line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Treex::Tool::Segment::RuleBased; |
2
|
|
|
|
|
|
|
$Treex::Tool::Segment::RuleBased::VERSION = '0.13095'; |
3
|
1
|
|
|
1
|
|
13842
|
use strict; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
33
|
|
4
|
1
|
|
|
1
|
|
3
|
use warnings; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
20
|
|
5
|
1
|
|
|
1
|
|
473
|
use utf8; |
|
1
|
|
|
|
|
8
|
|
|
1
|
|
|
|
|
3
|
|
6
|
1
|
|
|
1
|
|
180
|
use Moose; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
use Treex::Core::Common; |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
has use_paragraphs => ( |
10
|
|
|
|
|
|
|
is => 'ro', |
11
|
|
|
|
|
|
|
isa => 'Bool', |
12
|
|
|
|
|
|
|
default => 1, |
13
|
|
|
|
|
|
|
documentation => |
14
|
|
|
|
|
|
|
'Should paragraph boundaries be preserved as sentence boundaries?' |
15
|
|
|
|
|
|
|
. ' Paragraph boundary is defined as two or more consecutive newlines.', |
16
|
|
|
|
|
|
|
); |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
has use_lines => ( |
19
|
|
|
|
|
|
|
is => 'ro', |
20
|
|
|
|
|
|
|
isa => 'Bool', |
21
|
|
|
|
|
|
|
default => 0, |
22
|
|
|
|
|
|
|
documentation => |
23
|
|
|
|
|
|
|
'Should newlines in the text be preserved as sentence boundaries?' |
24
|
|
|
|
|
|
|
. '(But if you want to detect sentence boundaries just based on newlines' |
25
|
|
|
|
|
|
|
. ' and nothing else, use rather W2A::SegmentOnNewlines.)', |
26
|
|
|
|
|
|
|
); |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
has limit_words => ( |
29
|
|
|
|
|
|
|
is => 'ro', |
30
|
|
|
|
|
|
|
isa => 'Int', |
31
|
|
|
|
|
|
|
default => 250, |
32
|
|
|
|
|
|
|
documentation => |
33
|
|
|
|
|
|
|
'Should very long segments (longer than the given number of words) be split?' |
34
|
|
|
|
|
|
|
. 'The number of words is only approximate; detected by counting whitespace only,' |
35
|
|
|
|
|
|
|
. 'not by full tokenization. Set to zero to disable this function completely.', |
36
|
|
|
|
|
|
|
); |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
has detect_lists => ( |
39
|
|
|
|
|
|
|
is => 'ro', |
40
|
|
|
|
|
|
|
isa => 'Int', |
41
|
|
|
|
|
|
|
default => 100, |
42
|
|
|
|
|
|
|
documentation => |
43
|
|
|
|
|
|
|
'Minimum (approx.) number of words to toggle list detection, 0 = never, 1 = always.' |
44
|
|
|
|
|
|
|
); |
45
|
|
|
|
|
|
|
|
46
|
|
|
|
|
|
|
# Tokens that usually do not end a sentence even if they are followed by a period and a capital letter: |
47
|
|
|
|
|
|
|
# * single uppercase letters serve usually as first name initials |
48
|
|
|
|
|
|
|
# * in langauge-specific descendants consider adding |
49
|
|
|
|
|
|
|
# * period-ending items that never indicate sentence breaks |
50
|
|
|
|
|
|
|
# * titles before names of persons etc. |
51
|
|
|
|
|
|
|
# |
52
|
|
|
|
|
|
|
# Note, that we cannot write |
53
|
|
|
|
|
|
|
# sub get_unbreakers { return qr{...}; } |
54
|
|
|
|
|
|
|
# because we want the regex to be compiled just once, not on every method call. |
55
|
|
|
|
|
|
|
my $UNBREAKERS = qr{\p{Upper}}; |
56
