nhkore 0.3.3 → 0.3.8
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.yardopts +3 -0
- data/CHANGELOG.md +97 -2
- data/Gemfile +0 -18
- data/Gemfile.lock +89 -0
- data/README.md +58 -30
- data/Rakefile +68 -42
- data/bin/nhkore +4 -15
- data/lib/nhkore.rb +8 -20
- data/lib/nhkore/app.rb +231 -236
- data/lib/nhkore/article.rb +56 -53
- data/lib/nhkore/article_scraper.rb +308 -289
- data/lib/nhkore/cleaner.rb +20 -32
- data/lib/nhkore/cli/fx_cmd.rb +41 -53
- data/lib/nhkore/cli/get_cmd.rb +59 -70
- data/lib/nhkore/cli/news_cmd.rb +145 -154
- data/lib/nhkore/cli/search_cmd.rb +110 -120
- data/lib/nhkore/cli/sift_cmd.rb +111 -227
- data/lib/nhkore/datetime_parser.rb +328 -0
- data/lib/nhkore/defn.rb +48 -55
- data/lib/nhkore/dict.rb +26 -38
- data/lib/nhkore/dict_scraper.rb +31 -40
- data/lib/nhkore/entry.rb +43 -55
- data/lib/nhkore/error.rb +16 -21
- data/lib/nhkore/fileable.rb +10 -21
- data/lib/nhkore/lib.rb +6 -17
- data/lib/nhkore/missingno.rb +21 -33
- data/lib/nhkore/news.rb +61 -66
- data/lib/nhkore/polisher.rb +22 -34
- data/lib/nhkore/scraper.rb +75 -82
- data/lib/nhkore/search_link.rb +85 -78
- data/lib/nhkore/search_scraper.rb +89 -92
- data/lib/nhkore/sifter.rb +157 -171
- data/lib/nhkore/splitter.rb +19 -31
- data/lib/nhkore/user_agents.rb +28 -32
- data/lib/nhkore/util.rb +72 -101
- data/lib/nhkore/variator.rb +20 -32
- data/lib/nhkore/version.rb +4 -16
- data/lib/nhkore/word.rb +105 -99
- data/nhkore.gemspec +58 -65
- data/samples/looper.rb +71 -0
- data/test/nhkore/test_helper.rb +3 -15
- data/test/nhkore_test.rb +6 -18
- metadata +53 -30
@@ -0,0 +1,328 @@
|
|
1
|
+
# encoding: UTF-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
4
|
+
#--
|
5
|
+
# This file is part of NHKore.
|
6
|
+
# Copyright (c) 2020-2021 Jonathan Bradley Whited
|
7
|
+
#
|
8
|
+
# SPDX-License-Identifier: LGPL-3.0-or-later
|
9
|
+
#++
|
10
|
+
|
11
|
+
|
12
|
+
require 'attr_bool'
|
13
|
+
require 'date'
|
14
|
+
require 'time'
|
15
|
+
|
16
|
+
require 'nhkore/util'
|
17
|
+
|
18
|
+
|
19
|
+
module NHKore
|
20
|
+
###
|
21
|
+
# @author Jonathan Bradley Whited
|
22
|
+
# @since 0.3.4
|
23
|
+
###
|
24
|
+
class DatetimeParser
|
25
|
+
extend AttrBool::Ext
|
26
|
+
|
27
|
+
# Order matters!
|
28
|
+
FMTS = [
|
29
|
+
'%Y-%m-%d %H:%M',
|
30
|
+
'%Y-%m-%d %H',
|
31
|
+
'%Y-%m-%d',
|
32
|
+
'%m-%d %H:%M',
|
33
|
+
'%Y-%m %H:%M',
|
34
|
+
'%m-%d %H',
|
35
|
+
'%Y-%m %H',
|
36
|
+
'%m-%d',
|
37
|
+
'%Y-%m',
|
38
|
+
'%d %H:%M',
|
39
|
+
'%y %H:%M',
|
40
|
+
'%d %H',
|
41
|
+
'%Y %H',
|
42
|
+
'%H:%M',
|
43
|
+
'%d',
|
44
|
+
'%Y',
|
45
|
+
].freeze
|
46
|
+
|
47
|
+
def self.guess_year(year)
