feat(procedure): validate external links

Gemfile

@@ -8,6 +8,7 @@ gem 'active_link_to' # Automatically set a class on active links
 gem 'active_model_serializers'
 gem 'activestorage-openstack'
 gem 'active_storage_validations'
+gem 'addressable'
 gem 'administrate'
 gem 'administrate-field-enum' # Allow using Field::Enum in administrate
 gem 'after_party'

Gemfile.lock

@@ -800,6 +800,7 @@ DEPENDENCIES
   active_model_serializers
   active_storage_validations
   activestorage-openstack
+  addressable
   administrate
   administrate-field-enum
   after_party

app/graphql/schema.graphql

@@ -1016,7 +1016,7 @@ type DemarcheDescriptor {
   """
   URL pour commencer la démarche
   """
-  demarcheUrl: String
+  demarcheUrl: URL
 
   """
   Description de la démarche.
@@ -1039,7 +1039,7 @@ type DemarcheDescriptor {
   notice explicative de la démarche
   """
   notice: File
-  noticeUrl: String
+  noticeUrl: URL
 
   """
   Numero de la démarche.

app/graphql/types/demarche_descriptor_type.rb

@@ -25,10 +25,10 @@ Cela évite l’accès récursif aux dossiers."
 
     field :duree_conservation_dossiers, Int, "Durée de conservation des dossiers en mois.", null: false
 
-    field :demarche_url, String, "URL pour commencer la démarche", null: true
+    field :demarche_url, Types::URL, "URL pour commencer la démarche", null: true
     field :site_web_url, String, "URL où les usagers trouvent le lien vers la démarche", null: true
     field :dpo_url, String, "URL ou email pour contacter le Délégué à la Protection des Données (DPO)", null: true
-    field :notice_url, String, null: true
+    field :notice_url, Types::URL, null: true
     field :cadre_juridique_url, String, "URL du cadre juridique qui justifie le droit de collecter les données demandées dans la démarche", null: true
 
     field :opendata, Boolean, null: false

app/graphql/types/url.rb

@@ -3,12 +3,14 @@ module Types
     description "A valid URL, transported as a string"
 
     def self.coerce_input(input_value, context)
-      url = URI.parse(input_value)
-      if url.is_a?(URI::HTTP) || url.is_a?(URI::HTTPS)
+      url = Addressable::URI(input_value)
+      if uri.scheme.in?(['http', 'https'])
         url
       else
         raise GraphQL::CoercionError, "#{input_value.inspect} is not a valid URL"
       end
+    rescue Addressable::URI::InvalidURIError
+      raise GraphQL::CoercionError, "#{input_value.inspect} is not a valid URL"
     end
 
     def self.coerce_result(ruby_value, context)

app/helpers/procedure_helper.rb

@@ -48,7 +48,7 @@ module ProcedureHelper
   def url_or_email_to_lien_dpo(procedure)
     URI::MailTo.build([procedure.lien_dpo, "subject="]).to_s
   rescue URI::InvalidComponentError
-    uri = URI.parse(procedure.lien_dpo)
+    uri = Addressable::URI.parse(procedure.lien_dpo)
     return "//#{uri}" if uri.scheme.nil?
     uri.to_s
   end

app/jobs/cron/procedure_external_url_check_job.rb

@@ -0,0 +1,7 @@
+class Cron::ProcedureExternalURLCheckJob < Cron::CronJob
+  self.schedule_expression = "every week on monday at 1 am"
+
+  def perform
+    Procedure.with_external_urls.find_each { ::ProcedureExternalURLCheckJob.perform_later(_1) }
+  end
+end

app/jobs/procedure_external_url_check_job.rb

@@ -0,0 +1,33 @@
+class ProcedureExternalURLCheckJob < ApplicationJob
+  def perform(procedure)
+    procedure.validate
+
+    if procedure.lien_notice.present?
+      error = procedure.errors.find { _1.attribute == :lien_notice }
+      if error.present?
+        procedure.update!(lien_notice_error: error.message)
+      else
+        response = Typhoeus.get(procedure.lien_notice, followlocation: true)
+        if response.success?
+          procedure.update!(lien_notice_error: nil)
+        else
+          procedure.update!(lien_notice_error: "#{response.code} #{response.return_message}")
+        end
+      end
+    end
+
+    if procedure.lien_dpo.present? && !procedure.lien_dpo_email?
+      error = procedure.errors.find { _1.attribute == :lien_dpo }
+      if error.present?
+        procedure.update!(lien_dpo_error: error.message)
+      else
+        response = Typhoeus.get(procedure.lien_dpo, followlocation: true)
+        if response.success?
+          procedure.update!(lien_dpo_error: nil)
+        else
+          procedure.update!(lien_dpo_error: "#{response.code} #{response.return_message}")
+        end
+      end
+    end
+  end
+end

app/models/procedure.rb

@@ -234,6 +234,8 @@ class Procedure < ApplicationRecord
   scope :opendata,               -> { where(opendata: true) }
   scope :publiees_ou_closes,     -> { where(aasm_state: [:publiee, :close, :depubliee]) }
 
