
|
325
|
-
$node->set_text($value);
|
326
|
-
}
|
327
|
-
|
328
|
-
1;
|
data/parscit/lib/Omni/Omnidoc.pm
DELETED
@@ -1,153 +0,0 @@
|
|
1
|
-
package Omni::Omnidoc;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omnipage;
|
9
|
-
|
10
|
-
# Extern libraries
|
11
|
-
use XML::Twig;
|
12
|
-
use XML::Parser;
|
13
|
-
|
14
|
-
# Global variables
|
15
|
-
my $tag_list = $Omni::Config::tag_list;
|
16
|
-
my $att_list = $Omni::Config::att_list;
|
17
|
-
my $obj_list = $Omni::Config::obj_list;
|
18
|
-
|
19
|
-
# Temporary variables
|
20
|
-
my $tmp_content = undef;
|
21
|
-
my @tmp_pages = ();
|
22
|
-
|
23
|
-
###
|
24
|
-
# A whole document object in Omnipage xml: a document contains many pages
|
25
|
-
#
|
26
|
-
# Do Hoang Nhat Huy, 09 Jan 2011
|
27
|
-
###
|
28
|
-
# Initialization
|
29
|
-
sub new
|
30
|
-
{
|
31
|
-
my ($class) = @_;
|
32
|
-
|
33
|
-
# Lines: a paragraph can have multiple lines
|
34
|
-
my @pages = ();
|
35
|
-
|
36
|
-
# Class members
|
37
|
-
my $self = { '_self' => $obj_list->{ 'OMNIDOC' },
|
38
|
-
'_raw' => undef,
|
39
|
-
'_content' => undef,
|
40
|
-
'_pages' => \@pages };
|
41
|
-
|
42
|
-
bless $self, $class;
|
43
|
-
return $self;
|
44
|
-
}
|
45
|
-
|
46
|
-
#
|
47
|
-
sub set_raw
|
48
|
-
{
|
49
|
-
my ($self, $raw) = @_;
|
50
|
-
|
51
|
-
# Save the raw xml <para> ... </para>
|
52
|
-
$self->{ '_raw' } = $raw;
|
53
|
-
|
54
|
-
# At first, content is blank
|
55
|
-
$tmp_content = "";
|
56
|
-
# because there's no document
|
57
|
-
@tmp_pages = ();
|
58
|
-
|
59
|
-
# Parse the raw string
|
60
|
-
my $twig_roots = { $tag_list->{ 'DOCUMENT' } => 1 };
|
61
|
-
my $twig_handlers = { $tag_list->{ 'DOCUMENT' } => \&parse};
|
62
|
-
|
63
|
-
# XML::Twig
|
64
|
-
my $twig = new XML::Twig( twig_roots => $twig_roots,
|
65
|
-
twig_handlers => $twig_handlers,
|
66
|
-
pretty_print => 'indented' );
|
67
|
-
|
68
|
-
# Start the XML parsing
|
69
|
-
$twig->parse($raw);
|
70
|
-
$twig->purge;
|
71
|
-
|
72
|
-
# Copy information from temporary variables to class members
|
73
|
-
|
74
|
-
# Copy all pages
|
75
|
-
@{$self->{ '_pages' } } = @tmp_pages;
|
76
|
-
|
77
|
-
# Copy content
|
78
|
-
$self->{ '_content' } = $tmp_content;
|
79
|
-
}
|
80
|
-
|
81
|
-
sub get_raw
|
82
|
-
{
|
83
|
-
my ($self) = @_;
|
84
|
-
return $self->{ '_raw' };
|
85
|
-
}
|
86
|
-
|
87
|
-
sub parse
|
88
|
-
{
|
89
|
-
my ($twig, $node) = @_;
|
90
|
-
|
91
|
-
# Get <document> node attributes
|
92
|
-
|
93
|
-
# Check if there's any para
|
94
|
-
my @all_pages = $node->descendants( $tag_list->{ 'PAGE' } );
|
95
|
-
foreach my $pg (@all_pages)
|
96
|
-
{
|
97
|
-
my $page = new Omni::Omnipage();
|
98
|
-
|
99
|
-
# Set raw content
|
100
|
-
$page->set_raw($pg->sprint());
|
101
|
-
|
102
|
-
# Update page list
|
103
|
-
push @tmp_pages, $page;
|
104
|
-
|
105
|
-
# Update content
|
106
|
-
$tmp_content = $tmp_content . $page->get_content() . "\n";
|
107
|
-
}
|
108
|
-
}
|
109
|
-
|
110
|
-
sub get_name
|
111
|
-
{
|
112
|
-
my ($self) = @_;
|
113
|
-
return $self->{ '_self' };
|
114
|
-
}
|
115
|
-
|
116
|
-
sub get_objs_ref
|
117
|
-
{
|
118
|
-
my ($self) = @_;
|
119
|
-
return $self->{ '_pages' };
|
120
|
-
}
|
121
|
-
|
122
|
-
sub get_content
|
123
|
-
{
|
124
|
-
my ($self) = @_;
|
125
|
-
