text_alignment 0.2.8 → 0.2.9
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/text_alignment/approximate_fit.rb +7 -4
- data/lib/text_alignment/find_divisions.rb +112 -12
- 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: 8c1c45ed630cdd60291606b59e1944f0b854a689cfa0d281ae8b8879bf01e806
|
4
|
+
data.tar.gz: 9b33688f3a08f9110f556b4357fbe598d42d1147bc06933323cae4df7187341f
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 856a8fca63f80be4cea7f6beff85dcf475d9237d68bc96728bcfcc030397f414637d2a8e32b139a6fdbccfd8603d327738ff7f3f59d80dc9e61e55a11a04bf20
|
7
|
+
data.tar.gz: 6055f50827354461f194a50da74259e8d97473be9b085634df270964bb815d96d9f916bbad2ef3961ddd845291a80e13b1680f1c42bafeab156b0d59fa3ba952
|
@@ -8,7 +8,7 @@ module TextAlignment
|
|
8
8
|
SIGNATURE_NGRAM = 5
|
9
9
|
MIN_LENGTH_FOR_APPROXIMATION = 50
|
10
10
|
BUFFER_RATE = 0.1
|
11
|
-
TEXT_SIMILARITY_TRESHOLD = 0.
|
11
|
+
TEXT_SIMILARITY_TRESHOLD = 0.7
|
12
12
|
end
|
13
13
|
|
14
14
|
class << TextAlignment
|
@@ -28,6 +28,7 @@ class << TextAlignment
|
|
28
28
|
signature_ngrams = ngram_shared.select{|g| ngram2.count(g) == 1}
|
29
29
|
return nil, nil if signature_ngrams.empty? #raise "no signature ngram"
|
30
30
|
|
31
|
+
cache = {}
|
31
32
|
fit_begin, fit_end = nil, nil
|
32
33
|
signature_ngrams.each do |signature_ngram|
|
33
34
|
loc_signature_ngram_in_str1 = str1.index(signature_ngram)
|
@@ -42,13 +43,15 @@ class << TextAlignment
|
|
42
43
|
fit_end = loc_signature_ngram_in_str2 + offset_end + (offset_end * TextAlignment::BUFFER_RATE).to_i
|
43
44
|
fit_end = str2.length if fit_end > str2.length
|
44
45
|
|
46
|
+
next if cache.has_key?("#{fit_begin}-#{fit_end}")
|
45
47
|
text_similarity = text_similarity(str1, str2[fit_begin ... fit_end])
|
48
|
+
cache["#{fit_begin}-#{fit_end}"] = text_similarity
|
49
|
+
|
46
50
|
break if text_similarity > TextAlignment::TEXT_SIMILARITY_TRESHOLD
|
47
51
|
fit_begin, fit_end = nil, nil
|
48
52
|
end
|
49
|
-
|
50
|
-
return nil, nil
|
51
|
-
return fit_begin, fit_end
|
53
|
+
return fit_begin, fit_end if fit_begin && fit_end && fit_begin < fit_end
|
54
|
+
return nil, nil
|
52
55
|
end
|
53
56
|
|
54
57
|
private
|
@@ -8,7 +8,7 @@ module TextAlignment; end unless defined? TextAlignment
|
|
8
8
|
# to assume that there is no bag representation to this method
|
9
9
|
|
10
10
|
module TextAlignment
|
11
|
-
TextAlignment::SIMILARITY_THRESHOLD = 0.
|
11
|
+
TextAlignment::SIMILARITY_THRESHOLD = 0.7
|
12
12
|
end
|
13
13
|
|
14
14
|
class << TextAlignment
|
@@ -28,12 +28,102 @@ class << TextAlignment
|
|
28
28
|
target.tr!(characters_from, characters_to)
|
29
29
|
sources.each{|source| source[:text].tr!(characters_from, characters_to)}
|
30
30
|
|
31
|
+
# to process smaller ones first
|
31
32
|
sources.sort!{|s1, s2| s1[:text].size <=> s2[:text].size}
|
32
33
|
|
33
34
|
TextAlignment._find_divisions(target, sources)
|
34
35
|
end
|
35
36
|
|
36
|
-
def _find_divisions(
|
37
|
+
def _find_divisions(_target, _sources)
|
38
|
+
indice = []
|
39
|
+
history = []
|
40
|
+
cache = {}
|
41
|
+
target = _target.dup
|
42
|
+
sources = _sources.dup
|
43
|
+
until target.strip.empty? || sources.empty?
|
44
|
+
mode, cmp = nil, nil
|
45
|
+
candidates = []
|
46
|
+
sources.each_with_index do |source, i|
|
47
|
+
if target.size < source[:text].size
|
48
|
+
mode = :t_in_s
|
49
|
+
str1 = target
|
50
|
+
str2 = source[:text]
|
51
|
+
else
|
52
|
+
mode = :s_in_t
|
53
|
+
str1 = source[:text]
|
54
|
+
str2 = target
|
55
|
+
end
|
56
|
+
|
57
|
+
len1 = str1.length
|
58
|
+
len2 = str2.length
|
59
|
+
|
60
|
+
offset_begin, offset_end = if (len2 - len1) > len1 * (1 - TextAlignment::SIMILARITY_THRESHOLD)
|
61
|
+
approximate_fit(str1, str2)
|
62
|
+
else
|
63
|
+
# the whole target
|
64
|
+
[0, -1]
|
65
|
+
end
|
66
|
+
|
67
|
+
unless offset_begin.nil?
