Module: GoogleScholar

Defined in:
lib/rbbt/sources/gscholar.rb

Class Method Summary collapse

Class Method Details



8
9
10
11
12
13
14
15
16
17
18
 
# File 'lib/rbbt/sources/gscholar.rb', line 8

def self.citation_link(title)
  citation_link = nil

  # Get citation page
  user_agent.get("http://scholar.google.es/scholar?q='#{ title }'&hl=es&lr=&lr=") do |page|
    article = page.search('div[@class=gs_r]').first
    return nil if article.nil?

    return article.search('a').select{|link| link['href'] =~ /scholar\?cites/ && link.inner_html =~ /\d+$/ }.first
  end
end

.full_text_url(title) ⇒ Object 



20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
 
# File 'lib/rbbt/sources/gscholar.rb', line 20

def self.full_text_url(title)
  full_text_link = nil

  # Get page
  user_agent.get("http://scholar.google.es/scholar?q='#{ title }'&hl=es&lr=&lr=") do |page|
    article = page.search('div[@class=gs_r]').first
    return nil if article.nil?

    link =  article.search('a').select{ |link| 
      link['href'] =~ /\.pdf$/ || link['href'] =~ /type=pdf/
    }.first

    return nil if link.nil?

    return link['href']
  end
end

.number_cites(title) ⇒ Object 



39
40
41
42
43
44
45
46
47
 
# File 'lib/rbbt/sources/gscholar.rb', line 39

def self.number_cites(title)

  link = citation_link title
  return 0 if link.nil?

  link.inner_html =~ /(\d+)$/ 

  return $1.to_i
end

.user_agent ⇒ Object 



4
5
6
 
# File 'lib/rbbt/sources/gscholar.rb', line 4

def self.user_agent
  @@a ||= Mechanize.new
end