Add support for Postgres Range Types

[sirwolfgang]

This Add/Fixes support for Postgres Range types: #1486

Check the following boxes:

• Wrote good commit messages.
• Feature branch is up-to-date with master (if not - rebase it).
• Squashed related commits together.
• Added tests.
• Added an entry to the Changelog if the new
  code introduces user-observable changes.
• The PR relates to only one subject with a clear title
  and description in grammatically correct, complete sentences.

Related: #1487

[Copilot]

Pull request overview

This PR adds support for PostgreSQL range column types (daterange, numrange, tsrange, tstzrange, int4range, int8range) to PaperTrail's versioning system. Previously, these types were not properly serialized/deserialized when tracking changes to model attributes containing range values.

Key changes:

• Introduced PostgresRangeSerializer to handle serialization/deserialization of PostgreSQL range types
• Modified attribute serializers to detect and process range columns even when using JSON storage
• Refactored variable naming from klass/item_class to model_class for consistency

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
lib/paper_trail/type_serializers/postgres_range_serializer.rb	New serializer class that handles PostgreSQL range types with deserialize logic for Ruby-formatted range strings
lib/paper_trail/attribute_serializers/attribute_serializer_factory.rb	Added range type detection and factory method to instantiate PostgresRangeSerializer
lib/paper_trail/attribute_serializers/object_attribute.rb	Added logic to serialize range columns even in JSON mode; renamed variables to model_class
lib/paper_trail/attribute_serializers/object_changes_attribute.rb	Added logic to serialize range columns even in JSON mode; renamed variables to model_class
lib/paper_trail/attribute_serializers/cast_attribute_serializer.rb	Renamed variables from klass to model_class for consistency
spec/paper_trail/type_serializers/postgres_range_serializer_spec.rb	Comprehensive tests for deserializing various PostgreSQL range types
spec/paper_trail/attribute_serializers/object_attribute_spec.rb	Integration tests for range serialization/deserialization; updated test descriptions
CHANGELOG.md	Added entry documenting the fix for range type support

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The CHANGELOG entry states "Fixed errors when deserializing Range types from Ruby style strings to Postgres" but this is confusing. It should clarify whether it's deserializing FROM Ruby strings or FROM Postgres format strings. Based on the code, it appears to deserialize Ruby-style range strings (e.g., "1..5") to Ruby Range objects. Consider rewording to: "Fixed support for PostgreSQL range column types in object/object_changes serialization"

Suggested change

	- Fixed errors when deserializing Range types from Ruby style strings to Postgres
	- Fixed support for PostgreSQL range column types in object/object_changes serialization

[Copilot]

The test description says "serializes a postgres range into a ruby array" but it should say "serializes a postgres range into a ruby range" since ranges are being serialized as ranges, not arrays.

Suggested change

	it "serializes a postgres range into a ruby array" do
	it "serializes a postgres range into a ruby range" do

[Copilot]

The regex pattern /\.{2,3}/ will match 2 or more consecutive dots, which means it could match "....", ".....", etc. This should be /\.{2}|\.{3}/ or more simply /(\.\.\.?)/ to match only ".." or "..." exactly.

Suggested change

	delimiter = string[/\.{2,3}/]
	delimiter = string[/(\.\.\.?)/]

[github-actions]

This PR has been automatically marked as stale due to inactivity.
The resources of our volunteers are limited.
If this is something you are committed to continue working on, please address any concerns raised by review and/or ping us again.
Thank you for all your contributions.

[mattmenefee]

Thanks for working on this, @sirwolfgang! This is a real bug that affects anyone using PostgreSQL range columns with PaperTrail's JSON version storage. The approach mirrors the existing PostgresArraySerializer pattern nicely, and the test coverage across all six range types is solid.

A few observations and suggestions:

General Feedback

The approach is sound — routing range types through a custom serializer that understands Ruby-style string representations (1..5) rather than PostgreSQL bracket notation ([1,5)) is the right call, since PaperTrail's JSON columns store to_s output.

Specific Suggestions

PostgresRangeSerializer#deserialize_with_ar — edge cases

The delimiter = string[/\.{2,3}/] regex works correctly for well-formed Ruby range strings, but consider what happens if the input is somehow in PostgreSQL bracket format (e.g., [2024-01-01,2024-01-31)) — the delimiter match would return nil, and string.split(nil) would raise. A guard clause or fallback to the original Active Record deserializer might make this more robust:

def deserialize_with_ar(string)
  return nil if string.blank?

  delimiter = string[/\.{2,3}/]
  return @active_record_serializer.deserialize(string) unless delimiter

  range_start, range_end = string.split(delimiter, 2)
  # ...
end

attributes_to_serialize / changes_to_serialize — potential double-serialization

When a column is both encrypted and a range type, it would appear in both encrypted_to_serialize and columns_to_serialize. Since the merge is encrypted_to_serialize.merge(columns_to_serialize), the range version wins, which seems correct — but it's worth adding a brief comment noting this is intentional.

Test: description typo

# Line says "ruby array" but should say "ruby range"
it "serializes a postgres range into a ruby array" do

CHANGELOG entry

The current wording — "Fixed errors when deserializing Range types from Ruby style strings to Postgres" — is a bit ambiguous about the direction. Consider:

- [Fixed] Support for PostgreSQL range column types in version object serialization

Missing test coverage (nice-to-haves)

• nil range values
• Range with only one bound set (e.g., 1.. in Ruby 2.6+, or unbounded on one side in PostgreSQL)
• A full integration test that round-trips through reify (mirroring the reproduction script in #1486)

Overall this is good work and addresses a legitimate gap. Would love to see it merged!