CoCalc provides the best real-time collaborative environment for Jupyter Notebooks, LaTeX documents, and SageMath, scalable from individual users to large groups and classes!

1
require 'forwardable'
2

3
module Anemone
4
  class PageStore
5
    extend Forwardable
6

7
    def_delegators :@storage, :keys, :values, :size, :each
8

9
    def initialize(storage = {})
10
      @storage = storage
11
    end
12

13
    # We typically index the hash with a URI,
14
    # but convert it to a String for easier retrieval
15
    def [](index)
16
      @storage[index.to_s]
17
    end
18

19
    def []=(index, other)
20
      @storage[index.to_s] = other
21
    end
22

23
    def delete(key)
24
      @storage.delete key.to_s
25
    end
26

27
    def has_key?(key)
28
      @storage.has_key? key.to_s
29
    end
30

31
    def each_value
32
      each { |key, value| yield value }
33
    end
34

35
    def values
36
      result = []
37
      each { |key, value| result << value }
38
      result
39
    end
40

41
    def touch_key(key)
42
      self[key] = Page.new(key)
43
    end
44

45
    def touch_keys(keys)
46
      @storage.merge! keys.inject({}) { |h, k| h[k.to_s] = Page.new(k); h }
47
    end
48

49
    # Does this PageStore contain the specified URL?
50
    # HTTP and HTTPS versions of a URL are considered to be the same page.
51
    def has_page?(url)
52
      schemes = %w(http https)
53
      if schemes.include? url.scheme
54
        u = url.dup
55
        return schemes.any? { |s| u.scheme = s; has_key?(u) }
56
      end
57

58
      has_key? url
59
    end
60

61
    #
62
    # Use a breadth-first search to calculate the single-source
63
    # shortest paths from *root* to all pages in the PageStore
64
    #
65
    def shortest_paths!(root)
66
      root = URI(root) if root.is_a?(String)
67
      raise "Root node not found" if !has_key?(root)
68

69
      q = Queue.new
70

71
      q.enq root
72
      root_page = self[root]
73
      root_page.depth = 0
74
      root_page.visited = true
75
      self[root] = root_page
76
      while !q.empty?
77
        page = self[q.deq]
78
        page.links.each do |u|
79
          begin
80
            link = self[u]
81
            next if link.nil? || !link.fetched? || link.visited
82

83
            q << u unless link.redirect?
84
            link.visited = true
85
            link.depth = page.depth + 1
86
            self[u] = link
87

88
            if link.redirect?
89
              u = link.redirect_to
90
              redo
91
            end
92
          end
93
        end
94
      end
95

96
      self
97
    end
98

99
    #
100
    # Removes all Pages from storage where redirect? is true
101
    #
102
    def uniq!
103
      each_value { |page| delete page.url if page.redirect? }
104
      self
105
    end
106

107
    #
108
    # If given a single URL (as a String or URI), returns an Array of Pages which link to that URL
109
    # If given an Array of URLs, returns a Hash (URI => [Page, Page...]) of Pages linking to those URLs
110
    #
111
    def pages_linking_to(urls)
112
      unless urls.is_a?(Array)
113
        urls = [urls]
114
        single = true
115
      end
116

117
      urls.map! do |url|
118
        unless url.is_a?(URI)
119
          URI(url) rescue nil
120
        else
121
          url
122
        end
123
      end
124
      urls.compact
125

126
      links = {}
127
      urls.each { |url| links[url] = [] }
128
      values.each do |page|
129
        urls.each { |url| links[url] << page if page.links.include?(url) }
130
      end
131

132
      if single and !links.empty?
133
        return links[urls.first]
134
      else
135
        return links
136
      end
137
    end
138

139
    #
140
    # If given a single URL (as a String or URI), returns an Array of URLs which link to that URL
141
    # If given an Array of URLs, returns a Hash (URI => [URI, URI...]) of URLs linking to those URLs
142
    #
143
    def urls_linking_to(urls)
144
      unless urls.is_a?(Array)
145
        urls = [urls] unless urls.is_a?(Array)
146
        single = true
147
      end
148

149
      links = pages_linking_to(urls)
150
      links.each { |url, pages| links[url] = pages.map{|p| p.url} }
151

152
      if single and !links.empty?
153
        return links[urls.first]
154
      else
155
        return links
156
      end
157
    end
158

159
  end
160
end
161

162