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
@@ -1,336 +0,0 @@
|
|
1
|
-
package Omni::Omnitable;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omnicell;
|
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 = $Omni::Config::obj_list;
|
18
|
-
|
19
|
-
###
|
20
|
-
# A table object in Omnipage xml: a table contains cells with various objects
|
21
|
-
#
|
22
|
-
# Do Hoang Nhat Huy, 11 Feb 2011
|
23
|
-
###
|
24
|
-
# Initialization
|
25
|
-
sub new
|
26
|
-
{
|
27
|
-
my ($class) = @_;
|
28
|
-
|
29
|
-
# Objs: a table can have many cells
|
30
|
-
my @objs = ();
|
31
|
-
|
32
|
-
# Grid coordinates
|
33
|
-
my @grid_cols = ();
|
34
|
-
my @grid_rows = ();
|
35
|
-
|
36
|
-
# Content of all rows in the table
|
37
|
-
my @rcontent = ();
|
38
|
-
|
39
|
-
# Class members
|
40
|
-
my $self = { '_self' => $obj_list->{ 'OMNITABLE' },
|
41
|
-
'_raw' => undef,
|
42
|
-
'_content' => undef,
|
43
|
-
'_rcontent' => \@rcontent,
|
44
|
-
'_bottom' => undef,
|
45
|
-
'_top' => undef,
|
46
|
-
'_left' => undef,
|
47
|
-
'_right' => undef,
|
48
|
-
'_alignment' => undef,
|
49
|
-
'_grid_cols' => \@grid_cols,
|
50
|
-
'_grid_rows' => \@grid_rows,
|
51
|
-
'_objs' => \@objs };
|
52
|
-
|
53
|
-
bless $self, $class;
|
54
|
-
return $self;
|
55
|
-
}
|
56
|
-
|
57
|
-
#
|
58
|
-
sub set_raw
|
59
|
-
{
|
60
|
-
my ($self, $raw) = @_;
|
61
|
-
|
62
|
-
# Save the raw xml <table> ... </table>
|
63
|
-
$self->{ '_raw' } = $raw;
|
64
|
-
|
65
|
-
# Parse the raw string
|
66
|
-
my $twig_roots = { $tag_list->{ 'TABLE' } => 1 };
|
67
|
-
my $twig_handlers = { $tag_list->{ 'TABLE' } => sub { parse(@_, \$self); } };
|
68
|
-
|
69
|
-
# XML::Twig
|
70
|
-
my $twig = new XML::Twig( twig_roots => $twig_roots,
|
71
|
-
twig_handlers => $twig_handlers,
|
72
|
-
pretty_print => 'indented' );
|
73
|
-
|
74
|
-
# Start the XML parsing
|
75
|
-
$twig->parse($raw, \$self);
|
76
|
-
$twig->purge;
|
77
|
-
}
|
78
|
-
|
79
|
-
sub get_raw
|
80
|
-
{
|
81
|
-
my ($self) = @_;
|
82
|
-
return $self->{ '_raw' };
|
83
|
-
}
|
84
|
-
|
85
|
-
sub parse
|
86
|
-
{
|
87
|
-
my ($twig, $node, $self) = @_;
|
88
|
-
|
89
|
-
# At first, content is blank
|
90
|
-
my $tmp_content = "";
|
91
|
-
my @tmp_rcontent = ();
|
92
|
-
# because there's no object
|
93
|
-
my @tmp_objs = ();
|
94
|
-
# and no coordinate
|
95
|
-
my @tmp_grid_cols = ();
|
96
|
-
my @tmp_grid_rows = ();
|
97
|
-
|
98
|
-
# Get <table> node attributes
|
99
|
-
my $tmp_bottom = GetNodeAttr($node, $att_list->{ 'BOTTOM' });
|
100
|
-
my $tmp_top = GetNodeAttr($node, $att_list->{ 'TOP' });
|
101
|
-
my $tmp_left = GetNodeAttr($node, $att_list->{ 'LEFT' });
