datte 0.1.0 → 0.3.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/datte/data.txt +16020 -0
- data/lib/datte/date_parser.rb +26 -13
- data/lib/datte/datetime_table.rb +18 -10
- data/lib/datte/dattetime.rb +40 -15
- data/lib/datte/train.rb +203 -0
- data/lib/datte/version.rb +1 -1
- data/lib/datte.rb +1 -0
- data/spec/datte_spec.rb +8 -2
- metadata +3 -1
data/lib/datte/date_parser.rb
CHANGED
@@ -19,23 +19,36 @@ module Datte
|
|
19
19
|
end
|
20
20
|
end
|
21
21
|
|
22
|
-
|
22
|
+
ABSOLUTE_TIMES.each do |matcher|
|
23
|
+
if md = @body.match(matcher)
|
24
|
+
@date.update_time(md)
|
25
|
+
p @date
|
26
|
+
break
|
27
|
+
end
|
28
|
+
end
|
23
29
|
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
28
|
-
|
29
|
-
|
30
|
-
|
31
|
-
|
32
|
-
|
33
|
-
|
34
|
-
|
35
|
-
|
30
|
+
NOUNS.each do |matcher_s, method|
|
31
|
+
matcher = Regexp.new(matcher_s.to_s)
|
32
|
+
if md = @body.match(matcher)
|
33
|
+
eval(method)
|
34
|
+
break
|
35
|
+
end
|
36
|
+
end
|
37
|
+
|
38
|
+
AFTERS.each do |matcher|
|
39
|
+
if md = @body.match(matcher)
|
40
|
+
@date.after(md)
|
41
|
+
break
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
|
46
|
+
return @date.to_datetime
|
36
47
|
end
|
37
48
|
|
38
49
|
private
|
39
50
|
|
51
|
+
def next_day(day)
|
52
|
+
end
|
40
53
|
end
|
41
54
|
end
|
data/lib/datte/datetime_table.rb
CHANGED
@@ -37,17 +37,25 @@ module Datte
|
|
37
37
|
'(?<min>\d{1,2})分'
|
38
38
|
].map { |pattern| Regexp.compile(pattern) }.freeze
|
39
39
|
|
40
|
-
|
41
|
-
|
42
|
-
|
43
|
-
|
44
|
-
|
40
|
+
NOUNS = {
|
41
|
+
'明日|あした|あす': 'send(:next_day, 1)',
|
42
|
+
'明後日|あさって': 'send(:next_day, 2)',
|
43
|
+
'明々後日|しあさって': 'send(:next_day, 3)',
|
44
|
+
'今日|きょう': 'send(:next_day, 0)'
|
45
|
+
}
|
45
46
|
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
|
47
|
+
AFTERS = [
|
48
|
+
# 何年後
|
49
|
+
'(?<year>\d{1)年後',
|
50
|
+
# 何ヶ月後
|
51
|
+
'(?<month>\d{1,2}ヶ月後)',
|
52
|
+
# 何日後
|
53
|
+
'(?<day>\d{1,2})日後',
|
54
|
+
# 何時間後
|
55
|
+
'(?<hour>\d{1,2})時間後',
|
56
|
+
# 何分後
|
57
|
+
'(?<min>\d{1,2})分後'
|
58
|
+
].map { |pattern| Regexp.compile(pattern) }.freeze
|
51
59
|
|
52
60
|
class DatetimeTable
|
53
61
|
def initialize
|
data/lib/datte/dattetime.rb
CHANGED
@@ -2,22 +2,20 @@ module Datte
|
|
2
2
|
class Dattetime
|
3
3
|
|
4
4
|
DEFAULT_OPTIONS = {
|
5
|
-
force_update: false
|
5
|
+
force_update: false,
|
6
|
+
level: 1
|
6
7
|
}
|
7
8
|
|
8
|
-
attr_reader :year, :month, :day, :hour, :min
|
9
|
+
attr_reader :year, :month, :day, :hour, :min
|
9
10
|
|
10
11
|
def initialize(options = {})
|
11
12
|
@options = DEFAULT_OPTIONS.merge(options)
|
12
|
-
@date =
|
13
|
-
end
|
14
|
-
|
15
|
-
def to_s
|
16
|
-
@date.to_s
|
13
|
+
@date = DateTime.now
|
17
14
|
end
|
18
15
|
|
19
16
|
def to_datetime
|
20
|
-
|
17
|
+
return nil unless check_level?