|
48
|
+
if year < 1000
|
49
|
+
century = Util::JST_YEAR / 100 * 100 # 2120 -> 2100
|
50
|
+
millennium = Util::JST_YEAR / 1000 * 1000 # 2120 -> 2000
|
51
|
+
|
52
|
+
# If year <= 23 (2022 -> 23)...
|
53
|
+
if year <= ((Util::JST_YEAR % 100) + 1)
|
54
|
+
# Assume this century.
|
55
|
+
year = century + year
|
56
|
+
elsif year >= 100
|
57
|
+
# If (2000 + 150) <= 2201 (if current year is 2200)...
|
58
|
+
if (millennium + year) <= (Util::JST_YEAR + 1)
|
59
|
+
# Assume this millennium.
|
60
|
+
# So if the current year is 2200, and year is 150,
|
61
|
+
# then it will be 2000 + 150 = 2150.
|
62
|
+
else
|
63
|
+
# Assume previous millennium (2000 -> 1000),
|
64
|
+
# so year 999 will become 1999.
|
65
|
+
millennium -= 1000 if millennium >= 1000
|
66
|
+
end
|
67
|
+
|
68
|
+
year = millennium + year
|
69
|
+
else
|
70
|
+
# Assume previous century (2000 -> 1900).
|
71
|
+
century -= 100 if century >= 100
|
72
|
+
year = century + year
|
73
|
+
end
|
74
|
+
end
|
75
|
+
|
76
|
+
return year
|
77
|
+
end
|
78
|
+
|
79
|
+
def self.parse_range(value)
|
80
|
+
# Do not use unspace_web_str(), want spaces for formats.
|
81
|
+
value = Util.strip_web_str(Util.reduce_space(value))
|
82
|
+
values = value.split('...',2)
|
83
|
+
|
84
|
+
return nil if values.empty? # For '' or '...'
|
85
|
+
|
86
|
+
# For '2020...' or '...2020'.
|
87
|
+
if value.include?('...')
|
88
|
+
# values.length is always 2 because of 2 in split() above.
|
89
|
+
|
90
|
+
# For '2020...'.
|
91
|
+
if Util.empty_web_str?(values[1])
|
92
|
+
values[1] = :infinity
|
93
|
+
# For '...2020'.
|
94
|
+
elsif Util.empty_web_str?(values[0])
|
95
|
+
values[0] = :infinity
|
96
|
+
end
|
97
|
+
end
|
98
|
+
|
99
|
+
datetimes = [
|
100
|
+
DatetimeParser.new, # "From" date time
|
101
|
+
DatetimeParser.new, # "To" date time
|
102
|
+
]
|
103
|
+
|
104
|
+
values.each_with_index do |v,i|
|
105
|
+
dt = datetimes[i]
|
106
|
+
|
107
|
+
# Minimum/Maximum date time for '2020...' or '...2020'.
|
108
|
+
if v == :infinity
|
109
|
+
# "From" date time.
|
110
|
+
if i == 0
|
111
|
+
dt.min!
|
112
|
+
# "To" date time.
|
113
|
+
else
|
114
|
+
dt.max!
|
115
|
+
end
|
116
|
+
else
|
117
|
+
v = Util.strip_web_str(v)
|
118
|
+
|
119
|
+
FMTS.each_with_index do |fmt,j|
|
120
|
+
# If don't do this, "%d" values will be parsed using "%d %H".
|
121
|
+
# It seems as though strptime() ignores space.
|
122
|
+
raise ArgumentError if fmt.include?(' ') && !v.include?(' ')
|
123
|
+
|
124
|
+
# If don't do this, "%y..." values will be parsed using "%d...".
|
125
|
+
raise ArgumentError if fmt.start_with?('%d') && v.split(' ')[0].length > 2
|
126
|
+
|
127
|
+
dt.parse!(v,fmt)
