iGEL-ua_parser 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +50 -0
- data/Rakefile +12 -0
- data/apache_log_tester.rb +138 -0
- data/lib/ua_parser/user_agent.rb +492 -0
- data/lib/ua_parser/version.rb +85 -0
- data/lib/ua_parser.rb +2 -0
- data/test/crawler_test.rb +620 -0
- data/test/gecko_test.rb +493 -0
- data/test/other_test.rb +146 -0
- data/test/presto_test.rb +169 -0
- data/test/trident_test.rb +146 -0
- data/test/version_test.rb +130 -0
- data/test/webkit_test.rb +242 -0
- data/ua_parser.gemspec +32 -0
- metadata +81 -0
@@ -0,0 +1,85 @@
|
|
1
|
+
module UaParser
|
2
|
+
# This class stores the version number for comparison and printing out the
|
3
|
+
# major version
|
4
|
+
class Version
|
5
|
+
include Comparable
|
6
|
+
|
7
|
+
attr_reader :name
|
8
|
+
|
9
|
+
class DifferentAgentsNotComparable < RuntimeError
|
10
|
+
end
|
11
|
+
|
12
|
+
# Initializes a new Version object with the given version number (you should
|
13
|
+
# pass it as a String). Optional you may pass a user agent name as a Symbol,
|
14
|
+
# which allows #major to print out the version number in the common way.
|
15
|
+
def initialize(version, name = nil)
|
16
|
+
version = version.to_s if version.is_a?(Integer) || version.is_a?(Float)
|
17
|
+
raise TypeError, "version must be a String, Integer or Float, but was a #{version.class}" unless version.is_a?(String)
|
18
|
+
@tokens = version.split(/(\.|_|[a-z]+)/)
|
19
|
+
@tokens.delete_if { |t| t == "." || t == "_" }
|
20
|
+
insert_zeros_before = []
|
21
|
+
@tokens.each_index do |i|
|
22
|
+
if @tokens[i] =~ /\A[0-9]+\Z/
|
23
|
+
while @tokens[i] =~ /\A0[0-9]/ # Remember zeros, but chop them of for now
|
24
|
+
insert_zeros_before << i
|
25
|
+
@tokens[i] = @tokens[i][1..-1]
|
26
|
+
end
|
27
|
+
@tokens[i] = @tokens[i].to_i
|
28
|
+
end
|
29
|
+
end
|
30
|
+
insert_zeros_before.each do |i| # Insert zeros again
|
31
|
+
@tokens.insert(i, 0)
|
32
|
+
end
|
33
|
+
@version = version
|
34
|
+
@name = name
|
35
|
+
end
|
36
|
+
|
37
|
+
# Returns -1, if the given version is smaller than the current one, 0 if
|
38
|
+
# they are equal and 1 if the current version is bigger. Used by Comparable
|
39
|
+
# to provide funktions like <, >= or between?
|
40
|
+
def <=>(other)
|
41
|
+
raise TypeError unless other.is_a?(Version)
|
42
|
+
raise DifferentAgentsNotComparable, "Cannot compare #{@name} and #{other.name}" unless @name == other.name
|
43
|
+
i_sort(other)
|
44
|
+
end
|
45
|
+
|
46
|
+
# Returns the whole version number, as given by the by the user agent.
|
47
|
+
def full
|
48
|
+
@version
|
49
|
+
end
|
50
|
+
|
51
|
+
# Returns the major version, e.g. "3.0", "2.0b" or "1.5" for Firefox,
|
52
|
+
# "9.6" for Opera or "7.0" or "5.5" for Internet Explorer
|
53
|
+
def major
|
54
|
+
if @name == :opera || (@name == :netscape && self >= Version.new(5, :netscape))
|
55
|
+
return "#{@tokens[0]}.#{@tokens[1].to_s[0...1]}"
|
56
|
+
elsif [:internet_explorer, :navigator].include?(@name) && @tokens[1] == 0
|
57
|
+
return "#{@tokens[0]}"
|
58
|
+
elsif @tokens.size == 1
|
59
|
+
return "#{@tokens[0]}"
|
60
|
+
elsif @tokens[2].is_a?(String)
|
61
|
+
return "#{@tokens[0]}.#{@tokens[1]}#{@tokens[2]}"
|
62
|
+
else
|
63
|
+
return "#{@tokens[0]}.#{@tokens[1]}"
|
64
|
+
end
|
65
|
+
end
|
66
|
+
|
67
|
+
protected
|
68
|
+
# Required by i_sort
|
69
|
+
def tokens
|
70
|
+
@tokens
|
71
|
+
end
|
72
|
+
|
73
|
+
private
|
74
|
+
# Used to recursively compare the version numbers
|
75
|
+
def i_sort(other, deep = 0)
|
76
|
+
return 0 if @tokens[deep].nil? && other.tokens[deep].nil?