|
18
|
+
DateTime.new(y, m, d, h, mi, 0) rescue nil
|
21
19
|
end
|
22
20
|
|
23
21
|
# 年か月か日を更新
|
@@ -29,18 +27,29 @@ module Datte
|
|
29
27
|
end
|
30
28
|
|
31
29
|
# 時か分を更新
|
32
|
-
def update_time(
|
33
|
-
|
34
|
-
|
35
|
-
|
36
|
-
def after_date(year, month, day)
|
30
|
+
def update_time(md, options = @options)
|
31
|
+
op = @options[:force_update] ? '=' : '||='
|
32
|
+
eval("@hour #{op} hour!(md)")
|
33
|
+
eval("@min #{op} min!(md)")
|
37
34
|
end
|
38
35
|
|
39
|
-
#
|
40
|
-
def
|
36
|
+
# 何年後、何ヶ月後、何日後, 何時間後, 何分後
|
37
|
+
def after(md)
|
38
|
+
@date >> (md[:year].to_i * 12) if md.matched?(:year) # 何年後
|
39
|
+
@date >> md[:month].to_i if md.matched?(:month) # 何ヶ月後
|
40
|
+
@date + md[:day].to_i if md.matched?(:day) # 何日後
|
41
|
+
@date + Rational(md[:hour].to_i, 24) # 何時間後
|
42
|
+
@date + Rational(md[:hour].to_i, 24 * 60) # 何分後
|
41
43
|
end
|
42
44
|
|
43
45
|
private
|
46
|
+
|
47
|
+
def y; @year || now[:year] end
|
48
|
+
def m; @month || now[:month] end
|
49
|
+
def d; @day || now[:day] end
|
50
|
+
def h; @hour || now[:hour] end
|
51
|
+
def mi; @min || 0 end
|
52
|
+
|
44
53
|
def now
|
45
54
|
d = DateTime.now
|
46
55
|
{ year: d.year, month: d.month, day: d.day, hour: d.hour, min: d.min }
|
@@ -57,6 +66,22 @@ module Datte
|
|
57
66
|
def day!(md)
|
58
67
|
md.matched?(:day) ? md[:day].to_i : now[:day]
|
59
68
|
end
|
69
|
+
|
70
|
+
def hour!(md)
|
71
|
+
md.matched?(:hour) ? md[:hour].to_i : now[:hour]
|
72
|
+
end
|
73
|
+
|
74
|
+
def min!(md)
|
75
|
+
md.matched?(:min) ? md[:min].to_i : 0
|
76
|
+
end
|
77
|
+
|
78
|
+
def check_level?
|
79
|
+
counter = 0
|
80
|
+
[@year, @month, @day, @hour, @min].each do |check|
|
81
|
+
counter += 1 unless check.nil?
|
82
|
+
end
|
83
|
+
@options[:level] < counter
|
84
|
+
end
|
60
85
|
end
|
61
86
|
end
|
62
87
|
|
data/lib/datte/train.rb
ADDED
@@ -0,0 +1,203 @@
|
|
1
|
+
module Datte
|
2
|
+
class Train
|
3
|
+
FNAME = File.join(File.dirname(__FILE__), 'data.txt')
|
4
|
+
|
5
|
+
# SEE: http://qiita.com/Hironsan/items/326b66711eb4196aa9d4
|
6
|
+
|
7
|
+
def initialize(path = FNAME)
|
8
|
+
train_sents = corpus_read
|
9
|
+
x = corpus_read[0]
|
10
|
+
p x
|
11
|
+
p '==='
|
12
|
+
p sent2features(x)
|
13
|
+
#sent2features(corpus_read[0])[0]
|
14
|
+
#sent2features(train_sents[0])[0]
|
15
|
+
end
|
16
|
+
|
17
|
+
def x_train
|
18
|
+
train_sents.each do |s|
|
19
|
+
return sent2features(s)
|
20
|
+
end
|
21
|
+
end
|
22
|
+
|
23
|
+
def y_train
|
24
|
+
train_sents.each do |s|
|
25
|
+
return sent2labels(s)
|
26
|
+
end
|
27
|
+
end
|
28
|
+
|
29
|
+
def x_test
|
30
|
+
test_sents.each do |s|
|
31
|
+
return sent2features(s)
|
32
|
+
end
|
33
|
+
end
|
34
|
+
|
35
|
+
def y_test
|
36
|
+
test_sents.each do |s|
|
37
|
+
return sent2labels(s)
|
38
|
+
end
|
39
|
+
end
|
40
|
+
|
41
|
+
def train()
|
42
|
+
end
|
43
|
+
|
44
|
+
private
|
45
|
+
def hiragana?(ch)
|
46
|
+
0x3040 <= ch.ord && ch.ord <= 0x309F
|
47
|
+
end
|
48
|
+
|
49
|
+
def katakana(ch)
|
50
|
+
0x30A0 <= ch.ord && ch.ord <= 0x30FF
|
51
|
+
end
|
52
|
+
|
53
|
+
def space?(ch)
|
54
|
+
!(ch =~ /^\s*$/).nil?