|
128
|
+
|
129
|
+
break # No problem; this format worked
|
130
|
+
rescue ArgumentError
|
131
|
+
# Out of formats.
|
132
|
+
raise if j >= (FMTS.length - 1)
|
133
|
+
end
|
134
|
+
end
|
135
|
+
end
|
136
|
+
|
137
|
+
from = datetimes[0]
|
138
|
+
to = datetimes[1]
|
139
|
+
|
140
|
+
from.autofill!(:from,to)
|
141
|
+
to.autofill!(:to,from)
|
142
|
+
|
143
|
+
return [from.jst_time,to.jst_time]
|
144
|
+
end
|
145
|
+
|
146
|
+
attr_accessor :day
|
147
|
+
attr_accessor :hour
|
148
|
+
attr_accessor :min
|
149
|
+
attr_accessor :month
|
150
|
+
attr_accessor :sec
|
151
|
+
attr_accessor :year
|
152
|
+
|
153
|
+
attr_accessor? :has_day
|
154
|
+
attr_accessor? :has_hour
|
155
|
+
attr_accessor? :has_min
|
156
|
+
attr_accessor? :has_month
|
157
|
+
attr_accessor? :has_sec
|
158
|
+
attr_accessor? :has_year
|
159
|
+
|
160
|
+
attr_reader? :min_or_max
|
161
|
+
|
162
|
+
def initialize(year=nil,month=nil,day=nil,hour=nil,min=nil,sec=nil)
|
163
|
+
super()
|
164
|
+
|
165
|
+
set!(year,month,day,hour,min,sec)
|
166
|
+
|
167
|
+
self.has = false
|
168
|
+
@min_or_max = false
|
169
|
+
end
|
170
|
+
|
171
|
+
def autofill!(type,other)
|
172
|
+
case type
|
173
|
+
when :from
|
174
|
+
is_from = true
|
175
|
+
when :to
|
176
|
+
is_from = false
|
177
|
+
else
|
178
|
+
raise ArgumentError,"invalid type[#{type}]"
|
179
|
+
end
|
180
|
+
|
181
|
+
return self if @min_or_max
|
182
|
+
|
183
|
+
has_small = false
|
184
|
+
jst_now = Util.jst_now()
|
185
|
+
|
186
|
+
# Must be from smallest to biggest.
|
187
|
+
|
188
|
+
if @has_sec || other.has_sec?
|
189
|
+
@sec = other.sec unless @has_sec
|
190
|
+
has_small = true
|
191
|
+
else
|
192
|
+
if has_small
|
193
|
+
@sec = jst_now.sec
|
194
|
+
else
|
195
|
+
@sec = is_from ? 0 : 59
|
196
|
+
end
|
197
|
+
end
|
198
|
+
|
199
|
+
if @has_min || other.has_min?
|
200
|
+
@min = other.min unless @has_min
|
201
|
+
has_small = true
|
202
|
+
else
|
203
|
+
if has_small
|
204
|
+
@min = jst_now.min
|
205
|
+
else
|
206
|
+
@min = is_from ? 0 : 59
|
207
|
+
end
|
208
|
+
end
|
209
|
+
|
210
|
+
if @has_hour || other.has_hour?
|
211
|
+
@hour = other.hour unless @has_hour
|
212
|
+
has_small = true
|
213
|
+
else
|
214
|
+
if has_small
|
215
|
+
@hour = jst_now.hour
|
216
|
+
else
|
217
|
+
@hour = is_from ? 0 : 23
|
218
|
+
end
|
219
|
+
end
|
220
|
+
|
221
|
+
if @has_day || other.has_day?
|
222
|
+
@day = other.day unless @has_day
|
223
|
+
has_small = true
|
224
|
+
else
|
225
|
+
if has_small
|
226
|
+
@day = jst_now.day
|
227
|
+
else
|
228
|
+
@day = is_from ? 1 : :last_day
|
229
|
+
end
|
230
|
+
end
|
231
|
+
|
232
|
+
if @has_month || other.has_month?
|
233
|
+
@month = other.month unless @has_month
|
234
|
+
has_small = true
|
235
|
+
else
|
236
|
+
if has_small
|
237
|
+
@month = jst_now.month
|
238
|
+
else
|
239
|
+
@month = is_from ? 1 : 12
|
240
|
+
end
|
241
|
+
end
|
242
|
+
|
243
|
+
if @has_year || other.has_year?
|
244
|
+
@year = other.year unless @has_year
|
245
|
+
has_small = true # rubocop:disable Lint/UselessAssignment
|
246
|
+
else
|
247
|
+
if has_small
|
248
|
+
@year = jst_now.year
|
249
|
+
else
|
250
|
+
@year = is_from ? Util::MIN_SANE_YEAR : jst_now.year
|
251
|
+
end
|
252
|
+
end
|
253
|
+
|
254
|
+
# Must be after setting @year & @month.
|
255
|
+
if @day == :last_day
|
256
|
+
@day = Date.new(@year,@month,-1).day
|
257
|
+
end
|
258
|
+
|
259
|
+
return self
|
260
|
+
end
|
261
|
+
|
262
|
+
def max!