|
77
|
+
my_token = @tokens[deep] || 0
|
78
|
+
other_token = other.tokens[deep] || 0
|
79
|
+
return 1 if my_token.is_a?(Integer) && other_token.is_a?(String)
|
80
|
+
return -1 if my_token.is_a?(String) && other_token.is_a?(Integer)
|
81
|
+
return i_sort(other, deep + 1) if my_token == other_token
|
82
|
+
return my_token <=> other_token
|
83
|
+
end
|
84
|
+
end
|
85
|
+
end
|
data/lib/ua_parser.rb
ADDED
@@ -0,0 +1,620 @@
|
|
1
|
+
$:.unshift File.join(File.dirname(__FILE__),'..','lib')
|
2
|
+
|
3
|
+
require 'test/unit'
|
4
|
+
require 'ua_parser'
|
5
|
+
|
6
|
+
class CrawlerTest < Test::Unit::TestCase
|
7
|
+
include UaParser
|
8
|
+
|
9
|
+
def test_baiduspider
|
10
|
+
ua = UserAgent.new "Baiduspider+(+http://www.baidu.com/search/spider.htm)"
|
11
|
+
assert ua.known?
|
12
|
+
assert ua.bot?
|
13
|
+
assert !ua.browser?
|
14
|
+
assert !ua.feed_reader?
|
15
|
+
assert !ua.other?
|
16
|
+
assert_equal nil, ua.render_engine
|
17
|
+
assert_equal nil, ua.render_engine_version
|
18
|
+
assert_equal :baiduspider, ua.name
|
19
|
+
assert_equal nil, ua.version
|
20
|
+
assert_equal [], ua.emails
|
21
|
+
assert_equal nil, ua.email
|
22
|
+
assert_equal ["http://www.baidu.com/search/spider.htm"], ua.urls
|
23
|
+
assert_equal "http://www.baidu.com/search/spider.htm", ua.url
|
24
|
+
[:architecture, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
25
|
+
assert_raise NotImplementedError do
|
26
|
+
ua.method(method).call
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
|
31
|
+
def test_cuil_twiceler_0_9
|
32
|
+
ua = UserAgent.new "Mozilla/5.0 (Twiceler-0.9 http://www.cuil.com/twiceler/robot.html)"
|
33
|
+
assert ua.known?
|
34
|
+
assert ua.bot?
|
35
|
+
assert !ua.browser?
|
36
|
+
assert !ua.feed_reader?
|
37
|
+
assert !ua.other?
|
38
|
+
assert_equal nil, ua.render_engine
|
39
|
+
assert_equal nil, ua.render_engine_version
|
40
|
+
assert_equal :"twiceler", ua.name
|
41
|
+
assert_equal "0.9", ua.version.full
|
42
|
+
assert_equal "0.9", ua.version.major
|
43
|
+
assert_equal [], ua.emails
|
44
|
+
assert_equal nil, ua.email
|
45
|
+
assert_equal ["http://www.cuil.com/twiceler/robot.html"], ua.urls
|
46
|
+
assert_equal "http://www.cuil.com/twiceler/robot.html", ua.url
|
47
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
48
|
+
assert_raise NotImplementedError do
|
49
|
+
ua.method(method).call
|
50
|
+
end
|
51
|
+
end
|
52
|
+
end
|
53
|
+
|
54
|
+
def test_cuil_0_9_twiceler_wrong_domain
|
55
|
+
ua = UserAgent.new "Mozilla/5.0 (Twiceler-0.9 http://www.cuill.com/twiceler/robot.html)"
|
56
|
+
assert ua.known?
|
57
|
+
assert ua.bot?
|
58
|
+
assert !ua.browser?
|
59
|
+
assert !ua.feed_reader?
|
60
|
+
assert !ua.other?
|
61
|
+
assert_equal nil, ua.render_engine
|
62
|
+
assert_equal nil, ua.render_engine_version
|
63
|
+
assert_equal :"twiceler", ua.name
|
64
|
+
assert_equal "0.9", ua.version.major
|
65
|
+
assert_equal "0.9", ua.version.full
|
66
|
+
assert_equal [], ua.emails
|
67
|
+
assert_equal nil, ua.email
|
68
|
+
assert_equal ["http://www.cuill.com/twiceler/robot.html"], ua.urls
|
69
|
+
assert_equal "http://www.cuill.com/twiceler/robot.html", ua.url
|
70
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
71
|
+
assert_raise NotImplementedError do
|
72
|
+
ua.method(method).call
|
73
|
+
end
|
74
|
+
end
|
75
|
+
end
|
76
|
+
|
77
|
+
def test_gigabot_3_0
|
78
|
+
ua = UserAgent.new "Gigabot/3.0 (http://www.gigablast.com/spider.html)"
|
79
|
+
assert ua.known?
