fixed review part 2

lib/tasks/web_crawler.rake

@@ -50,38 +50,39 @@ namespace :crawler do
         end
         end
         # benefits, description, requirement, other_info
         # benefits, description, requirement, other_info
         job_detail_rows = job_page.css('section.job-detail-content div.detail-row')
         job_detail_rows = job_page.css('section.job-detail-content div.detail-row')
-        benefits, description, requirement, other_info = []
+        benefits, description, requirement, other_info = ''
         job_detail_rows.each do |detail_row|
         job_detail_rows.each do |detail_row|
           detail_title = detail_row.css('.detail-title').text.strip
           detail_title = detail_row.css('.detail-title').text.strip
+          detail_content = detail_row.css(':not(h3.detail-title)')
           case detail_title
           case detail_title
           when 'Phúc lợi'
           when 'Phúc lợi'
-            benefits = detail_row.css(':not(h3.detail-title)').map(&:text).map(&:squish)[1..-1].reject(&:blank?).join('---')
+            benefits = get_detail_text(detail_content)
           when 'Mô tả Công việc'
           when 'Mô tả Công việc'
-            description = detail_row.css(':not(h3.detail-title)').map(&:text).map(&:squish)[1..-1].reject(&:blank?).join('---')
+            description = get_detail_text(detail_content)
           when 'Yêu Cầu Công Việc'
           when 'Yêu Cầu Công Việc'
-            requirement = detail_row.css(':not(h3.detail-title)').map(&:text).map(&:squish)[1..-1].reject(&:blank?).join('---')
+            requirement = get_detail_text(detail_content)
           when 'Thông tin khác'
           when 'Thông tin khác'
-            other_info = detail_row.css(':not(h3.detail-title)').map(&:text).map(&:squish)[1..-1].reject(&:blank?).join('---')
+            other_info = get_detail_text(detail_content)
           end
           end
         end
         end
         # Company
         # Company
         company_name = job_page.css('div.job-desc a.job-company-name').text
         company_name = job_page.css('div.job-desc a.job-company-name').text
-        # Cities
-        cities = job_page.css('.job-detail-content .detail-box .map p a').map(&:text)
         company_object = Company.find_or_create_by(name: company_name)
         company_object = Company.find_or_create_by(name: company_name)
-        job_object = Job.create({ title: job_title,
-                                  job_type: job_type,
-                                  salary: salary,
-                                  experience: experience,
-                                  position: level,
-                                  expiration_date: expiration_date,
-                                  description: description,
-                                  benefit: benefits,
-                                  requirement: requirement,
-                                  other_info: other_info,
-                                  company_id: company_object.id })
+        job_object = Job.find_or_create_by({  title: job_title,
+                                              job_type: job_type,
+                                              salary: salary,
+                                              experience: experience,
+                                              position: level,
+                                              expiration_date: expiration_date,
+                                              description: description,
+                                              benefit: benefits,
+                                              requirement: requirement,
+                                              other_info: other_info,
+                                              company_id: company_object.id })
         industry_objects = industries.map { |industry| Industry.find_or_create_by(name: industry) }
         industry_objects = industries.map { |industry| Industry.find_or_create_by(name: industry) }
         job_object.industries << industry_objects
         job_object.industries << industry_objects
+        # Cities
+        cities = job_page.css('.job-detail-content .detail-box .map p a').map(&:text)
         city_objects = cities.map { |city| City.find_or_create_by(name: city) }
         city_objects = cities.map { |city| City.find_or_create_by(name: city) }
         job_object.cities << city_objects
         job_object.cities << city_objects
       rescue URI::InvalidURIError => e
       rescue URI::InvalidURIError => e
@@ -112,15 +113,19 @@ namespace :crawler do
     list_location = parsed_page.css('div.main-jobs-by-location ul li')
     list_location = parsed_page.css('div.main-jobs-by-location ul li')
     list_location.each do |city|
     list_location.each do |city|
       city_name = city.text
       city_name = city.text
-      region = City.regions[:international]
+      region = :international
       if city_name.include?('Việc làm tại')
       if city_name.include?('Việc làm tại')
         city_name = city_name.remove('Việc làm tại').strip
         city_name = city_name.remove('Việc làm tại').strip
-        region = City.regions[:vietnam]
+        region = :vietnam
       end
       end
-      City.create(
+      City.find_or_create_by(
         name: city_name,
         name: city_name,
         region: region
         region: region
       )
       )
     end
     end
   end
   end
+
+  def get_detail_text(arr)
+    arr.map(&:text).map(&:squish)[1..-1].reject(&:blank?).join('---')
+  end
 end
 end