link_thumbnailer 3.3.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +19 -0
- data/.rspec +2 -0
- data/.ruby-version +1 -0
- data/.travis.yml +5 -0
- data/CHANGELOG.md +334 -0
- data/Gemfile +12 -0
- data/LICENSE.txt +22 -0
- data/README.md +210 -0
- data/Rakefile +9 -0
- data/lib/generators/link_thumbnailer/install_generator.rb +17 -0
- data/lib/generators/templates/initializer.rb +89 -0
- data/lib/link_thumbnailer.rb +38 -0
- data/lib/link_thumbnailer/configuration.rb +72 -0
- data/lib/link_thumbnailer/exceptions.rb +11 -0
- data/lib/link_thumbnailer/grader.rb +43 -0
- data/lib/link_thumbnailer/graders/base.rb +39 -0
- data/lib/link_thumbnailer/graders/html_attribute.rb +48 -0
- data/lib/link_thumbnailer/graders/length.rb +37 -0
- data/lib/link_thumbnailer/graders/link_density.rb +20 -0
- data/lib/link_thumbnailer/graders/position.rb +13 -0
- data/lib/link_thumbnailer/image_comparator.rb +26 -0
- data/lib/link_thumbnailer/image_comparators/base.rb +19 -0
- data/lib/link_thumbnailer/image_comparators/size.rb +13 -0
- data/lib/link_thumbnailer/image_parser.rb +62 -0
- data/lib/link_thumbnailer/image_validator.rb +32 -0
- data/lib/link_thumbnailer/model.rb +20 -0
- data/lib/link_thumbnailer/models/description.rb +37 -0
- data/lib/link_thumbnailer/models/favicon.rb +27 -0
- data/lib/link_thumbnailer/models/image.rb +56 -0
- data/lib/link_thumbnailer/models/title.rb +22 -0
- data/lib/link_thumbnailer/models/video.rb +44 -0
- data/lib/link_thumbnailer/models/website.rb +54 -0
- data/lib/link_thumbnailer/page.rb +43 -0
- data/lib/link_thumbnailer/parser.rb +15 -0
- data/lib/link_thumbnailer/processor.rb +128 -0
- data/lib/link_thumbnailer/railtie.rb +6 -0
- data/lib/link_thumbnailer/response.rb +39 -0
- data/lib/link_thumbnailer/scraper.rb +62 -0
- data/lib/link_thumbnailer/scrapers/base.rb +69 -0
- data/lib/link_thumbnailer/scrapers/default/base.rb +12 -0
- data/lib/link_thumbnailer/scrapers/default/description.rb +49 -0
- data/lib/link_thumbnailer/scrapers/default/favicon.rb +38 -0
- data/lib/link_thumbnailer/scrapers/default/images.rb +78 -0
- data/lib/link_thumbnailer/scrapers/default/title.rb +27 -0
- data/lib/link_thumbnailer/scrapers/default/videos.rb +18 -0
- data/lib/link_thumbnailer/scrapers/opengraph/base.rb +45 -0
- data/lib/link_thumbnailer/scrapers/opengraph/description.rb +12 -0
- data/lib/link_thumbnailer/scrapers/opengraph/favicon.rb +17 -0
- data/lib/link_thumbnailer/scrapers/opengraph/image.rb +107 -0