return $self->{ '_content' };
|
126
|
-
}
|
127
|
-
|
128
|
-
# Support functions
|
129
|
-
sub GetNodeAttr
|
130
|
-
{
|
131
|
-
my ($node, $attr) = @_;
|
132
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
133
|
-
}
|
134
|
-
|
135
|
-
sub SetNodeAttr
|
136
|
-
{
|
137
|
-
my ($node, $attr, $value) = @_;
|
138
|
-
$node->set_att($attr, $value);
|
139
|
-
}
|
140
|
-
|
141
|
-
sub GetNodeText
|
142
|
-
{
|
143
|
-
my ($node) = @_;
|
144
|
-
return $node->text;
|
145
|
-
}
|
146
|
-
|
147
|
-
sub SetNodeText
|
148
|
-
{
|
149
|
-
my ($node, $value) = @_;
|
150
|
-
$node->set_text($value);
|
151
|
-
}
|
152
|
-
|
153
|
-
1;
|
@@ -1,223 +0,0 @@
|
|
1
|
-
package Omni::Omniframe;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omnipara;
|
9
|
-
use Omni::Omnitable;
|
10
|
-
|
11
|
-
# Extern libraries
|
12
|
-
use XML::Twig;
|
13
|
-
use XML::Parser;
|
14
|
-
|
15
|
-
# Global variables
|
16
|
-
my $tag_list = $Omni::Config::tag_list;
|
17
|
-
my $att_list = $Omni::Config::att_list;
|
18
|
-
my $obj_list = $Omni::Config::obj_list;
|
19
|
-
|
20
|
-
###
|
21
|
-
# A frame object in Omnipage xml: a frame contains paragraphs
|
22
|
-
# (this is my observation and can be invalid under close scrunity
|
23
|
-
# of new "evidence")
|
24
|
-
#
|
25
|
-
# Do Hoang Nhat Huy, 23 Feb 2011
|
26
|
-
###
|
27
|
-
# Initialization
|
28
|
-
sub new
|
29
|
-
{
|
30
|
-
my ($class) = @_;
|
31
|
-
|
32
|
-
# Objs: paragraphs
|
33
|
-
my @objs = ();
|
34
|
-
|
35
|
-
# Class members
|
36
|
-
my $self = { '_self' => $obj_list->{ 'OMNIFRAME' },
|
37
|
-
'_raw' => undef,
|
38
|
-
'_content' => undef,
|
39
|
-
'_bottom' => undef,
|
40
|
-
'_top' => undef,
|
41
|
-
'_left' => undef,
|
42
|
-
'_right' => undef,
|
43
|
-
'_objs' => \@objs };
|
44
|
-
|
45
|
-
bless $self, $class;
|
46
|
-
return $self;
|
47
|
-
}
|
48
|
-
|
49
|
-
sub set_raw
|
50
|
-
{
|
51
|
-
my ($self, $raw) = @_;
|
52
|
-
|
53
|
-
# Save the raw xml <column> ... </column>
|
54
|
-
$self->{ '_raw' } = $raw;
|
55
|
-
|
56
|
-
# Parse the raw string
|
57
|
-
my $twig_roots = { $tag_list->{ 'FRAME' } => 1 };
|
58
|
-
my $twig_handlers = { $tag_list->{ 'FRAME' } => sub { parse(@_, \$self); } };
|
59
|
-
|
60
|
-
# XML::Twig
|
61
|
-
my $twig = new XML::Twig( twig_roots => $twig_roots,
|
62
|
-
twig_handlers => $twig_handlers,
|
63
|
-
pretty_print => 'indented' );
|
64
|
-
|
65
|
-
# Start the XML parsing
|
66
|
-
$twig->parse($raw, \$self);
|
67
|
-
$twig->purge;
|
68
|
-
}
|
69
|
-
|
70
|
-
sub get_raw
|
71
|
-
{
|
72
|
-
my ($self) = @_;
|
73
|
-
return $self->{ '_raw' };
|
74
|
-
}
|
75
|
-
|
76
|
-
sub parse
|
77
|
-
{
|
78
|
-
my ($twig, $node, $self) = @_;
|
79
|
-
|
80
|
-
# At first, content is blank
|
81
|
-
my $tmp_content = "";
|
82
|
-
# because there's no object
|
83
|
-
my @tmp_objs = ();
|
84
|
-
|
85
|
-
# Get <column> node attributes
|
86
|
-
my $tmp_bottom = GetNodeAttr($node, $att_list->{ 'BOTTOM' });
|
87
|
-
my $tmp_top = GetNodeAttr($node, $att_list->{ 'TOP' });
|
88
|
-
my $tmp_left = GetNodeAttr($node, $att_list->{ 'LEFT' });
|
89
|
-
my $tmp_right = GetNodeAttr($node, $att_list->{ 'RIGHT' });
|
90
|
-
|
91
|
-
# Check if there's any paragraph, dd, table, or picture
|
92
|
-
# The large number of possible children is due to the
|
