指定pl.Object仍报‘struct orders must remain the same’，如何让Polars存字典为Object？

问题场景 #

使用数据库fetchall()结果创建DataFrame，指定dict_col为pl.Object类型以保留原始字典结构，但触发如下报错：

thread '<unnamed>' panicked at 'called `Result::unwrap()` on an `Err` value: ComputeError(Borrowed("struct orders must remain the same"))', /Users/runner/work/polars/polars/polars/polars-core/src/frame/row.rs:457:86
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
Traceback (most recent call last):
  File:
    df = pl.DataFrame(
  File "venv/lib/python3.9/site-packages/polars/internals/dataframe/frame.py", line 299, in __init__
    self._df = sequence_to_pydf(
  File "venv/lib/python3.9/site-packages/polars/internals/construction.py", line 612, in sequence_to_pydf
    pydf = PyDataFrame.read_rows(data, infer_schema_length)
pyo3_runtime.PanicException: called `Result::unwrap()` on an `Err` value: ComputeError(Borrowed("struct orders must remain the same"))

报错原因 #

Polars 0.14.18版本中，通过行数据（如fetchall()返回结果）创建DataFrame时，会自动尝试将字典解析为struct类型——即使已指定pl.Object类型。若不同行的字典键顺序不一致，就会触发"struct orders must remain the same"错误。

解决方案 #

绕过行读取阶段的自动解析逻辑，通过单独创建pl.Series再组合为DataFrame：

# 获取原始数据库返回数据
rows = c.fetchall()

# 拆分各列数据
primary_keys = [row[0] for row in rows]
string_cols = [row[1] for row in rows]
dict_cols = [row[2] for row in rows]

# 分别创建指定类型的Series，再组合成DataFrame
df = pl.DataFrame({
    "primary_key": pl.Series(primary_keys, dtype=pl.Int32),
    "string_col": pl.Series(string_cols, dtype=pl.Object),
    "dict_col": pl.Series(dict_cols, dtype=pl.Object)
})

这种方式会直接将字典作为Python原生对象存储在pl.Object列中，避免Polars自动解析为struct，从而解决报错。

内容的提问来源于stack exchange，提问作者ldrg