Class: Wovnrb::URL

Inherits:
Object
  • Object
show all
Defined in:
lib/wovnrb/services/url.rb

Overview

URL utility ported from html-swapper

Defined Under Namespace

Modules: FileExtension

Class Method Summary collapse

Class Method Details

.absolute_path?(href) ⇒ Boolean

Returns:

  • (Boolean) 


22
23
24
 
# File 'lib/wovnrb/services/url.rb', line 22

def self.absolute_path?(href)
  href.match?(%r{^/})
end

.absolute_url?(href) ⇒ Boolean

Returns:

  • (Boolean) 


18
19
20
 
# File 'lib/wovnrb/services/url.rb', line 18

def self.absolute_url?(href)
  href =~ %r{^(https?:)?//}i
end

.change_protocol(uri, new_protocol) ⇒ Object

Returns copy of uri [Addressable::URI].

Parameters:

  • uri (Addressable::URI)
  • new_protocol (String | nil)

Returns:

  • copy of uri [Addressable::URI]



99
100
101
102
103
 
# File 'lib/wovnrb/services/url.rb', line 99

def self.change_protocol(uri, new_protocol)
  result = uri.dup
  result.scheme = new_protocol
  result
end

.file?(href_with_query_and_hash) ⇒ Boolean

Returns:

  • (Boolean) 


112
113
114
115
116
117
118
119
 
# File 'lib/wovnrb/services/url.rb', line 112

def self.file?(href_with_query_and_hash)
  href = remove_query_and_hash(href_with_query_and_hash)
  img_files = %r{^(https?://)?.*(\.(#{FileExtension::IMG_FILES}))((\?|#).*)?$}io
  audio_files = %r{^(https?://)?.*(\.(#{FileExtension::AUDIO_FILES}))((\?|#).*)?$}io
  video_files = %r{^(https?://)?.*(\.(#{FileExtension::VIDEO_FILES}))((\?|#).*)?$}io
  doc_files = %r{^(https?://)?.*(\.(#{FileExtension::DOC_FILES}))((\?|#).*)?$}io
  href.match?(img_files) || href.match?(audio_files) || href.match?(video_files) || href.match?(doc_files)
end

.host_with_port(parsed_uri) ⇒ Object 



42
43
44
45
46
47
48
 
# File 'lib/wovnrb/services/url.rb', line 42

def self.host_with_port(parsed_uri)
  if parsed_uri.port
    "#{parsed_uri.host}:#{parsed_uri.port}"
  else
    parsed_uri.host.to_s
  end
end

.join_paths(*paths) ⇒ Object 



83
84
85
86
87
88
89
90
91
92
93
94
 
# File 'lib/wovnrb/services/url.rb', line 83

def self.join_paths(*paths)
  paths.inject('') do |left, right|
    case [left.end_with?('/'), right.start_with?('/')]
    when [true, true]
      left + right[1..-1]
    when [false, false]
      left + (right.blank? ? right : "/#{right}")
    else
      left + right
    end
  end
end

.normalize_path_slash(original_path, new_path) ⇒ Object

if original path does not end in slash, remove it from new path if original path ends in slash, add it to new path



127
128
129
130
131
132
133
134
 
# File 'lib/wovnrb/services/url.rb', line 127

def self.normalize_path_slash(original_path, new_path)
  if !original_path.end_with?('/') && new_path.end_with?('/')
    new_path = new_path.chomp('/')
  elsif original_path.end_with?('/') && !new_path.end_with?('/')
    new_path += '/'
  end
  new_path
end

.normalize_url(href) ⇒ Object

TODO: Maybe this should be applied to all get_attribute calls rather than just href



12
13
14
15
16
 
# File 'lib/wovnrb/services/url.rb', line 12

def self.normalize_url(href)
  return nil unless href

  href.delete("\u200b").strip
end

.path_and_query(parsed_uri) ⇒ Object

Parameters:

  • parsed_uri (Addressable::URI) 


31
32
33
 
# File 'lib/wovnrb/services/url.rb', line 31

def self.path_and_query(parsed_uri)
  parsed_uri.path + (parsed_uri.query ? "?#{parsed_uri.query}" : '')
end

.path_and_query_and_hash(parsed_uri) ⇒ Object 



35
36
37
38
39
40
 
# File 'lib/wovnrb/services/url.rb', line 35

def self.path_and_query_and_hash(parsed_uri)
  uri = parsed_uri.path
  uri += "?#{parsed_uri.query}" if parsed_uri.query
  uri += "##{parsed_uri.fragment}" if parsed_uri.fragment
  uri
end

.prepend_path(url, dir) ⇒ Object

Set the path lang to



68
69
70
 
# File 'lib/wovnrb/services/url.rb', line 68

def self.prepend_path(url, dir)
  url.sub(%r{(.+\.[^/]+)(/|$)}, "\\1/#{dir}\\2")
end

.prepend_path_slash(path) ⇒ Object 



76
77
78
79
80
81
 
# File 'lib/wovnrb/services/url.rb', line 76

def self.prepend_path_slash(path)
  path ||= ''
  return path if path.starts_with?('/')

  "/#{path}"
end

.relative_path?(href) ⇒ Boolean

Returns:

  • (Boolean) 


26
27
28
 
# File 'lib/wovnrb/services/url.rb', line 26

def self.relative_path?(href)
  !absolute_url?(href) && !absolute_path?(href)
end

.remove_query_and_hash(href) ⇒ Object 



121
122
123
 
# File 'lib/wovnrb/services/url.rb', line 121

def self.remove_query_and_hash(href)
  href.gsub(/[#?].*/, '')
end

.resolve_absolute_path(base_url, href) ⇒ Object 



58
59
60
61
62
63
64
65
 
# File 'lib/wovnrb/services/url.rb', line 58

def self.resolve_absolute_path(base_url, href)
  normalized_uri = resolve_absolute_uri(base_url, href)
  path = normalized_uri.path
  query = normalized_uri.query ? "?#{normalized_uri.query}" : ''
  fragment = normalized_uri.fragment ? "##{normalized_uri.fragment}" : ''

  path + query + fragment
end

.resolve_absolute_uri(base_url, href) ⇒ Object 



50
51
52
53
54
55
56
 
# File 'lib/wovnrb/services/url.rb', line 50

def self.resolve_absolute_uri(base_url, href)
  # This resolves ./../ and also handles href already being absolute
  Addressable::URI.join(base_url, href)
rescue Addressable::URI::InvalidURIError, ArgumentError => e
  Rollbar.warning('Failed to resolve absolute URI', original_error: e, base_url: base_url, href: href)
  raise
end

.trim_slashes(path) ⇒ Object 



72
73
74
 
# File 'lib/wovnrb/services/url.rb', line 72

def self.trim_slashes(path)
  path.gsub(%r{^/|/$}, '')
end

.valid_protocol?(href) ⇒ Boolean

Returns:

  • (Boolean) 


105
106
107
108
109
110
 
# File 'lib/wovnrb/services/url.rb', line 105

def self.valid_protocol?(href)
  scheme_matches = /^\s*(?<scheme>[a-zA-Z]+):/.match(href)
  scheme = scheme_matches ? scheme_matches[:scheme] : nil

  scheme.nil? || %w[http https].include?(scheme)
end