text_alignment 0.3.9 → 0.3.15
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/bin/align_annotations +45 -1
- data/lib/text_alignment/anchor_finder.rb +22 -25
- data/lib/text_alignment/text_alignment.rb +12 -10
- data/lib/text_alignment/version.rb +1 -1
- metadata +2 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 8252929e7b74251db493ad991332d0da8a92f35441e2a442d05b6fb29139d657
|
4
|
+
data.tar.gz: c9e41127fb231e4db2843696c1029b35711449c5bf5104158da301d03b817131
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 1fc8da7324d71cf25edbec9765ab512928323079472736ea4e294abb12dfafc87f55d71cb49c371470811775bd489d3c91cce4a787b99faa305f2f326dc80c77
|
7
|
+
data.tar.gz: f694c99216b59dd693a6acdfffc727fe74b5c189b4b9583b31fb7e6394319a3176de76237a142dac3770bcd1fbbc467d5d1e97a7225ed993a21246a66de8b2ec
|
data/bin/align_annotations
CHANGED
@@ -103,6 +103,35 @@ target_annotations = if source_annotations.class == Array
|
|
103
103
|
else
|
104
104
|
alignment = TextAlignment::TextAlignment.new(source_annotations[:text], target_text)
|
105
105
|
|
106
|
+
# verification
|
107
|
+
source_text = source_annotations[:text]
|
108
|
+
puts "=====BEGIN"
|
109
|
+
(0 ... source_text.length).each do |p|
|
110
|
+
t = alignment.transform_begin_position(p)
|
111
|
+
if t.nil?
|
112
|
+
print source_text[p]
|
113
|
+
else
|
114
|
+
print '.'
|
115
|
+
end
|
116
|
+
end
|
117
|
+
puts
|
118
|
+
puts "=====END"
|
119
|
+
|
120
|
+
puts "=====BEGIN"
|
121
|
+
(0 .. source_text.length).each do |p|
|
122
|
+
t = alignment.transform_end_position(p)
|
123
|
+
if t.nil?
|
124
|
+
print source_text[p]
|
125
|
+
else
|
126
|
+
print '.'
|
127
|
+
end
|
128
|
+
end
|
129
|
+
puts
|
130
|
+
puts "=====END"
|
131
|
+
|
132
|
+
pp alignment
|
133
|
+
|
134
|
+
exit
|
106
135
|
# alignment.block_alignments.each do |a|
|
107
136
|
# if a[:alignment].nil? || a[:alignment] == :empty
|
108
137
|
# # p [a[:source], a[:target]]
|
@@ -119,7 +148,22 @@ else
|
|
119
148
|
# end
|
120
149
|
# exit
|
121
150
|
|
151
|
+
# verification of source denotations
|
152
|
+
puts "[Invalid source denotations]"
|
153
|
+
source_annotations[:denotations] do |d|
|
154
|
+
p d unless d[:span][:begin] && d[:span][:end] && d[:span][:begin] < d[:span][:end] && d[:span][:begin] >= 0 && d[:span][:end] < source_text.length
|
155
|
+
end
|
156
|
+
puts "====="
|
157
|
+
puts
|
158
|
+
|
122
159
|
denotations = alignment.transform_hdenotations(source_annotations[:denotations])
|
160
|
+
puts "[Invalid transformation]"
|
161
|
+
denotations.each do |d|
|
162
|
+
p d unless d[:span][:begin] && d[:span][:end] && d[:span][:begin] < d[:span][:end] && d[:span][:begin] >= 0 && d[:span][:end] < target_text.length
|
163
|
+
end
|
164
|
+
puts "====="
|
165
|
+
puts
|
166
|
+
|
123
167
|
lost_annotations += alignment.lost_annotations if alignment.lost_annotations
|
124
168
|
|
125
169
|
source_annotations.merge({text:target_text, denotations:denotations})
|
@@ -159,7 +203,7 @@ if lost_annotations
|
|
159
203
|
warn "#{lost_annotations.length}"
|
160
204
|
end
|
161
205
|
|
162
|
-
puts target_annotations.to_json
|
206
|
+
#puts target_annotations.to_json
|
163
207
|
|
164
208
|
# denotations = anns1[:denotations]
|
165
209
|
|
@@ -3,9 +3,9 @@ require 'string-similarity'
|
|
3
3
|
|
4
4
|
module TextAlignment; end unless defined? TextAlignment
|
5
5
|
|
6
|
-
TextAlignment::SIZE_NGRAM =
|
7
|
-
TextAlignment::SIZE_WINDOW =
|
8
|
-
TextAlignment::TEXT_SIMILARITY_TRESHOLD = 0.
