Note persistent id issue

backend/app/lib/job_runners/ns2_remover_runner.rb

@@ -0,0 +1,65 @@
+include JSONModel
+
+class NS2RemoverRunner < JobRunner
+  register_for_job_type('ns2_remover_job',
+                        :create_permissions => :administer_system,
+                        :cancel_permissions => :administer_system)
+
+  def run
+    begin
+      modified_records = []
+
+      job_data = @json.job
+
+      RequestContext.open(:repo_id => @job.repo_id) do
+        count = 0
+        Note.each do |n|
+          parent = NotePersistentId.where(:note_id => n[:id]).first
+          next unless ['resource', 'archival_object', 'digital_object', 'digital_object_component'].include?(parent[:parent_type])
+          if n.notes.lit.include?(' ns2:')
+            replaced = n.notes.lit.gsub('ns2:', '')
+            if replaced != n[:notes].lit
+              count += 1
+              parent_repo = parent[:parent_type].capitalize.constantize.where(id: parent[:parent_id]).get(:repo_id)
+              if job_data['dry_run']
+                changes = <<~CHANGES
+                  Note:
+                    #{n[:notes].lit}
+                  would become:
+                    #{replaced}\n
+                CHANGES
+                @job.write_output(changes)
+              else
+                n.update(:notes => replaced.to_sequel_blob)
+              end
+              if parent_repo == @job.repo_id
+                modified_records << JSONModel(parent[:parent_type].to_sym).uri_for(parent[:parent_id], :repo_id => @job.repo_id)
+              else
+                modified_records << JSONModel(parent[:parent_type].to_sym).uri_for(parent[:parent_id], :repo_id => parent_repo)
+              end
+            end
+          end
+        end
+
+        @job.write_output("#{count} note(s)#{' would be' if job_data['dry_run']} modified.")
+        @job.write_output("================================")
+      end
+
+      if job_data['dry_run']
+        @job.write_output("Dry run complete, no records modified.")
+      elsif modified_records.empty?
+        @job.write_output("All done, no records modified.")
+      else
+        @job.write_output("All done, logging modified records.")
+      end
+
+      self.success!
+
+      @job.record_created_uris(modified_records.uniq) unless job_data['dry_run']
+    rescue Exception => e
+      @job.write_output(e.message)
+      @job.write_output(e.backtrace)
+      raise e
+    end
+  end
+end

backend/spec/model_ns2_remover_job_spec.rb

@@ -0,0 +1,82 @@
+require 'spec_helper'
+
+def ns2_remover_job(dry_run = false)
+  build(
+    :json_job,
+    :job_type => 'ns2_remover_job',
+    :job => build(:json_ns2_remover_job, dry_run: dry_run)
+  )
+end
+
+def create_messy_note
+  content =<<~CONTENT
+    <p>I am a <extref ns2:actuate="onRequest" ns2:show="new" 
+    ns2:title="oh no" ns2:href="http://example.com/">bad, bad note</extref>.<p>
+  CONTENT
+  note = build(:json_note_bibliography,
+    :content => [content],
+    :persistent_id => "123456")
+
+  note
+end
+
+describe "ns2 remover job model" do
+
+  let(:user) { create_nobody_user }
+  let(:resource) {
+    create(:json_resource, {
+           :notes => [create_messy_note] })
+  }
+
+  it "can create a ns2 remover job" do
+    json = ns2_remover_job
+    job = Job.create_from_json(
+      json,
+      :repo_id => $repo_id,
+      :user => user
+    )
+
+    jr = JobRunner.for(job)
+    jr.run
+    job.refresh
+
+    expect(job).not_to be_nil
+    expect(job.job_type).to eq('ns2_remover_job')
+    expect(job.owner.username).to eq('nobody')
+  end
+
+  it "deletes 'ns2:' strings in notes" do
+    expect(Resource.to_jsonmodel(resource.id).notes[0]['content']).to include(/ ns2:/)
+
+    json = ns2_remover_job
+    job = Job.create_from_json(
+      json,
+      :repo_id => $repo_id,
+      :user => user
+    )
+
+    jr = JobRunner.for(job)
+    jr.run
+    job.refresh
+
+    expect(Resource.to_jsonmodel(resource.id).notes[0]['content']).not_to include(/ ns2:/)
+  end
+
+  it "won't delete 'ns2:' strings in notes when running dry run" do
+    expect(Resource.to_jsonmodel(resource.id).notes[0]['content']).to include(/ ns2:/)
+
+    json = ns2_remover_job(true)
+    job = Job.create_from_json(
+      json,
+      :repo_id => $repo_id,
+      :user => user
+    )
+
+    jr = JobRunner.for(job)
+    jr.run
+    job.refresh
+
+    expect(Resource.to_jsonmodel(resource.id).notes[0]['content']).to include(/ ns2:/)
+  end
+
+end