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
sub unbreakers { |
58
|
|
|
|
|
|
|
return $UNBREAKERS; |
59
|
|
|
|
|
|
|
} |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
# Characters that can appear after period (or other end-sentence symbol) |
62
|
|
|
|
|
|
|
sub closings { |
63
|
|
|
|
|
|
|
return '"”»)'; |
64
|
|
|
|
|
|
|
} |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
# Characters that can appear before the first word of a sentence |
67
|
|
|
|
|
|
|
sub openings { |
68
|
|
|
|
|
|
|
return '"“«('; |
69
|
|
|
|
|
|
|
} |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
# Contextual rules for "un-breaking" (to be overridden) |
72
|
|
|
|
|
|
|
sub apply_contextual_rules { |
73
|
|
|
|
|
|
|
my ($self, $text) = @_; |
74
|
|
|
|
|
|
|
return $text; |
75
|
|
|
|
|
|
|
} |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
sub get_segments { |
78
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
# Pre-processing |
81
|
|
|
|
|
|
|
$text = $self->apply_contextual_rules($text); |
82
|
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
my $unbreakers = $self->unbreakers; |
84
|
|
|
|
|
|
|
$text =~ s/\b($unbreakers)\./$1<<>>/g; |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
# two newlines usually separate paragraphs |
87
|
|
|
|
|
|
|
if ( $self->use_paragraphs ) { |
88
|
|
|
|
|
|
|
$text =~ s/([^.!?])\n\n+/$1<<>>/gsm; |
89
|
|
|
|
|
|
|
} |
90
|
|
|
|
|
|
|
|
91
|
|
|
|
|
|
|
if ( $self->use_lines ) { |
92
|
|
|
|
|
|
|
$text =~ s/\n/<<>>/gsm; |
93
|
|
|
|
|
|
|
} |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
# Normalize whitespaces |
96
|
|
|
|
|
|
|
$text =~ s/\s+/ /gsm; |
97
|
|
|
|
|
|
|
|
98
|
|
|
|
|
|
|
# This is the main work |
99
|
|
|
|
|
|
|
$text = $self->split_at_terminal_punctuation($text); |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
# Post-processing |
102
|
|
|
|
|
|
|
$text =~ s/<<>>/\n/gsmx; |
103
|
|
|
|
|
|
|
$text =~ s/<<>>/./gsxm; |
104
|
|
|
|
|
|
|
$text =~ s/\s+$//gsxm; |
105
|
|
|
|
|
|
|
$text =~ s/^\s+//gsxm; |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
# try to separate various list items (e.g. TV programmes, calendars) |
108
|
|
|
|
|
|
|
my @segs = map { $self->split_at_list_items($_) } split /\n/, $text; |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
# handle segments that are too long |
111
|
|
|
|
|
|
|
return map { $self->segment_too_long($_) ? $self->handle_long_segment($_) : $_ } @segs; |
112
|
|
|
|
|
|
|
} |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
sub split_at_terminal_punctuation { |
115
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
116
|
|
|
|
|
|
|
my ( $openings, $closings ) = ( $self->openings, $self->closings ); |
117
|
|
|
|
|
|
|
$text =~ s{ |
118
|
|
|
|
|
|
|
([.?!]) # $1 = end-sentence punctuation |
119
|
|
|
|
|
|
|
([$closings]?) # $2 = optional closing quote/bracket |
120
|
|
|
|
|
|
|
\s # space |
121
|
|
|
|
|
|
|
([$openings]?\p{Upper}) # $3 = uppercase letter (optionally preceded by opening quote) |
122
|
|
|
|
|
|
|
}{$1$2\n$3}gsxm; |
123
|
|
|
|
