Class: Wikipedia::Page

Inherits:
Object
  • Object
show all
Defined in:
lib/wikipedia/page.rb

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(json) ⇒ Page

Returns a new instance of Page.



5
6
7
8
9
 
# File 'lib/wikipedia/page.rb', line 5

def initialize(json)
  require 'json'
  @json = json
  @data = JSON.parse(json)
end

Instance Attribute Details

#json ⇒ Object (readonly)

Returns the value of attribute json.



3
4
5
 
# File 'lib/wikipedia/page.rb', line 3

def json
  @json
end

Class Method Details

.sanitize(s) ⇒ Object

rubocop:disable Metrics/MethodLength rubocop:disable Metrics/AbcSize



110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
 
# File 'lib/wikipedia/page.rb', line 110

def self.sanitize(s)
  return unless s

  # Transform punctuation templates
  # Em dash (https://en.wikipedia.org/wiki/Template:Em_dash)
  s.gsub!(/\{\{(em dash|emdash)\}\}/i, '—')
  # En dash (https://en.wikipedia.org/wiki/Template:En_dash)
  s.gsub!(/\{\{(en dash|ndash|nsndns)\}\}/i, '–')
  # Spaced en dashes (https://en.wikipedia.org/wiki/Template:Spaced_en_dash_space)
  s.gsub!(/\{\{(spaced e?n\s?dash( space)?|snds?|spndsp|sndashs|spndashsp)\}\}/i, ' – ')
  # Bold middot
  s.gsub!(/\{\{(·|dot|middot|\,)\}\}/i, '&nbsp;<b>&middot;</b>')
  # Bullets
  s.gsub!(/\{\{(•|bull(et)?)\}\}/i, '&nbsp;&bull;')
  # Forward Slashes (https://en.wikipedia.org/wiki/Template:%5C)
  s.gsub!(/\{\{\\\}\}/i, '&nbsp;/')

  # Transform language specific blocks
  s.gsub!(/\{\{lang[\-\|]([a-z]+)\|([^\|\{\}]+)(\|[^\{\}]+)?\}\}/i, '<span lang="\1">\2</span>')

  # Parse Old Style Date template blocks
  # Old Style Dates (https://en.wikipedia.org/wiki/Template:OldStyleDate)
  s.gsub!(/\{\{OldStyleDate\|([^\|]*)\|([^\|]*)\|([^\|]*)\}\}/i, '\1 [<abbr title="Old Style">O.S.</abbr> \3] \2')
  # Old Style Dates with different years (https://en.wikipedia.org/wiki/Template:OldStyleDateDY)
  s.gsub!(/\{\{OldStyleDateDY\|([^\|]*)\|([^\|]*)\|([^\|]*)\}\}/i, '\1 \2 [<abbr title="Old Style">O.S.</abbr> \3]')
  # Old Style Dates with no year (https://en.wikipedia.org/wiki/Template:OldStyleDateNY)
  s.gsub!(/\{\{OldStyleDateNY\|([^\|]*)\|([^\|]*)\}\}/i, '\1 [<abbr title="Old Style">O.S.</abbr> \2]')

  # strip anything else inside curly braces!
  s.gsub!(/\{\{[^\{\}]+?\}\}[\;\,]?/, '') while s =~ /\{\{[^\{\}]+?\}\}[\;\,]?/

  # strip info box
  s.sub!(/^\{\|[^\{\}]+?\n\|\}\n/, '')

  # strip internal links
  s.gsub!(/\[\[([^\]\|]+?)\|([^\]\|]+?)\]\]/, '\2')
  s.gsub!(/\[\[([^\]\|]+?)\]\]/, '\1')

  # strip images and file links
  s.gsub!(/\[\[Image:(.*?(?=\]\]))??\]\]/, '')
  s.gsub!(/\[\[File:(.*?(?=\]\]))??\]\]/, '')

  # convert bold/italic to html
  s.gsub!(/'''''(.+?)'''''/, '<b><i>\1</i></b>')
  s.gsub!(/'''(.+?)'''/, '<b>\1</b>')
  s.gsub!(/''(.+?)''/, '<i>\1</i>')

  # misc
  s.gsub!(/(\d)<ref[^<>]*>[\s\S]*?<\/ref>(\d)/, '\1&nbsp;&ndash;&nbsp;\2')
  s.gsub!(/<ref[^<>]*>[\s\S]*?<\/ref>/, '')
  s.gsub!(/<ref(.*?(?=\/>))??\/>/, '')
  s.gsub!(/<!--[^>]+?-->/, '')
  s.gsub!(/\(\s+/, '(')
  s.gsub!('  ', ' ')
  s.strip!