|
102
|
-
my $tmp_right = GetNodeAttr($node, $att_list->{ 'RIGHT' });
|
103
|
-
my $tmp_alignment = GetNodeAttr($node, $att_list->{ 'ALIGN' });
|
104
|
-
|
105
|
-
# A table contains <cell> and <gridtable> tag
|
106
|
-
my $cell_tag = $tag_list->{ 'CELL' };
|
107
|
-
my $grid_tag = $tag_list->{ 'GRID' };
|
108
|
-
my $grid_col_tag = $tag_list->{ 'GRID-COL' };
|
109
|
-
my $grid_row_tag = $tag_list->{ 'GRID-ROW' };
|
110
|
-
|
111
|
-
my $child = undef;
|
112
|
-
# Get the first child which is tha <gridtable>
|
113
|
-
$child = $node->first_child( $grid_tag );
|
114
|
-
|
115
|
-
# Grid table found, a formatted table
|
116
|
-
if (defined $child)
|
117
|
-
{
|
118
|
-
# Get the first grid coordinate
|
119
|
-
$child = $child->first_child();
|
120
|
-
|
121
|
-
# Extract the grid coodinates
|
122
|
-
while (defined $child)
|
123
|
-
{
|
124
|
-
my $xpath = $child->path();
|
125
|
-
|
126
|
-
# if this child is a <gridCol> tag
|
127
|
-
if ($xpath =~ m/\/$grid_col_tag$/)
|
128
|
-
{
|
129
|
-
push @tmp_grid_cols, GetNodeText( $child );
|
130
|
-
}
|
131
|
-
# if this child is a <gridRow> tag
|
132
|
-
elsif ($xpath =~ m/\/$grid_row_tag$/)
|
133
|
-
{
|
134
|
-
push @tmp_grid_rows, GetNodeText( $child );
|
135
|
-
}
|
136
|
-
|
137
|
-
# Little brother
|
138
|
-
if ($child->is_last_child)
|
139
|
-
{
|
140
|
-
last;
|
141
|
-
}
|
142
|
-
else
|
143
|
-
{
|
144
|
-
$child = $child->next_sibling();
|
145
|
-
}
|
146
|
-
}
|
147
|
-
}
|
148
|
-
|
149
|
-
# All cells
|
150
|
-
my @all_cells = $node->descendants( $cell_tag );
|
151
|
-
foreach my $cell (@all_cells)
|
152
|
-
{
|
153
|
-
my $obj = new Omni::Omnicell();
|
154
|
-
|
155
|
-
# Set raw content
|
156
|
-
$obj->set_raw($cell->sprint());
|
157
|
-
|
158
|
-
# Update cell list
|
159
|
-
push @tmp_objs, $obj;
|
160
|
-
|
161
|
-
}
|
162
|
-
|
163
|
-
# Unformatted table
|
164
|
-
if ((scalar(@tmp_grid_cols) == 0) || (scalar(@tmp_grid_rows) == 0))
|
165
|
-
{
|
166
|
-
# Just append cell content
|
167
|
-
foreach my $cell (@tmp_objs) { $tmp_content = $tmp_content . $cell->get_content() . "\n"; }
|
168
|
-
}
|
169
|
-
# Formatted table
|
170
|
-
else
|
171
|
-
{
|
172
|
-
# Table content
|
173
|
-
my @content_matrix = ();
|
174
|
-
|
175
|
-
# Matrix initialization
|
176
|
-
for(my $i = 0; $i < scalar(@tmp_grid_rows); $i++)
|
177
|
-
{
|
178
|
-
# Empty row
|
179
|
-
my @row = ();
|
180
|
-
# Update the row
|
181
|
-
for(my $j = 0; $j < scalar(@tmp_grid_cols); $j++) { push @row, ""; }
|
182
|
-
# Save the row
|
183
|
-
push @content_matrix, [ @row ];
|
184
|
-
}
|
185
|
-
|
186
|
-
# Update table content
|
187
|
-