|
|
|
return $text; |
124
|
|
|
|
|
|
|
} |
125
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
sub handle_long_segment { |
127
|
|
|
|
|
|
|
my ( $self, $seg ) = @_; |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
# split at some other dividing punctuation characters (poems, unending speech) |
130
|
|
|
|
|
|
|
my @split = map { $self->segment_too_long($_) ? $self->split_at_dividing_punctuation($_) : $_ } $seg; |
131
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
# split at any punctuation |
133
|
|
|
|
|
|
|
@split = map { $self->segment_too_long($_) ? $self->split_at_any_punctuation($_) : $_ } @split; |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
# split hard if still too long |
136
|
|
|
|
|
|
|
return map { $self->segment_too_long($_) ? $self->split_hard($_) : $_ } @split; |
137
|
|
|
|
|
|
|
} |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
# Return 1 if the segment is too long |
140
|
|
|
|
|
|
|
sub segment_too_long { |
141
|
|
|
|
|
|
|
my ( $self, $seg ) = @_; |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
# skip everything if the limit is infinity |
144
|
|
|
|
|
|
|
return 0 if ( $self->limit_words == 0 ); |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
# return 1 if the number of space-separated segments exceeds the limit |
147
|
|
|
|
|
|
|
my $wc = () = $seg =~ m/\s+/g; |
148
|
|
|
|
|
|
|
return 1 if ( $wc >= $self->limit_words ); |
149
|
|
|
|
|
|
|
return 0; |
150
|
|
|
|
|
|
|
} |
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
# "Non-final" punctuation that could divide segments (NB: single dot excluded due to abbreviations) |
153
|
|
|
|
|
|
|
my $DIV_PUNCT = qr{(!|\.\.+|\?|\*|[–—-](\s*[–—-])+|;)}; |
154
|
|
|
|
|
|
|
|
155
|
|
|
|
|
|
|
sub split_at_dividing_punctuation { |
156
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
157
|
|
|
|
|
|
|
|
158
|
|
|
|
|
|
|
my $closings = $self->closings; |
159
|
|
|
|
|
|
|
$text =~ s/($DIV_PUNCT\s*[$closings]?,?)/$1\n/g; |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
return split /\n/, $self->_join_too_short_segments($text); |
162
|
|
|
|
|
|
|
} |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
# Universal list types (currently only semicolon-separated lists, to be overridden in language-specific blocks) |
165
|
|
|
|
|
|
|
my $LIST_TYPES = [ |
166
|
|
|
|
|
|
|
{ |
167
|
|
|
|
|
|
|
name => ';', # a label for the list type (just for debugging) |
168
|
|
|
|
|
|
|
sep => ';\h+', # separator regexp |
169
|
|
|
|
|
|
|
sel_sep => undef, # separator regexp used only for the selection of this list (sep used if not set) |
170
|
|
|
|
|
|
|
type => 'e', # type of separator (ending: e / staring: s) |
171
|
|
|
|
|
|
|
max => 400, # maximum average list-item length (overrides the default) |
172
|
|
|
|
|
|
|
min => 30, # minimum average list-item length (overrides the default) |
173
|
|
|
|
|
|
|
# negative pre-context, not used if not set (here: skip semicolons separating just numbers) |
174
|
|
|
|
|
|
|
neg_pre => '[0-9]\h*(?=;\h*[0-9]+(?:[^\.0-9]|\.[0-9]|$))', |
