Class: EmailCrawler::Runner

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

Constant Summary collapse

MAX_CONCURRENCY = 
50

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(google_website) {|_self| ... } ⇒ Runner

Returns a new instance of Runner.

Yields:

  • (_self)

Yield Parameters:



19
20
21
22
 
# File 'lib/email_crawler.rb', line 19

def initialize(google_website)
  @google_website = google_website
  yield(self)
end

Instance Attribute Details

#blacklisted_domains=(value) ⇒ Object (writeonly)

Sets the attribute blacklisted_domains

Parameters:

  • value —

    the value to set the attribute blacklisted_domains to.



16
17
18
 
# File 'lib/email_crawler.rb', line 16

def blacklisted_domains=(value)
  @blacklisted_domains = value
end

#logger=(value) ⇒ Object

Sets the attribute logger

Parameters:

  • value —

    the value to set the attribute logger to.



16
17
18
 
# File 'lib/email_crawler.rb', line 16

def logger=(value)
  @logger = value
end

#max_concurrency=(value) ⇒ Object (writeonly)

Sets the attribute max_concurrency

Parameters:

  • value —

    the value to set the attribute max_concurrency to.



16
17
18
 
# File 'lib/email_crawler.rb', line 16

def max_concurrency=(value)
  @max_concurrency = value
end

#max_links=(value) ⇒ Object (writeonly)

Sets the attribute max_links

Parameters:

  • value —

    the value to set the attribute max_links to.



16
17
18
 
# File 'lib/email_crawler.rb', line 16

def max_links=(value)
  @max_links = value
end

#max_results=(value) ⇒ Object (writeonly)

Sets the attribute max_results

Parameters:

  • value —

    the value to set the attribute max_results to.



16
17
18
 
# File 'lib/email_crawler.rb', line 16

def max_results=(value)
  @max_results = value
end

Instance Method Details

#run(q) ⇒ Object 



24
25
26
27
28
29
30
31
32
33
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
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
 
# File 'lib/email_crawler.rb', line 24

def run(q)
  urls = Scraper.new(@google_website,
                     max_results: @max_results,
                     blacklisted_domains: @blacklisted_domains).
                 search_result_urls_for(q)
  urls.each { |url| logger.info "#{url}" }
  queue = Thread::Queue.new
  urls.each { |url| queue.push(url) }
  links_by_url = ThreadSafe::Array.new

  threads = (1..[urls.length, @max_concurrency].min).map do |i|
    Thread.new(i) do |i|
      url = begin
              queue.pop(true)
            rescue ThreadError; end

      while url
        logger.info "[Thread ##{i}] grabbing page links for '#{url}'.."
        links = PageLinks.for(url, max_links: @max_links, logger: logger)
        links_by_url << [url, links]

        url = begin
                queue.pop(true)
              rescue ThreadError; end
      end
    end
  end
  threads.each(&:join)
  logger.debug "links_by_url: #{links_by_url.inspect}"

  links_by_url.each { |arr| queue.push(arr) }
  emails_by_url = ThreadSafe::Hash.new
  threads = (1..[links_by_url.length, @max_concurrency].min).map do |i|
    Thread.new(i) do |i|
      arr = begin
              queue.pop(true)
            rescue ThreadError; end

      while arr
        url, links = arr
        logger.info "[Thread ##{i}] scanning for emails on page '#{url}' (#{links.length} links)"
        emails = EmailScanner.new(logger).scan(links)
        emails_by_url[url] = emails

        arr = begin
                queue.pop(true)
              rescue ThreadError; end
      end
    end
  end
  threads.each(&:join)
  logger.debug "emails_by_url: #{emails_by_url.inspect}"

  read_emails = Set.new
  CSV.generate do |csv|
    csv << %w(Email Domain URL)
    csv << []

    emails_by_url.each do |url, emails_by_link|
      email_count = emails_by_link.inject(0) { |sum, arr| sum += arr.last.length }
      logger.info "#{url} (#{email_count} emails)"

      emails_by_link.each do |link, emails|
        emails.each do |email|
          csv << [email, url, link] if read_emails.add?(email)
        end
      end
    end
  end
end