- data/lib/link_thumbnailer/scrapers/opengraph/images.rb +18 -0
- data/lib/link_thumbnailer/scrapers/opengraph/title.rb +12 -0
- data/lib/link_thumbnailer/scrapers/opengraph/video.rb +115 -0
- data/lib/link_thumbnailer/scrapers/opengraph/videos.rb +18 -0
- data/lib/link_thumbnailer/uri.rb +20 -0
- data/lib/link_thumbnailer/version.rb +5 -0
- data/lib/link_thumbnailer/video_parser.rb +47 -0
- data/link_thumbnailer.gemspec +29 -0
- data/spec/configuration_spec.rb +61 -0
- data/spec/fixture_spec.rb +114 -0
- data/spec/fixtures/bar.png +2907 -0
- data/spec/fixtures/default_from_body.html +13 -0
- data/spec/fixtures/default_from_meta.html +12 -0
- data/spec/fixtures/foo.png +0 -0
- data/spec/fixtures/google_shift_jis.html +6 -0
- data/spec/fixtures/google_utf8.html +6 -0
- data/spec/fixtures/og_not_valid_example.html +12 -0
- data/spec/fixtures/og_valid_example.html +18 -0
- data/spec/fixtures/og_valid_multi_image_example.html +13 -0
- data/spec/fixtures/og_valid_multi_video_example.html +13 -0
- data/spec/grader_spec.rb +27 -0
- data/spec/graders/base_spec.rb +14 -0
- data/spec/graders/html_attribute_spec.rb +50 -0
- data/spec/graders/length_spec.rb +93 -0
- data/spec/graders/link_density_spec.rb +52 -0
- data/spec/graders/position_spec.rb +49 -0
- data/spec/image_comparators/size_spec.rb +58 -0
- data/spec/image_validator_spec.rb +37 -0
- data/spec/model_spec.rb +27 -0
- data/spec/models/description_spec.rb +66 -0
- data/spec/models/favicon_spec.rb +12 -0
- data/spec/models/image_spec.rb +95 -0
- data/spec/models/title_spec.rb +26 -0
- data/spec/models/video_spec.rb +49 -0
- data/spec/models/website_spec.rb +51 -0
- data/spec/page_spec.rb +28 -0
- data/spec/processor_spec.rb +410 -0
- data/spec/response_spec.rb +62 -0
- data/spec/scraper_spec.rb +70 -0
- data/spec/scrapers/base_spec.rb +69 -0
- data/spec/scrapers/opengraph/base_spec.rb +96 -0
- data/spec/spec_helper.rb +11 -0
- data/spec/uri_spec.rb +44 -0
- data/spec/video_parser_spec.rb +148 -0
- metadata +271 -0
data/Rakefile
ADDED
@@ -0,0 +1,17 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module LinkThumbnailer
|
4
|
+
module Generators
|
5
|
+
class InstallGenerator < ::Rails::Generators::Base
|
6
|
+
|
7
|
+
source_root File.expand_path('../../templates', __FILE__)
|
8
|
+
|
9
|
+
desc 'Creates a LinkThumbnailer initializer for your application.'
|
10
|
+
|
11
|
+
def copy_initializer
|
12
|
+
template 'initializer.rb', 'config/initializers/link_thumbnailer.rb'