93
|
-
# ambiguous structure of the Omnipage XML
|
94
|
-
my $para_tag = $tag_list->{ 'PARA' };
|
95
|
-
my $table_tag = $tag_list->{ 'TABLE' };
|
96
|
-
|
97
|
-
# Get the first child in the body text
|
98
|
-
my $child = $node->first_child();
|
99
|
-
|
100
|
-
while (defined $child)
|
101
|
-
{
|
102
|
-
my $xpath = $child->path();
|
103
|
-
|
104
|
-
# if this child is <para>
|
105
|
-
if ($xpath =~ m/\/$para_tag$/)
|
106
|
-
{
|
107
|
-
my $para = new Omni::Omnipara();
|
108
|
-
|
109
|
-
# Set raw content
|
110
|
-
$para->set_raw($child->sprint());
|
111
|
-
|
112
|
-
# Update paragraph list
|
113
|
-
push @tmp_objs, $para;
|
114
|
-
|
115
|
-
# Update content
|
116
|
-
$tmp_content = $tmp_content . $para->get_content() . "\n";
|
117
|
-
}
|
118
|
-
elsif ($xpath =~ m/\/$table_tag$/)
|
119
|
-
{
|
120
|
-
my $table = new Omni::Omnitable();
|
121
|
-
|
122
|
-
# Set raw content
|
123
|
-
$table->set_raw($child->sprint());
|
124
|
-
|
125
|
-
# Update paragraph list
|
126
|
-
push @tmp_objs, $table;
|
127
|
-
|
128
|
-
# Update content
|
129
|
-
$tmp_content = $tmp_content . $table->get_content() . "\n";
|
130
|
-
}
|
131
|
-
|
132
|
-
# Little brother
|
133
|
-
if ($child->is_last_child)
|
134
|
-
{
|
135
|
-
last;
|
136
|
-
}
|
137
|
-
else
|
138
|
-
{
|
139
|
-
$child = $child->next_sibling();
|
140
|
-
}
|
141
|
-
}
|
142
|
-
|
143
|
-
# Copy information from temporary variables to class members
|
144
|
-
$$self->{ '_bottom' } = $tmp_bottom;
|
145
|
-
$$self->{ '_top' } = $tmp_top;
|
146
|
-
$$self->{ '_left' } = $tmp_left;
|
147
|
-
$$self->{ '_right' } = $tmp_right;
|
148
|
-
|
149
|
-
# Copy all objects
|
150
|
-
@{$$self->{ '_objs' } } = @tmp_objs;
|
151
|
-
|
152
|
-
# Copy content
|
153
|
-
$$self->{ '_content' } = $tmp_content;
|
154
|
-
}
|
155
|
-
|
156
|
-
sub get_name
|
157
|
-
{
|
158
|
-
my ($self) = @_;
|
159
|
-
return $self->{ '_self' };
|
160
|
-
}
|
161
|
-
|
162
|
-
sub get_objs_ref
|
163
|
-
{
|
164
|
-
my ($self) = @_;
|
165
|
-
return $self->{ '_objs' };
|
166
|
-
}
|
167
|
-
|
168
|
-
sub get_content
|
169
|
-
{
|
170
|
-
my ($self) = @_;
|
171
|
-
return $self->{ '_content' };
|
172
|
-
}
|
173
|
-
|
174
|
-
sub get_bottom_pos
|
175
|
-
{
|
176
|
-
my ($self) = @_;
|
177
|
-
return $self->{ '_bottom' };
|
178
|
-
}
|
179
|
-
|
180
|
-
sub get_top_pos
|
181
|
-
{
|
182
|
-
my ($self) = @_;
|
183
|
-
return $self->{ '_top' };
|
184
|
-
}
|
185
|
-
|
186
|
-
sub get_left_pos
|
187
|
-
{
|
188
|
-
my ($self) = @_;
|
189
|
-
return $self->{ '_left' };
|
190
|
-
}
|
191
|
-
|
192
|
-
sub get_right_pos
|
193
|
-
{
|
194
|
-
my ($self) = @_;
|
195
|
-
return $self->{ '_right' };
|
196
|
-
}
|
197
|
-
|
198
|
-
# Support functions
|
199
|
-
sub GetNodeAttr
|
200
|
-
{
|
201
|
-
my ($node, $attr) = @_;
|
202
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
203
|
-
}
|
204
|
-
|
205
|
-
sub SetNodeAttr
|
206
|
-
{
|
207
|
-
my ($node, $attr, $value) = @_;
|
208
|
-
$node->set_att($attr, $value);
|
209
|
-
}
|
210
|
-
|
211
|
-
sub GetNodeText
|
212
|
-
{
|
213
|
-
my ($node) = @_;
|
214
|
-
return $node->text;
|
215
|
-
}
|
216
|
-
|
217
|
-
sub SetNodeText
|
218
|
-
{
|
219
|
-
my ($node, $value) = @_;
|
220
|
-
$node->set_text($value);
|
221
|
-
}
|
222
|
-
|
223
|
-
1;
|