|
80
|
+
assert ua.bot?
|
81
|
+
assert !ua.browser?
|
82
|
+
assert !ua.feed_reader?
|
83
|
+
assert !ua.other?
|
84
|
+
assert_equal nil, ua.render_engine
|
85
|
+
assert_equal nil, ua.render_engine_version
|
86
|
+
assert_equal :gigabot, ua.name
|
87
|
+
assert_equal "3.0", ua.version.full
|
88
|
+
assert_equal "3.0", ua.version.major
|
89
|
+
assert_equal [], ua.emails
|
90
|
+
assert_equal nil, ua.email
|
91
|
+
assert_equal ["http://www.gigablast.com/spider.html"], ua.urls
|
92
|
+
assert_equal "http://www.gigablast.com/spider.html", ua.url
|
93
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
94
|
+
assert_raise NotImplementedError do
|
95
|
+
ua.method(method).call
|
96
|
+
end
|
97
|
+
end
|
98
|
+
end
|
99
|
+
|
100
|
+
def test_gonzo_1
|
101
|
+
ua = UserAgent.new "gonzo1[P] +http://www.suchen.de/faq.html"
|
102
|
+
assert ua.known?
|
103
|
+
assert ua.bot?
|
104
|
+
assert !ua.browser?
|
105
|
+
assert !ua.feed_reader?
|
106
|
+
assert !ua.other?
|
107
|
+
assert_equal nil, ua.render_engine
|
108
|
+
assert_equal nil, ua.render_engine_version
|
109
|
+
assert_equal :gonzo, ua.name
|
110
|
+
assert_equal "1", ua.version.full
|
111
|
+
assert_equal "1", ua.version.major
|
112
|
+
assert_equal [], ua.emails
|
113
|
+
assert_equal nil, ua.email
|
114
|
+
assert_equal ["http://www.suchen.de/faq.html"], ua.urls
|
115
|
+
assert_equal "http://www.suchen.de/faq.html", ua.url
|
116
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
117
|
+
assert_raise NotImplementedError do
|
118
|
+
ua.method(method).call
|
119
|
+
end
|
120
|
+
end
|
121
|
+
end
|
122
|
+
|
123
|
+
def test_gonzo_2
|
124
|
+
ua = UserAgent.new "gonzo2[P] +http://www.suchen.de/faq.html"
|
125
|
+
assert ua.known?
|
126
|
+
assert ua.bot?
|
127
|
+
assert !ua.browser?
|
128
|
+
assert !ua.feed_reader?
|
129
|
+
assert !ua.other?
|
130
|
+
assert_equal nil, ua.render_engine
|
131
|
+
assert_equal nil, ua.render_engine_version
|
132
|
+
assert_equal :gonzo, ua.name
|
133
|
+
assert_equal "2", ua.version.full
|
134
|
+
assert_equal "2", ua.version.major
|
135
|
+
assert_equal [], ua.emails
|
136
|
+
assert_equal nil, ua.email
|
137
|
+
assert_equal ["http://www.suchen.de/faq.html"], ua.urls
|
138
|
+
assert_equal "http://www.suchen.de/faq.html", ua.url
|
139
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
140
|
+
assert_raise NotImplementedError do
|
141
|
+
ua.method(method).call
|
142
|
+
end
|
143
|
+
end
|
144
|
+
end
|
145
|
+
|
146
|
+
def test_googlebot_2_1
|
147
|
+
ua = UserAgent.new "Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)"
|
148
|
+
assert ua.known?
|
149
|
+
assert ua.bot?
|
150
|
+
assert !ua.browser?
|
151
|
+
assert !ua.feed_reader?
|
152
|
+
assert !ua.other?
|
153
|
+
assert_equal nil, ua.render_engine
|
154
|
+
assert_equal nil, ua.render_engine_version
|
155
|
+
assert_equal :googlebot, ua.name
|
156
|
+
assert_equal "2.1", ua.version.full
|
157
|
+
assert_equal "2.1", ua.version.major
|
158
|
+
assert_equal [], ua.emails
|
159
|
+
assert_equal nil, ua.email
|
160
|
+
assert_equal ["http://www.google.com/bot.html"], ua.urls
|
161
|
+
assert_equal "http://www.google.com/bot.html", ua.url
|
162
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
163
|
+
assert_raise NotImplementedError do
|
164
|
+
ua.method(method).call
|
165
|
+
end
|
166
|
+
end
|
167
|
+
end
|
168
|
+
|
169
|
+
def test_googlebot_image_1_0
|
170
|
+
ua = UserAgent.new "Googlebot-Image/1.0"
|
171
|
+
assert ua.known?
