Class: Anemone::PageHash

Inherits:
Hash
  • Object
show all
Defined in:
lib/anemone/page_hash.rb

Instance Method Summary collapse

Instance Method Details

#[](index) ⇒ Object

We typically index the hash with a URI, but convert it to a String for easier retrieval



6
7
8
 
# File 'lib/anemone/page_hash.rb', line 6

def [](index)
  super(index.to_s)
end

#[]=(index, other) ⇒ Object 



10
11
12
 
# File 'lib/anemone/page_hash.rb', line 10

def []=(index, other)
  super(index.to_s, other)
end

#has_key?(key) ⇒ Boolean

Returns:

  • (Boolean) 


14
15
16
 
# File 'lib/anemone/page_hash.rb', line 14

def has_key?(key)
  super(key.to_s)
end

#has_page?(url) ⇒ Boolean

Does this PageHash contain the specified URL? HTTP and HTTPS versions of a URL are considered to be the same page.

Returns:

  • (Boolean) 


20
21
22
23
24
25
26
27
28
 
# File 'lib/anemone/page_hash.rb', line 20

def has_page?(url)
  schemes = %w(http https)
  if schemes.include? url.scheme
    u = url.dup
    return schemes.any? { |s| u.scheme = s; has_key?(u) }
  end

  has_key?(url)
end

#pages_linking_to(urls) ⇒ Object

If given a single URL (as a String or URI), returns an Array of Pages which link to that URL If given an Array of URLs, returns a Hash (URI => [Page, Page…]) of Pages linking to those URLs



93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
 
# File 'lib/anemone/page_hash.rb', line 93

def pages_linking_to(urls)
  unless urls.is_a?(Array)
    urls = [urls] unless urls.is_a?(Array)
    single = true
  end

  urls.map! do |url|
    if url.is_a?(String)
      URI(url) rescue nil
    else
      url
    end
  end
  urls.compact

  links = {}
  urls.each { |url| links[url] = [] }
  values.each do |page|
    urls.each { |url| links[url] << page if page.links.include?(url) }
  end

  if single and !links.empty?
    return links.first
  else
    return links
  end
end

#shortest_paths!(root) ⇒ Object

Use a breadth-first search to calculate the single-source shortest paths from root to all pages in the PageHash



34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
 
# File 'lib/anemone/page_hash.rb', line 34

def shortest_paths!(root)
  root = URI(root) if root.is_a?(String)
  raise "Root node not found" if !has_key?(root)
  
  each_value {|p| p.visited = false if p}
  
  q = Queue.new
  
  q.enq(root)
  self[root].depth = 0
  self[root].visited = true
  while(!q.empty?)
    url = q.deq
    
    next if !has_key?(url)
    
    page = self[url]
    
    page.links.each do |u|
      next if !has_key?(u) or self[u].nil?
      link = self[u]
      aliases = [link].concat(link.aliases.map {|a| self[a] })
      
      aliases.each do |node|
        if node.depth.nil? or page.depth + 1 < node.depth
          node.depth = page.depth + 1
        end
      end
      
      q.enq(self[u].url) if !self[u].visited
      self[u].visited = true
    end
  end
  
  self
end

#uniq ⇒ Object

Returns a new PageHash by removing redirect-aliases for each non-redirect Page



75
76
77
78
79
80
81
82
83
84
85
86
87
 
# File 'lib/anemone/page_hash.rb', line 75

def uniq
  results = PageHash.new
  each do |url, page|
    #if none of the aliases of this page have been added, and this isn't a redirect page, add this page
    page_added = page.aliases.inject(false) { |r, a| r ||= results.has_key? a}
    if !page.redirect? and !page_added
      results[url] = page.clone 
      results[url].aliases = []
    end
  end
  
  results
end

#urls_linking_to(urls) ⇒ Object

If given a single URL (as a String or URI), returns an Array of URLs which link to that URL If given an Array of URLs, returns a Hash (URI => [URI, URI…]) of URLs linking to those URLs



125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
 
# File 'lib/anemone/page_hash.rb', line 125

def urls_linking_to(urls)
  unless urls.is_a?(Array)
    urls = [urls] unless urls.is_a?(Array)
    single = true
  end

  links = pages_linking_to(urls)
  links.each { |url, pages| links[url] = pages.map{|p| p.url} }

  if single and !links.empty?
    return links.first
  else
    return links
  end	  
end