  # create paragraphs
  sections = s.split("\n\n")
  s =
    if sections.size > 1
      sections.map { |paragraph| "<p>#{paragraph.strip}</p>" }.join("\n")
    else
      "<p>#{s}</p>"
    end

  s
end

Instance Method Details

#categories ⇒ Object 



51
52
53
 
# File 'lib/wikipedia/page.rb', line 51

def categories
  page['categories'].map { |c| c['title'] } if page['categories']
end

#content ⇒ Object 



15
16
17
 
# File 'lib/wikipedia/page.rb', line 15

def content
  page['revisions'].first['*'] if page['revisions']
end

#coordinates ⇒ Object 



87
88
89
 
# File 'lib/wikipedia/page.rb', line 87

def coordinates
  page['coordinates'].first.values if page['coordinates']
end

#editurl ⇒ Object 



39
40
41
 
# File 'lib/wikipedia/page.rb', line 39

def editurl
  page['editurl']
end

#extlinks ⇒ Object 



59
60
61
 
# File 'lib/wikipedia/page.rb', line 59

def extlinks
  page['extlinks'].map { |c| c['*'] } if page['extlinks']
end

#fullurl ⇒ Object 



35
36
37
 
# File 'lib/wikipedia/page.rb', line 35

def fullurl
  page['fullurl']
end

#image_descriptionurl ⇒ Object 



71
72
73
 
# File 'lib/wikipedia/page.rb', line 71

def image_descriptionurl
  page['imageinfo'].first['descriptionurl'] if page['imageinfo']
end

#image_descriptionurls ⇒ Object 



79
80
81
 
# File 'lib/wikipedia/page.rb', line 79

def image_descriptionurls
  image_metadata.map(&:image_descriptionurl) unless image_metadata.nil?
end

#image_metadata ⇒ Object 



95
96
97
98
99
100
101
102
 
# File 'lib/wikipedia/page.rb', line 95

def image_metadata
  unless @cached_image_metadata
    return if images.nil?
    filtered = images.select { |i| i =~ /:.+\.(jpg|jpeg|png|gif|svg)$/i && !i.include?('LinkFA-star') }
    @cached_image_metadata = filtered.map { |title| Wikipedia.find_image(title) }
  end
  @cached_image_metadata || []
end

#image_url ⇒ Object 



67
68
69
 
# File 'lib/wikipedia/page.rb', line 67

def image_url
  page['imageinfo'].first['url'] if page['imageinfo']
end

#image_urls ⇒ Object 



75
76
77
 
# File 'lib/wikipedia/page.rb', line 75

def image_urls
  image_metadata.map(&:image_url) unless image_metadata.nil?
end

#images ⇒ Object 



63
64
65
 
# File 'lib/wikipedia/page.rb', line 63

def images
  page['images'].map { |c| c['title'] } if page['images']
end

#links ⇒ Object 



55
56
57
 
# File 'lib/wikipedia/page.rb', line 55

def links
  page['links'].map { |c| c['title'] } if page['links']
end

#main_image_url ⇒ Object 



83
84
85
 
# File 'lib/wikipedia/page.rb', line 83

def main_image_url
  page['thumbnail']['source'].sub(/\/thumb/, '').sub(/\/[^\/]*$/, '') if page['thumbnail']
end

#page ⇒ Object 



11
12
13
 
# File 'lib/wikipedia/page.rb', line 11

def page
  @data['query']['pages'].values.first if @data['query']['pages']
end

#raw_data ⇒ Object 



91
92
93
 
# File 'lib/wikipedia/page.rb', line 91

def raw_data
  @data
end

#redirect? ⇒ Boolean

Returns:

  • (Boolean) 


23
24
25
 
# File 'lib/wikipedia/page.rb', line 23

def redirect?
  content && content.match(/\#REDIRECT\s*\[\[(.*?)\]\]/i)
end

#redirect_title ⇒ Object 



27
28
29
 
# File 'lib/wikipedia/page.rb', line 27

def redirect_title
  redirect?[1] rescue nil
end

#sanitized_content ⇒ Object 



19
20
21
 
# File 'lib/wikipedia/page.rb', line 19

def sanitized_content
  self.class.sanitize(content)
end

#summary ⇒ Object 



47
48
49
 
# File 'lib/wikipedia/page.rb', line 47

def summary
  page['extract'].split('==')[0].strip if page['extract'] && page['extract'] != ''
end

#templates ⇒ Object 



104
105
106
 
# File 'lib/wikipedia/page.rb', line 104

def templates
  page['templates'].map { |c| c['title'] } if page['templates']
end

#text ⇒ Object 



43
44
45
 
# File 'lib/wikipedia/page.rb', line 43

def text
  page['extract']
end

#title ⇒ Object 



31
32
33
 
# File 'lib/wikipedia/page.rb', line 31

def title
  page['title']
end