|
172
|
+
assert ua.bot?
|
173
|
+
assert !ua.browser?
|
174
|
+
assert !ua.feed_reader?
|
175
|
+
assert !ua.other?
|
176
|
+
assert_equal nil, ua.render_engine
|
177
|
+
assert_equal nil, ua.render_engine_version
|
178
|
+
assert_equal :"googlebot-image", ua.name
|
179
|
+
assert_equal "1.0", ua.version.full
|
180
|
+
assert_equal "1.0", ua.version.major
|
181
|
+
assert_equal [], ua.emails
|
182
|
+
assert_equal nil, ua.email
|
183
|
+
assert_equal [], ua.urls
|
184
|
+
assert_equal nil, ua.url
|
185
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
186
|
+
assert_raise NotImplementedError do
|
187
|
+
ua.method(method).call
|
188
|
+
end
|
189
|
+
end
|
190
|
+
end
|
191
|
+
|
192
|
+
def test_mediapartners_google
|
193
|
+
ua = UserAgent.new "Mediapartners-Google"
|
194
|
+
assert ua.known?
|
195
|
+
assert ua.bot?
|
196
|
+
assert !ua.browser?
|
197
|
+
assert !ua.feed_reader?
|
198
|
+
assert !ua.other?
|
199
|
+
assert_equal nil, ua.render_engine
|
200
|
+
assert_equal nil, ua.render_engine_version
|
201
|
+
assert_equal :"mediapartners-google", ua.name
|
202
|
+
assert_equal nil, ua.version
|
203
|
+
assert_equal [], ua.emails
|
204
|
+
assert_equal nil, ua.email
|
205
|
+
assert_equal [], ua.urls
|
206
|
+
assert_equal nil, ua.url
|
207
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
208
|
+
assert_raise NotImplementedError do
|
209
|
+
ua.method(method).call
|
210
|
+
end
|
211
|
+
end
|
212
|
+
end
|
213
|
+
|
214
|
+
def test_mediapartners_google_21
|
215
|
+
ua = UserAgent.new "Mediapartners-Google/2.1"
|
216
|
+
assert ua.known?
|
217
|
+
assert ua.bot?
|
218
|
+
assert !ua.browser?
|
219
|
+
assert !ua.feed_reader?
|
220
|
+
assert !ua.other?
|
221
|
+
assert_equal nil, ua.render_engine
|
222
|
+
assert_equal nil, ua.render_engine_version
|
223
|
+
assert_equal :"mediapartners-google", ua.name
|
224
|
+
assert_equal "2.1", ua.version.full
|
225
|
+
assert_equal "2.1", ua.version.major
|
226
|
+
assert_equal [], ua.emails
|
227
|
+
assert_equal nil, ua.email
|
228
|
+
assert_equal [], ua.urls
|
229
|
+
assert_equal nil, ua.url
|
230
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
231
|
+
assert_raise NotImplementedError do
|
232
|
+
ua.method(method).call
|
233
|
+
end
|
234
|
+
end
|
235
|
+
end
|
236
|
+
|
237
|
+
def test_mj12bot_1_2_3
|
238
|
+
ua = UserAgent.new "Mozilla/5.0 (compatible; MJ12bot/v1.2.3; http://www.majestic12.co.uk/bot.php?+)"
|
239
|
+
assert ua.known?
|
240
|
+
assert ua.bot?
|
241
|
+
assert !ua.browser?
|
242
|
+
assert !ua.feed_reader?
|
243
|
+
assert !ua.other?
|
244
|
+
assert_equal nil, ua.render_engine
|
245
|
+
assert_equal nil, ua.render_engine_version
|
246
|
+
assert_equal :mj12bot, ua.name
|
247
|
+
assert_equal "1.2.3", ua.version.full
|
248
|
+
assert_equal "1.2", ua.version.major
|
249
|
+
assert_equal [], ua.emails
|
250
|
+
assert_equal nil, ua.email
|
251
|
+
assert_equal ["http://www.majestic12.co.uk/bot.php?+"], ua.urls
|
252
|
+
assert_equal "http://www.majestic12.co.uk/bot.php?+", ua.url
|
253
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
254
|
+
assert_raise NotImplementedError do
|
255
|
+
ua.method(method).call
|
256
|
+
end
|
257
|
+
end
|
258
|
+
end
|
259
|
+
|
260
|
+
def test_msnbot_10
|
261
|
+
ua = UserAgent.new "msnbot/1.0 (+http://search.msn.com/msnbot.htm)"
|
262
|
+
assert ua.known?
|
263
|
+
assert ua.bot?
|
264
|
+
assert !ua.browser?
