rename namespace, use length instead of count,...

lib/tasks/web_crawler.rake

@@ -2,18 +2,17 @@ require 'open-uri'
 require 'csv'
 require 'csv'
 require 'zip'
 require 'zip'
 
 
-namespace :job do
+namespace :crawler do
   desc 'importjob'
   desc 'importjob'
 
 
-  task web_job_crawler: :environment do
-    parsed_page ||= Nokogiri::HTML(HTTParty.get('https://careerbuilder.vn/viec-lam/tat-ca-viec-lam-vi.html').body)
-    job_item = parsed_page.css('div.job-item')
-    (0..job_item.count - 1).each do |item|
-      job_link = job_item[item].css('div.title a').attribute('href').text
-      unparsed_job_link = HTTParty.get(job_link)
-      parsed_job_link ||= Nokogiri::HTML(unparsed_job_link.body)
-      job_desc = parsed_job_link.css('div.job-desc')
-      job_detail = parsed_job_link.css('section.job-detail-content')
+  task jobs: :environment do
+    parsed_page = Nokogiri::HTML(HTTParty.get('https://careerbuilder.vn/viec-lam/tat-ca-viec-lam-vi.html').body)
+    jobs_item = parsed_page.css('div.job-item .job_link')
+    (0..jobs_item.length - 1).each do |item|
+      job_link = jobs_item[item].attribute('href').text
+      job_page = Nokogiri::HTML(HTTParty.get(job_link).body)
+      job_desc = job_page.css('div.job-desc')
+      job_detail = job_page.css('section.job-detail-content')
       # title - company
       # title - company
       title = job_desc.css('h1.title').text
       title = job_desc.css('h1.title').text
       company = job_desc.css('a.job-company-name').text
       company = job_desc.css('a.job-company-name').text
@@ -24,7 +23,7 @@ namespace :job do
       # city, update_at, industry, type, salary, experience, level, expiration_date
       # city, update_at, industry, type, salary, experience, level, expiration_date
       update_at, industry, type, salary, experience, level, expiration_date = ''
       update_at, industry, type, salary, experience, level, expiration_date = ''
       city = city_box.text
       city = city_box.text
-      (0..info_box_item.count - 1).each do |part|
+      (0..info_box_item.length - 1).each do |part|
         info = info_box_item[part].text
         info = info_box_item[part].text
         if info.include?(key = 'Ngày cập nhật')
         if info.include?(key = 'Ngày cập nhật')
           update_at = info.squish.remove(key).strip
           update_at = info.squish.remove(key).strip
@@ -47,13 +46,13 @@ namespace :job do
       benefit_list = []
       benefit_list = []
       other_info_list = []
       other_info_list = []
       benefits = job_detail.css('ul.welfare-list li')
       benefits = job_detail.css('ul.welfare-list li')
-      (0..benefits.count - 1).each do |part|
+      (0..benefits.length - 1).each do |part|
         benefit = benefits[part].text.strip
         benefit = benefits[part].text.strip
         benefit_list << benefit
         benefit_list << benefit
       end
       end
       # description, requirement
       # description, requirement
       description, requirement = ''
       description, requirement = ''
-      (0..job_detail_row.count - 1).each do |part|
+      (0..job_detail_row.length - 1).each do |part|
         job_detail_text = job_detail_row[part].text
         job_detail_text = job_detail_row[part].text
         if job_detail_text.include?('Mô tả Công việc')
         if job_detail_text.include?('Mô tả Công việc')
           description = job_detail_text.partition('Mô tả Công việc').last.squish.strip
           description = job_detail_text.partition('Mô tả Công việc').last.squish.strip
@@ -63,7 +62,7 @@ namespace :job do
       end
       end
       # benefit
       # benefit
       other_info = job_detail.css('div.content_fck ul li')
       other_info = job_detail.css('div.content_fck ul li')
-      (0..other_info.count - 1).each do |part|
+      (0..other_info.length - 1).each do |part|
         info = other_info[part].text.squish.strip
         info = other_info[part].text.squish.strip
         other_info_list << info
         other_info_list << info
       end
       end
@@ -89,11 +88,11 @@ namespace :job do
   end
   end
 
 
   desc 'crawler industry form CareerBuilder'
   desc 'crawler industry form CareerBuilder'
-  task web_industry_crawler: :environment do
+  task industries: :environment do
     parsed_page ||= Nokogiri::HTML(HTTParty.get('https://careerbuilder.vn/tim-viec-lam.html').body)
     parsed_page ||= Nokogiri::HTML(HTTParty.get('https://careerbuilder.vn/tim-viec-lam.html').body)
     list_job = parsed_page.css('div.list-of-working-positions ul.list-jobs li a')
     list_job = parsed_page.css('div.list-of-working-positions ul.list-jobs li a')
     industry_list = []
     industry_list = []
-    (0..list_job.count - 1).each do |part|
+    (0..list_job.length - 1).each do |part|
       industry = list_job[part].text.squish.strip
       industry = list_job[part].text.squish.strip
       industry_list << industry
       industry_list << industry
     end
     end
@@ -101,11 +100,11 @@ namespace :job do
   end
   end
 
 
   desc 'crawler city form CareerBuilder'
   desc 'crawler city form CareerBuilder'
-  task web_city_crawler: :environment do
+  task cities: :environment do
     parsed_page ||= Nokogiri::HTML(HTTParty.get('https://careerbuilder.vn/tim-viec-lam.html').body)
     parsed_page ||= Nokogiri::HTML(HTTParty.get('https://careerbuilder.vn/tim-viec-lam.html').body)
     list_location = parsed_page.css('div.main-jobs-by-location ul li')
     list_location = parsed_page.css('div.main-jobs-by-location ul li')
     city_list = []
     city_list = []
-    (0..list_location.count - 1).each do |part|
+    (0..list_location.length - 1).each do |part|
       city_item = list_location[part].text
       city_item = list_location[part].text
       region = 1
       region = 1
       if city_item.include?(key = 'Việc làm tại')
       if city_item.include?(key = 'Việc làm tại')