feat(python): support a custom license_header in generated files

[devin-ai-integration]

Description

Requested by a customer who must ship a license notice in every file of their published SDK repo. Today the only knob is whitelabel, which just swaps the Fern notice text; there is no way to add a header.

Adds a license_header custom config to the Python SDK generator. The configured text is emitted as a comment block above the existing auto-generated notice in every generated Python file:

config:
  license_header: |
    Copyright 2024 Acme, Inc.
    Licensed under the Apache License, Version 2.0.

# Copyright 2024 Acme, Inc.
# Licensed under the Apache License, Version 2.0.

# This file was auto-generated by Fern from our API Definition.

Lines that already start with # are passed through, so pre-commented text works too. Output is byte-identical to before when the option is omitted.

Changes Made

• New codegen/file_header.py centralizes header construction (Fern / whitelabel notice + optional license text); WriterImpl and the hardcoded copy in Project._create_package_path_init_files both now use it, so __init__.py files at every level get the header too.
• Threaded license_header from custom_config through AbstractGenerator → Project → SourceFileImpl/NodeWriterImpl/ModuleManager.
• Added license_header to SDKCustomConfig (required since it forbids extra fields).

Testing

• Unit tests added/updated — tests/codegen/test_file_header.py
• Manual testing completed — new exhaustive/license-header seed fixture (combined with package_path so the intermediate __init__.py path is covered); diffing it against exhaustive/package-path shows only the added header lines. poetry run mypy . clean, poetry run pre-commit run -a clean, full exhaustive seed run 36/36 passing.

Link to Devin session: https://app.devin.ai/sessions/37b1ba4f0fc247f8913d7d4e0cb21551
Open in Devin Desktop: https://app.devin.ai/desktop/session/37b1ba4f0fc247f8913d7d4e0cb21551?variant=devin

---

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[nitpickybot]

AI Review Summary

Adds a license_header custom config threaded through the Python generator's codegen layer, with a shared get_file_header helper. Implementation is straightforward; two issues worth addressing: the license header is applied to snippet writers (which likely shouldn't carry it), and the _as_comment handling of \r\n / trailing whitespace is slightly fragile. Also the config is read via raw dict access in abstract_generator rather than the typed SDKCustomConfig.

• 🟡 1 warning(s)
• 🔵 2 suggestion(s)

---

To request another review, comment /ai-review on this pull request.

[nitpickybot]

🔵 suggestion

split("\n") leaves a trailing \r on CRLF input, producing # foo\r lines. Also line.startswith("#") misses indented comments (e.g. # foo), which is probably fine but worth being deliberate about. Consider splitlines():

Suggested change

	return "\n".join(
	line if line.startswith("#") else (f"# {line}" if line.strip() else "#")
	for line in text.rstrip("\n").split("\n")
	)
	return "\n".join(
	line if line.lstrip().startswith("#") else (f"# {line}" if line.strip() else "#")
	for line in text.rstrip().splitlines()
	)

[nitpickybot]

🟡 warning

NodeWriterImpl is also used for snippet generation (should_format_as_snippet / should_include_header=False paths, and Project snippet writers). Confirm that generated code snippets in snippet.json / reference.md don't end up prefixed with the customer's license text — headers there would be noise. If snippets already pass should_include_header=False, ignore; the to_str guard covers it.

[nitpickybot]

🔵 suggestion

Raw dict access bypasses the typed SDKCustomConfig.license_header you just added, so a non-string value (e.g. a list or int in YAML) flows straight into _as_comment and blows up with an unclear AttributeError. The import_paths block above has the same smell, but for a new field consider validating the type here (or reading from the parsed custom config) so a misconfigured license_header yields a sensible error.

Note also this is generic to all generators (abstract_generator), while the config field is only declared on SDKCustomConfig — for pydantic-based generators with extra="forbid", other generators' configs will reject license_header before this code runs.

[devin-ai-integration]

Devin Review found 2 potential issues.

[github-actions]

SDK Generation Benchmark Results

Comparing PR branch against median of 5 nightly run(s) on main (latest: 2026-09-01T04:06:34Z).

Full benchmark table (click to expand)

Generator	Spec	main (generator)	main (E2E)	PR (generator)	Delta
python-sdk	square	149s (n=5)	254s (n=5)	141s	-8s (-5.4%)

main (generator): generator-only time via --skip-scripts (includes Docker image build, container startup, IR parsing, and code generation — this is the same Docker-based flow customers use via fern generate). main (E2E): full customer-observable time including build/test scripts (nightly baseline, informational). Delta is computed against generator-only baseline.
⚠️ = generation exited with a non-zero exit code (timing may not reflect a successful run).
Baseline from nightly runs on main (latest: 2026-09-01T04:06:34Z). Trigger benchmark-baseline to refresh.
Last updated: 2026-09-01 14:59 UTC