|
265
|
+
assert !ua.feed_reader?
|
266
|
+
assert !ua.other?
|
267
|
+
assert_equal nil, ua.render_engine
|
268
|
+
assert_equal nil, ua.render_engine_version
|
269
|
+
assert_equal :msnbot, ua.name
|
270
|
+
assert_equal "1.0", ua.version.full
|
271
|
+
assert_equal "1.0", ua.version.major
|
272
|
+
assert_equal [], ua.emails
|
273
|
+
assert_equal nil, ua.email
|
274
|
+
assert_equal ["http://search.msn.com/msnbot.htm"], ua.urls
|
275
|
+
assert_equal "http://search.msn.com/msnbot.htm", ua.url
|
276
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
277
|
+
assert_raise NotImplementedError do
|
278
|
+
ua.method(method).call
|
279
|
+
end
|
280
|
+
end
|
281
|
+
end
|
282
|
+
|
283
|
+
def test_msnbot_11
|
284
|
+
ua = UserAgent.new "msnbot/1.1 (+http://search.msn.com/msnbot.htm)"
|
285
|
+
assert ua.known?
|
286
|
+
assert ua.bot?
|
287
|
+
assert !ua.browser?
|
288
|
+
assert !ua.feed_reader?
|
289
|
+
assert !ua.other?
|
290
|
+
assert_equal nil, ua.render_engine
|
291
|
+
assert_equal nil, ua.render_engine_version
|
292
|
+
assert_equal :msnbot, ua.name
|
293
|
+
assert_equal "1.1", ua.version.full
|
294
|
+
assert_equal "1.1", ua.version.major
|
295
|
+
assert_equal [], ua.emails
|
296
|
+
assert_equal nil, ua.email
|
297
|
+
assert_equal ["http://search.msn.com/msnbot.htm"], ua.urls
|
298
|
+
assert_equal "http://search.msn.com/msnbot.htm", ua.url
|
299
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
300
|
+
assert_raise NotImplementedError do
|
301
|
+
ua.method(method).call
|
302
|
+
end
|
303
|
+
end
|
304
|
+
end
|
305
|
+
|
306
|
+
def test_msnbot_media_10
|
307
|
+
ua = UserAgent.new "msnbot-media/1.0 (+http://search.msn.com/msnbot.htm)"
|
308
|
+
assert ua.known?
|
309
|
+
assert ua.bot?
|
310
|
+
assert !ua.browser?
|
311
|
+
assert !ua.feed_reader?
|
312
|
+
assert !ua.other?
|
313
|
+
assert_equal nil, ua.render_engine
|
314
|
+
assert_equal nil, ua.render_engine_version
|
315
|
+
assert_equal :"msnbot-media", ua.name
|
316
|
+
assert_equal "1.0", ua.version.full
|
317
|
+
assert_equal "1.0", ua.version.major
|
318
|
+
assert_equal [], ua.emails
|
319
|
+
assert_equal nil, ua.email
|
320
|
+
assert_equal ["http://search.msn.com/msnbot.htm"], ua.urls
|
321
|
+
assert_equal "http://search.msn.com/msnbot.htm", ua.url
|
322
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
323
|
+
assert_raise NotImplementedError do
|
324
|
+
ua.method(method).call
|
325
|
+
end
|
326
|
+
end
|
327
|
+
end
|
328
|
+
|
329
|
+
def test_msnbot_media_11
|
330
|
+
ua = UserAgent.new "msnbot-media/1.1 (+http://search.msn.com/msnbot.htm)"
|
331
|
+
assert ua.known?
|
332
|
+
assert ua.bot?
|
333
|
+
assert !ua.browser?
|
334
|
+
assert !ua.feed_reader?
|
335
|
+
assert !ua.other?
|
336
|
+
assert_equal nil, ua.render_engine
|
337
|
+
assert_equal nil, ua.render_engine_version
|
338
|
+
assert_equal :"msnbot-media", ua.name
|
339
|
+
assert_equal "1.1", ua.version.major
|
340
|
+
assert_equal "1.1", ua.version.full
|
341
|
+
assert_equal [], ua.emails
|
342
|
+
assert_equal nil, ua.email
|
343
|
+
assert_equal ["http://search.msn.com/msnbot.htm"], ua.urls
|
344
|
+
assert_equal "http://search.msn.com/msnbot.htm", ua.url
|
345
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
346
|
+
assert_raise NotImplementedError do
|
347
|
+
ua.method(method).call
|
348
|
+
end
|
349
|
+
end
|
350
|
+
end
|
351
|
+
|
352
|
+
def test_seekbot_1_0
|
353
|
+
ua = UserAgent.new "Seekbot/1.0 (http://www.seekbot.net/bot.html) HTTPFetcher/2.2"
|
354
|
+
assert ua.known?
