Class: EmailCrawler::PageLinks

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

Constant Summary collapse 

100
SLEEP_TIME = 
0.5
MAX_RETRIES = 
5

Constants included from MechanizeHelper

MechanizeHelper::READ_TIMEOUT

Class Method Summary collapse

Instance Method Summary collapse

Methods included from MechanizeHelper

#get, #new_agent

Constructor Details

#initialize(url, logger = Logger.new("/dev/null")) ⇒ PageLinks

Returns a new instance of PageLinks.



9
10
11
12
13
14
15
16
17
18
19
 
# File 'lib/email_crawler/page_links.rb', line 9

def initialize(url, logger = Logger.new("/dev/null"))
  @url = url
  uri = URI(url)
  scheme_and_host = if uri.host
                      "#{uri.scheme}://#{uri.host}"
                    else
                      url[%r(\A(https?://([^/]+))), 1]
                    end
  @domain = Regexp.new("#{scheme_and_host}/", true)
  @logger = logger
end

Class Method Details

.for(url, max_links: MAX_LINKS, logger: Logger.new("/dev/null")) ⇒ Object 



21
22
23
 
# File 'lib/email_crawler/page_links.rb', line 21

def self.for(url, max_links: MAX_LINKS, logger: Logger.new("/dev/null"))
  new(url, logger).fetch_links(max_links)
end

Instance Method Details



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
 
# File 'lib/email_crawler/page_links.rb', line 25

def fetch_links(max_links = MAX_LINKS)
  queue, links = Set.new([@url]), Set.new([@url])
  retries = 0

  until queue.empty?
    current_link = queue.first
    @logger.info "current_link: #{current_link}"

    begin
      page = get(current_link)
    rescue Net::HTTP::Persistent::Error => err
      @logger.warn err.inspect
      page = nil

      if retries < MAX_RETRIES
        retries += 1
        @logger.debug "Retry ##{retries}"
        agent.shutdown
        Thread.current[:agent] = nil
        sleep(SLEEP_TIME)
        retry
      else
        @logger.error "Giving up grabbing link for '#{@url}' after #{retries} retries"
        break
      end
    rescue URI::InvalidComponentError => err
      @logger.warn err.inspect
    else
      retries = 0
    end

    if page
      new_links = page.links_with(href: @domain).map(&:href)
      new_links.reject! { |link| links.include?(link) }
      @logger.debug "found: #{new_links.length} new link(s)"
      new_links.each { |link| queue << link }
      links << current_link

      if links.length == max_links
        break
      else
        sleep(SLEEP_TIME)
      end
    end

    queue.delete(current_link)
  end

  links.to_a
end