foreach my $cell (@tmp_objs)
|
188
|
-
{
|
189
|
-
my $row_index = $cell->get_grid_row_from();
|
190
|
-
my $col_index = $cell->get_grid_col_from();
|
191
|
-
|
192
|
-
# Check index and update
|
193
|
-
if (($row_index < scalar(@content_matrix)) && ($col_index < scalar(@{ $content_matrix[ $row_index ] })))
|
194
|
-
{
|
195
|
-
my $cell_content = undef;
|
196
|
-
# Get content
|
197
|
-
$cell_content = $cell->get_content();
|
198
|
-
# Trim
|
199
|
-
$cell_content =~ s/^\s+|\s+$//g;
|
200
|
-
# Remove blank line
|
201
|
-
$cell_content =~ s/\n\s*\n/\n/g;
|
202
|
-
|
203
|
-
$content_matrix[ $row_index ][ $col_index ] = $cell_content;
|
204
|
-
}
|
205
|
-
}
|
206
|
-
|
207
|
-
# Save content
|
208
|
-
foreach my $row (@content_matrix)
|
209
|
-
{
|
210
|
-
# This is used to handle the case in which a cell have multiple lines
|
211
|
-
my @lines = ();
|
212
|
-
# Foreach cell in the row, get its content
|
213
|
-
foreach my $cell (@{ $row })
|
214
|
-
{
|
215
|
-
my @local_lines = split /\n/, $cell;
|
216
|
-
for (my $i = 0; $i < scalar(@local_lines); $i++)
|
217
|
-
{
|
218
|
-
if ($i == scalar(@lines)) { push @lines, ""; }
|
219
|
-
$lines[ $i ] = $lines[ $i ] . $local_lines[ $i ] . "\t";
|
220
|
-
}
|
221
|
-
}
|
222
|
-
|
223
|
-
my $row_content = "";
|
224
|
-
# Add a new row to the table content and the row content
|
225
|
-
foreach my $line (@lines)
|
226
|
-
{
|
227
|
-
$row_content = $row_content . $line . "\n";
|
228
|
-
$tmp_content = $tmp_content . $line . "\n";
|
229
|
-
}
|
230
|
-
|
231
|
-
# Save row content
|
232
|
-
push @tmp_rcontent, $row_content;
|
233
|
-
}
|
234
|
-
}
|
235
|
-
|
236
|
-
# Copy information from temporary variables to class members
|
237
|
-
$$self->{ '_bottom' } = $tmp_bottom;
|
238
|
-
$$self->{ '_top' } = $tmp_top;
|
239
|
-
$$self->{ '_left' } = $tmp_left;
|
240
|
-
$$self->{ '_right' } = $tmp_right;
|
241
|
-
$$self->{ '_alignment' } = $tmp_alignment;
|
242
|
-
|
243
|
-
# Copy all cells
|
244
|
-
@{$$self->{ '_objs' } } = @tmp_objs;
|
245
|
-
|
246
|
-
# Copy all grid columns
|
247
|
-
@{$$self->{ '_grid_cols' } } = @tmp_grid_cols;
|
248
|
-
# Copy all grid rows
|
249
|
-
@{$$self->{ '_grid_rows' } } = @tmp_grid_rows;
|
250
|
-
|
251
|
-
# Copy content
|
252
|
-
$$self->{ '_content' } = $tmp_content;
|
253
|
-
# Copy row content
|
254
|
-
@{ $$self->{ '_rcontent' } } = @tmp_rcontent;
|
255
|
-
}
|
256
|
-
|
257
|
-
sub get_name
|
258
|
-
{
|
259
|
-
my ($self) = @_;
|
260
|
-
return $self->{ '_self' };
|
261
|
-
}
|
262
|
-
|
263
|
-
sub get_objs_ref
|
264
|
-
{
|
265
|
-
my ($self) = @_;
|
266
|
-
return $self->{ '_objs' };
|
267
|
-
}
|
268