|
355
|
+
assert ua.bot?
|
356
|
+
assert !ua.browser?
|
357
|
+
assert !ua.feed_reader?
|
358
|
+
assert !ua.other?
|
359
|
+
assert_equal nil, ua.render_engine
|
360
|
+
assert_equal nil, ua.render_engine_version
|
361
|
+
assert_equal :seekbot, ua.name
|
362
|
+
assert_equal "1.0", ua.version.major
|
363
|
+
assert_equal "1.0", ua.version.full
|
364
|
+
assert_equal [], ua.emails
|
365
|
+
assert_equal nil, ua.email
|
366
|
+
assert_equal ["http://www.seekbot.net/bot.html"], ua.urls
|
367
|
+
assert_equal "http://www.seekbot.net/bot.html", ua.url
|
368
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
369
|
+
assert_raise NotImplementedError do
|
370
|
+
ua.method(method).call
|
371
|
+
end
|
372
|
+
end
|
373
|
+
end
|
374
|
+
|
375
|
+
def test_speedy_spider
|
376
|
+
ua = UserAgent.new "Speedy Spider (http://www.entireweb.com/about/search_tech/speedy_spider/)"
|
377
|
+
assert ua.known?
|
378
|
+
assert ua.bot?
|
379
|
+
assert !ua.browser?
|
380
|
+
assert !ua.feed_reader?
|
381
|
+
assert !ua.other?
|
382
|
+
assert_equal nil, ua.render_engine
|
383
|
+
assert_equal nil, ua.render_engine_version
|
384
|
+
assert_equal :"speedy spider", ua.name
|
385
|
+
assert_equal nil, ua.version
|
386
|
+
assert_equal [], ua.emails
|
387
|
+
assert_equal nil, ua.email
|
388
|
+
assert_equal ["http://www.entireweb.com/about/search_tech/speedy_spider/"], ua.urls
|
389
|
+
assert_equal "http://www.entireweb.com/about/search_tech/speedy_spider/", ua.url
|
390
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
391
|
+
assert_raise NotImplementedError do
|
392
|
+
ua.method(method).call
|
393
|
+
end
|
394
|
+
end
|
395
|
+
end
|
396
|
+
|
397
|
+
def test_unknown_bot
|
398
|
+
ua = UserAgent.new "iGELbot"
|
399
|
+
assert !ua.known?
|
400
|
+
assert ua.bot?
|
401
|
+
assert !ua.browser?
|
402
|
+
assert !ua.feed_reader?
|
403
|
+
assert !ua.other?
|
404
|
+
assert_equal nil, ua.render_engine
|
405
|
+
assert_equal nil, ua.render_engine_version
|
406
|
+
assert_equal :unknown_bot, ua.name
|
407
|
+
assert_equal nil, ua.version
|
408
|
+
assert_equal [], ua.emails
|
409
|
+
assert_equal nil, ua.email
|
410
|
+
assert_equal [], ua.urls
|
411
|
+
assert_equal nil, ua.url
|
412
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
413
|
+
assert_raise NotImplementedError do
|
414
|
+
ua.method(method).call
|
415
|
+
end
|
416
|
+
end
|
417
|
+
end
|
418
|
+
|
419
|
+
def test_unknown_bot_with_emails
|
420
|
+
ua = UserAgent.new "iGELbot (a_sf-32@sub-dom3.x.de; +info@x.de"
|
421
|
+
assert !ua.known?
|
422
|
+
assert ua.bot?
|
423
|
+
assert !ua.browser?
|
424
|
+
assert !ua.feed_reader?
|
425
|
+
assert !ua.other?
|
426
|
+
assert_equal nil, ua.render_engine
|
427
|
+
assert_equal nil, ua.render_engine_version
|
428
|
+
assert_equal :unknown_bot, ua.name
|
429
|
+
assert_equal nil, ua.version
|
430
|
+
assert_equal ["a_sf-32@sub-dom3.x.de", "info@x.de"], ua.emails
|
431
|
+
assert_equal "a_sf-32@sub-dom3.x.de", ua.email
|
432
|
+
assert_equal [], ua.urls
|
433
|
+
assert_equal nil, ua.url
|
434
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
435
|
+
assert_raise NotImplementedError do
|
436
|
+
ua.method(method).call
|
437
|
+
end
|
438
|
+
end
|
439
|
+
end
|
440
|
+
|
441
|
+
def test_unknown_bot_with_urls
|
442
|
+
ua = UserAgent.new "iGELbot (+http://xa2.a-z.43x.com:93/afzde.php?if=asdf&rm=3;x=hello+r%C3%BCdiger; www.heise.de)"
|
443
|
+
assert !ua.known?
