Merge pull request #766 from alphagov/fix-organogram-filenames

Fix organogram filenames

Author: kentsanggds

lib/tasks/fix_existing_organogram_filenames.rake

@@ -0,0 +1,6 @@
+require_relative 'update_organogram_filenames.rb'
+
+desc "Fix existing organogram filenames"
+task "fix_existing_organogram_filenames" => :environment do
+  UpdateOrganogramFilenames.new.call
+end

lib/tasks/update_organogram_filenames.rb

@@ -0,0 +1,54 @@
+require 'csv'
+
+class UpdateOrganogramFilenames
+  def initialize
+    @old_urls = []
+    @new_urls = []
+  end
+
+  def call
+    csv_file = "./lib/tasks/old_new_urls.csv"
+
+    parse_csv(csv_file)
+    msg = replace_urls
+    puts msg
+  end
+
+  def parse_csv(csv_file)
+    puts "Parsing CSV '#{csv_file}'"
+    CSV.foreach(csv_file) do |row|
+      @old_urls << row.first
+      @new_urls << row.last.strip
+    end
+  end
+
+  def replace_urls
+    puts "Searching for urls containing '-posts-'..."
+
+    if @old_urls.empty?
+      "No urls to process"
+    else
+      Link.all.each do |link|
+        if link.url.include? "-posts-"
+          index = @old_urls.index(link.url)
+          if index != nil
+            puts "From dataset: " + link.dataset.name
+            puts "Replace url '" + link.url + "' with '" + @new_urls[index] + "'"
+            link.url = @new_urls[index]
+            link.save(validate: false)
+
+            if Link.where(url: @new_urls[index]).empty?
+              puts "Url replacement failed"
+            else
+              puts "Url successfully replaced"
+            end
+          else
+            puts "WARNING: " + link.url + " not found"
+          end
+          puts "==============="
+        end
+      end
+      "Update complete"
+    end
+  end
+end

spec/lib/tasks/update_organogram_filenames_spec.rb

@@ -0,0 +1,38 @@
+require 'rails_helper'
+require_relative '../../../lib/tasks/update_organogram_filenames.rb'
+
+describe UpdateOrganogramFilenames do
+  before do
+    @link1 = FactoryBot.create(:link, url: "https://amazonaws.com/datagovuk/dataset/resources/organogram-senior-posts-2019-06-06T11-18-26Z.csv")
+    @link2 = FactoryBot.create(:link, url: "https://amazonaws.com/datagovuk/dataset/resources/organogram-junior-posts-2019-06-06T11-18-30Z.csv")
+    File.write("sample_urls.csv", "https://amazonaws.com/datagovuk/dataset/resources/organogram-senior-posts-2019-06-06T11-18-26Z.csv, https://amazonaws.com/datagovuk/dataset/resources/2019-06-06T11-18-26Z-organogram-senior.csv\nhttps://amazonaws.com/datagovuk/dataset/resources/organogram-junior-posts-2019-06-06T11-18-30Z.csv, https://amazonaws.com/datagovuk/dataset/resources/2019-06-06T11-18-26Z-organogram-junior.csv")
+  end
+
+  context "when parsing a CSV file" do
+    it "should set old_urls and new_urls array if it contains data" do
+      update_organogram_filenames = UpdateOrganogramFilenames.new
+      update_organogram_filenames.parse_csv("sample_urls.csv")
+      expect(update_organogram_filenames.instance_eval { @old_urls }.length).to be(2)
+      expect(update_organogram_filenames.instance_eval { @new_urls }.length).to be(2)
+    end
+
+    it "should abort and return a message if it's empty" do
+      File.write("sample_urls.csv", "")
+      update_organogram_filenames = UpdateOrganogramFilenames.new
+      update_organogram_filenames.parse_csv("sample_urls.csv")
+      expect(update_organogram_filenames.replace_urls).to eql("No urls to process")
+    end
+  end
+
+  context "when a url containing '-posts-' is found" do
+    it "replaces the url with the correct url" do
+      update_organogram_filenames = UpdateOrganogramFilenames.new
+      update_organogram_filenames.parse_csv("sample_urls.csv")
+
+      update_organogram_filenames.replace_urls
+
+      expect(Link.all[0].url).to eql("https://amazonaws.com/datagovuk/dataset/resources/2019-06-06T11-18-26Z-organogram-senior.csv")
+      expect(Link.all[1].url).to eql("https://amazonaws.com/datagovuk/dataset/resources/2019-06-06T11-18-26Z-organogram-junior.csv")
+    end
+  end
+end