|
13
|
+
end
|
14
|
+
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
@@ -0,0 +1,89 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
# Use this hook to configure LinkThumbnailer bahaviors.
|
4
|
+
LinkThumbnailer.configure do |config|
|
5
|
+
# Numbers of redirects before raising an exception when trying to parse given url.
|
6
|
+
#
|
7
|
+
# config.redirect_limit = 3
|
8
|
+
|
9
|
+
# Set user agent
|
10
|
+
#
|
11
|
+
# config.user_agent = 'link_thumbnailer'
|
12
|
+
|
13
|
+
# Enable or disable SSL verification
|
14
|
+
#
|
15
|
+
# config.verify_ssl = true
|
16
|
+
|
17
|
+
# The amount of time in seconds to wait for a connection to be opened.
|
18
|
+
# If the HTTP object cannot open a connection in this many seconds,
|
19
|
+
# it raises a Net::OpenTimeout exception.
|
20
|
+
#
|
21
|
+
# See http://www.ruby-doc.org/stdlib-2.1.1/libdoc/net/http/rdoc/Net/HTTP.html#open_timeout
|
22
|
+
#
|
23
|
+
# config.http_open_timeout = 5
|
24
|
+
|
25
|
+
# List of blacklisted urls you want to skip when searching for images.
|
26
|
+
#
|
27
|
+
# config.blacklist_urls = [
|
28
|
+
# %r{^http://ad\.doubleclick\.net/},
|
29
|
+
# %r{^http://b\.scorecardresearch\.com/},
|
30
|
+
# %r{^http://pixel\.quantserve\.com/},
|
31
|
+
# %r{^http://s7\.addthis\.com/}
|
32
|
+
# ]
|
33
|
+
|
34
|
+
# List of attributes you want LinkThumbnailer to fetch on a website.
|
35
|
+
#
|
36
|
+
# config.attributes = [:title, :images, :description, :videos, :favicon]
|
37
|
+
|
38
|
+
# List of procedures used to rate the website description. Add you custom class
|
39
|
+
# here. See wiki for more details on how to build your own graders.
|
40
|
+
#
|
41
|
+
# config.graders = [
|
42
|
+
# ->(description) { ::LinkThumbnailer::Graders::Length.new(description) },
|
43
|
+
# ->(description) { ::LinkThumbnailer::Graders::HtmlAttribute.new(description, :class) },
|
44
|
+
# ->(description) { ::LinkThumbnailer::Graders::HtmlAttribute.new(description, :id) },
|
45
|
+
# ->(description) { ::LinkThumbnailer::Graders::Position.new(description, weight: 3) },
|
46
|
+
# ->(description) { ::LinkThumbnailer::Graders::LinkDensity.new(description) }
|
47
|
+
# ]
|
48
|
+
|
49
|
+
# Minimum description length for a website.
|
50
|
+
#
|
51
|
+
# config.description_min_length = 25
|
52
|
+
|
53
|
+
# Regex of words considered positive to rate website description.
|
54
|
+
#
|
55
|
+
# config.positive_regex = /article|body|content|entry|hentry|main|page|pagination|post|text|blog|story/i
|
56
|
+
|
57
|
+
# Regex of words considered negative to rate website description.
|
58
|
+
#
|
59
|
+
# config.negative_regex = /combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget|modal/i
|
60
|
+
|
61
|
+
# Numbers of images to fetch. Fetching too many images will be slow.
|
62
|
+
# Note that LinkThumbnailer will only sort fetched images between each other.
|
63
|
+
# Meaning that they could be a "better" image on the page.
|
64
|
+
#
|
65
|
+
# config.image_limit = 5
|
66
|
+
|
67
|
+
# Whether you want LinkThumbnailer to return image size and type or not.
|
68
|
+
# Setting this value to false will increase performance since for each images, LinkThumbnailer
|
69
|
+
# does not have to fetch its size and type.
|
70
|
+
#
|
71
|
+
# config.image_stats = true
|
72
|
+
|
73
|
+
# Whether you want LinkThumbnailer to raise an exception if the Content-Type of the HTTP request
|
74
|
+
# is not an html or xml.
|
75
|
+
#
|
76
|
+
# config.raise_on_invalid_format = false
|
77
|
+
|
78
|
+
# Sets number of concurrent http connections that can be opened to fetch images informations such as size and type.
|
79
|
+
#
|
80
|
+
# config.max_concurrency = 20
|
81
|
+
|
82
|
+
# Defines the strategies to use to scrap the website. See the [Open Graph Protocol](http://ogp.me/) for more information.
