Merge pull request #1183 from DaanVanVugt/feature/han_ingestor

han material ingestor

Author: fbacall

lib/ingestors/ingestor_factory.rb

@@ -36,6 +36,7 @@ def self.taxila_ingestors
         Ingestors::Taxila::DccIngestor,
         Ingestors::Taxila::SenseIngestor,
         Ingestors::Taxila::VuMaterialIngestor,
+        Ingestors::Taxila::HanIngestor,
       ]
     end
 

lib/ingestors/taxila/han_ingestor.rb

@@ -0,0 +1,45 @@
+require 'open-uri'
+require 'csv'
+require 'nokogiri'
+
+module Ingestors
+  module Taxila
+    class HanIngestor < Ingestor
+      def self.config
+        {
+          key: 'han_material',
+          title: 'HAN Materials API',
+          category: :materials
+        }
+      end
+
+      def read(url)
+        begin
+          process_han(url)
+        rescue Exception => e
+          @messages << "#{self.class.name} failed with: #{e.message}"
+        end
+
+        # finished
+        nil
+      end
+
+      private
+
+      def process_han(_url)
+        url = 'https://www.han.nl/studeren/scholing-voor-werkenden/laboratorium/'
+
+        material_page = Nokogiri::HTML5.parse(open_url(url.to_s, raise: true)).css('#content > .section--cards-skinny > .section-wrapper > .section__content > .cards-skinny > .cards-skinny-wrapper > .cards-skinny__item > .card-skinny > .card-skinny__content')
+        material_page.each_with_index do |el, _idx|
+          material = OpenStruct.new
+          material.title = el.css('.card-skinny__content__title').first.text
+          material.url = "https://www.han.nl#{el.css('.card-skinny__content__buttons > .buttons > .buttons__button > a').first.get_attribute('href')}"
+          material.description = el.css('.card-skinny__content__body').first.text
+          add_material(material)
+        rescue Exception => e
+          @messages << "Extract event fields failed with: #{e.message}"
+        end
+      end
+    end
+  end
+end

test/unit/ingestors/taxila/han_ingestor_test.rb

@@ -0,0 +1,49 @@
+require 'test_helper'
+
+class HanIngestorTest < ActiveSupport::TestCase
+  setup do
+    @user = users(:regular_user)
+    @content_provider = content_providers(:portal_provider)
+    mock_ingestions
+    mock_timezone # System time zone should not affect test result
+  end
+
+  teardown do
+    reset_timezone
+  end
+
+  test 'can ingest materials from han' do
+    source = @content_provider.sources.build(
+      url: 'https://www.han.nl/studeren/scholing-voor-werkenden/laboratorium/',
+      method: 'han',
+      enabled: true
+    )
+
+    ingestor = Ingestors::Taxila::HanIngestor.new
+
+    # check materials don't exist
+    new_title = 'Synthetiseren en Karakteriseren van Moleculen'
+    new_url = 'https://www.han.nl/opleidingen/module/synthetiseren-karakteriseren-moleculen/'
+    refute Material.where(title: new_title, url: new_url).any?
+
+    # run task
+    assert_difference('Material.count', 27) do
+      freeze_time(2019) do
+        VCR.use_cassette('ingestors/han') do
+          ingestor.read(source.url)
+          ingestor.write(@user, @content_provider)
+        end
+      end
+    end
+
+    # check event does exist
+    material = Material.where(title: new_title, url: new_url).first
+    assert material
+    assert_equal new_title, material.title
+    assert_equal new_url, material.url
+
+    # check other fields
+    # assert_equal 'Software Carpentry lessons introduce basic lab skills for research computing. They cover three core topics: the Unix shell, version control with Git, and a programming language (Python or R). ',
+    #              material.description
+  end
+end