|
6
|
+
TextAlignment::SIZE_NGRAM = 10 unless defined? TextAlignment::SIZE_NGRAM
|
7
|
+
TextAlignment::SIZE_WINDOW = 30 unless defined? TextAlignment::SIZE_WINDOW
|
8
|
+
TextAlignment::TEXT_SIMILARITY_TRESHOLD = 0.8 unless defined? TextAlignment::TEXT_SIMILARITY_TRESHOLD
|
9
9
|
|
10
10
|
class TextAlignment::AnchorFinder
|
11
11
|
|
@@ -23,39 +23,35 @@ class TextAlignment::AnchorFinder
|
|
23
23
|
|
24
24
|
# current position in s1
|
25
25
|
@beg_s1 = 0
|
26
|
+
@end_s2_prev = 0
|
26
27
|
end
|
27
28
|
|
28
29
|
def get_next_anchor
|
29
30
|
# find the position of an anchor ngram in s1 and s2
|
30
|
-
@beg_s2 = nil
|
31
31
|
while @beg_s1 < (@s1.length - @size_ngram)
|
32
|
-
|
33
|
-
|
34
|
-
|
35
|
-
|
36
|
-
|
37
|
-
|
38
|
-
|
39
|
-
break unless @beg_s2.nil?
|
40
|
-
@beg_s1 += 1
|
41
|
-
end
|
32
|
+
anchor = @s1[@beg_s1, @size_ngram]
|
33
|
+
|
34
|
+
search_position = 0
|
35
|
+
# search_position = @end_s2_prev
|
36
|
+
while @beg_s2 = @s2.index(anchor, search_position)
|
37
|
+
# if both the begining points are sufficiantly close to the end points of the last match
|
38
|
+
break if @end_s1_prev && (@beg_s1 - @end_s1_prev < 5) && (@beg_s2 >= @end_s2_prev) && (@beg_s2 - @end_s2_prev < 5)
|
42
39
|
|
43
|
-
|
44
|
-
|
40
|
+
left_window_s1, left_window_s2 = get_left_windows
|
41
|
+
break if left_window_s1 && (text_similarity(left_window_s1, left_window_s2) > TextAlignment::TEXT_SIMILARITY_TRESHOLD)
|
45
42
|
|
46
|
-
|
47
|
-
|
43
|
+
right_window_s1, right_window_s2 = get_right_windows
|
44
|
+
break if right_window_s2 && (text_similarity(right_window_s1, right_window_s2) > TextAlignment::TEXT_SIMILARITY_TRESHOLD)
|
48
45
|
|
49
|
-
|
50
|
-
|
46
|
+
search_position = @beg_s2 + 1
|
47
|
+
end
|
51
48
|
|
52
|
-
|
53
|
-
break if right_window_s2 && text_similarity(right_window_s1, left_window_s2) > TextAlignment::TEXT_SIMILARITY_TRESHOLD
|
49
|
+
break unless @beg_s2.nil?
|
54
50
|
|
55
51
|
@beg_s1 += 1
|
56
52
|
end
|
57
53
|
|
58
|
-
return nil if @
|
54
|
+
return nil if @beg_s1 >= (@s1.length - @size_ngram)
|
59
55
|
|
60
56
|
# extend the block
|
61
57
|
b1 = @beg_s1
|
@@ -67,7 +63,7 @@ class TextAlignment::AnchorFinder
|
|
67
63
|
|
68
64
|
e1 = @beg_s1 + @size_ngram
|
69
65
|
e2 = @beg_s2 + @size_ngram
|
70
|
-
while @s1[e1] == @s2[e2]
|
66
|
+
while @s1[e1] && @s1[e1] == @s2[e2]
|
71
67
|
e1 += 1; e2 += 1
|
72
68
|
end
|
73
69
|
|
@@ -113,7 +109,7 @@ class TextAlignment::AnchorFinder
|
|
113
109
|
end
|
114
110
|
|
115
111
|
def get_right_windows
|
116
|
-
return if (@beg_s1 + @size_ngram
|
112
|
+
return if (@beg_s1 + @size_ngram > (@s1.length - @size_window)) || (@beg_s2 + @size_ngram > (@s2.length - @size_window))
|
117
113
|
|
118
114
|
window_s1 = ''
|
119
115
|
loc = @beg_s1 + @size_ngram
|
@@ -143,6 +139,7 @@ class TextAlignment::AnchorFinder
|
|
143
139
|
end
|
144
140
|
|
145
141
|
def text_similarity(str1, str2, ngram_order = 2)
|
142
|
+
return 0 if str1.nil? || str2.nil?