|
83
|
+
#
|
84
|
+
# config.scrapers = [:opengraph, :default]
|
85
|
+
|
86
|
+
# Sets the default encoding.
|
87
|
+
#
|
88
|
+
# config.encoding = 'utf-8'
|
89
|
+
end
|
@@ -0,0 +1,38 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require 'json'
|
4
|
+
require 'link_thumbnailer/version'
|
5
|
+
require 'link_thumbnailer/configuration'
|
6
|
+
require 'link_thumbnailer/exceptions'
|
7
|
+
require 'link_thumbnailer/page'
|
8
|
+
|
9
|
+
module LinkThumbnailer
|
10
|
+
|
11
|
+
class << self
|
12
|
+
|
13
|
+
attr_reader :page
|
14
|
+
|
15
|
+
def generate(url, options = {})
|
16
|
+
@page = ::LinkThumbnailer::Page.new(url, options)
|
17
|
+
|
18
|
+
page.generate
|
19
|
+
end
|
20
|
+
|
21
|
+
end
|
22
|
+
|
23
|
+
end
|
24
|
+
|
25
|
+
begin
|
26
|
+
require 'rails'
|
27
|
+
rescue LoadError
|
28
|
+
end
|
29
|
+
|
30
|
+
$stderr.puts <<-EOC if !defined?(Rails)
|
31
|
+
warning: no framework detected.
|
32
|
+
|
33
|
+
Your Gemfile might not be configured properly.
|
34
|
+
---- e.g. ----
|
35
|
+
Rails:
|
36
|
+
gem 'link_thumbnailer'
|
37
|
+
|
38
|
+
EOC
|
@@ -0,0 +1,72 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module LinkThumbnailer
|
4
|
+
|
5
|
+
# Access point for the gem configurations.
|
6
|
+
#
|
7
|
+
# @return [LinkThumbnailer::Configuration] a configuration instance.
|
8
|
+
def self.config
|
9
|
+
@config ||= Configuration.new
|
10
|
+
end
|
11
|
+
|
12
|
+
# Configure hook used in the gem initializer. Convinient way to set all the
|
13
|
+
# gem configurations.
|
14
|
+
#
|
15
|
+
# @example inside config/initializers/link_thumbnaler.rb
|
16
|
+
# LinkThumbnailer.configure do |config|
|
17
|
+
# config.user_agent = 'link_thumbnailer'
|
18
|
+
# end
|
19
|
+
#
|
20
|
+
# @return [void]
|
21
|
+
def self.configure
|
22
|
+
yield config if block_given?
|
23
|
+
end
|
24
|
+
|
25
|
+
class Configuration
|
26
|
+
|
27
|
+
attr_accessor :redirect_limit, :blacklist_urls, :user_agent,
|
28
|
+
:verify_ssl, :http_open_timeout, :http_read_timeout, :attributes,
|
29
|
+
:graders, :description_min_length, :positive_regex, :negative_regex,
|
30
|
+
:image_limit, :image_stats, :raise_on_invalid_format, :max_concurrency,
|
31
|
+
:scrapers, :http_override_headers, :encoding
|
32
|
+
|
33
|
+
alias_method :http_timeout, :http_open_timeout
|
34
|
+
alias_method :http_timeout=, :http_open_timeout=
|
35
|
+
|
36
|
+
# Create a new instance.
|
37
|
+
#
|
38
|
+
# @return [LinkThumbnailer::Configuration]
|
39
|
+
def initialize
|
40
|
+
@redirect_limit = 3
|
41
|
+
@user_agent = 'link_thumbnailer'
|
42
|
+
@verify_ssl = true
|
43
|
+
@http_open_timeout = 5
|
44
|
+
@http_read_timeout = 5
|
45
|
+
@blacklist_urls = [
|
46
|
+
%r{^http://ad\.doubleclick\.net/},
|
47
|
+
%r{^http://b\.scorecardresearch\.com/},
|
48
|
+
%r{^http://pixel\.quantserve\.com/},
|
49
|
+
%r{^http://s7\.addthis\.com/}
|
50
|
+
]
|
51
|
+
@attributes = [:title, :images, :description, :videos, :favicon]
|
52
|
+
@graders = [
|
53
|
+
->(description) { ::LinkThumbnailer::Graders::Length.new(description) },
|
54
|
+
->(description) { ::LinkThumbnailer::Graders::HtmlAttribute.new(description, :class) },
|
55
|
+
->(description) { ::LinkThumbnailer::Graders::HtmlAttribute.new(description, :id) },
|
56
|
+
->(description) { ::LinkThumbnailer::Graders::Position.new(description, weigth: 3) },
|
57
|
+
->(description) { ::LinkThumbnailer::Graders::LinkDensity.new(description) },
|
58
|
+
]
|
59
|
+
@description_min_length = 50
