Class: Myimdb::Scraper::Imdb

Inherits:
Base
  • Object
show all
Defined in:
lib/myimdb/scraper/imdb.rb

Constant Summary

Constants included from HandleExceptions

HandleExceptions::EXCEPTIONS_ENABLED

Instance Method Summary collapse

Methods inherited from Base

all, #name, #summary, #to_hash

Methods included from StringExtensions

#scraper_unescape_html, #strip_useless_chars

Methods included from HandleExceptions

included

Constructor Details

#initialize(url) ⇒ Imdb

Returns a new instance of Imdb.



4
5
6
 
# File 'lib/myimdb/scraper/imdb.rb', line 4

def initialize(url)
  @url = url
end

Instance Method Details

#directors ⇒ Object 



8
9
10
 
# File 'lib/myimdb/scraper/imdb.rb', line 8

def directors
  document.css('#overview-top .txt-block h4:contains("Director") ~ a:not(:contains(" more "))').collect{ |a| a.text }
end

#directors_with_url ⇒ Object 



12
13
14
 
# File 'lib/myimdb/scraper/imdb.rb', line 12

def directors_with_url
  document.css('#overview-top .txt-block h4:contains("Director") ~ a:not(:contains(" more "))').collect{ |a| {:name=> a.text, :url=> "http://www.imdb.com#{a['href']}" } }
end

#genres ⇒ Object 



32
33
34
 
# File 'lib/myimdb/scraper/imdb.rb', line 32

def genres
  document.css('.see-more.inline.canwrap h4:contains("Genres:") ~ a:not(:contains(" more "))').collect{ |a| a.text }
end

#image ⇒ Object 



52
53
54
 
# File 'lib/myimdb/scraper/imdb.rb', line 52

def image
  document.css('#img_primary img').first['src']
end

#plot ⇒ Object 



40
41
42
 
# File 'lib/myimdb/scraper/imdb.rb', line 40

def plot
  strip_useless_chars(document.css('.article h2:contains("Storyline") ~ p').inner_text).gsub(/Written by.*/, '')
end

#rating ⇒ Object 



24
25
26
 
# File 'lib/myimdb/scraper/imdb.rb', line 24

def rating
  document.css(".star-box-details span[itemprop='ratingValue']").inner_text.strip.to_f
end

#release_date ⇒ Object 



48
49
50
 
# File 'lib/myimdb/scraper/imdb.rb', line 48

def release_date
  Date.parse(document.css("a[title='See all release dates']").inner_text)
end

#tagline ⇒ Object 



36
37
38
 
# File 'lib/myimdb/scraper/imdb.rb', line 36

def tagline
  strip_useless_chars(document.css('.txt-block h4:contains("Taglines:")').first.parent.inner_text).gsub(/Taglines |See more/, '') rescue nil
end

#votes ⇒ Object 



28
29
30
 
# File 'lib/myimdb/scraper/imdb.rb', line 28

def votes
  document.css(".star-box a[href='ratings']").inner_text.strip.split(' ').first.sub(',', '').to_i
end

#writers ⇒ Object 



16
17
18
 
# File 'lib/myimdb/scraper/imdb.rb', line 16

def writers
  document.css('#overview-top .txt-block h4:contains("Writer") ~ a:not(:contains(" more "))').collect{ |a| a.text }
end

#writers_with_url ⇒ Object 



20
21
22
 
# File 'lib/myimdb/scraper/imdb.rb', line 20

def writers_with_url
  document.css('#overview-top .txt-block h4:contains("Writer") ~ a:not(:contains(" more "))').collect{ |a| {:name=> a.text, :url=> "http://www.imdb.com#{a['href']}" } }
end

#year ⇒ Object 



44
45
46
 
# File 'lib/myimdb/scraper/imdb.rb', line 44

def year
  release_date.year
end