implement crawler data

app/assets/javascripts/datas.js

-// Place all the behaviors and hooks related to the matching controller here.
-// All this logic will automatically be available in application.js.

app/assets/stylesheets/datas.scss

-// Place all the styles related to the datas controller here.
-// They will automatically be included in application.css.
-// You can use Sass (SCSS) here: http://sass-lang.com/

app/controllers/datas_controller.rb → app/data/crawler.rb

-class DatasController < ApplicationController
-  require 'nokogiri'
-  require 'open-uri'
+require 'nokogiri'
+require 'open-uri'
 
+class Crawler
   BASE_CAREERBUILDER_URL = "https://careerbuilder.vn"
   LIST_URL = "#{BASE_CAREERBUILDER_URL}/viec-lam"
 
-  def index
-    @datas = crawl_data
-  end
-
-  def crawl_data
-    links = get_link[0..4]
+  def self.crawl_job_infomation(job_links)
+    links = job_links
     job_details = []
 
     links.each do |link|
@@ -18,10 +14,33 @@ class DatasController < ApplicationController
       params = {}
       doc = Nokogiri::HTML(open(link))
 
-      # get company information
-      params[:name] = doc.css('.top-job .top-job-info h1').text
-      params[:company_name] = doc.css('.top-job .top-job-info .tit_company').text
-      params[:updated_date] = doc.css('.top-job .datepost').text
+      if doc.css('#template_vantai').blank? &&
+         doc.css('#template_1').blank? &&
+         doc.css('#template_2').blank? &&
+         doc.css('#template_3').blank? &&
+         doc.css('#template_4').blank?
+
+        params = use_template_default(doc, link)
+        job_details << params
+      end
+    end
+    return job_details
+  end
+
+  def self.use_template_default(doc, link)
+    params = {}
+    params_company = {}
+    description = []
+
+    # get job's name
+    params[:name] = doc.css('.top-job .top-job-info h1').text.strip
+    params[:updated_date] = doc.css('.datepost').text.split(':')[1].strip
+
+    # get company infoimation
+    params_company = crawl_company_infomation(doc)
+    params[:company_name] = params_company[:name]
+    params[:company_location] = params_company[:location]
+    params[:company_description] = params_company[:description]
 
     # get employment information
     doc.css('.MyJobLeft .box2Detail .DetailJobNew li').children.each do |child|
@@ -29,7 +48,7 @@ class DatasController < ApplicationController
       next if info.blank?
       case info[0].strip.upcase
       when 'NƠI LÀM VIỆC'
-          params[:city] = info[1]
+        params[:city] = info[1].strip
       when 'CẤP BẬC'
         params[:level] = info[1]
       when 'KINH NGHIỆM'
@@ -45,17 +64,30 @@ class DatasController < ApplicationController
 
     # get job description
     doc.css('.MarBot20').children.each do |child|
-        params[:description] = child.to_html
+      description << child.to_html
     end
+    params[:description] = description.join("")
 
-      job_details << params
+    # original_link
+    params[:original_link] = link
+
+    return params
   end
-    return job_details
+
+  def self.crawl_company_infomation(doc)
+    # get company information
+    params = {}
+
+    params[:name] = doc.css('.box1Detail .TitleDetailNew span').text
+    params[:location] = doc.css('.box1Detail .TitleDetailNew label')[0].text
+    params[:description] = doc.css('.desc_company p').text
+    return params
   end
 
-  def get_link
+  def self.get_job_link
     url = "#{LIST_URL}/tat-ca-viec-lam-trang-#{1}-vi.html"
     doc = Nokogiri::HTML(open(url))
-    return doc.css('.brief .jobtitle .job a').map { |a| a['href'] }.compact.uniq
+    links = doc.css('.gird_standard .brief .jobtitle .job a').map { |a| a['href'] }.compact.uniq
+    return links.delete_if{|link| link.include?('–')}
   end
 end

app/data/import.rb

+require "crawler.rb"
+
+class ImportData
+  def job
+    
+  end
+end

app/helpers/datas_helper.rb

-module DatasHelper
-end

app/models/city.rb

 class City < ApplicationRecord
-  belongs_to :country
+  belongs_to :country, optional: true
   has_many :companies
   has_many :jobs
 end

app/models/job.rb

@@ -6,4 +6,27 @@ class Job < ApplicationRecord
   has_many :favorite_jobs
   has_many :people_who_liked, through: :favorite_jobs, class_name: 'User', source: :user
   has_and_belongs_to_many :industries
+
+  def self.create_new_jobs(arr_jobs)
+    arr_jobs.each do |item|
+      job = Job.new(name:          item[:name],
+                    salary:        item[:salary],
+                    description:   item[:description],
+                    level:         item[:level],
+                    original_link: item[:original_link],
+                    experience:    item[:experience],
+                    expiry_date:   item[:expiry_date],
+                    updated_date:  item[:updated_date])
+      job.city = City.find_or_initialize_by(name: (item[:city] ||= '').split(':')[0]) #TODO fix this line
+      job.company = Company.find_or_initialize_by(name: item[:company_name])
+      job.company.location = item[:company_location]
+      job.company.description = item[:company_description]
+      job.company.city = job.city
+      job.save
+    end
+  end
+
+  def self.filter_link_exist(links)
+    return links - Job.all.map{|job| job.original_link}
+  end
 end

app/views/datas/index.html.erb

-<%= @datas %>

db/migrate/20171004094208_add_original_link_to_jobs.rb

+class AddOriginalLinkToJobs < ActiveRecord::Migration[5.1]
+  def change
+    add_column :jobs, :original_link, :string
+  end
+end

db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 20171002033456) do
+ActiveRecord::Schema.define(version: 20171004094208) do
 
   create_table "apply_jobs", force: :cascade, options: "ENGINE=InnoDB DEFAULT CHARSET=utf8" do |t|
     t.bigint "job_id"
@@ -82,6 +82,7 @@ ActiveRecord::Schema.define(version: 20171002033456) do
     t.datetime "updated_date"
     t.datetime "updated_at", null: false
     t.datetime "created_at", null: false
+    t.string "original_link"
     t.index ["city_id"], name: "index_jobs_on_city_id"
     t.index ["company_id"], name: "index_jobs_on_company_id"
   end

lib/tasks/data.rake

+ENV["RAILS_ENV"] ||= "production"
+require "./app/data/crawler.rb"
+
+namespace :data do
+  task insert_job: :environment do |t|
+    links = Crawler.get_job_link
+    links = Job.filter_link_exist(links)
+    @data = Crawler.crawl_job_infomation(links)
+    Job.create_new_jobs(@data)
+  end
+end

test/controllers/datas_controller_test.rb

-require 'test_helper'
-
-class DatasControllerTest < ActionDispatch::IntegrationTest
-  test "should get index" do
-    get datas_index_url
-    assert_response :success
-  end
-
-end