|
60
|
+
@positive_regex = /article|body|content|entry|hentry|main|page|pagination|post|text|blog|story/i
|
61
|
+
@negative_regex = /combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget|modal/i
|
62
|
+
@image_limit = 5
|
63
|
+
@image_stats = true
|
64
|
+
@raise_on_invalid_format = false
|
65
|
+
@max_concurrency = 20
|
66
|
+
@scrapers = [:opengraph, :default]
|
67
|
+
@http_override_headers = { 'Accept-Encoding' => 'none' }
|
68
|
+
@encoding = 'utf-8'
|
69
|
+
end
|
70
|
+
|
71
|
+
end
|
72
|
+
end
|
@@ -0,0 +1,11 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module LinkThumbnailer
|
4
|
+
Exceptions = Class.new(StandardError)
|
5
|
+
RedirectLimit = Class.new(Exceptions)
|
6
|
+
BadUriFormat = Class.new(Exceptions)
|
7
|
+
FormatNotSupported = Class.new(Exceptions)
|
8
|
+
ScraperInvalid = Class.new(Exceptions)
|
9
|
+
HTTPError = Class.new(Exceptions)
|
10
|
+
SyntaxError = Class.new(Exceptions)
|
11
|
+
end
|
@@ -0,0 +1,43 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require 'delegate'
|
4
|
+
require 'link_thumbnailer/graders/base'
|
5
|
+
require 'link_thumbnailer/graders/length'
|
6
|
+
require 'link_thumbnailer/graders/html_attribute'
|
7
|
+
require 'link_thumbnailer/graders/link_density'
|
8
|
+
require 'link_thumbnailer/graders/position'
|
9
|
+
|
10
|
+
module LinkThumbnailer
|
11
|
+
class Grader < ::SimpleDelegator
|
12
|
+
|
13
|
+
attr_reader :config, :description
|
14
|
+
|
15
|
+
def initialize(description)
|
16
|
+
@config = ::LinkThumbnailer.page.config
|
17
|
+
@description = description
|
18
|
+
|
19
|
+
super(config)
|
20
|
+
end
|
21
|
+
|
22
|
+
# For given description, computes probabilities returned by each graders by multipying them together.
|
23
|
+
#
|
24
|
+
# @return [Float] the probability for the given description to be considered good
|
25
|
+
def call
|
26
|
+
probability = 1.0
|
27
|
+
|
28
|
+
graders.each do |lambda|
|
29
|
+
instance = lambda.call(description)
|
30
|
+
probability *= instance.call.to_f ** instance.weight
|
31
|
+
end
|
32
|
+
|
33
|
+
probability
|
34
|
+
end
|
35
|
+
|
36
|
+
private
|
37
|
+
|
38
|
+
def graders
|
39
|
+
config.graders
|
40
|
+
end
|
41
|
+
|
42
|
+
end
|
43
|
+
end
|
@@ -0,0 +1,39 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require 'delegate'
|
4
|
+
|
5
|
+
module LinkThumbnailer
|
6
|
+
module Graders
|
7
|
+
class Base < ::SimpleDelegator
|
8
|
+
|
9
|
+
attr_reader :config, :description, :options
|
10
|
+
|
11
|
+
def initialize(description, options = {})
|
12
|
+
@config = ::LinkThumbnailer.page.config
|
13
|
+
@description = description
|
14
|
+
@options = options
|
15
|
+
|
16
|
+
super(config)
|
17
|
+
end
|
18
|
+
|
19
|
+
def call
|
20
|
+
fail NotImplementedError
|
21
|
+
end
|
22
|
+
|
23
|
+
def weight
|
24
|
+
options.fetch(:weigth, 1)
|
25
|
+
end
|
26
|
+
|
27
|
+
private
|
28
|
+
|
29
|
+
def node
|
30
|
+
description.node
|
31
|
+
end
|
32
|
+
|
33
|
+
def text
|
34
|
+
description.text
|
35
|
+
end
|
36
|
+
|
37
|
+
end
|
38
|
+
end
|
39
|
+
end
|
@@ -0,0 +1,48 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module LinkThumbnailer
|
4
|
+
module Graders
|
5
|
+
class HtmlAttribute < ::LinkThumbnailer::Graders::Base
|
6
|
+
|
7
|
+
attr_reader :attribute_name
|
8
|
+
|
9
|
+
def initialize(description, attribute_name)
|
10
|
+
super(description)
|
11
|
+
@attribute_name = attribute_name.to_sym
|
12
|
+
end
|
13
|
+
|
14
|
+
def call
|
15
|
+
return 1.0 if positive?
