class Rumba::Crawler::Parser

Constants

SK

Service Keys

Public Instance Methods

create_object(name, node, template) click to toggle source
# File lib/rumba/crawler/parser.rb, line 37
def create_object(name, node, template)
  object = send(name)
  template.reject{|key, _| SK.include?(key)}.each do |key, value|
    if value.is_a? Array
      object.send("#{key}=", parse_multi(node, value.first))
    else
      object.send("#{key}=", parse_node(node, value, key))
    end
  end
  return object
end
get_content(node, template) click to toggle source
# File lib/rumba/crawler/parser.rb, line 49
def get_content(node, template)
  if template['regexp']
    /#{template['regexp']}/i.match(node.content).to_s
  else
    node.content
  end
end
get_node(doc, template) click to toggle source
# File lib/rumba/crawler/parser.rb, line 57
def get_node(doc, template)
  if template.is_a?(String)
    doc.css(template)
  elsif template['root']
    @doc.css(template['root']).css(template['css'])
  else
    doc.css(template['css'])
  end
end
leaf_node?(template) click to toggle source
# File lib/rumba/crawler/parser.rb, line 67
def leaf_node?(template)
  template.is_a?(String) || template.reject{|key, _| SK.include?(key)}.empty?
end
parse_multi(doc, template) click to toggle source
# File lib/rumba/crawler/parser.rb, line 18
def parse_multi(doc, template)
  result = []
  template.each do |key, value|
    get_node(doc, value).each do |node|
      result << create_object(key, node, value)
    end
  end
  return result
end
parse_node(doc, template, name) click to toggle source
# File lib/rumba/crawler/parser.rb, line 28
def parse_node(doc, template, name)
  node = get_node(doc, template).first
  if leaf_node?(template)
    get_content(node, template)
  else
    create_object(name, node, template)
  end
end
process(response, template) click to toggle source
# File lib/rumba/crawler/parser.rb, line 8
def process(response, template)
  template = JSON.parse(template)
  @doc = Nokogiri::HTML(response)
  if template.is_a? Array
    parse_multi(@doc, template.first)
  else
    parse_node(@doc, template, template.keys.first)
  end
end