|
55
|
+
end
|
56
|
+
|
57
|
+
def integer?(ch)
|
58
|
+
Integer(ch)
|
59
|
+
Integer(ch)
|
60
|
+
true
|
61
|
+
rescue ArgumentError
|
62
|
+
false
|
63
|
+
end
|
64
|
+
|
65
|
+
def lower?(ch)
|
66
|
+
ch == ch.downcase
|
67
|
+
end
|
68
|
+
|
69
|
+
def upper?(ch)
|
70
|
+
ch == ch.upcase
|
71
|
+
end
|
72
|
+
|
73
|
+
def chara_type(ch)
|
74
|
+
if space?(ch) then 'ZSPACE'
|
75
|
+
elsif integer?(ch) then 'ZDIGIT'
|
76
|
+
elsif lower?(ch) then 'ZLLET'
|
77
|
+
elsif upper?(ch) then 'ZULET'
|
78
|
+
elsif hiragana?(ch) then 'HIRAG'
|
79
|
+
elsif katakana?(ch) then 'KATAK'
|
80
|
+
else 'OTHER'
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
def chara_types(str)
|
85
|
+
types = str.each_char.to_a.map do |ch|
|
86
|
+
chara_type(ch)
|
87
|
+
end
|
88
|
+
types.uniq.sort().join('-')
|
89
|
+
end
|
90
|
+
|
91
|
+
def extract_pos(morph)
|
92
|
+
idx = morph.index('*')
|
93
|
+
morph[1, idx-1].join('-')
|
94
|
+
end
|
95
|
+
|
96
|
+
def word2features(sent, i)
|
97
|
+
word = sent[i][0]
|
98
|
+
chtype = chara_types(sent[i][0])
|
99
|
+
postag = extract_pos(sent[i])
|
100
|
+
|
101
|
+
features = [
|
102
|
+
'bias',
|
103
|
+
'word=' + word,
|
104
|
+
'type=' + chtype,
|
105
|
+
'pos_tag=' + postag
|
106
|
+
]
|
107
|
+
|
108
|
+
if i >= 2
|
109
|
+
word2 = sent[i-2][0]
|
110
|
+
chtype2 = chara_types(sent[i-2][0])
|
111
|
+
postag2 = extract_pos(sent[i-2])
|
112
|
+
iobtag2 = sent[i-2][-1]
|
113
|
+
features.push(*[
|
114
|
+
'-2:word=' + word2,
|
115
|
+
'-2:type=' + chtype2,
|
116
|
+
'-2:postag=' + postag2,
|
117
|
+
'-2:iobtag=' + iobtag2
|
118
|
+
])
|
119
|
+
else
|
120
|
+
features.push('BOS')
|
121
|
+
end
|
122
|
+
|
123
|
+
if i >= 1
|
124
|
+
word1 = sent[i-1][0]
|
125
|
+
chtype1 = chara_types(sent[i-1][0])
|
126
|
+
postag1 = extract_pos(sent[i-1])
|
127
|
+
iobtag1 = sent[i-1][-1]
|
128
|
+
features.push(*[
|
129
|
+
'-1:word=' + word1,
|
130
|
+
'-1:type=' + chtype1,
|
131
|
+
'-1:postag=' + postag1,
|
132
|
+
'-1:iobtag=' + iobtag1
|
133
|
+
])
|
134
|
+
else
|
135
|
+
features.push('BOS')
|
136
|
+
end
|
137
|
+
|
138
|
+
if i < sent.length - 1
|
139
|
+
word1 = sent[i+1][0]
|
140
|
+
chtype1 = chara_types(sent[i+1][0])
|
141
|
+
postag1 = extract_pos(sent[i+1])
|
142
|
+
features.push(*[
|
143
|
+
'+1:word=' + word1,
|
144
|
+
'+1:type=' + chtype1,
|
145
|
+
'+1:postag=' + postag1
|
146
|
+
])
|
147
|
+
else
|
148
|
+
features.push('EOS')
|
149
|
+
end
|
150
|
+
|
151
|
+
if i < sent.length - 2
|
152
|
+
word2 = sent[i+2][0]
|
153
|
+
chtype2 = chara_types(sent[i+2][0])
|
154
|
+
postag2 = extract_pos(sent[i+2])