common/db/migrations/160_persistent_ids_for_notes_created_during_migrations.rb

@@ -0,0 +1,31 @@
+require 'json'
+
+Sequel.migration do
+  $stderr.puts "Assigning note persistent ids for agent contact notes"
+  up do
+    suspect_fks = [
+      :rights_statement_id,
+      :lang_material_id,
+      :agent_topic_id,
+      :agent_place_id,
+      :agent_occupation_id,
+      :agent_function_id,
+      :agent_gender_id,
+      :used_language_id,
+      :agent_contact_id
+    ]
+
+    self[:note]
+      .left_join(:note_persistent_id, Sequel.qualify(:note, :id) => Sequel.qualify(:note_persistent_id, :note_id))
+      .filter(Sequel.qualify(:note_persistent_id, :persistent_id) => nil)
+      .select(Sequel.qualify(:note, :id), :notes, *suspect_fks)
+      .each do |row|
+      notes = JSON.parse(row[:notes].to_s)
+      next unless notes["persistent_id"]
+      row.reject! { |k, v| v.nil? }
+      parent_type = row.keys.last.to_s.sub(/_id$/, '')
+      parent_id = row.values.last
+      self[:note_persistent_id].insert(note_id: row[:id], persistent_id: notes["persistent_id"], parent_type: parent_type, parent_id: parent_id)
+    end
+  end
+end

common/locales/en.yml

@@ -2494,6 +2494,7 @@ en:
       unknown_job_type: Unknown job type
       bulk_import_job: Load via Spreadsheet
       top_container_linker_job: Top Container Linker
+      ns2_remover_job: ns2 Namespace Remover
     status: Status
     status_completed: Completed
     status_queued: Queued
@@ -2574,6 +2575,14 @@ en:
   trim_whitespace_job:
     description: Removes leading and trailing whitespace from titles for resources, accessions, digital objects, archival objects, and digital object components.
 
+  ns2_remover_job:
+      description: >
+        Archivists' Toolit-era EADs imported into ASpace may have vestigal "ns2" namespace references pre-pended to various EAD tag attributes inside note subrecords.
+        This may produce validation errors on EAD export from ArchivesSpace, as the root <ead> tag doesn't declare the "ns2" namespace.
+        This job bulk deletes "ns2" references within Notes subrecords attached to resource, archival_object, digital_object, and digital_object_component
+        records. As this job works across all repositories, it is restricted to users with <b>System Administrator permissions only</b>.
+      dry_run: Dry run?
+
   advanced_search:
     type:
       text: Text

common/schemas/ns2_remover_job.rb

@@ -0,0 +1,16 @@
+{
+  :schema => {
+    "$schema" => "http://www.archivesspace.org/archivesspace.json",
+    "version" => 1,
+    "type" => "object",
+
+    "properties" => {
+
+      "dry_run" => {
+        "type" => "boolean",
+        "default" => false
+      }
+
+    }
+  }
+}

common/spec/lib/factory_bot_helpers.rb

@@ -890,6 +890,10 @@ def few_or_none(key)
     {}
   end
 
+  factory :json_ns2_remover_job, class: JSONModel(:ns2_remover_job) do
+    dry_run { false }
+  end
+
   factory :json_group, class: JSONModel(:group) do
     group_code { generate(:alphanumstr) }
     description { generate(:generic_description) }

frontend/app/helpers/resolver_helper.rb

@@ -1,7 +1,7 @@
 module ResolverHelper
 
-  def resolve_readonly_link_to(label, uri)
-    link_to label, :controller => :resolver, :action => :resolve_readonly, :uri => uri
+  def resolve_readonly_link_to(label, uri, active = true)
+    link_to_if active, label, :controller => :resolver, :action => :resolve_readonly, :uri => uri
   end
 
