Refactor and expand Paper validations

app/models/paper.rb

@@ -1,12 +1,23 @@
 require 'elasticsearch/model'
 require 'json'
+require 'parseable_date_validator'
 
 class Paper < ActiveRecord::Base
   include Elasticsearch::Model
   include Elasticsearch::Model::Callbacks
 
-  validates_presence_of :body, :content, :name , :originator, :paper_type, :reference, :url, :published_at
-  validates :url, uniqueness: true
+  validates :name,         presence: true, length: { maximum: 1000 }
+  validates :url,          presence: true,
+                           length: { maximum: 1000 },
+                           uniqueness: true, # TODO use unique index instead
+                           url: true
+  validates :reference,    presence: true, length: { maximum: 100 }
+  validates :body,         presence: true, length: { maximum: 100 }
+  validates :content,      presence: true, length: { maximum: 100_000 }
+  validates :originator,   presence: true, length: { maximum: 300 }
+  validates :paper_type,   presence: true, length: { maximum: 50 }
+  validates :published_at, presence: true, parseable_date: true
+  validates :resolution,   length: { maximum: 30_000 }
 
   settings index: { number_of_shards: 1 } do
     mappings dynamic: false do

config/locales/de.yml

@@ -1,2 +1,5 @@
 de:
-  hello: 'Hallo'
+  errors:
+    messages:
+      unparseable_date: "ist kein gültiges Datum"
+      url: "ist keine gültige URL"

config/locales/en.yml

@@ -0,0 +1,5 @@
+en:
+  errors:
+    messages:
+      unparseable_date: "is not a parseable date"
+      url: "is not a valid URL"

lib/parseable_date_validator.rb

@@ -0,0 +1,12 @@
+class ParseableDateValidator < ActiveModel::EachValidator
+
+  def validate_each(record, attribute, value)
+    raw_value = record.read_attribute_before_type_cast(attribute)
+    return nil if raw_value.nil?
+    Date.parse(raw_value)
+    nil
+  rescue ArgumentError => e
+    record.errors[attribute] << (options[:message] || I18n.t("errors.messages.unparseable_date"))
+  end
+
+end

test/factories/papers.rb

@@ -0,0 +1,14 @@
+FactoryGirl.define do
+  factory :paper do
+    name         Faker::Lorem.sentence
+    url          Faker::Internet.url("example-ris.de")
+    reference    "MyString"
+    body         "leipzig"
+    published_at "2015-07-20 21:16:53"
+    scraped_at   "2015-07-20 21:16:53"
+    paper_type   "Verwaltungsstandpunkt"
+    originator   Faker::Name.name
+    resolution   Faker::Lorem.paragraph(3)
+    content      "-------------------------------\n\n "  # oh well...
+  end
+end

test/models/paper_test.rb

@@ -3,14 +3,50 @@ require 'test_helper'
 class PaperTest < ActiveSupport::TestCase
 
   context "Validations" do
-    should validate_presence_of(:body)
-    should validate_presence_of(:content)
     should validate_presence_of(:name)
-    should validate_presence_of(:originator)
-    should validate_presence_of(:paper_type)
-    should validate_presence_of(:reference)
+    should validate_length_of(:name).is_at_most(1000)
+
     should validate_presence_of(:url)
-    should validate_presence_of(:published_at)
+    should validate_length_of(:url).is_at_most(1000)
+    context "URL uniqueness" do
+      subject { FactoryGirl.build(:paper) }
+      should validate_uniqueness_of(:url)
+    end
+    should "validate url format sane" do
+      expected_error = "ist keine gültige URL"
+      paper = FactoryGirl.build(:paper, url: "wtf")
+      assert_not paper.valid?, "Expected paper to not be valid with invalid URL"
+      assert_not paper.errors[:url].empty?
+      assert paper.errors[:url].include?(expected_error), "Expected #{paper.errors[:url]} to include \"#{expected_error}\""
+    end
+
+    should validate_presence_of(:reference)
+    should validate_length_of(:reference).is_at_most(100)
+
+    should validate_presence_of(:body)
+    should validate_length_of(:body).is_at_most(100)
+
+    should validate_presence_of(:content)
+    should validate_length_of(:content).is_at_most(100_000)
+
+    should validate_presence_of(:originator)
+    should validate_length_of(:originator).is_at_most(300)
+
+    should validate_presence_of(:paper_type)
+    should validate_length_of(:paper_type).is_at_most(50)
+
+    context "published_at" do
+      should validate_presence_of(:published_at)
+      should "validate date is parseable" do
+        expected_error = "ist kein gültiges Datum"
+        paper = FactoryGirl.build(:paper, published_at: "fubar")
+        assert_not paper.valid?
+        assert_not paper.errors[:published_at].empty?
+        assert paper.errors[:published_at].include?(expected_error), "Expected #{paper.errors[:published_at]} to include \"#{expected_error}\""
+      end
+    end
+
+    should validate_length_of(:resolution).is_at_most(30_000)
   end
 
 end