|
16
|
+
return 0.0 if negative?
|
17
|
+
1.0
|
18
|
+
end
|
19
|
+
|
20
|
+
private
|
21
|
+
|
22
|
+
def attribute
|
23
|
+
node[attribute_name]
|
24
|
+
end
|
25
|
+
|
26
|
+
def attribute?
|
27
|
+
attribute && !attribute.empty?
|
28
|
+
end
|
29
|
+
|
30
|
+
def negative?
|
31
|
+
attribute? && attribute =~ negative_regex
|
32
|
+
end
|
33
|
+
|
34
|
+
def positive?
|
35
|
+
attribute? && attribute =~ positive_regex
|
36
|
+
end
|
37
|
+
|
38
|
+
def negative_regex
|
39
|
+
config.negative_regex
|
40
|
+
end
|
41
|
+
|
42
|
+
def positive_regex
|
43
|
+
config.positive_regex
|
44
|
+
end
|
45
|
+
|
46
|
+
end
|
47
|
+
end
|
48
|
+
end
|
@@ -0,0 +1,37 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module LinkThumbnailer
|
4
|
+
module Graders
|
5
|
+
class Length < ::LinkThumbnailer::Graders::Base
|
6
|
+
|
7
|
+
def call
|
8
|
+
return 0.0 if too_short?
|
9
|
+
|
10
|
+
y / get_gaussian_value_for(ideal_description_length)
|
11
|
+
end
|
12
|
+
|
13
|
+
private
|
14
|
+
|
15
|
+
def get_gaussian_value_for(x)
|
16
|
+
Math.sqrt(2.0 * Math::PI ** 2) * Math.exp(-(x - ideal_description_length) ** 2 / 2.0 * 0.005 ** 2)
|
17
|
+
end
|
18
|
+
|
19
|
+
def x
|
20
|
+
text.length
|
21
|
+
end
|
22
|
+
|
23
|
+
def y
|
24
|
+
get_gaussian_value_for(x)
|
25
|
+
end
|
26
|
+
|
27
|
+
def ideal_description_length
|
28
|
+
options.fetch(:ideal_description_length, 120).to_f
|
29
|
+
end
|
30
|
+
|
31
|
+
def too_short?
|
32
|
+
text.length < config.description_min_length
|
33
|
+
end
|
34
|
+
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
@@ -0,0 +1,20 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module LinkThumbnailer
|
4
|
+
module Graders
|
5
|
+
class LinkDensity < ::LinkThumbnailer::Graders::Base
|
6
|
+
|
7
|
+
def call
|
8
|
+
return 0.0 if text.length == 0
|
9
|
+
1.0 - (links.count.to_f / text.length.to_f)
|
10
|
+
end
|
11
|
+
|
12
|
+
private
|
13
|
+
|
14
|
+
def links
|
15
|
+
node.css('a').map(&:text).compact.reject(&:empty?)
|
16
|
+
end
|
17
|
+
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|