Class: Bookbinder::Spider

Inherits:
Object
  • Object
show all
Defined in:
lib/bookbinder/spider.rb,
lib/bookbinder/stabilimentum.rb

Defined Under Namespace

Classes: Result, Stabilimentum

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(app_dir: nil) ⇒ Spider

Returns a new instance of Spider.



34
35
36
 
# File 'lib/bookbinder/spider.rb', line 34

def initialize(app_dir: nil)
  @app_dir = app_dir || raise('Spiders must be initialized with an app directory.')
end

Class Method Details

.prepend_location(location, url) ⇒ Object 



30
31
32
 
# File 'lib/bookbinder/spider.rb', line 30

def self.prepend_location(location, url)
  "#{URI(location).path} => #{url}"
end

Instance Method Details

#generate_sitemap(target_host, port, streams, broken_link_exclusions: /(?!.*)/) ⇒ Object 



38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
 
# File 'lib/bookbinder/spider.rb', line 38

def generate_sitemap(target_host, port, streams,
                     broken_link_exclusions: /(?!.*)/)
  temp_host = "localhost:#{port}"

  sieve = Sieve.new domain: "http://#{temp_host}"
  links = crawl_from "http://#{temp_host}/index.html", sieve
  broken_links, working_links = links
  sitemap_links = substitute_hostname(temp_host, target_host, working_links)
  public_broken_links = broken_links.reject {|l| l.match(broken_link_exclusions)}
  announce_broken_links(public_broken_links, streams)
  Result.new(
    public_broken_links,
    SitemapGenerator.new.generate(sitemap_links), app_dir
  )
end