|
-
|
269
|
-
sub get_content
|
270
|
-
{
|
271
|
-
my ($self) = @_;
|
272
|
-
return $self->{ '_content' };
|
273
|
-
}
|
274
|
-
|
275
|
-
sub get_row_content
|
276
|
-
{
|
277
|
-
my ($self) = @_;
|
278
|
-
return $self->{ '_rcontent' };
|
279
|
-
}
|
280
|
-
|
281
|
-
sub get_bottom_pos
|
282
|
-
{
|
283
|
-
my ($self) = @_;
|
284
|
-
return $self->{ '_bottom' };
|
285
|
-
}
|
286
|
-
|
287
|
-
sub get_top_pos
|
288
|
-
{
|
289
|
-
my ($self) = @_;
|
290
|
-
return $self->{ '_top' };
|
291
|
-
}
|
292
|
-
|
293
|
-
sub get_left_pos
|
294
|
-
{
|
295
|
-
my ($self) = @_;
|
296
|
-
return $self->{ '_left' };
|
297
|
-
}
|
298
|
-
|
299
|
-
sub get_right_pos
|
300
|
-
{
|
301
|
-
my ($self) = @_;
|
302
|
-
return $self->{ '_right' };
|
303
|
-
}
|
304
|
-
|
305
|
-
sub get_alignment
|
306
|
-
{
|
307
|
-
my ($self) = @_;
|
308
|
-
return $self->{ '_alignment' };
|
309
|
-
}
|
310
|
-
|
311
|
-
# Support functions
|
312
|
-
sub GetNodeAttr
|
313
|
-
{
|
314
|
-
my ($node, $attr) = @_;
|
315
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
316
|
-
}
|
317
|
-
|
318
|
-
sub SetNodeAttr
|
319
|
-
{
|
320
|
-
my ($node, $attr, $value) = @_;
|
321
|
-
$node->set_att($attr, $value);
|
322
|
-
}
|
323
|
-
|
324
|
-
sub GetNodeText
|
325
|
-
{
|
326
|
-
my ($node) = @_;
|
327
|
-
return $node->text;
|
328
|
-
}
|
329
|
-
|
330
|
-
sub SetNodeText
|
331
|
-
{
|
332
|
-
my ($node, $value) = @_;
|
333
|
-
$node->set_text($value);
|
334
|
-
}
|
335
|
-
|
336
|
-
1;
|
@@ -1,162 +0,0 @@
|
|
1
|
-
package Omni::Omniword;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omnirun;
|
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_bottom = undef;
|
22
|
-
my $tmp_top = undef;
|
23
|
-
my $tmp_left = undef;
|
24
|
-
my $tmp_right = undef;
|
25
|
-
|
26
|
-
###
|
27
|
-
# A word object in Omnipage xml: basic blocks of the xml
|
28
|
-
#
|
29
|
-
# Do Hoang Nhat Huy, 07 Jan 2011
|
30
|
-
###
|
31
|
-
# Initialization
|
32
|
-
sub new
|
33
|
-
{
|
34
|
-
my ($class) = @_;
|
35
|
-
|
36
|
-
# Class members
|
37
|
-
my $self = { '_self' => $obj_list->{ 'OMNIWORD' },
|
38
|
-
'_raw' => undef,
|
39
|
-
'_content' => undef,
|
40
|
-
'_bottom' => undef,
|
41
|
-
'_top' => undef,
|
42
|
-
'_left' => undef,
|
43
|
-
'_right' => undef };
|
44
|
-
|
45
|
-
bless $self, $class;
|
46
|
-
return $self;
|
47
|
-
}
|
48
|
-
|
49
|
-
#
|
50
|
-
sub set_raw
|
51
|
-
{
|
52
|
-
my ($self, $raw) = @_;
|
53
|
-
|
54
|
-
# Save the raw xml <wd> ... </wd>
|
55
|
-
$self->{ '_raw' } = $raw;
|
56
|
-
|
57
|
-
# Parse the raw string
|
58
|
-
my $twig_roots = { $tag_list->{ 'WORD' } => 1 };
|
59
|
-
