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
@@ -1,232 +0,0 @@
|
|
1
|
-
package Omni::Omnipara;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omniword;
|
9
|
-
use Omni::Omnirun;
|
10
|
-
use Omni::Omniline;
|
11
|
-
|
12
|
-
# Extern libraries
|
13
|
-
use XML::Twig;
|
14
|
-
use XML::Parser;
|
15
|
-
|
16
|
-
# Global variables
|
17
|
-
my $tag_list = $Omni::Config::tag_list;
|
18
|
-
my $att_list = $Omni::Config::att_list;
|
19
|
-
my $obj_list = $Omni::Config::obj_list;
|
20
|
-
|
21
|
-
# Temporary variables
|
22
|
-
my $tmp_content = undef;
|
23
|
-
my $tmp_bottom = undef;
|
24
|
-
my $tmp_top = undef;
|
25
|
-
my $tmp_left = undef;
|
26
|
-
my $tmp_right = undef;
|
27
|
-
my $tmp_language = undef;
|
28
|
-
my $tmp_alignment = undef;
|
29
|
-
my $tmp_spaceb = undef;
|
30
|
-
my @tmp_lines = ();
|
31
|
-
|
32
|
-
###
|
33
|
-
# A para object in Omnipage xml: a paragraph contains zero or many lines
|
34
|
-
#
|
35
|
-
# Do Hoang Nhat Huy, 09 Jan 2011
|
36
|
-
###
|
37
|
-
# Initialization
|
38
|
-
sub new
|
39
|
-
{
|
40
|
-
my ($class) = @_;
|
41
|
-
|
42
|
-
# Lines: a paragraph can have multiple lines
|
43
|
-
my @lines = ();
|
44
|
-
|
45
|
-
# Class members
|
46
|
-
my $self = { '_self' => $obj_list->{ 'OMNIPARA' },
|
47
|
-
'_raw' => undef,
|
48
|
-
'_content' => undef,
|
49
|
-
'_bottom' => undef,
|
50
|
-
'_top' => undef,
|
51
|
-
'_left' => undef,
|
52
|
-
'_right' => undef,
|
53
|
-
'_language' => undef,
|
54
|
-
'_alignment' => undef,
|
55
|
-
'_spaceb' => undef,
|
56
|
-
'_lines' => \@lines };
|
57
|
-
|
58
|
-
bless $self, $class;
|
59
|
-
return $self;
|
60
|
-
}
|
61
|
-
|
62
|
-
#
|
63
|
-
sub set_raw
|
64
|
-
{
|
65
|
-
my ($self, $raw) = @_;
|
66
|
-
|
67
|
-
# Save the raw xml <para> ... </para>
|
68
|
-
$self->{ '_raw' } = $raw;
|
69
|
-
|
70
|
-
# Parse the raw string
|
71
|
-
my $twig_roots = { $tag_list->{ 'PARA' } => 1 };
|
72
|
-
my $twig_handlers = { $tag_list->{ 'PARA' } => \&parse};
|
73
|
-
|
74
|
-
# XML::Twig
|
75
|
-
my $twig = new XML::Twig( twig_roots => $twig_roots,
|
76
|
-
twig_handlers => $twig_handlers,
|
77
|
-
pretty_print => 'indented' );
|
78
|
-
|
79
|
-
# Start the XML parsing
|
80
|
-
$twig->parse($raw);
|
81
|
-
$twig->purge;
|
82
|
-
|
83
|
-
# Copy information from temporary variables to class members
|
84
|
-
$self->{ '_bottom' } = $tmp_bottom;
|
85
|
-
$self->{ '_top' } = $tmp_top;
|
86
|
-
$self->{ '_left' } = $tmp_left;
|
87
|
-
$self->{ '_right' } = $tmp_right;
|
88
|
-
$self->{ '_language' } = $tmp_language;
|
89
|
-
$self->{ '_alignment' } = $tmp_alignment;
|
90
|
-
$self->{ '_spaceb' } = $tmp_spaceb;
|
91
|
-
|
92
|
-
# Copy all lines
|
93
|
-
@{$self->{ '_lines' } } = @tmp_lines;
|
94
|
-
|
95
|
-
# Copy content
|
