Hacktoberfest 2026：メンテナが10月に向けて印を付けた、オープンで初心者向けの issue。 Hacktoberfest の issue を見る

`dumps(sort_keys=True)` only sorts the top level of a parsed document (nested tables / inline tables / AoT keep original order)

オープン
  • 何か問題がありますか？
#614 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
5/5
見積もり時間
1週間以上
初心者へのやさしさ
35/100
issue の種類
バグ
明瞭さ
おおむね明確
活発さ
活発
技術スタック
python
領域
tooling

調査の方向性

Start with the reproduction, then read api.py:60-64 and items.py:114, 130-150 to confirm how parsed and plain-dict inputs differ. The issue still needs a decision between recursive sorting with possible formatting loss and documenting top-level-only behavior; done means implementing the chosen direction and covering nested tables, inline tables, and arrays of tables.

索引モデルが issue の本文から書いたものです。

説明

Summary

For a document produced by tomlkit.parse() / loads(), dumps(doc, sort_keys=True) sorts only the top-level keys. Nested tables, inline tables, and arrays-of-tables keep their original (insertion) order. The same option applied to a plain dict sorts recursively. So the output depends on whether the input was parsed or built from scratch, which is surprising for a documented, explicit option.

Reproduction (copy-paste)

import tomlkit

doc = tomlkit.parse("""
[zeta]
b = 2
a = 1

[alpha]
b = 2
a = 1

top = 0
""")

print(tomlkit.dumps(doc, sort_keys=True))
# [alpha]
# b = 2
# a = 1
#
# top = 0
#
# [zeta]
# b = 2
# a = 1
#     -> top level IS sorted (alpha before zeta), but inside each table
#        `b` still comes before `a`.

print(tomlkit.dumps({"zeta": {"b": 2, "a": 1}, "alpha": {"b": 2, "a": 1}, "top": 0}, sort_keys=True))
# top = 0
#
# [alpha]
# a = 1
# b = 2
#
# [zeta]
# a = 1
# b = 2
#     -> fully sorted, including nested keys.

Same split for inline tables and arrays of tables:

print(tomlkit.dumps(tomlkit.parse("[t]\ninl = { z = 1, a = 2 }\n"), sort_keys=True))
# inl = { z = 1, a = 2 }      <- NOT sorted
print(tomlkit.dumps({"t": {"inl": {"z": 1, "a": 2}}}, sort_keys=True))
# inl = { a = 2, z = 1 }      <- sorted

print(tomlkit.dumps(tomlkit.parse("[[t]]\nz = 1\na = 2\n"), sort_keys=True))
# z = 1
# a = 2                        <- NOT sorted
print(tomlkit.dumps({"t": [{"z": 1, "a": 2}]}, sort_keys=True))
# a = 2
# z = 1                        <- sorted

Expected

sort_keys=True sorts recursively, consistent with the plain-dict path and with the documented "sort the keys in alphabetic order".

Actual

Only the top level of a parsed document is sorted; nested structures keep their original order.

Root cause

dumps() (api.py:60-64) routes parsed documents through item(data, _sort_keys=True). In item() (items.py:114), the top-level TOMLDocument/Container is a dict (not an Item), so it is rebuilt in the isinstance(value, dict) branch (items.py:139-150), which sorts. But every nested value in a parsed document is already a Table / InlineTable / AoT, and item() returns it unchanged at the early guard if isinstance(value, Item): return value (items.py:130-131). Those items are never rebuilt, so their keys keep their original order. The plain-dict path has no such guard (nested plain dicts are rebuilt recursively), which is why it sorts fully.

This is the residual of #466 ("sort_keys does not (always?) work"), which fixed the top level; the nested levels were not covered.

Scope note / open question

A fix would need to rebuild nested Table / InlineTable / AoT items when _sort_keys is set. Because tomlkit's whole purpose is lossless round-tripping (preserving comments, blank lines, key order, formatting), rebuilding nested tables would drop that formatting. So this may be a deliberate trade-off rather than a plain bug — filing here to get a decision:

  • (a) sort recursively and accept loss of nested formatting, or
  • (b) document that sort_keys applies only to the top level of parsed documents.

Happy to take either once there's a direction.

Environment

tomlkit master @ 8c959b5 (0.15.1+), Python 3.12.

主要言語
Python
スター
850
フォーク
163
平均マージ
13分
マージ済み PR（30日）
2

環境構築

このプロジェクトの環境構築ファイルはまだ確認していません。まず README を読み、一般的な手順ははじめてのコントリビューションガイドを参照してください。

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

python-poetry/tomlkit のほかの issue

python-poetry/tomlkit の issue をすべて見る

似ている issue

Python の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。