 

frontend/app/views/jobs/_form.html.erb

@@ -96,6 +96,16 @@
   <br/>
 <% end %>
 
+<% define_template("ns2_remover_job", jsonmodel_definition(:ns2_remover_job)) do |form| %>
+  <p><%= I18n.t("ns2_remover_job.description") %></p>
+
+  <div class="col-sm-9 checkbox">
+    <%= form.label_and_boolean "dry_run" %>
+  </div>
+  <br/>
+  <br/>
+<% end %>
+
 <% define_template("job", jsonmodel_definition(:job)) do |form| %>
 
   <input id="job_type" name="job[job_type]" type="hidden" value="<%= params['job_type'] %>">
@@ -186,7 +196,7 @@
 
 <%# Now create a template for all job types not handled above - eg from plugins %>
 <% job_types.keys.each do |type| %>
-  <% next if ['container_labels_job', 'find_and_replace_job', 'print_to_pdf_job', 'import_job', 'report_job', 'generate_slugs_job', 'generate_arks_job', 'bulk_import_job'].include?(type) %>
+  <% next if ['container_labels_job', 'find_and_replace_job', 'print_to_pdf_job', 'import_job', 'report_job', 'generate_slugs_job', 'generate_arks_job', 'ns2_remover_job', 'bulk_import_job'].include?(type) %>
 
   <% define_template(type, jsonmodel_definition(type.intern)) do |form| %>
 

frontend/app/views/jobs/_job_records.html.erb

@@ -10,7 +10,17 @@
          end
       %>
       <div>
-        <span class="asicon icon-<%= result["record"]["_resolved"]["jsonmodel_type"] %>" title="<%= I18n.t("#{result["record"]["_resolved"]["jsonmodel_type"]}._singular") %>"></span> <%= resolve_readonly_link_to record_title, result["record"]["ref"] %>
+        <span class="asicon icon-<%= result["record"]["_resolved"]["jsonmodel_type"] %>"
+              title="<%= I18n.t("#{result["record"]["_resolved"]["jsonmodel_type"]}._singular") %>">
+        </span>
+        <% if result["record"]["ref"].include?(current_repo.uri) %>
+         <%= resolve_readonly_link_to record_title, result["record"]["ref"] %>
+        <% else %>
+          <%= resolve_readonly_link_to "#{record_title} (#{result["record"]["ref"]})", result["record"]["ref"], false %>
+          <span class="label label-warning">
+            <%= I18n.t("job._frontend.external_repo") %>
+          </span>
+        <% end %>
       </div>
     <% end %>
   </div>

frontend/config/locales/en.yml

@@ -1400,6 +1400,7 @@ en:
   job:
     _frontend:
       audit_data: Audit Data
+      external_repo: Record belongs to a different repository
       actions:
         create: Create Job
         save: Start Job

frontend/spec/helpers/resolver_helper_spec.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+require 'rails_helper'
+
+describe ResolverHelper do
+  let(:url) { '/repositories/2/resources/1' }
+  let(:label) { 'A Test Link' }
+
+  describe '#resolve_readonly_link_to' do
+
+    it 'generates a readonly link' do
+      link = resolve_readonly_link_to(label, url)
+
+      expect(link).to have_link(label,
+                                href: '/resolve/readonly?uri=%2Frepositories%2F2%2Fresources%2F1')
+    end
+
+    it 'does not generate a link if active is false' do
+      inactive_link = resolve_readonly_link_to(label, url, false)
+
+      expect(inactive_link).not_to have_link(label)
+      expect(inactive_link).to have_text(label)
+    end
+  end
+
+  describe '#resolve_edit_link_to' do
+
+    it 'generates an edit link' do
+      link = resolve_edit_link_to(label, url)
+      expect(link).to have_link(label,
+                                href: '/resolve/edit?uri=%2Frepositories%2F2%2Fresources%2F1')
+    end
+  end
+end