175
|
|
|
|
|
|
|
}, |
176
|
|
|
|
|
|
|
]; |
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
# Language-specific blocks should override this method and provide usual list types for the given language |
179
|
|
|
|
|
|
|
sub list_types { |
180
|
|
|
|
|
|
|
return @{$LIST_TYPES}; |
181
|
|
|
|
|
|
|
} |
182
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
my $MAX_AVG_ITEM_LEN = 400; # default maximum average list item length, in characters |
184
|
|
|
|
|
|
|
my $MIN_AVG_ITEM_LEN = 30; # default minimum average list item length, in characters |
185
|
|
|
|
|
|
|
my $MIN_LIST_ITEMS = 3; # minimum number of items in a list |
186
|
|
|
|
|
|
|
my $PRIORITY = 2.5; # multiple of list items a lower-rank list type must have over a higher-rank type |
187
|
|
|
|
|
|
|
|
188
|
|
|
|
|
|
|
sub split_at_list_items { |
189
|
|
|
|
|
|
|
|
190
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
191
|
|
|
|
|
|
|
|
192
|
|
|
|
|
|
|
# skip this if list detection is turned off |
193
|
|
|
|
|
|
|
return $text if ( $self->detect_lists == 0 ); |
194
|
|
|
|
|
|
|
|
195
|
|
|
|
|
|
|
# skip too short lines |
196
|
|
|
|
|
|
|
my $wc = () = $text =~ m/\s+/g; |
197
|
|
|
|
|
|
|
return $text if ( $self->detect_lists > $wc ); |
198
|
|
|
|
|
|
|
|
199
|
|
|
|
|
|
|
my @list_types = $self->list_types; |
200
|
|
|
|
|
|
|
my $sel_list_type; |
201
|
|
|
|
|
|
|
my $sel_len; |
202
|
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
# find out which list type is the best for the given text |
204
|
|
|
|
|
|
|
for ( my $i = 0; $i < @list_types; ++$i ) { |
205
|
|
|
|
|
|
|
|
206
|
|
|
|
|
|
|
my $cur_list_type = $list_types[$i]; |
207
|
|
|
|
|
|
|
my $sep = $cur_list_type->{sel_sep} || $cur_list_type->{sep}; |
208
|
|
|
|
|
|
|
my $neg = $cur_list_type->{neg_pre}; |
209
|
|
|
|
|
|
|
my $min = $cur_list_type->{min} || $MIN_AVG_ITEM_LEN; |
210
|
|
|
|
|
|
|
my $max = $cur_list_type->{max} || $MAX_AVG_ITEM_LEN; |
211
|
|
|
|
|
|
|
|
212
|
|
|
|
|
|
|
my $items = () = $text =~ m/$sep/gi; |
213
|
|
|
|
|
|
|
|
214
|
|
|
|
|
|
|
# count number of items; exclude negative pre-context matches, if negative pre-context is specified |
215
|
|
|
|
|
|
|
my $false = 0; |
216
|
|
|
|
|
|
|
$false = () = $text =~ m/$neg(?=$sep)/gi if ($neg); |
217
|
|
|
|
|
|
|
$items -= $false; |
218
|
|
|
|
|
|
|
|
219
|
|
|
|
|
|
|
my $len = $items > 0 ? ( length($text) / $items ) : 'NaN'; |
220
|
|
|
|
|
|
|
|
221
|
|
|
|
|
|
|
# test if this type overrides the previously set one |
222
|
|
|
|
|
|
|
if ( $items >= $MIN_LIST_ITEMS && $len < $max && $len > $min && ( !$sel_len || $len * $PRIORITY < $sel_len ) ) { |
223
|
|
|
|
|
|
|
$sel_list_type = $cur_list_type; |
224
|
|
|
|
|
|
|
$sel_len = $len; |
225
|
|
|
|
|
|
|
} |
226
|
|
|
|
|
|
|
} |
227
|
|
|
|
|
|
|
|
228
|
|
|
|
|
|
|
# return if no list type found |
229
|
|
|
|
|
|
|
return $text if ( !$sel_list_type ); |
230
|
|
|
|
|
|
|
|
231
|
|
|
|
|
|
|
# list type detected, split by the given list type |
232
|
|
|
|
|
|
|
my $sep = $sel_list_type->{sep}; |
233
|
|
|
|
|
|
|
