fix(internal): encode nested multipart fields with bracket notation - #339
Closed
leoplct wants to merge 1 commit into
Closed
fix(internal): encode nested multipart fields with bracket notation#339leoplct wants to merge 1 commit into
leoplct wants to merge 1 commit into
Conversation
The OpenAI API expects nested fields in multipart/form-data bodies to be
flattened deepObject-style (e.g. `expires_after[anchor]=created_at`),
but nested hashes were serialized as a single JSON part, which the API
rejects with "Additional properties are not allowed ('expires_after'
was unexpected)".
Flatten nested hashes and arrays inside multipart bodies using bracket
notation, matching openai-python's `_serialize_multipartform`
(qs "brackets" array format). Top-level primitive arrays keep the
existing repeated-key encoding.
Fixes openai#212
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Nc2s6tXUGssbrFoafV34Rj
Contributor
|
Thanks for the clear reproduction and for grounding the approach in openai-python’s global brackets format. The recursive nested-object work here is an important part of the complete fix. We’ve incorporated that behavior into #348 and combined it with the missing bracket names for top-level primitive and file arrays, plus scalar compatibility and broader wire-format coverage. I’m closing this PR in favor of #348 so the consolidated implementation can be reviewed and landed as one change. Appreciate the thoughtful work here. |
jbeckwith-oai
added a commit
to fallintoplace/openai-ruby
that referenced
this pull request
Aug 10, 2026
## Summary - encode multipart arrays recursively as `field[]`, including both primitive and file arrays - flatten nested multipart objects with bracket notation, including arrays of nested objects - preserve scalar field names and streaming file behavior - escape both `name` and `filename` `Content-Disposition` parameters - keep the fix entirely in the openai-ruby runtime; no generated model, OpenAPI, or Castiron changes are needed ## Root cause The shared multipart writer repeated top-level primitive and file arrays using the scalar field name and serialized nested hashes as JSON parts. The API's form parser expects the same global brackets format used by openai-python and by the transformed OpenAPI examples. Before: - `timestamp_granularities=word` and `timestamp_granularities=segment` - duplicate scalar `image` file parts - one JSON `expires_after` part After: - `timestamp_granularities[]=word` and `timestamp_granularities[]=segment` - duplicate `image[]` file parts, while a single image remains `image` - `expires_after[anchor]=created_at` and `expires_after[seconds]=3600` The implementation replaces the old top-level array special case with one recursive field writer. It preserves insertion order, array order, duplicate values, and streaming file contents without materializing a flattened body. This consolidates and supersedes openai#249, openai#311, and openai#339. Fixes openai#201. Fixes openai#212. Fixes openai#253. ## Validation - `mise exec ruby@4.0.6 -- bundle exec ruby -Itest test/openai/internal/util_test.rb` — 45 runs, 178 assertions - `mise exec ruby@4.0.6 -- ./scripts/test` — 522 runs, 1,709 assertions - `mise exec ruby@4.0.6 -- bundle exec rake lint` — 2,596 files, no RuboCop offenses; Sorbet clean; 1,211 RBS files valid - `mise exec ruby@4.0.6 -- bundle exec rake build:gem` — built `openai-0.78.0.gem` - RBI and RBS formatting plus `git diff --check` - thermo-nuclear maintainability review and a separate correctness/readability/architecture/security/performance review, repeated after findings until clean
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The OpenAI API expects nested fields in multipart/form-data bodies to be flattened deepObject-style (e.g.
expires_after[anchor]=created_at), but nested hashes were serialized as a single JSON part, which the API rejects with "Additional properties are not allowed ('expires_after' was unexpected)".Flatten nested hashes and arrays inside multipart bodies using bracket notation, matching openai-python's
_serialize_multipartform(qs "brackets" array format). Top-level primitive arrays keep the existing repeated-key encoding.Fixes #212