|
146
143
|
String::Similarity.cosine(str1, str2, ngram:ngram_order)
|
147
144
|
end
|
148
145
|
|
@@ -6,6 +6,8 @@ module TextAlignment; end unless defined? TextAlignment
|
|
6
6
|
|
7
7
|
TextAlignment::SIGNATURE_NGRAM = 7 unless defined? TextAlignment::SIGNATURE_NGRAM
|
8
8
|
TextAlignment::BUFFER_RATE = 0.1 unless defined? TextAlignment::BUFFER_RATE
|
9
|
+
TextAlignment::BUFFER_MIN = 10 unless defined? TextAlignment::BUFFER_MIN
|
10
|
+
|
9
11
|
|
10
12
|
class TextAlignment::TextAlignment
|
11
13
|
attr_reader :block_alignments
|
@@ -70,7 +72,7 @@ class TextAlignment::TextAlignment
|
|
70
72
|
@block_alignments << {source:{begin:0, end:e1}, target:{begin:0, end:e2}, alignment: :empty}
|
71
73
|
else
|
72
74
|
len_min = [_str1.length, _str2.length].min
|
73
|
-
len_buffer = (len_min * (1 + TextAlignment::BUFFER_RATE)).to_i
|
75
|
+
len_buffer = (len_min * (1 + TextAlignment::BUFFER_RATE)).to_i + TextAlignment::BUFFER_MIN
|
74
76
|
b1 = _str1.length < len_buffer ? 0 : e1 - len_buffer
|
75
77
|
b2 = _str2.length < len_buffer ? 0 : e2 - len_buffer
|
76
78
|
|
@@ -135,9 +137,11 @@ class TextAlignment::TextAlignment
|
|
135
137
|
@block_alignments << {source:{begin:b1, end:str1.length}, target:{begin:b2, end:str2.length}, alignment: :empty}
|
136
138
|
else
|
137
139
|
len_min = [_str1.length, _str2.length].min
|
138
|
-
len_buffer = (len_min * (1 + TextAlignment::BUFFER_RATE)).to_i
|
140
|
+
len_buffer = (len_min * (1 + TextAlignment::BUFFER_RATE)).to_i + TextAlignment::BUFFER_MIN
|
139
141
|
e1 = _str1.length < len_buffer ? str1.length : b1 + len_buffer
|
140
|
-
e2 = _str2.length < len_buffer ?
|
142
|
+
e2 = _str2.length < len_buffer ? str2.length : b2 + len_buffer
|
143
|
+
_str1 = str1[b1 ... e1]
|
144
|
+
_str2 = str2[b2 ... e2]
|
141
145
|
|
142
146
|
alignment = TextAlignment::MixedAlignment.new(_str1.downcase, _str2.downcase, mappings)
|
143
147
|
if alignment.similarity < 0.6
|
@@ -166,7 +170,8 @@ class TextAlignment::TextAlignment
|
|
166
170
|
if begin_position == block_alignment[:source][:begin]
|
167
171
|
block_alignment[:target][:begin]
|
168
172
|
else
|
169
|
-
raise "lost annotation"
|
173
|
+
# raise "lost annotation"
|
174
|
+
nil
|
170
175
|
end
|
171
176
|
else
|
172
177
|
block_alignment[:alignment].transform_begin_position(begin_position - block_alignment[:source][:begin]) + block_alignment[:target][:begin]
|
@@ -183,7 +188,8 @@ class TextAlignment::TextAlignment
|
|
183
188
|
if end_position == block_alignment[:source][:end]
|
184
189
|
block_alignment[:target][:end]
|
185
190
|
else
|
186
|
-
raise "lost annotation"
|
191
|
+
# raise "lost annotation"
|
192
|
+
nil
|
187
193
|
end
|
188
194
|
else
|
189
195
|
block_alignment[:alignment].transform_end_position(end_position - block_alignment[:source][:begin]) + block_alignment[:target][:begin]
|
@@ -199,13 +205,9 @@ class TextAlignment::TextAlignment
|
|
199
205
|
end
|
200
206
|
|
201
207
|
def transform_denotations!(denotations)
|
202
|
-
puts "hereherehere========"
|
203
208
|
return nil if denotations.nil?
|
204
209
|
@lost_annotations = []
|
205
210
|
|
206
|
-
pp denotations
|
207
|
-
puts "-----"
|
208
|
-
|
209
211
|
denotations.each do |d|
|
210
212
|
begin
|
211
213
|
d.begin = transform_begin_position(d.begin);
|
@@ -217,7 +219,7 @@ class TextAlignment::TextAlignment
|
|
217
219
|
end
|
218
220
|
end
|
219
221
|
|
220
|
-
|
222
|
+
@lost_annotations
|
221
223
|
end
|
222
224
|
|
223
225
|
def transform_hdenotations(hdenotations)
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: text_alignment
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.3.
|
4
|
+
version: 0.3.15
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Jin-Dong Kim
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2020-
|
11
|
+
date: 2020-09-17 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: ruby-dictionary
|