Class: Apollo::Crawler::HackerNewsCrawler

Inherits:
BaseCrawler
  • Object
show all
Defined in:
lib/apollo_crawler/crawler/hacker_news_crawler.rb

Constant Summary collapse

@@MATCHER_ITEM = 
"(//td[@class = 'title']/a)[not(position() > last() -1)]"

Instance Method Summary collapse

Methods inherited from BaseCrawler

create_metadoc, #enqueue_url, #etl, fetch, #fetch_document, #initialize, name_re, #process_url, try_get_doc, try_get_url, #url_processed?

Constructor Details

This class inherits a constructor from Apollo::Crawler::BaseCrawler

Instance Method Details

#extract_data(doc) ⇒ Object 



36
37
38
39
40
41
42
43
44
45
46
47
48
 
# File 'lib/apollo_crawler/crawler/hacker_news_crawler.rb', line 36

def extract_data(doc)
  res = doc.xpath(@@MATCHER_ITEM).map { |node|
    url = BaseCrawler.try_get_url(self.url, node['href']).to_s
    next if url.nil?

    { 
      :text => node.text,
      :link => url
    }
  }

  return res
end

#extract_links(doc) ⇒ Object 



50
51
52
53
54
55
56
57
58
59
60
61
 
# File 'lib/apollo_crawler/crawler/hacker_news_crawler.rb', line 50

def extract_links(doc)
  res = doc.xpath("(//td[@class = 'title']/a)[last()]").map { |node|
    url = BaseCrawler.try_get_url(self.url, node['href']).to_s
    next if url.nil?

    { 
      :link => url
    }
  }
  
  return res.uniq
end

#name ⇒ Object 



28
29
30
 
# File 'lib/apollo_crawler/crawler/hacker_news_crawler.rb', line 28

def name()
  return "Hacker News"
end

#url ⇒ Object 



32
33
34
 
# File 'lib/apollo_crawler/crawler/hacker_news_crawler.rb', line 32

def url()
  return "http://news.ycombinator.com/"
end