From 23ec819e708cac7fec6a0f68f845ce86a46ac87c Mon Sep 17 00:00:00 2001
From: David Taylor <david@taylorhq.com>
Date: Fri, 19 Aug 2022 11:30:46 +0100
Subject: [PATCH] DEV: Introduce debugging rake task for PostRevision issues
 (#17998)

See https://meta.discourse.org/t/232885 for context. This rake task enumerates problematic records and prints out a sample of each failure.
---
 lib/tasks/revisions.rake | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100644 lib/tasks/revisions.rake

diff --git a/lib/tasks/revisions.rake b/lib/tasks/revisions.rake
new file mode 100644
index 0000000000000..722d67471decb
--- /dev/null
+++ b/lib/tasks/revisions.rake
@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+desc 'Print debug information about post_revisions which cannot be deserialized'
+task 'revisions:debug_deserialization' => :environment do
+  puts "Checking #{PostRevision.count} PostRevision records in batches of 1000... this may take some time..."
+  sleep 1
+
+  counts = Hash.new(0)
+  examples = Hash.new()
+
+  PostRevision.find_each do |revision|
+    revision.modifications
+  rescue Psych::DisallowedClass => e
+    class_name = e.message.sub("Tried to load unspecified class: ", "")
+    counts[class_name] += 1
+    examples[class_name] ||= revision
+    puts "#{Discourse.base_url}/p/#{revision.post_id} (revision number:#{revision.number} id:#{revision.id}) #{e.message}"
+  end
+
+  puts
+  puts "Done"
+  puts
+
+  puts "---- Summary ----"
+  puts "Checked records: #{PostRevision.count}"
+  counts.each_pair { |k, v| puts "  #{k}: #{v}" }
+
+  puts
+  puts "---- Examples ----"
+  puts
+  examples.each_pair do |class_name, revision|
+    puts "-- BEGIN Example #{class_name} --"
+    puts
+    puts revision.modifications_before_type_cast
+    puts
+    puts "-- END Example #{class_name} --"
+  end
+end