|
444
|
+
assert ua.bot?
|
445
|
+
assert !ua.browser?
|
446
|
+
assert !ua.feed_reader?
|
447
|
+
assert !ua.other?
|
448
|
+
assert_equal nil, ua.render_engine
|
449
|
+
assert_equal nil, ua.render_engine_version
|
450
|
+
assert_equal :unknown_bot, ua.name
|
451
|
+
assert_equal nil, ua.version
|
452
|
+
assert_equal [], ua.emails
|
453
|
+
assert_equal nil, ua.email
|
454
|
+
assert_equal ["http://xa2.a-z.43x.com:93/afzde.php?if=asdf&rm=3;x=hello+r%c3%bcdiger", "www.heise.de"], ua.urls
|
455
|
+
assert_equal "http://xa2.a-z.43x.com:93/afzde.php?if=asdf&rm=3;x=hello+r%c3%bcdiger", ua.url
|
456
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
457
|
+
assert_raise NotImplementedError do
|
458
|
+
ua.method(method).call
|
459
|
+
end
|
460
|
+
end
|
461
|
+
end
|
462
|
+
|
463
|
+
def test_unknown_crawler
|
464
|
+
ua = UserAgent.new "iGELcrawler"
|
465
|
+
assert !ua.known?
|
466
|
+
assert ua.bot?
|
467
|
+
assert !ua.browser?
|
468
|
+
assert !ua.feed_reader?
|
469
|
+
assert !ua.other?
|
470
|
+
assert_equal nil, ua.render_engine
|
471
|
+
assert_equal nil, ua.render_engine_version
|
472
|
+
assert_equal :unknown_bot, ua.name
|
473
|
+
assert_equal nil, ua.version
|
474
|
+
assert_equal [], ua.emails
|
475
|
+
assert_equal nil, ua.email
|
476
|
+
assert_equal [], ua.urls
|
477
|
+
assert_equal nil, ua.url
|
478
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
479
|
+
assert_raise NotImplementedError do
|
480
|
+
ua.method(method).call
|
481
|
+
end
|
482
|
+
end
|
483
|
+
end
|
484
|
+
|
485
|
+
def test_unknown_spider
|
486
|
+
ua = UserAgent.new "iGELspider"
|
487
|
+
assert !ua.known?
|
488
|
+
assert ua.bot?
|
489
|
+
assert !ua.browser?
|
490
|
+
assert !ua.feed_reader?
|
491
|
+
assert !ua.other?
|
492
|
+
assert_equal nil, ua.render_engine
|
493
|
+
assert_equal nil, ua.render_engine_version
|
494
|
+
assert_equal :unknown_bot, ua.name
|
495
|
+
assert_equal nil, ua.version
|
496
|
+
assert_equal [], ua.emails
|
497
|
+
assert_equal nil, ua.email
|
498
|
+
assert_equal [], ua.urls
|
499
|
+
assert_equal nil, ua.url
|
500
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
501
|
+
assert_raise NotImplementedError do
|
502
|
+
ua.method(method).call
|
503
|
+
end
|
504
|
+
end
|
505
|
+
end
|
506
|
+
|
507
|
+
def test_yahoo_slurp
|
508
|
+
ua = UserAgent.new "Mozilla/5.0 (compatible; Yahoo! Slurp; http://help.yahoo.com/help/us/ysearch/slurp)"
|
509
|
+
assert ua.known?
|
510
|
+
assert ua.bot?
|
511
|
+
assert !ua.browser?
|
512
|
+
assert !ua.feed_reader?
|
513
|
+
assert !ua.other?
|
514
|
+
assert_equal nil, ua.render_engine
|
515
|
+
assert_equal nil, ua.render_engine_version
|
516
|
+
assert_equal :"yahoo! slurp", ua.name
|
517
|
+
assert_equal nil, ua.version
|
518
|
+
assert_equal [], ua.emails
|
519
|
+
assert_equal nil, ua.email
|
520
|
+
assert_equal ["http://help.yahoo.com/help/us/ysearch/slurp"], ua.urls
|
521
|
+
assert_equal "http://help.yahoo.com/help/us/ysearch/slurp", ua.url
|
522
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
523
|
+
assert_raise NotImplementedError do
|
524
|
+
ua.method(method).call
|
525
|
+
end
|
526
|
+
end
|
527
|
+
end
|
528
|
+
|
529
|
+
def test_yahoo_slurp_3_0
|
530
|
+
ua = UserAgent.new "Mozilla/5.0 (compatible; Yahoo! Slurp/3.0; http://help.yahoo.com/help/us/ysearch/slurp)"
|
531
|
+
assert ua.known?
|
532
|
+
assert ua.bot?