+  scope :with_external_urls,     -> { where.not(lien_notice: [nil, '']).or(where.not(lien_dpo: [nil, ''])) }
+
   scope :publiques,              -> do
     publiees_ou_closes
       .opendata
@@ -294,7 +296,12 @@ class Procedure < ApplicationRecord
   validates :libelle, presence: true, allow_blank: false, allow_nil: false
   validates :description, presence: true, allow_blank: false, allow_nil: false
   validates :administrateurs, presence: true
+
   validates :lien_site_web, presence: true, if: :publiee?
+  validates :lien_notice, url: { no_local: true, allow_blank: true }
+  validates :lien_dpo, format: { with: Devise.email_regexp, message: "n'est pas valide" }, if: :lien_dpo_email?
+  validates :lien_dpo, url: { no_local: true, allow_blank: true }, unless: :lien_dpo_email?
+
   validates :draft_types_de_champ_public,
     'types_de_champ/no_empty_block': true,
     'types_de_champ/no_empty_drop_down': true,
@@ -320,7 +327,6 @@ class Procedure < ApplicationRecord
                                                     less_than_or_equal_to: 60
                                                   }
 
-  validates :lien_dpo, email_or_link: true, allow_nil: true
   validates_with MonAvisEmbedValidator
 
   validates_associated :draft_revision, on: :publication
@@ -978,6 +984,10 @@ class Procedure < ApplicationRecord
     update!(routing_enabled: self.groupe_instructeurs.active.many?)
   end
 
+  def lien_dpo_email?
+    lien_dpo.present? && lien_dpo.match?(/@/)
+  end
+
   private
 
   def validate_auto_archive_on_in_the_future

app/validators/email_or_link_validator.rb

@@ -1,7 +0,0 @@
-class EmailOrLinkValidator < ActiveModel::EachValidator
-  def validate_each(record, attribute, value)
-    URI.parse(value)
-  rescue URI::InvalidURIError
-    record.errors.add(attribute, :invalid_uri_or_email)
-  end
-end

app/validators/url_validator.rb

@@ -0,0 +1,69 @@
+require 'active_model'
+require 'active_support/i18n'
+require 'public_suffix'
+require 'addressable/uri'
+
+# Most of this code is borowed from https://github.com/perfectline/validates_url
+
+class URLValidator < ActiveModel::EachValidator
+  RESERVED_OPTIONS = [:schemes, :no_local]
+
+  def initialize(options)
+    options.reverse_merge!(schemes: ['http', 'https'])
+    options.reverse_merge!(message: :url)
+    options.reverse_merge!(no_local: false)
+    options.reverse_merge!(public_suffix: false)
+    options.reverse_merge!(accept_array: false)
+
+    super(options)
+  end
+
+  def validate_each(record, attribute, value)
+    message = options.fetch(:message)
+    schemes = [*options.fetch(:schemes)].map(&:to_s)
+
+    if value.respond_to?(:each)
+      # Error out if we're not allowing arrays
+      if !options.include?(:accept_array) || !options.fetch(:accept_array)
+        record.errors.add(attribute, message, **filtered_options(value))
+      end
+
+      # We have to manually handle `:allow_nil` and `:allow_blank` since it's not caught by
+      # ActiveRecord's own validators. We do that by just removing all the nil's if we want to
+      # allow them so it's not passed on later.
+      value = value.compact if options.include?(:allow_nil) && options.fetch(:allow_nil)
+      value = value.compact_blank if options.include?(:allow_blank) && options.fetch(:allow_blank)
+
+      result = value.flat_map { validate_url(record, attribute, _1, message, schemes) }
+      errors = result.compact
+
+      return errors.any? ? errors.first : true
+    end
+
+    validate_url(record, attribute, value, message, schemes)
+  end
+
+  protected
+
+  def filtered_options(value)
+    filtered = options.except(*RESERVED_OPTIONS)
+    filtered[:value] = value
+    filtered
+  end
+
+  def validate_url(record, attribute, value, message, schemes)
+    uri = Addressable::URI.parse(value)
+    host = uri && uri.host
+    scheme = uri && uri.scheme
+
+    valid_scheme = host && scheme && schemes.include?(scheme)
+    valid_no_local = !options.fetch(:no_local) || (host && host.include?('.'))
+    valid_suffix = !options.fetch(:public_suffix) || (host && PublicSuffix.valid?(host, default_rule: nil))
+
+    unless valid_scheme && valid_no_local && valid_suffix
+      record.errors.add(attribute, message, **filtered_options(value))
+    end
+  rescue Addressable::URI::InvalidURIError
+    record.errors.add(attribute, message, **filtered_options(value))
+  end
+end

config/locales/en.yml

@@ -335,7 +335,6 @@ en:
           updated_at: "Updated on %{datetime}"
         edit:
           autosave: Your file is automatically saved after each modification. You can close the window at any time and pick up where you left off later.
-          notice: "Download the notice of the procedure"
           pending_correction:
             confirm_label: I certify that I have made all corrections requested by the administration.
         messages:
@@ -569,6 +568,7 @@ en:
       messages:
         not_a_phone: 'Invalid phone number'
         not_a_rna: 'Invalid RNA number'
+        url: 'is not a valid link'
       models:
         attestation_template:
           attributes:

config/locales/fr.yml

@@ -335,7 +335,6 @@ fr:
           updated_at: "Modifié le %{datetime}"
         edit:
           autosave: Votre dossier est enregistré automatiquement après chaque modification. Vous pouvez à tout moment fermer la fenêtre et reprendre plus tard là où vous en étiez.
-          notice: Télécharger le guide de la démarche
           pending_correction:
             confirm_label: Je certifie avoir effectué toutes les corrections demandées par l’administration.
         messages:
@@ -572,6 +571,7 @@ fr:
       messages:
         not_a_phone: 'Numéro de téléphone invalide'
         not_a_rna: 'Numéro RNA invalide'
+        url: 'n’est pas un lien valide'
       models:
         attestation_template:
           attributes: