web_stat 0.3.17 → 0.4.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.ruby-version +1 -1
- data/Dockerfile +1 -1
- data/Gemfile.lock +15 -13
- data/README.md +0 -23
- data/lib/web_stat/config/web_stat.yml +5 -1
- data/lib/web_stat/fetch.rb +14 -11
- data/lib/web_stat/version.rb +1 -1
- data/spec/web_stat/configure_spec.rb +17 -2
- data/web_stat.gemspec +2 -0
- metadata +31 -3
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 871e9eb97dc238635bd6a46571e86f6a9548c0ca32b6bf3576017ad71c81394f
|
4
|
+
data.tar.gz: 7ed7e9750fc2030f486c14f4de9441ba2c2c98a8357565bf708e6d43f7ae620e
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 963dca0c991935d084ba40d2ff35de87c55d1584f9d39f3c54c7df0d89aa701ddef21014c1ea6b4a0a83d118b9abd6daf98cd084663992d0a3ae09c9b92e4267
|
7
|
+
data.tar.gz: df4c99e70693d004d312219dd2550cd7a3b39e4eb64ac101cd0654a886ee0d6a4d667a82dbdb15574f825aa4cc71fb3d34d9e10f080a977fe026c6f7b7d6006e
|
data/.ruby-version
CHANGED
@@ -1 +1 @@
|
|
1
|
-
|
1
|
+
3.0.0
|
data/Dockerfile
CHANGED
@@ -1,5 +1,5 @@
|
|
1
1
|
# Define base image, you can use --build-arg
|
2
|
-
ARG base_image="newsdict/rails:ubuntu20.10_nvmv0.
|
2
|
+
ARG base_image="newsdict/rails:ubuntu20.10_nvmv0.37.0_nodev15.2.1_rubyv3.0.0_sasscv2.4.0_ffiv1.13.1_chromedriver"
|
3
3
|
FROM $base_image
|
4
4
|
|
5
5
|
# Set locale
|
data/Gemfile.lock
CHANGED
@@ -1,16 +1,18 @@
|
|
1
1
|
PATH
|
2
2
|
remote: .
|
3
3
|
specs:
|
4
|
-
web_stat (0.
|
4
|
+
web_stat (0.4.2)
|
5
5
|
bundler (>= 2.0.2)
|
6
6
|
cld (>= 0.8.0)
|
7
7
|
mechanize (>= 2.7)
|
8
8
|
natto (>= 1.1.2)
|
9
9
|
nokogiri (>= 1.10.4)
|
10
10
|
pdf-reader (= 2.4.0)
|
11
|
+
rexml (>= 3.2.4)
|
11
12
|
ruby-readability (>= 0.7)
|
12
13
|
sanitize (>= 5.0.0)
|
13
14
|
selenium-webdriver (= 3.142.7)
|
15
|
+
webrick (>= 1.7.0)
|
14
16
|
|
15
17
|
GEM
|
16
18
|
remote: https://rubygems.org/
|
@@ -24,14 +26,13 @@ GEM
|
|
24
26
|
cld (0.8.0)
|
25
27
|
ffi
|
26
28
|
coderay (1.1.3)
|
27
|
-
connection_pool (2.2.3)
|
28
29
|
crack (0.4.3)
|
29
30
|
safe_yaml (~> 1.0.0)
|
30
31
|
crass (1.0.6)
|
31
32
|
diff-lcs (1.3)
|
32
33
|
domain_name (0.5.20190701)
|
33
34
|
unf (>= 0.0.5, < 1.0.0)
|
34
|
-
ffi (1.
|
35
|
+
ffi (1.14.2)
|
35
36
|
guess_html_encoding (0.0.11)
|
36
37
|
hashdiff (1.0.1)
|
37
38
|
hashery (2.1.2)
|
@@ -49,16 +50,14 @@ GEM
|
|
49
50
|
method_source (1.0.0)
|
50
51
|
mime-types (3.3.1)
|
51
52
|
mime-types-data (~> 3.2015)
|
52
|
-
mime-types-data (3.2020.