|
263
|
+
@min_or_max = true
|
264
|
+
|
265
|
+
# Ex: 2020-12-31 23:59:59
|
266
|
+
return set!(Util::JST_YEAR,12,31,23,59,59)
|
267
|
+
end
|
268
|
+
|
269
|
+
def min!
|
270
|
+
@min_or_max = true
|
271
|
+
|
272
|
+
# Ex: 1924-01-01 00:00:00
|
273
|
+
return set!(Util::MIN_SANE_YEAR,1,1,0,0,0)
|
274
|
+
end
|
275
|
+
|
276
|
+
def parse!(value,fmt)
|
277
|
+
value = Time.strptime(value,fmt,&self.class.method(:guess_year))
|
278
|
+
|
279
|
+
@has_day = fmt.include?('%d')
|
280
|
+
@has_hour = fmt.include?('%H')
|
281
|
+
@has_min = fmt.include?('%M')
|
282
|
+
@has_month = fmt.include?('%m')
|
283
|
+
@has_sec = fmt.include?('%S')
|
284
|
+
@has_year = fmt.include?('%Y')
|
285
|
+
|
286
|
+
@day = value.day if @has_day
|
287
|
+
@hour = value.hour if @has_hour
|
288
|
+
@min = value.min if @has_min
|
289
|
+
@month = value.month if @has_month
|
290
|
+
@sec = value.sec if @has_sec
|
291
|
+
@year = value.year if @has_year
|
292
|
+
|
293
|
+
return self
|
294
|
+
end
|
295
|
+
|
296
|
+
def set!(year=nil,month=nil,day=nil,hour=nil,min=nil,sec=nil)
|
297
|
+
@year = year
|
298
|
+
@month = month
|
299
|
+
@day = day
|
300
|
+
@hour = hour
|
301
|
+
@min = min
|
302
|
+
@sec = sec
|
303
|
+
|
304
|
+
return self
|
305
|
+
end
|
306
|
+
|
307
|
+
def has=(value)
|
308
|
+
@has_day = value
|
309
|
+
@has_hour = value
|
310
|
+
@has_min = value
|
311
|
+
@has_month = value
|
312
|
+
@has_sec = value
|
313
|
+
@has_year = value
|
314
|
+
end
|
315
|
+
|
316
|
+
def jst_time
|
317
|
+
return Util.jst_time(time)
|
318
|
+
end
|
319
|
+
|
320
|
+
def time
|
321
|
+
return Time.new(@year,@month,@day,@hour,@min,@sec)
|
322
|
+
end
|
323
|
+
|
324
|
+
def to_s
|
325
|
+
return "#{@year}-#{@month}-#{@day} #{@hour}:#{@min}:#{@sec}"
