Feature/crawl subjects

lib/mwcrawler.rb

@@ -9,6 +9,7 @@
 require 'mwcrawler/classes'
 require 'mwcrawler/courses'
 require 'mwcrawler/departments'
+require 'mwcrawler/subjects'
 require 'mwcrawler/helpers'
 require 'mwcrawler/crawler'
 

lib/mwcrawler/crawler.rb

@@ -18,6 +18,11 @@ class Crawler
       end
     end
 
+    def subjects(department, options = { log: false })
+      Options.init(options)
+      Subjects.scrap department, options
+    end
+
     def semester
       page = Helpers.set_crawler(nil, 'graduacao/default.aspx', exact: true)
       page.css("a[title='Período Atual'] span").first.text

lib/mwcrawler/subjects.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+module Mwcrawler
+  # Scraps Subjects by department
+  module Subjects
+    def self.scrap(department_or_id, options)
+      if options[:by_id]
+        subject_by_id(department_or_id)
+      elsif options[:by_department]
+        subject_by_department(department_or_id)
+      else
+        raise ArgumentError, 'second argument not specified. You can find a subject by department code or id'
+      end
+    end
+
+    private_class_method def self.subject_by_department(department)
+      page = Helpers.set_crawler(department, 'graduacao/oferta_dis.aspx?cod=', exact: true)
+      scrap_row(department, page)
+    end
+
+    private_class_method def self.subject_by_id(id)
+      page = Helpers.set_crawler(id, 'graduacao/oferta_dados.aspx?cod=', exact: true)
+      row_init_by_id(page)
+    end
+
+    private_class_method def self.row_init_by_id(page)
+      { code: page.css('#datatable')[0].css('tr:nth-child(2) td').text.to_i,
+        name: page.css('#datatable')[0].css('tr:nth-child(3) td').text,
+        department: page.css('#datatable tr:first-child a').first['href'].scan(/\d+/)[0].to_i,
+        level: 'graduação' }
+    end
+
+    private_class_method def self.scrap_row(dep_code, page)
+      subjects = []
+      length = page.css('#datatable tr td:nth-child(1)').count
+      length.times do |i|
+        subjects << row_init_by_department(page, dep_code, i)
+      end
+      subjects
+    end
+
+    private_class_method def self.row_init_by_department(page, dep_code, index)
+      { code: page.css('#datatable tr td:nth-child(1)').map(&:text)[index].to_i,
+        name: page.css('#datatable tr td:nth-child(2)').map(&:text)[index],
+        department: dep_code.to_i,
+        level: 'graduação' }
+    end
+  end
+end

spec/subjects_spec.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+RSpec.describe Mwcrawler::Subjects do
+  context 'Subjects by department: CIC' do
+    before :all do
+      VCR.use_cassette('subjects/subjects_cic') do
+        @subjects = Mwcrawler::Crawler.new.subjects('116', by_department: true)
+      end
+    end
+
+    it { expect(@subjects).to be_a_kind_of Array }
+
+    it { expect(@subjects).not_to be nil }
+
+    it { expect(@subjects.first.keys).to include(:code, :department, :name, :level) }
+
+    it { expect(@subjects.first).to include(code: be_integer) }
+
+    it { expect(@subjects.first).to include(code: be_nonzero) }
+
+    it { expect(@subjects.first).to include(department: be_integer) }
+
+    it { expect(@subjects.first).to include(department: be_nonzero) }
+  end
+
+  context 'crawls subject by id' do
+    before :all do
+      VCR.use_cassette('subjects/subject_116441') do
+        @subject = Mwcrawler::Crawler.new.subjects('116441', by_id: true)
+      end
+    end
+
+    it { expect(@subject).to be_a_kind_of Hash }
+
+    it { expect(@subject.keys).to include(:code, :department, :name, :level) }
+
+    it { expect(@subject).to include(code: be_integer) }
+
+    it { expect(@subject).to include(code: be_nonzero) }
+
+    it { expect(@subject).to include(department: be_integer) }
+
+    it { expect(@subject).to include(department: be_nonzero) }
+  end
+end