|
533
|
+
assert !ua.browser?
|
534
|
+
assert !ua.feed_reader?
|
535
|
+
assert !ua.other?
|
536
|
+
assert_equal nil, ua.render_engine
|
537
|
+
assert_equal nil, ua.render_engine_version
|
538
|
+
assert_equal :"yahoo! slurp", ua.name
|
539
|
+
assert_equal "3.0", ua.version.full
|
540
|
+
assert_equal "3.0", ua.version.major
|
541
|
+
assert_equal [], ua.emails
|
542
|
+
assert_equal nil, ua.email
|
543
|
+
assert_equal ["http://help.yahoo.com/help/us/ysearch/slurp"], ua.urls
|
544
|
+
assert_equal "http://help.yahoo.com/help/us/ysearch/slurp", ua.url
|
545
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
546
|
+
assert_raise NotImplementedError do
|
547
|
+
ua.method(method).call
|
548
|
+
end
|
549
|
+
end
|
550
|
+
end
|
551
|
+
|
552
|
+
def test_yeti_0_01
|
553
|
+
ua = UserAgent.new "Yeti/0.01 (nhn/1noon, yetibot@naver.com, check robots.txt daily and follow it)"
|
554
|
+
assert ua.known?
|
555
|
+
assert ua.bot?
|
556
|
+
assert !ua.browser?
|
557
|
+
assert !ua.feed_reader?
|
558
|
+
assert !ua.other?
|
559
|
+
assert_equal nil, ua.render_engine
|
560
|
+
assert_equal nil, ua.render_engine_version
|
561
|
+
assert_equal :yeti, ua.name
|
562
|
+
assert_equal "0.01", ua.version.full
|
563
|
+
assert_equal "0.0", ua.version.major
|
564
|
+
assert_equal ["yetibot@naver.com"], ua.emails
|
565
|
+
assert_equal "yetibot@naver.com", ua.email
|
566
|
+
assert_equal [], ua.urls
|
567
|
+
assert_equal nil, ua.url
|
568
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
569
|
+
assert_raise NotImplementedError do
|
570
|
+
ua.method(method).call
|
571
|
+
end
|
572
|
+
end
|
573
|
+
end
|
574
|
+
|
575
|
+
def test_yeti_1_0
|
576
|
+
ua = UserAgent.new "Yeti/1.0 (+http://help.naver.com/robots/)"
|
577
|
+
assert ua.known?
|
578
|
+
assert ua.bot?
|
579
|
+
assert !ua.browser?
|
580
|
+
assert !ua.feed_reader?
|
581
|
+
assert !ua.other?
|
582
|
+
assert_equal nil, ua.render_engine
|
583
|
+
assert_equal nil, ua.render_engine_version
|
584
|
+
assert_equal :yeti, ua.name
|
585
|
+
assert_equal "1.0", ua.version.full
|
586
|
+
assert_equal "1.0", ua.version.major
|
587
|
+
assert_equal [], ua.emails
|
588
|
+
assert_equal nil, ua.email
|
589
|
+
assert_equal ["http://help.naver.com/robots/"], ua.urls
|
590
|
+
assert_equal "http://help.naver.com/robots/", ua.url
|
591
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
592
|
+
assert_raise NotImplementedError do
|
593
|
+
ua.method(method).call
|
594
|
+
end
|
595
|
+
end
|
596
|
+
end
|
597
|
+
|
598
|
+
def test_yeti_1_0_with_company
|
599
|
+
ua = UserAgent.new "Yeti/1.0 (NHN Corp.; http://help.naver.com/robots/)"
|
600
|
+
assert ua.known?
|
601
|
+
assert ua.bot?
|
602
|
+
assert !ua.browser?
|
603
|
+
assert !ua.feed_reader?
|
604
|
+
assert !ua.other?
|
605
|
+
assert_equal nil, ua.render_engine
|
606
|
+
assert_equal nil, ua.render_engine_version
|
607
|
+
assert_equal :yeti, ua.name
|
608
|
+
assert_equal "1.0", ua.version.full
|
609
|
+
assert_equal "1.0", ua.version.major
|
610
|
+
assert_equal [], ua.emails
|
611
|
+
assert_equal nil, ua.email
|
612
|
+
assert_equal ["http://help.naver.com/robots/"], ua.urls
|
613
|
+
assert_equal "http://help.naver.com/robots/", ua.url
|
614
|
+
[:architecture, :dotnet_versions, :os, :os_type, :os_version, :ui_lang, :ui_lang_country, :vendor].each do |method|
|
615
|
+
assert_raise NotImplementedError do
|
616
|
+
ua.method(method).call
|
617
|
+
end
|
618
|
+
end
|
619
|
+
end
|
620
|
+
end
|