|
326
|
+
end
|
327
|
+
end
|
328
|
+
end
|
data/lib/nhkore/defn.rb
CHANGED
@@ -1,23 +1,11 @@
|
|
1
|
-
#!/usr/bin/env ruby
|
2
1
|
# encoding: UTF-8
|
3
2
|
# frozen_string_literal: true
|
4
3
|
|
5
4
|
#--
|
6
5
|
# This file is part of NHKore.
|
7
|
-
# Copyright (c) 2020 Jonathan Bradley Whited
|
8
|
-
#
|
9
|
-
#
|
10
|
-
# it under the terms of the GNU Lesser General Public License as published by
|
11
|
-
# the Free Software Foundation, either version 3 of the License, or
|
12
|
-
# (at your option) any later version.
|
13
|
-
#
|
14
|
-
# NHKore is distributed in the hope that it will be useful,
|
15
|
-
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
16
|
-
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
17
|
-
# GNU Lesser General Public License for more details.
|
18
|
-
#
|
19
|
-
# You should have received a copy of the GNU Lesser General Public License
|
20
|
-
# along with NHKore. If not, see <https://www.gnu.org/licenses/>.
|
6
|
+
# Copyright (c) 2020-2021 Jonathan Bradley Whited
|
7
|
+
#
|
8
|
+
# SPDX-License-Identifier: LGPL-3.0-or-later
|
21
9
|
#++
|
22
10
|
|
23
11
|
|
@@ -29,75 +17,80 @@ require 'nhkore/word'
|
|
29
17
|
|
30
18
|
module NHKore
|
31
19
|
###
|
32
|
-
# @author Jonathan Bradley Whited
|
20
|
+
# @author Jonathan Bradley Whited
|
33
21
|
# @since 0.2.0
|
34
22
|
###
|
35
23
|
class Defn
|
36
24
|
attr_reader :hyoukis
|
37
25
|
attr_accessor :text
|
38
26
|
attr_reader :words
|
39
|
-
|
40
|
-
def initialize
|
27
|
+
|
28
|
+
def initialize
|
41
29
|
super()
|
42
|
-
|
30
|
+
|
43
31
|
@hyoukis = []
|
44
|
-
@text = ''.dup
|
32
|
+
@text = ''.dup
|
45
33
|
@words = []
|
46
34
|
end
|
47
|
-
|
35
|
+
|
48
36
|
# If no data, don't raise errors; don't care if have a definition or not.
|
49
37
|
def self.scrape(hash,missingno: nil,url: nil)
|
50
|
-
defn = Defn.new
|
51
|
-
|
38
|
+
defn = Defn.new
|
39
|
+
|
52
40
|
hyoukis = hash['hyouki']
|
53
|
-
|
54
|
-
|
55
|
-
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
defn.hyoukis << hyouki
|
60
|
-
end
|
41
|
+
|
42
|
+
hyoukis&.each() do |hyouki|
|
43
|
+
next if hyouki.nil?
|
44
|
+
next if (hyouki = Util.strip_web_str(hyouki)).empty?
|
45
|
+
|
46
|
+
defn.hyoukis << hyouki
|
61
47
|
end
|
62
|
-
|
48
|
+
|
63
49
|
def_str = hash['def']
|
64
|
-
|
50
|
+
|
65
51
|
if Util.empty_web_str?(def_str)
|
66
|
-
return defn.hyoukis.empty?
|
52
|
+
return defn.hyoukis.empty? ? nil : defn
|
67
53
|
end
|
68
|
-
|
54
|
+
|
69
55
|
doc = Nokogiri::HTML(def_str)
|
70
|
-
doc = doc.css('body') # Auto-added by Nokogiri
|
71
|
-
|
72
|
-
doc.children.each
|
73
|
-
name = Util.unspace_web_str(child.name).downcase
|
74
|
-
|
56
|
+
doc = doc.css('body') # Auto-added by Nokogiri.
|
57
|
+
|
58
|
+
doc.children.each do |child|
|
59
|
+
name = Util.unspace_web_str(child.name).downcase if child.respond_to?(:name)
|
60
|
+
|
75
61
|
is_text = false
|
76
|
-
|
77
|
-
|
62
|
+
words = []
|
63
|
+
|
78
64
|
if name == 'ruby'
|
79
|
-
|
80
|
-
|
81
|
-
|
65
|
+
# Returns an array.
|
66
|
+
words = Word.scrape_ruby_tag(child,missingno: missingno,url: url)
|
67
|
+
elsif child.respond_to?(:text) # Don't do child.text?(), just want content.
|
68
|
+
words << Word.scrape_text_node(child,url: url)
|
82
69
|
is_text = true
|
83
70
|
end
|
84
|
-
|
85
|
-
|
71
|
+
|
72
|
+
# All word-scraping methods can return nil,
|
73
|
+
# so remove all nils for empty?() check.
|
74
|
+
words = words&.compact
|
75
|
+
|
76
|
+
if words.nil? || words.empty?
|
86
77
|
defn.text << Util.reduce_jpn_space(child.text) if is_text
|
87
78
|
else
|
88
|
-
|
89
|
-
|
79
|
+
words.each do |word|
|
80
|
+
defn.text << Util.reduce_jpn_space(word.word)
|
81
|
+
defn.words << word unless Util.empty_web_str?(word.word)
|
82
|
+
end
|
90
83
|
end
|
91
84
|
end
|
92
|
-
|
93
|
-
return nil if defn.hyoukis.empty?
|
94
|
-
|
85
|
+
|
86
|
+
return nil if defn.hyoukis.empty? && defn.words.empty?
|
87
|
+
|
95
88
|
defn.text = Util.strip_web_str(defn.text)
|
96
|
-
|
89
|
+
|
97
90
|
return defn
|
98
91
|
end
|
99
|
-
|
100
|
-
def to_s
|
92
|
+
|
93
|
+
def to_s
|
101
94
|
return @text
|
102
95
|
end
|
103
96
|
end
|