|
155
|
+
features.push(*[
|
156
|
+
'+2:word=' + word2,
|
157
|
+
'+2:type=' + chtype2,
|
158
|
+
'+2:postag=' + postag2
|
159
|
+
])
|
160
|
+
else
|
161
|
+
features.push('EOS')
|
162
|
+
end
|
163
|
+
|
164
|
+
return features
|
165
|
+
end
|
166
|
+
|
167
|
+
def sent2features(sent)
|
168
|
+
(0..(sent.length)).to_a.map do |i|
|
169
|
+
return word2features(sent, i)
|
170
|
+
end
|
171
|
+
end
|
172
|
+
|
173
|
+
def sent2labels(sent)
|
174
|
+
sent.map do |morph|
|
175
|
+
morph[-1]
|
176
|
+
end
|
177
|
+
end
|
178
|
+
|
179
|
+
def sent2tokens(sent)
|
180
|
+
sent.map do |morph|
|
181
|
+
morph[0]
|
182
|
+
end
|
183
|
+
end
|
184
|
+
|
185
|
+
def corpus_read
|
186
|
+
sents, sent = [], []
|
187
|
+
|
188
|
+
File.open(FNAME, 'r') do |file|
|
189
|
+
file.each_line do |line|
|
190
|
+
if line == "\n"
|
191
|
+
sents.push(sent)
|
192
|
+
sent = []
|
193
|
+
next
|
194
|
+
end
|
195
|
+
morph_info = line.strip().split(' ')
|
196
|
+
sent.push(morph_info)
|
197
|
+
end
|
198
|
+
end
|
199
|
+
sents
|
200
|
+
end
|
201
|
+
|
202
|
+
end
|
203
|
+
end
|
data/lib/datte/version.rb
CHANGED
data/lib/datte.rb
CHANGED
@@ -4,3 +4,4 @@ require File.join(File.dirname(__FILE__), 'datte', 'parser')
|
|
4
4
|
require File.join(File.dirname(__FILE__), 'datte', 'date_parser')
|
5
5
|
require File.join(File.dirname(__FILE__), 'datte', 'dattetime')
|
6
6
|
require File.join(File.dirname(__FILE__), 'datte', 'datetime_table')
|
7
|
+
require File.join(File.dirname(__FILE__), 'datte', 'train')
|
data/spec/datte_spec.rb
CHANGED
@@ -5,7 +5,13 @@ describe Datte do
|
|
5
5
|
expect(Datte::VERSION).not_to be nil
|
6
6
|
end
|
7
7
|
|
8
|
-
|
9
|
-
|
8
|
+
describe "::Parser" do
|
9
|
+
datte = Datte::Parser.new
|
10
|
+
|
11
|
+
it "2016/11/1日に遊ぼー" do
|
12
|
+
body = "2016/11/1日に遊ぼー"
|
13
|
+
d = DateTime.now
|
14
|
+
expect(datte.parse_date(body)).to eq(DateTime.new(2016, 11, 1, d.hour, 0, 0))
|
15
|
+
end
|
10
16
|
end
|
11
17
|
end
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: datte
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.
|
4
|
+
version: 0.3.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- pokohide
|
@@ -73,10 +73,12 @@ files:
|
|
73
73
|
- bin/setup
|
74
74
|
- datte.gemspec
|
75
75
|
- lib/datte.rb
|
76
|
+
- lib/datte/data.txt
|
76
77
|
- lib/datte/date_parser.rb
|
77
78
|
- lib/datte/datetime_table.rb
|
78
79
|
- lib/datte/dattetime.rb
|
79
80
|
- lib/datte/parser.rb
|
81
|
+
- lib/datte/train.rb
|
80
82
|
- lib/datte/version.rb
|
81
83
|
- spec/datte_spec.rb
|
82
84
|
- spec/spec_helper.rb
|