Start redacting sender name and princial name in defamations

CHANGELOG.md

@@ -1,5 +1,9 @@
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/). It uses [CalVer](https://calver.org/) as of May 2019.
 
+## [22.08c](https://github.com/berkmancenter/lumendatabase/releases/tag/2022.08c) - 2022-08-30
+### Changed
+* Started redacting sender and principal names in defamations.
+
 ## [22.08b](https://github.com/berkmancenter/lumendatabase/releases/tag/2022.08b) - 2022-08-26
 ### Changed
 * Started catching more 404 errors.

app/models/defamation.rb

@@ -33,4 +33,34 @@ def principal_name
   def hide_identities?
     (recipient_name =~ REDACTION_REGEX).present?
   end
+
+  def auto_redact
+    custom_redactors = [
+      InstanceRedactor::PhoneNumberRedactor.new,
+      InstanceRedactor::SSNRedactor.new,
+      InstanceRedactor::EmailRedactor.new,
+      InstanceRedactor::EntityNameRedactor.new
+    ]
+
+    instance_redactor = InstanceRedactor.new(
+      custom_redactors,
+      {
+        entity_name: principal&.name || sender&.name
+      }
+    )
+
+    instance_redactor.redact(self)
+
+    works.each do |work|
+      instance_redactor.redact(work, %w[description])
+
+      work.infringing_urls.each do |url|
+        instance_redactor.redact(url, %w[url])
+      end
+
+      work.copyrighted_urls.each do |url|
+        instance_redactor.redact(url, %w[url])
+      end
+    end
+  end
 end

app/models/instance_redactor.rb

@@ -1,8 +1,14 @@
 class InstanceRedactor
-  def initialize(redactors = [PhoneNumberRedactor.new,
-                              SSNRedactor.new,
-                              EmailRedactor.new])
+  def initialize(
+    redactors = [
+      PhoneNumberRedactor.new,
+      SSNRedactor.new,
+      EmailRedactor.new
+    ],
+    options = {}
+  )
     @redactors = redactors
+    @options = options
   end
 
   def redact(instance, field_or_fields = Notice::REDACTABLE_FIELDS)
@@ -29,7 +35,7 @@ def redact_all(instance_ids,
 
   def redact_field(instance, field, text)
     new_text = redactors.inject(text) do |result, redactor|
-      redactor.redact(result)
+      redactor.redact(result, @options)
     end
 
     return unless new_text != text
@@ -88,7 +94,7 @@ def initialize(string_or_regex)
       @string_or_regex = string_or_regex
     end
 
-    def redact(content)
+    def redact(content, _options = {})
       content.gsub(@string_or_regex) { |s| STOP_WORDS.include?(s) ? s : mask }
     end
 
@@ -98,7 +104,7 @@ def mask
   end
 
   class PhoneNumberRedactor
-    def redact(text)
+    def redact(text, _options = {})
       redactor = ContentRedactor.new(
         /(\()?\b(\(?\d{3}\)?.?)?  # optional area code
          (\d{3}[^\d]?\d{4})|      # phone number, optional single-char separator
@@ -111,7 +117,7 @@ def redact(text)
   end
 
   class SSNRedactor
-    def redact(text)
+    def redact(text, _options = {})
       redactor = ContentRedactor.new(
         /\b(\d{3})\D?(\d{2})\D?(\d{4})\b/x
       )
@@ -121,7 +127,7 @@ def redact(text)
   end
 
   class EmailRedactor
-    def redact(text)
+    def redact(text, _options = {})
       redactor = ContentRedactor.new(
         /\S+@\S+\.\S+[^.\s]/i
       )
@@ -131,15 +137,16 @@ def redact(text)
   end
 
   class EntityNameRedactor
-    def initialize(name)
-      match = name.gsub(/[-*+?\d]/, ' ').strip.split(/\s+/)
-      separator = (name =~ /[a-z]/mi ? '[^a-z]' : '\s')
+    def redact(text, options = {})
+      return text unless options[:entity_name].present?
+
+      match = options[:entity_name].gsub(/[^'0-9A-Za-z ]/, '').gsub(/[-*+?\d]/, ' ').strip.split(/\s+/)
+      separator = (options[:entity_name] =~ /[a-z]/mi ? '[^a-z]' : '\s')
       @regex_base = "(?:#{match.join('|')})(?:#{separator}*(?:#{match.join('|')}))*"
       @regex = /#{@regex_base}/mi
-    end
 