my $twig_handlers = { $tag_list->{ 'WORD' } => \&parse};
|
60
|
-
|
61
|
-
# XML::Twig
|
62
|
-
my $twig = new XML::Twig( twig_roots => $twig_roots,
|
63
|
-
twig_handlers => $twig_handlers,
|
64
|
-
pretty_print => 'indented' );
|
65
|
-
|
66
|
-
# Start the XML parsing
|
67
|
-
$twig->parse($raw);
|
68
|
-
$twig->purge;
|
69
|
-
|
70
|
-
# Copy information from temporary variables to class members
|
71
|
-
$self->{ '_bottom' } = $tmp_bottom;
|
72
|
-
$self->{ '_top' } = $tmp_top;
|
73
|
-
$self->{ '_left' } = $tmp_left;
|
74
|
-
$self->{ '_right' } = $tmp_right;
|
75
|
-
|
76
|
-
# Copy content
|
77
|
-
$self->{ '_content' } = $tmp_content;
|
78
|
-
}
|
79
|
-
|
80
|
-
sub get_raw
|
81
|
-
{
|
82
|
-
my ($self) = @_;
|
83
|
-
return $self->{ '_raw' };
|
84
|
-
}
|
85
|
-
|
86
|
-
sub parse
|
87
|
-
{
|
88
|
-
my ($twig, $node) = @_;
|
89
|
-
|
90
|
-
# Get <run> node attributes
|
91
|
-
$tmp_bottom = GetNodeAttr($node, $att_list->{ 'BOTTOM' });
|
92
|
-
$tmp_top = GetNodeAttr($node, $att_list->{ 'TOP' });
|
93
|
-
$tmp_left = GetNodeAttr($node, $att_list->{ 'LEFT' });
|
94
|
-
$tmp_right = GetNodeAttr($node, $att_list->{ 'RIGHT' });
|
95
|
-
|
96
|
-
# Get the word's content
|
97
|
-
$tmp_content = GetNodeText($node);
|
98
|
-
$tmp_content =~ s/^\s+|\s+$//g;
|
99
|
-
}
|
100
|
-
|
101
|
-
sub get_name
|
102
|
-
{
|
103
|
-
my ($self) = @_;
|
104
|
-
return $self->{ '_self' };
|
105
|
-
}
|
106
|
-
|
107
|
-
sub get_content
|
108
|
-
{
|
109
|
-
my ($self) = @_;
|
110
|
-
return $self->{ '_content' };
|
111
|
-
}
|
112
|
-
|
113
|
-
sub get_bottom_pos
|
114
|
-
{
|
115
|
-
my ($self) = @_;
|
116
|
-
return $self->{ '_bottom' };
|
117
|
-
}
|
118
|
-
|
119
|
-
sub get_top_pos
|
120
|
-
{
|
121
|
-
my ($self) = @_;
|
122
|
-
return $self->{ '_top' };
|
123
|
-
}
|
124
|
-
|
125
|
-
sub get_left_pos
|
126
|
-
{
|
127
|
-
my ($self) = @_;
|
128
|
-
return $self->{ '_left' };
|
129
|
-
}
|
130
|
-
|
131
|
-
sub get_right_pos
|
132
|
-
{
|
133
|
-
my ($self) = @_;
|
134
|
-
return $self->{ '_right' };
|
135
|
-
}
|
136
|
-
|
137
|
-
# Support functions
|
138
|
-
sub GetNodeAttr
|
139
|
-
{
|
140
|
-
my ($node, $attr) = @_;
|
141
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
142
|
-
}
|
143
|
-
|
144
|
-
sub SetNodeAttr
|
145
|
-
{
|
146
|
-
my ($node, $attr, $value) = @_;
|
147
|
-
$node->set_att($attr, $value);
|
148
|
-
}
|
149
|
-
|
150
|
-
sub GetNodeText
|
151
|
-
{
|
152
|
-
my ($node) = @_;
|
153
|
-
return $node->text;
|
154
|
-
}
|
155
|
-
|
156
|
-
sub SetNodeText
|
157
|
-
{
|
158
|
-
my ($node, $value) = @_;
|
159
|
-
$node->set_text($value);
|
160
|
-
}
|
161
|
-
|
162
|
-
1;
|