|
53
|
-
mini_portile2 (2.4.0)
|
53
|
+
mime-types-data (3.2020.1104)
|
54
54
|
natto (1.2.0)
|
55
55
|
ffi (>= 1.9.0)
|
56
56
|
net-http-digest_auth (1.4.1)
|
57
|
-
net-http-persistent (
|
58
|
-
|
59
|
-
|
60
|
-
|
61
|
-
nokogumbo (2.0.2)
|
57
|
+
net-http-persistent (2.9.4)
|
58
|
+
nokogiri (1.11.1-x86_64-linux)
|
59
|
+
racc (~> 1.4)
|
60
|
+
nokogumbo (2.0.4)
|
62
61
|
nokogiri (~> 1.8, >= 1.8.4)
|
63
62
|
ntlm-http (0.1.1)
|
64
63
|
pdf-reader (2.4.0)
|
@@ -74,7 +73,9 @@ GEM
|
|
74
73
|
byebug (~> 11.0)
|
75
74
|
pry (~> 0.13.0)
|
76
75
|
public_suffix (4.0.5)
|
76
|
+
racc (1.5.2)
|
77
77
|
rake (13.0.1)
|
78
|
+
rexml (3.2.4)
|
78
79
|
rspec (3.9.0)
|
79
80
|
rspec-core (~> 3.9.0)
|
80
81
|
rspec-expectations (~> 3.9.0)
|
@@ -94,14 +95,14 @@ GEM
|
|
94
95
|
nokogiri (>= 1.6.0)
|
95
96
|
rubyzip (2.3.0)
|
96
97
|
safe_yaml (1.0.5)
|
97
|
-
sanitize (5.2.
|
98
|
+
sanitize (5.2.2)
|
98
99
|
crass (~> 1.0.2)
|
99
100
|
nokogiri (>= 1.8.0)
|
100
101
|
nokogumbo (~> 2.0)
|
101
102
|
selenium-webdriver (3.142.7)
|
102
103
|
childprocess (>= 0.5, < 4.0)
|
103
104
|
rubyzip (>= 1.2.2)
|
104
|
-
ttfunk (1.
|
105
|
+
ttfunk (1.7.0)
|
105
106
|
unf (0.1.4)
|
106
107
|
unf_ext
|
107
108
|
unf_ext (0.0.7.7)
|
@@ -109,6 +110,7 @@ GEM
|
|
109
110
|
addressable (>= 2.3.6)
|
110
111
|
crack (>= 0.3.2)
|
111
112
|
hashdiff (>= 0.4.0, < 2.0.0)
|
113
|
+
webrick (1.7.0)
|
112
114
|
webrobots (0.1.2)
|
113
115
|
|
114
116
|
PLATFORMS
|
@@ -123,4 +125,4 @@ DEPENDENCIES
|
|
123
125
|
webmock (>= 3.8.3)
|
124
126
|
|
125
127
|
BUNDLED WITH
|
126
|
-
2.
|
128
|
+
2.2.4
|
data/README.md
CHANGED
@@ -1,30 +1,7 @@
|
|
1
|
-
# !!!!! Precautions when using with Rails !!!!!
|
2
|
-
|
3
|
-
Write this line your Gemfile.
|
4
|
-
```
|
5
|
-
gem "final_redirect_url", :git => "git@github.com:yubele/final_redirect_url"
|
6
|
-
```
|
7
|
-
|
8
1
|
# WebStat
|
9
2
|
|
10
3
|
Fetch the web pages and stat.