96
|
-
$self->{ '_content' } = $tmp_content;
|
97
|
-
}
|
98
|
-
|
99
|
-
sub get_raw
|
100
|
-
{
|
101
|
-
my ($self) = @_;
|
102
|
-
return $self->{ '_raw' };
|
103
|
-
}
|
104
|
-
|
105
|
-
sub parse
|
106
|
-
{
|
107
|
-
my ($twig, $node) = @_;
|
108
|
-
|
109
|
-
# At first, content is blank
|
110
|
-
$tmp_content = "";
|
111
|
-
# because there's no line
|
112
|
-
@tmp_lines = ();
|
113
|
-
|
114
|
-
# Get <para> node attributes
|
115
|
-
$tmp_bottom = GetNodeAttr($node, $att_list->{ 'BOTTOM' });
|
116
|
-
$tmp_top = GetNodeAttr($node, $att_list->{ 'TOP' });
|
117
|
-
$tmp_left = GetNodeAttr($node, $att_list->{ 'LEFT' });
|
118
|
-
$tmp_right = GetNodeAttr($node, $att_list->{ 'RIGHT' });
|
119
|
-
$tmp_language = GetNodeAttr($node, $att_list->{ 'LANGUAGE' });
|
120
|
-
$tmp_alignment = GetNodeAttr($node, $att_list->{ 'ALIGN' });
|
121
|
-
$tmp_spaceb = GetNodeAttr($node, $att_list->{ 'SPACEB' });
|
122
|
-
|
123
|
-
# Check if there's any bullet
|
124
|
-
my $bullet = $node->first_child( $tag_list->{ 'BULLET' } );
|
125
|
-
my $has_bullet = (defined $bullet) ? 1 : 0;
|
126
|
-
|
127
|
-
# Check if there's any line
|
128
|
-
my @all_lines = $node->descendants( $tag_list->{ 'LINE' } );
|
129
|
-
foreach my $ln (@all_lines)
|
130
|
-
{
|
131
|
-
my $line = new Omni::Omniline();
|
132
|
-
|
133
|
-
# Set raw content
|
134
|
-
$line->set_raw($ln->sprint());
|
135
|
-
|
136
|
-
# Set bullet if needed
|
137
|
-
if ($has_bullet == 1) { $line->set_bullet('true'); }
|
138
|
-
|
139
|
-
# Update line list
|
140
|
-
push @tmp_lines, $line;
|
141
|
-
|
142
|
-
# Update content
|
143
|
-
$tmp_content = $tmp_content . $line->get_content() . "\n";
|
144
|
-
}
|
145
|
-
}
|
146
|
-
|
147
|
-
sub get_name
|
148
|
-
{
|
149
|
-
my ($self) = @_;
|
150
|
-
return $self->{ '_self' };
|
151
|
-
}
|
152
|
-
|
153
|
-
sub get_objs_ref
|
154
|
-
{
|
155
|
-
my ($self) = @_;
|
156
|
-
return $self->{ '_lines' };
|
157
|
-
}
|
158
|
-
|
159
|
-
sub get_content
|
160
|
-
{
|
161
|
-
my ($self) = @_;
|
162
|
-
return $self->{ '_content' };
|
163
|
-
}
|
164
|
-
|
165
|
-
sub get_bottom_pos
|
166
|
-
{
|
167
|
-
my ($self) = @_;
|
168
|
-
return $self->{ '_bottom' };
|
169
|
-
}
|
170
|
-
|
171
|
-
sub get_top_pos
|
172
|
-
{
|
173
|
-
my ($self) = @_;
|
174
|
-
return $self->{ '_top' };
|
175
|
-
}
|
176
|
-
|
177
|
-
sub get_left_pos
|
178
|
-
{
|
179
|
-
my ($self) = @_;
|
180
|
-
return $self->{ '_left' };
|
181
|
-
}
|
182
|
-
|
183
|
-
sub get_right_pos
|
184
|
-
{
|
185
|
-
my ($self) = @_;
|
186
|
-
return $self->{ '_right' };
|
187
|
-
}
|
188
|
-
|
189
|
-
sub get_language
|
190
|
-
{
|
191
|
-
my ($self) = @_;
|
192
|
-
return $self->{ '_language' };
|
193
|
-
}
|
194
|
-
|
195
|
-
sub get_alignment
|
196
|