|
68
|
+
key = str1 + ' _:_ ' + str2[offset_begin .. offset_end]
|
69
|
+
cmp = if cache.has_key? key
|
70
|
+
cache[key]
|
71
|
+
else
|
72
|
+
cmp = TextAlignment::LCSComparison.new(str1, str2[offset_begin .. offset_end])
|
73
|
+
end
|
74
|
+
cache[key] = cmp
|
75
|
+
|
76
|
+
if (cmp.similarity > TextAlignment::SIMILARITY_THRESHOLD) && ((len1 - (cmp.str1_match_final - cmp.str1_match_initial + 1)) < len1 * (1 - TextAlignment::SIMILARITY_THRESHOLD))
|
77
|
+
candidates << {idx:i, offset:offset_begin, mode:mode, cmp:cmp}
|
78
|
+
end
|
79
|
+
end
|
80
|
+
end
|
81
|
+
|
82
|
+
# return remaining target and sources if m.nil?
|
83
|
+
break if candidates.empty?
|
84
|
+
|
85
|
+
choice = candidates.max{|a, b| a[:cmp].similarity <=> a[:cmp].similarity}
|
86
|
+
m = choice[:idx]
|
87
|
+
mode = choice[:mode]
|
88
|
+
|
89
|
+
index = if mode == :t_in_s
|
90
|
+
{divid:sources[m][:divid], region:[0, target.size]}
|
91
|
+
else # :s_in_t
|
92
|
+
cmp = choice[:cmp]
|
93
|
+
offset = choice[:offset]
|
94
|
+
{divid:sources[m][:divid], region:[cmp.str2_match_initial + offset, cmp.str2_match_final + offset + 1]}
|
95
|
+
end
|
96
|
+
|
97
|
+
target = target[0 ... index[:region][0]] + target[index[:region][1] .. -1]
|
98
|
+
history << index[:region].dup
|
99
|
+
|
100
|
+
before_begin = index[:region][0]
|
101
|
+
before_end = index[:region][1]
|
102
|
+
|
103
|
+
rhistory = history.reverse
|
104
|
+
rhistory.shift
|
105
|
+
rhistory.each do |h|
|
106
|
+
gap = h[1] - h[0]
|
107
|
+
index[:region][0] += gap if index[:region][0] >= h[0]
|
108
|
+
index[:region][1] += gap if index[:region][1] > h[0]
|
109
|
+
end
|
110
|
+
|
111
|
+
indice << index
|
112
|
+
|
113
|
+
sources.delete_at(m)
|
114
|
+
end
|
115
|
+
|
116
|
+
unless target.strip.empty? && sources.empty?
|
117
|
+
index = {divid:nil}
|
118
|
+
index[:remaining_target] = target unless target.strip.empty?
|
119
|
+
index[:remaining_sources] = sources.collect{|s| s[:divid]} unless sources.empty?
|
120
|
+
indice << index
|
121
|
+
end
|
122
|
+
|
123
|
+
indice
|
124
|
+
end
|
125
|
+
|
126
|
+
def _find_divisions_old(target, sources)
|
37
127
|
mode, m, c, offset_begin = nil, nil, nil, nil
|
38
128
|
|
39
129
|
sources.each_with_index do |source, i|
|
@@ -88,6 +178,7 @@ class << TextAlignment
|
|
88
178
|
return [index] + more_index
|
89
179
|
end
|
90
180
|
end
|
181
|
+
|
91
182
|
end
|
92
183
|
|
93
184
|
if __FILE__ == $0
|
@@ -98,20 +189,29 @@ if __FILE__ == $0
|
|
98
189
|
|
99
190
|
sources = JSON.parse File.read(ARGV[1]), :symbolize_names => true
|
100
191
|
div_index = TextAlignment::find_divisions(target_text, sources)
|
192
|
+
pp div_index
|
101
193
|
|
102
194
|
# str1 = File.read(ARGV[0]).strip
|
103
195
|
# str2 = File.read(ARGV[1]).strip
|
104
196
|
# div_index = TextAlignment::find_divisions(str1, [str2])
|
105
197
|
|
106
|
-
puts "target length: #{target_text.length}"
|
107
|
-
div_index.each do |i|
|
108
|
-
|
109
|
-
|
110
|
-
|
111
|
-
|
112
|
-
|
113
|
-
|
114
|
-
|
115
|
-
|
198
|
+
# puts "target length: #{target_text.length}"
|
199
|
+
# div_index.each do |i|
|
200
|
+
# unless i[:divid].nil?
|
201
|
+
# puts "[Div: #{i[:divid]}] (#{i[:region][0]}, #{i[:region][1]})"
|
202
|
+
# puts target_text[i[:region][0] ... i[:region][1]]
|
203
|
+
# puts "=========="
|
204
|
+
# else
|
205
|
+
# p i
|
206
|
+
# end
|
207
|
+
|
208
|
+
# # if i[0] >= 0
|
209
|
+
# # puts "[Div: #{i[0]}] (#{i[1][0]}, #{i[1][1]})"
|
210
|
+
# # puts target_text[i[1][0] ... i[1][1]]
|
211
|
+
# # puts "=========="
|
212
|
+
# # else
|
213
|
+
# # p i
|
214
|
+
# # end
|
215
|
+
# end
|
116
216
|
end
|
117
217
|
end
|
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: 0.2.
|
4
|
+
version: 0.2.9
|
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: 2020-07-
|
11
|
+
date: 2020-07-10 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: ruby-dictionary
|