|
11
4
|
|
12
|
-
## Requirements
|
13
|
-
|
14
|
-
- [MeCab _0.996_](http://taku910.github.io/mecab/#download)
|
15
|
-
- add runtime dependency
|
16
|
-
- "bundler", "~> 2.0"
|
17
|
-
- "nokogiri", "~> 1.10"
|
18
|
-
- "mechanize", "~> 2.7"
|
19
|
-
- "ruby-readability", "~> 0.7"
|
20
|
-
- "final_redirect_url", "~> 0.1.0"
|
21
|
-
- "natto", "~> 1.1.2"
|
22
|
-
- add development dependency
|
23
|
-
- "rake", "~> 10.0"
|
24
|
-
- "rspec", "~> 3.0"
|
25
|
-
- "rake", "~> 10.0"
|
26
|
-
- "rspec", "~> 3.0"
|
27
|
-
|
28
5
|
### Install mecab
|
29
6
|
|
30
7
|
$ sudo apt install mecab-ipadic-utf8 libmecab
|
@@ -1,7 +1,7 @@
|
|
1
1
|
development: &development
|
2
2
|
# Minimum number of characters to detect meta title
|
3
3
|
min_length_of_meta_title: 10
|
4
|
-
# Split regular expression for titles
|
4
|
+
# Split regular expression for titles
|
5
5
|
regex_to_sprit_title: '\||-|:|||:|〜|\~| – '
|
6
6
|
# User Agent
|
7
7
|
user_agent: "web_stat gem agent"
|
@@ -14,6 +14,10 @@ development: &development
|
|
14
14
|
- '//img/@src'
|
15
15
|
userdic: ""
|
16
16
|
use_chromedirver: false
|
17
|
+
thumbnail_regex:
|
18
|
+
youtube:
|
19
|
+
- '%r{^https://www.youtube.com/watch\?v=([^&]+)}'
|
20
|
+
- 'http://img.youtube.com/vi/\1/default.jpg'
|
17
21
|
test:
|
18
22
|
<<: *development
|
19
23
|
production:
|
data/lib/web_stat/fetch.rb
CHANGED
@@ -1,11 +1,5 @@
|
|
1
1
|
module WebStat
|
2
2
|
class Fetch
|
3
|
-
THUMBNAIL_REGEXS = {
|
4
|
-
:youtube => [
|
5
|
-
%r{^https://www.youtube.com/watch\?v=([^&]+)},
|
6
|
-
'http://img.youtube.com/vi/\1/default.jpg'
|
7
|
-
]
|
8
|
-
}
|
9
3
|
attr_accessor :url, :html, :nokogiri, :userdic, :status
|
10
4
|
# Get title
|
11
5
|
# @return [String] title
|
@@ -40,7 +34,7 @@ module WebStat
|
|
40
34
|
end
|
41
35
|
# Get main section
|
42
36
|
def content
|
43
|
-
Sanitize.clean(Readability::Document.new(@nokogiri.at('body')).content)
|
37
|
+
Sanitize.clean(Readability::Document.new(@nokogiri.at('body').to_s).content)
|
44
38
|
end
|
45
39
|
|
46
40
|
# Get temporary path of image
|
@@ -54,12 +48,12 @@ module WebStat
|
|
54
48
|
end
|
55
49
|
end
|
56
50
|
# If there is a thumbnail rule, apply it.
|
57
|
-
|
51
|
+
WebStat::Configure.get["thumbnail_regex"].each do |provider, v|
|
58
52
|
if @url.match(v[0])
|
59
53
|
return @url.gsub(v[0], v[1])
|
60
54
|
end
|
61
55
|
end
|
62
|
-
readability_content = ::Nokogiri::HTML(Readability::Document.new(@nokogiri.at('body')).content)
|
56
|
+
readability_content = ::Nokogiri::HTML(Readability::Document.new(@nokogiri.at('body').to_s).content)
|
63
57
|
if (path.nil? || path.empty?) && readability_content.xpath('//img').first
|
64
58
|
path = readability_content.xpath('//img').first.attr('src')
|
65
59
|
end
|
@@ -83,7 +77,7 @@ module WebStat
|
|
83
77
|
File.open(tmp_file, "w+b") do |_file|
|
84
78
|
if image.class == Mechanize::File
|
85
79
|
_file.puts(image.body)
|
86
|
-
elsif image.respond_to?(:
|
80
|
+
elsif image.respond_to?(:body_io)
|
87
81
|
_file.puts(image.body_io.read)
|
88
82
|
end
|
89
83
|
end
|
@@ -102,7 +96,16 @@ module WebStat
|
|
102
96
|
raise Mechanize::RobotsDisallowedError.new(url)
|
103
97
|
end
|
104
98
|
if WebStat::Configure.get["use_chromedirver"]
|
105
|
-
|
99
|
+
begin
|
100
|
+
body = WebStat::WebDriverHelper.get_source(url)
|
101
|
+
rescue Selenium::WebDriver::Error::UnknownError => e
|
102
|
+
document = mech.agent.get(url, [], nil, { 'Accept-Language' => 'ja'})
|
103
|
+
if document.class == Mechanize::File
|
104
|
+
body = document.body
|
105
|
+
else
|
106
|
+
body = document.body.encode('UTF-8', document.encoding)