-    def redact(text)
       return text if @regex_base.blank?
+
       redactor = ContentRedactor.new(@regex)
 
       redactor.redact(text)

spec/integration/faceted_notice_search_spec.rb

@@ -59,7 +59,6 @@
       end
 
       open_dropdown_for_facet('date_received_facet')
-      page.save_screenshot(full: true)
       facet = page.find('ol.date_received_facet li:nth-child(3)').text
 
       within_faceted_search_results_for('title', :date_received_facet, facet) do

spec/integration/submit_notice_via_api_spec.rb

@@ -282,6 +282,102 @@
     expect(curb.response_code).to eq 400
   end
 
+  context 'redacting defamations' do
+    scenario 'sender name is redacted if there is no principal provided' do
+      original_text = <<-BODY
+        My name is Tonny Bastet, do something please!!!
+      BODY
+      redacted_text = <<-BODY
+        My name is [REDACTED], do something please!!!
+      BODY
+      original_infringing_url = 'http://disney.com/tonny_bastet.mp4'
+      redacted_infringing_url = 'http://disney.com/[REDACTED].mp4'
+
+      parameters = request_hash(
+        default_notice_hash(
+          type: 'Defamation',
+          body: original_text,
+          entity_notice_roles_attributes: [
+            {
+              name: 'recipient',
+              entity_attributes: {
+                name: 'The Googs'
+              }
+            },
+            {
+              name: 'sender',
+              entity_attributes: {
+                name: 'Tonny Bastet'
+              }
+            }
+          ],
+          works_attributes: [
+            {
+              description: original_text,
+              infringing_urls_attributes: [
+                {
+                  url: original_infringing_url
+                }
+              ]
+            }
+          ]
+        )
+      )
+
+      submit_and_test_defamation_redaction(parameters, original_text, redacted_text, original_infringing_url, redacted_infringing_url)
+    end
+
+    scenario 'principal name is redacted if there is principal provided' do
+      original_text = <<-BODY
+        My name not John Bastet, it's Tonny Kokosh, do something please!!!
+      BODY
+      redacted_text = <<-BODY
+        My name not John Bastet, it's [REDACTED], do something please!!!
+      BODY
+      original_infringing_url = 'http://disney.com/tonny_kokosh.mp4'
+      redacted_infringing_url = 'http://disney.com/[REDACTED].mp4'
+
+      parameters = request_hash(
+        default_notice_hash(
+          type: 'Defamation',
+          body: original_text,
+          entity_notice_roles_attributes: [
+            {
+              name: 'recipient',
+              entity_attributes: {
+                name: 'The Googs'
+              }
+            },
+            {
+              name: 'sender',
+              entity_attributes: {
+                name: 'John Bastet'
+              }
+            },
+            {
+              name: 'principal',
+              entity_attributes: {
+                name: 'Tonny Kokosh'
+              }
+            }
+          ],
+          works_attributes: [
+            {
+              description: original_text,
+              infringing_urls_attributes: [
+                {
+                  url: original_infringing_url
+                }
+              ]
+            }
+          ]
+        )
+      )
+
+      submit_and_test_defamation_redaction(parameters, original_text, redacted_text, original_infringing_url, redacted_infringing_url)
+    end
+  end
+
   private
 
   def original_document_file(notice)
@@ -353,4 +449,17 @@ def request_hash(notice_hash, user = create(:user, :submitter))
       authentication_token: user.authentication_token
     }
   end
+
+  def submit_and_test_defamation_redaction(parameters, original_text, redacted_text, original_infringing_url, redacted_infringing_url)
+    post_api('/notices', parameters)
+
+    notice = Notice.last
+
+    expect(notice.body).to eq redacted_text
+    expect(notice.body_original).to eq original_text
+    expect(notice.works.first.description).to eq redacted_text
+    expect(notice.works.first.description_original).to eq original_text
+    expect(notice.works.first.infringing_urls.first.url).to eq redacted_infringing_url
+    expect(notice.works.first.infringing_urls.first.url_original).to eq original_infringing_url
+  end
 end

spec/support/notice_actions.rb

@@ -1,10 +1,10 @@
 require 'support/sign_in'
 
 module NoticeActions
-  def submit_recent_notice(title = 'A title')
+  def submit_recent_notice(title = 'A title', type = 'DMCA')
     sign_in(create(:user, :submitter))
 
-    visit '/notices/new?type=DMCA'
+    visit "/notices/new?type=#{type}"
 
     fill_in 'Title', with: title
     fill_in 'Date received', with: Time.now