-
{
|
197
|
-
my ($self) = @_;
|
198
|
-
return $self->{ '_alignment' };
|
199
|
-
}
|
200
|
-
|
201
|
-
sub get_space_before
|
202
|
-
{
|
203
|
-
my ($self) = @_;
|
204
|
-
return $self->{ '_spaceb' };
|
205
|
-
}
|
206
|
-
|
207
|
-
# Support functions
|
208
|
-
sub GetNodeAttr
|
209
|
-
{
|
210
|
-
my ($node, $attr) = @_;
|
211
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
212
|
-
}
|
213
|
-
|
214
|
-
sub SetNodeAttr
|
215
|
-
{
|
216
|
-
my ($node, $attr, $value) = @_;
|
217
|
-
$node->set_att($attr, $value);
|
218
|
-
}
|
219
|
-
|
220
|
-
sub GetNodeText
|
221
|
-
{
|
222
|
-
my ($node) = @_;
|
223
|
-
return $node->text;
|
224
|
-
}
|
225
|
-
|
226
|
-
sub SetNodeText
|
227
|
-
{
|
228
|
-
my ($node, $value) = @_;
|
229
|
-
$node->set_text($value);
|
230
|
-
}
|
231
|
-
|
232
|
-
1;
|
data/parscit/lib/Omni/Omnirun.pm
DELETED
@@ -1,303 +0,0 @@
|
|
1
|
-
package Omni::Omnirun;
|
2
|
-
|
3
|
-
# Configuration
|
4
|
-
use strict;
|
5
|
-
|
6
|
-
# Local libraries
|
7
|
-
use Omni::Config;
|
8
|
-
use Omni::Omniword;
|
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_font_face = undef;
|
22
|
-
my $tmp_font_family = undef;
|
23
|
-
my $tmp_font_pitch = undef;
|
24
|
-
my $tmp_font_size = undef;
|
25
|
-
my $tmp_spacing = undef;
|
26
|
-
my $tmp_su_script = undef; # sub-script or super-script
|
27
|
-
my $tmp_underline = undef;
|
28
|
-
my $tmp_bold = undef;
|
29
|
-
my $tmp_italic = undef;
|
30
|
-
my @tmp_words = ();
|
31
|
-
|
32
|
-
###
|
33
|
-
# A run object in Omnipage xml: a run contains zero or many words
|
34
|
-
#
|
35
|
-
# Do Hoang Nhat Huy, 07 Jan 2011
|
36
|
-
###
|
37
|
-
# Initialization
|
38
|
-
sub new
|
39
|
-
{
|
40
|
-
my ($class) = @_;
|
41
|
-
|
42
|
-
# Words: a run can have multiple words
|
43
|
-
my @words = ();
|
44
|
-
|
45
|
-
# Class members
|
46
|
-
my $self = { '_self' => $obj_list->{ 'OMNIRUN' },
|
47
|
-
'_raw' => undef,
|
48
|
-
'_content' => undef,
|
49
|
-
'_font_face' => undef,
|
50
|
-
'_font_family' => undef,
|
51
|
-
'_font_pitch' => undef,
|
52
|
-
'_font_size' => undef,
|
53
|
-
'_spacing' => undef,
|
54
|
-
'_su_script' => undef, # sub-script or super-script
|
55
|
-
'_underline' => undef,
|
56
|
-
'_bold' => undef,
|
57
|
-
'_italic' => undef,
|
58
|
-
'_words' => \@words };
|
59
|
-
|
60
|
-
bless $self, $class;
|
61
|
-
return $self;
|
62
|
-
}
|
63
|
-
|
64
|
-
#
|
65
|
-
sub set_raw
|
66
|
-
{
|
67
|
-
my ($self, $raw) = @_;
|
68
|
-
|
69
|
-
# Save the raw xml <run> ... </run>
|
70
|
-
$self->{ '_raw' } = $raw;
|
71
|
-
|
72
|
-
# Parse the raw string
|
73
|
-
my $twig_roots = { $tag_list->{ 'RUN' } => 1 };
|
74
|
-
my $twig_handlers = { $tag_list->{ 'RUN' } => \&parse};
|
75
|
-
|
76
|
-
# XML::Twig
|
77
|
-
my $twig= new XML::Twig( twig_roots => $twig_roots,
|
78
|
-
twig_handlers => $twig_handlers,
|
79
|
-
pretty_print => 'indented' );
|
80
|
-
|
81
|
-
# Start the XML parsing
|
82
|
-
$twig->parse($raw);
|
83
|
-
$twig->purge;
|
84
|
-
|
85
|
-
# Copy information from temporary variables to class members
|
86
|
-
$self->{ '_font_face' } = $tmp_font_face;
|
87
|
-
$self->{ '_font_family' } = $tmp_font_family;
|
88
|
-
$self->{ '_font_pitch' } = $tmp_font_pitch;
|
89
|
-
$self->{ '_font_size' } = $tmp_font_size;
|
90
|
-
$self->{ '_spacing' } = $tmp_spacing;
|
91
|
-
$self->{ '_su_script' } = $tmp_su_script;
|
92
|
-
$self->{ '_underline' } = $tmp_underline;
|
93
|
-
$self->{ '_bold' } = $tmp_bold;
|
94
|
-
$self->{ '_italic' } = $tmp_italic;
|
95
|
-
|
96
|
-
# Copy all words
|
97
|
-
@{ $self->{ '_words' } } = @tmp_words;
|
98
|
-
|
99
|
-
# Copy content
|
100
|
-
$self->{ '_content' } = $tmp_content;
|
101
|
-
}
|
102
|
-
|
103
|
-
sub get_raw
|
104
|
-
{
|
105
|
-
my ($self) = @_;
|
106
|
-
return $self->{ '_raw' };
|
107
|
-
}
|
108
|
-
|
109
|
-
sub parse
|
110
|
-
{
|
111
|
-
my ($twig, $node) = @_;
|
112
|
-
|
113
|
-
# Get <run> node attributes
|
114
|
-
$tmp_font_face = GetNodeAttr($node, $att_list->{ 'FONTFACE' });
|
115
|
-
$tmp_font_family = GetNodeAttr($node, $att_list->{ 'FONTFAMILY' });
|
116
|
-
$tmp_font_pitch = GetNodeAttr($node, $att_list->{ 'FONTPITCH' });
|
117
|
-
$tmp_font_size = GetNodeAttr($node, $att_list->{ 'FONTSIZE' });
|
118
|
-
$tmp_spacing = GetNodeAttr($node, $att_list->{ 'SPACING' });
|
119
|
-
$tmp_su_script = GetNodeAttr($node, $att_list->{ 'SUSCRIPT' }); # sub-script or super-script
|
120
|
-
$tmp_underline = GetNodeAttr($node, $att_list->{ 'UNDERLINE' });
|
121
|
-
$tmp_bold = GetNodeAttr($node, $att_list->{ 'BOLD' });
|
122
|
-
$tmp_italic = GetNodeAttr($node, $att_list->{ 'ITALIC' });
|
123
|
-
|
124
|
-
# At first, content is blank
|
125
|
-
$tmp_content = "";
|
126
|
-
# because there's no word
|
127
|
-
@tmp_words = ();
|
128
|
-
|
129
|
-
# Check if there's any child
|
130
|
-
my $child = $node->first_child();
|
131
|
-
|
132
|
-
# Has some child
|
133
|
-
# #PCDATA$ is the returned path from XML::Twig if $child is data content
|
134
|
-
if ((defined $child) && ($child->path() =~ m/#PCDATA$/))
|
135
|
-
{
|
136
|
-
my $content = undef;
|
137
|
-
$content = GetNodeText($node);
|
138
|
-
$content =~ s/^\s+|\s+$//g;
|
139
|
-
|
140
|
-
# Save the content
|
141
|
-
$tmp_content = $tmp_content . $content;
|
142
|
-
}
|
143
|
-
else
|
144
|
-
{
|
145
|
-
# Some type of separator
|
146
|
-
my $space_tag = $tag_list->{ 'SPACE' };
|
147
|
-
my $tab_tag = $tag_list->{ 'TAB' };
|
148
|
-
my $newline_tag = $tag_list->{ 'NEWLINE' };
|
149
|
-
my $word_tag = $tag_list->{ 'WORD' };
|
150
|
-
|
151
|
-
# Get every word in the <run> together with <space> and <tab> ...
|
152
|
-
my $obj = $node->first_child();
|
153
|
-
while (defined $obj)
|
154
|
-
{
|
155
|
-
my $xpath = $obj->path();
|
156
|
-
|
157
|
-
# if this child is <wd>
|
158
|
-
if ($xpath =~ m/\/$word_tag$/)
|
159
|
-
{
|
160
|
-
my $word = new Omni::Omniword();
|
161
|
-
|
162
|
-
# Set raw content
|
163
|
-
$word->set_raw($obj->sprint);
|
164
|
-
|
165
|
-
# Update word list
|
166
|
-
push @tmp_words, $word;
|
167
|
-
|
168
|
-
# Update content
|
169
|
-
$tmp_content = $tmp_content . $word->get_content;
|
170
|
-
}
|
171
|
-
# if this child is <space>
|
172
|
-
elsif ($xpath =~ m/\/$space_tag$/)
|
173
|
-
{
|
174
|
-
$tmp_content = $tmp_content . " ";
|
175
|
-
}
|
176
|
-
# if this child is <tab>
|
177
|
-
elsif ($xpath =~ m/\/$tab_tag$/)
|
178
|
-
{
|
179
|
-
$tmp_content = $tmp_content . "\t";
|
180
|
-
}
|
181
|
-
# if this child is <nl>
|
182
|
-
#elsif ($xpath =~ m/\/$newline_tag$/)
|
183
|
-
#{
|
184
|
-
# $tmp_content = $tmp_content . "\n";
|
185
|
-
#}
|
186
|
-
|
187
|
-
# Little brother
|
188
|
-
if ($obj->is_last_child)
|
189
|
-
{
|
190
|
-
last;
|
191
|
-
}
|
192
|
-
else
|
193
|
-
{
|
194
|
-
$obj = $obj->next_sibling();
|
195
|
-
}
|
196
|
-
}
|
197
|
-
}
|
198
|
-
}
|
199
|
-
|
200
|
-
sub add_word
|
201
|
-
{
|
202
|
-
my ($self, $word) = @_;
|
203
|
-
push @{ $self->{ '_words' } }, $word;
|
204
|
-
}
|
205
|
-
|
206
|
-
sub get_name
|
207
|
-
{
|
208
|
-
my ($self) = @_;
|
209
|
-
return $self->{ '_self' };
|
210
|
-
}
|
211
|
-
|
212
|
-
sub get_objs_ref
|
213
|
-
{
|
214
|
-
my ($self) = @_;
|
215
|
-
return $self->{ '_words' };
|
216
|
-
}
|
217
|
-
|
218
|
-
sub get_content
|
219
|
-
{
|
220
|
-
my ($self) = @_;
|
221
|
-
return $self->{ '_content' };
|
222
|
-
}
|
223
|
-
|
224
|
-
sub get_font_face
|
225
|
-
{
|
226
|
-
my ($self) = @_;
|
227
|
-
return $self->{ '_font_face' };
|
228
|
-
}
|
229
|
-
|
230
|
-
sub get_font_family
|
231
|
-
{
|
232
|
-
my ($self) = @_;
|
233
|
-
return $self->{ '_font_family' };
|
234
|
-
}
|
235
|
-
|
236
|
-
sub get_font_pitch
|
237
|
-
{
|
238
|
-
my ($self) = @_;
|
239
|
-
return $self->{ '_font_pitch' };
|
240
|
-
}
|
241
|
-
|
242
|
-
sub get_font_size
|
243
|
-
{
|
244
|
-
my ($self) = @_;
|
245
|
-
return $self->{ '_font_size' };
|
246
|
-
}
|
247
|
-
|
248
|
-
sub get_spacing
|
249
|
-
{
|
250
|
-
my ($self) = @_;
|
251
|
-
return $self->{ '_spacing' };
|
252
|
-
}
|
253
|
-
|
254
|
-
sub get_suscript
|
255
|
-
{
|
256
|
-
my ($self) = @_;
|
257
|
-
return $self->{ '_su_script' };
|
258
|
-
}
|
259
|
-
|
260
|
-
sub get_underline
|
261
|
-
{
|
262
|
-
my ($self) = @_;
|
263
|
-
return $self->{ '_underline' };
|
264
|
-
}
|
265
|
-
|
266
|
-
sub get_bold
|
267
|
-
{
|
268
|
-
my ($self) = @_;
|
269
|
-
return $self->{ '_bold' };
|
270
|
-
}
|
271
|
-
|
272
|
-
sub get_italic
|
273
|
-
{
|
274
|
-
my ($self) = @_;
|
275
|
-
return $self->{ '_italic' };
|
276
|
-
}
|
277
|
-
|
278
|
-
# Support functions
|
279
|
-
sub GetNodeAttr
|
280
|
-
{
|
281
|
-
my ($node, $attr) = @_;
|
282
|
-
return ($node->att($attr) ? $node->att($attr) : "");
|
283
|
-
}
|
284
|
-
|
285
|
-
sub SetNodeAttr
|
286
|
-
{
|
287
|
-
my ($node, $attr, $value) = @_;
|
288
|
-
$node->set_att($attr, $value);
|
289
|
-
}
|
290
|
-
|
291
|
-
sub GetNodeText
|
292
|
-
{
|
293
|
-
my ($node) = @_;
|
294
|
-
return $node->text;
|
295
|
-
}
|
296
|
-
|
297
|
-
sub SetNodeText
|
298
|
-
{
|
299
|
-
my ($node, $value) = @_;
|
300
|
-
$node->set_text($value);
|
301
|
-
}
|
302
|
-
|
303
|
-
1;
|