my $neg = $sel_list_type->{neg_pre}; |
234
|
|
|
|
|
|
|
my $name = $sel_list_type->{name}; |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
# protect negative pre-context, if any is specified |
237
|
|
|
|
|
|
|
$text =~ s/($neg)(?=$sep)/$1<<>>/gi if ($neg); |
238
|
|
|
|
|
|
|
|
239
|
|
|
|
|
|
|
# split at the given list type |
240
|
|
|
|
|
|
|
if ( $sel_list_type->{type} eq 'e' ) { |
241
|
|
|
|
|
|
|
$text =~ s/(?>>)($sep)/$1\n/gi; |
242
|
|
|
|
|
|
|
} |
243
|
|
|
|
|
|
|
else { |
244
|
|
|
|
|
|
|
$text =~ s/(?>>)($sep)/\n$1/gi; |
245
|
|
|
|
|
|
|
} |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
# remove negative pre-context protection |
248
|
|
|
|
|
|
|
$text =~ s/<<>>//g; |
249
|
|
|
|
|
|
|
|
250
|
|
|
|
|
|
|
# delete too short splits |
251
|
|
|
|
|
|
|
$text = $self->_join_too_short_segments($text); |
252
|
|
|
|
|
|
|
|
253
|
|
|
|
|
|
|
# return the split result |
254
|
|
|
|
|
|
|
return split /\n/, $text; |
255
|
|
|
|
|
|
|
} |
256
|
|
|
|
|
|
|
|
257
|
|
|
|
|
|
|
sub _join_too_short_segments { |
258
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
259
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
$text =~ s/^\n//; |
261
|
|
|
|
|
|
|
$text =~ s/\n$//; |
262
|
|
|
|
|
|
|
$text =~ s/\n(?=\h*(\S+(\h+\S+){0,2})?\h*(\n|$))/ /g; |
263
|
|
|
|
|
|
|
return $text; |
264
|
|
|
|
|
|
|
} |
265
|
|
|
|
|
|
|
|
266
|
|
|
|
|
|
|
sub split_at_any_punctuation { |
267
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
268
|
|
|
|
|
|
|
|
269
|
|
|
|
|
|
|
my $closings = $self->closings; |
270
|
|
|
|
|
|
|
|
271
|
|
|
|
|
|
|
# prefer punctuation followed by a letter |
272
|
|
|
|
|
|
|
$text =~ s/([,;!?–—-]+\s*[$closings]?)\s+(\p{Alpha})/$1\n$2/g; |
273
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
# delete too short splits |
275
|
|
|
|
|
|
|
$text = $self->_join_too_short_segments($text); |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
my @split = split /\n/, $text; |
278
|
|
|
|
|
|
|
|
279
|
|
|
|
|
|
|
# split at any punctuation if the text is still too long |
280
|
|
|
|
|
|
|
return map { |
281
|
|
|
|
|
|
|
$_ =~ s/([,;!?–—-]+\s*[$closings]?)/$1\n/g if ( $self->segment_too_long($_) ); |
282
|
|
|
|
|
|
|
split /\n/, $self->_join_too_short_segments($_) |
283
|
|
|
|
|
|
|
} @split; |
284
|
|
|
|
|
|
|
} |
285
|
|
|
|
|
|
|
|
286
|
|
|
|
|
|
|
sub split_hard { |
287
|
|
|
|
|
|
|
my ( $self, $text ) = @_; |
288
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
my @tokens = split /(\s+)/, $text; |
290
|
|
|
|
|
|
|
my @result; |
291
|
|
|
|
|
|
|
my $pos = 0; |
292
|
|
|
|
|
|
|
|
293
|
|
|
|
|
|
|
while ( $pos < @tokens ) { |
294
|
|
|
|
|
|
|
my $limit = $pos + $self->limit_words * 2 - 1; |
295
|
|
|
|
|
|
|
$limit = @tokens - 1 if ( $limit > @tokens - 1 ); |
296
|
|
|
|
|
|
|
push @result, join( '', @tokens[ $pos .. $limit ] ); |
297
|
|
|
|
|
|
|
$pos = $limit + 1; |
298
|
|
|
|
|
|
|
} |
299
|
|
|
|
|
|
|
return @result; |
300
|
|
|
|
|
|
|
} |
301
|
|
|
|
|
|
|
|
302
|
|
|
|
|
|
|
1; |
303
|
|
|
|
|
|
|
|
304
|
|
|
|
|
|
|
__END__ |