|
107
|
+
end
|
108
|
+
end
|
106
109
|
@status = 200
|
107
110
|
else
|
108
111
|
document = mech.get(url, [], nil, { 'Accept-Language' => 'ja'})
|
data/lib/web_stat/version.rb
CHANGED
@@ -3,11 +3,26 @@ RSpec.describe WebStat::Configure do
|
|
3
3
|
configure = WebStat::Configure.get
|
4
4
|
expect(configure).not_to eq nil
|
5
5
|
end
|
6
|
-
|
6
|
+
|
7
7
|
it "Readable Config" do
|
8
8
|
config = WebStat::Configure.get
|
9
|
-
|
9
|
+
|
10
10
|
expect(config["min_length_of_meta_title"]).to eq 10
|
11
11
|
expect(config["regex_to_sprit_title"]).to eq '\||-|:|||:|〜|\~| – '
|
12
12
|
end
|
13
|
+
|
14
|
+
it "Get thumbnail_regex.youtube." do
|
15
|
+
config = WebStat::Configure.get
|
16
|
+
expect(config["thumbnail_regex"]["yotube"].nil?).to eq true
|
17
|
+
expect(config["thumbnail_regex"]["youtube"].count).to eq 2
|
18
|
+
end
|
19
|
+
|
20
|
+
it "Match youtube url." do
|
21
|
+
sample_url = "https://www.youtube.com/watch?v=aChpsuUffUM"
|
22
|
+
WebStat::Configure.get["thumbnail_regex"].each do |provider, v|
|
23
|
+
if sample_url.match(v[0])
|
24
|
+
expect(sample_url.gsub(v[0], v[1])).to eq 'http://img.youtube.com/vi/aChpsuUffUM/default.jpg'
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
13
28
|
end
|
data/web_stat.gemspec
CHANGED
@@ -29,6 +29,8 @@ Gem::Specification.new do |spec|
|
|
29
29
|
spec.add_runtime_dependency "cld", ">= 0.8.0"
|
30
30
|
spec.add_runtime_dependency "selenium-webdriver", "= 3.142.7"
|
31
31
|
spec.add_runtime_dependency "pdf-reader", "2.4.0"
|
32
|
+
spec.add_runtime_dependency "webrick", ">= 1.7.0"
|
33
|
+
spec.add_runtime_dependency "rexml", ">= 3.2.4"
|
32
34
|
|
33
35
|
spec.add_development_dependency "rake", ">= 10.0"
|
34
36
|
spec.add_development_dependency "rspec", ">= 3.0"
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: web_stat
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.
|
4
|
+
version: 0.4.2
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- yusuke abe
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date:
|
11
|
+
date: 2021-01-31 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: bundler
|
@@ -136,6 +136,34 @@ dependencies:
|
|
136
136
|
- - '='
|
137
137
|
- !ruby/object:Gem::Version
|
138
138
|
version: 2.4.0
|
139
|
+
- !ruby/object:Gem::Dependency
|
140
|
+
name: webrick
|
141
|
+
requirement: !ruby/object:Gem::Requirement
|
142
|
+
requirements:
|
143
|
+
- - ">="
|
144
|
+
- !ruby/object:Gem::Version
|
145
|
+
version: 1.7.0
|
146
|
+
type: :runtime
|
147
|
+
prerelease: false
|
148
|
+
version_requirements: !ruby/object:Gem::Requirement
|
149
|
+
requirements:
|
150
|
+
- - ">="
|
151
|
+
- !ruby/object:Gem::Version
|
152
|
+
version: 1.7.0
|
153
|
+
- !ruby/object:Gem::Dependency
|
154
|
+
name: rexml
|
155
|
+
requirement: !ruby/object:Gem::Requirement
|
156
|
+
requirements:
|
157
|
+
- - ">="
|
158
|
+
- !ruby/object:Gem::Version
|
159
|
+
version: 3.2.4
|
160
|
+
type: :runtime
|
161
|
+
prerelease: false
|
162
|
+
version_requirements: !ruby/object:Gem::Requirement
|
163
|
+
requirements:
|
164
|
+
- - ">="
|
165
|
+
- !ruby/object:Gem::Version
|
166
|
+
version: 3.2.4
|
139
167
|
- !ruby/object:Gem::Dependency
|
140
168
|
name: rake
|
141
169
|
requirement: !ruby/object:Gem::Requirement
|
@@ -274,7 +302,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
274
302
|
- !ruby/object:Gem::Version
|
275
303
|
version: '0'
|
276
304
|
requirements: []
|
277
|
-
rubygems_version: 3.
|
305
|
+
rubygems_version: 3.2.3
|
278
306
|
signing_key:
|
279
307
|
specification_version: 4
|
280
308
|
summary: Get the status of the web pages.
|