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

cursor.description's type_code is always None, even though col_types is already available internally

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

まだ誰も着手していません。

評価

難易度
3/5
見積もり時間
1〜2日
初心者へのやさしさ
72/100
issue の種類
バグ
明瞭さ
明確に書かれている
活発さ
活発
技術スタック
python
領域
database, databases

調査の方向性

Read src/crate/client/cursor.py's Cursor.description property and the type definitions and _resolve() helper in src/crate/client/converter.py. Check how col_types is populated and how existing cursor tests cover description; add coverage for the type_code value when column types are available. Done means the public description exposes each column's type code without requiring a configured Converter.

索引モデルが issue の本文から書いたものです。

説明

Summary

`Cursor.description` (PEP 249) always returns `None` for every field after the column name, including `type_code`:

https://github.com/crate/crate-python/blob/main/src/crate/client/cursor.py

```python
@property
def description(self):
if self._closed:
return None

description = []
for col in self._result["cols"]:
    description.append((col, None, None, None, None, None, None))
return tuple(description)

```

This isn't because the driver doesn't have the type information. `self._result["col_types"]` is already populated from the same HTTP response (CrateDB's HTTP SQL interface column types) and is already used internally, e.g. in `Cursor._convert_rows()`:

```python
def _convert_rows(self):
if not ("col_types" in self._result and self._result["col_types"]):
raise ValueError(...)
types = self.result["col_types"]
converters = [self.converter.get(type) for type in types]
...
```

and `converter.py` already has a complete, documented mapping from these wire type codes to a `DataType` enum (`TIMESTAMP_WITH_TZ = 11`, `TIMESTAMP_WITHOUT_TZ = 15`, etc.) — so the information isn't missing, it's just not propagated into the public `description` property.

Why this matters

`type_code` is part of the standard DB-API 2.0 `description` contract (PEP 249). A consumer that only has a `Cursor` object (e.g. a driver-agnostic tool built against SQLAlchemy's cursor/dialect layer, or any generic DB-API client) has no supported way to find out a column's CrateDB type without either enabling a `Converter` (which changes value representation, not just exposing the type) or reaching into the private `cursor._result["col_types"]` attribute directly, which isn't part of the public API and could change without notice.

We actually hit this downstream in Apache Superset: `CrateEngineSpec.fetch_data()` has to read `cursor._result.get("col_types", [])` directly to tell timestamp columns apart from plain numeric ones, specifically because `cursor.description`'s `type_code` is always `None`:
https://github.com/apache/superset/blob/master/superset/db_engine_specs/crate.py#L98-L123

Suggested fix

In `Cursor.description`, populate the `type_code` slot (index 1 of the 7-tuple) from `self._result["col_types"]` (resolved through the existing `DataType` enum / `_resolve()` helper in `converter.py`) instead of hardcoding `None`, when `col_types` is available. This would bring `description` in line with the PEP 249 contract without needing a `Converter` to be configured, and let consumers stop reaching into `cursor._result` directly.

Environment
  • crate-python: current `main` (same code present going back through recent tags)
  • Verified by reading `src/crate/client/cursor.py` and `src/crate/client/converter.py` directly, cross-checked against the CrateDB HTTP SQL docs' column-types table.
主要言語
Python
スター
85
フォーク
35
平均マージ
2日 13時間
マージ済み PR（30日）
2

環境構築

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。

crate/crate-python のほかの issue

crate/crate-python 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。