module ZhSieve

Constants

BASE_URL
VERSION
ZL_URI

Public Class Methods

crawl_answer(options) click to toggle source
# File lib/zhSieve.rb, line 9
def self.crawl_answer(options)
        question_id = "#{options[:question_id]}"
        question_uri = "/question/#{question_id}"
        answer_id = "#{options[:answer_id]}"
        answer_uri = "#answer-#{answer_id}"
        search_uri = "#{BASE_URL}#{question_uri}#{answer_uri}"
        agent = Mechanize.new
        agent.user_agent = 'Chrome/53.0.2785.143'
        agent.max_history = 1
        # Dir.chdir(File.dirname(__FILE__))
        agent.cookie_jar.load_cookiestxt("./cookies.txt")
        search_page = agent.get("#{search_uri}")
        haha = HTMLPage.new(contents:search_page,question_id:question_id,answer_id:answer_id).answerMarkdown
end
crawl_zl_article(options) click to toggle source
# File lib/zhSieve.rb, line 28
def self.crawl_zl_article(options)
        article_id = "#{options[:article_id]}"
        search_uri = "#{ZL_URI}#{article_id}"
        agent = Mechanize.new
        agent.user_agent = 'Chrome/53.0.2785.143'
        agent.max_history = 1
        # Dir.chdir(File.dirname(__FILE__))
        agent.cookie_jar.load_cookiestxt("./cookies.txt")
        search_page = agent.get("#{search_uri}")
        haha = HTMLPage.new(contents:search_page,article_id:article_id).articleMarkdown
end
crawl_zl_people(options) click to toggle source
# File lib/zhSieve.rb, line 24
def self.crawl_zl_people(options)

end