BUG: read_csv() with engine="pyarrow" converts numeric string even when dtype=str is specified

[LawrenceLiu023]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd

demo_df: pd.DataFrame = pd.DataFrame(
    {
        "sample_name": ["sample_1", "sample_2"],
        "case_id": ["00001", "00002"],
        "sample_id": ["00001001", "00001002"],
        "sample_type": ["T", "T"],
    }
)
demo_df.to_csv("bug_demo.csv", header=True, index=False, sep=",", encoding="utf-8")
pyarrow_df: pd.DataFrame = pd.read_csv(
    "bug_demo.csv",
    header=0,
    index_col=None,
    sep=",",
    encoding="utf-8",
    engine="pyarrow",
    dtype=str,
)
print("This is read by pyarrow engine.")
print(pyarrow_df)
c_df: pd.DataFrame = pd.read_csv(
    "bug_demo.csv",
    header=0,
    index_col=None,
    sep=",",
    encoding="utf-8",
    engine="c",
    dtype=str,
)
print("This is read by c engine.")
print(c_df)
python_df: pd.DataFrame = pd.read_csv(
    "bug_demo.csv",
    header=0,
    index_col=None,
    sep=",",
    encoding="utf-8",
    engine="python",
    dtype=str,
)
print("This is read by python engine.")
print(python_df)

Issue Description

I have encountered an issue with the read_csv() function in pandas when using the pyarrow engine. Even when specifying dtype=str, pure numeric strings are being converted to numeric type. Additionally, pure numeric strings starting with multiple zeros lose the leading zeros in the resulting DataFrame. This behavior is unexpected as I would like to preserve the original format of the numeric strings as text.

Expected Behavior

The example demonstrates reading a CSV file with different engines. It is expected that pyarrow engine should get the same DataFrame as c engine and python engine when using dtype=str. It should output the following texts.

# This is read by pyarrow engine.
  sample_name case_id sample_id sample_type
0    sample_1       1      1001           T
1    sample_2       2      1002           T
# This is read by c engine.
  sample_name case_id sample_id sample_type
0    sample_1   00001  00001001           T
1    sample_2   00002  00001002           T
# This is read by python engine.
  sample_name case_id sample_id sample_type
0    sample_1   00001  00001001           T
1    sample_2   00002  00001002           T

Installed Versions

INSTALLED VERSIONS

commit : d9cdd2e
python : 3.12.2.final.0
python-bits : 64
OS : Linux
OS-release : 6.5.0-26-generic
Version : #26~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Tue Mar 12 10:22:43 UTC 2
machine : x86_64
processor :
byteorder : little
LC_ALL : None
LANG : None
LOCALE : en_US.UTF-8

pandas : 2.2.2
numpy : 1.26.4
pytz : 2024.1
dateutil : 2.9.0
setuptools : 69.5.1
pip : 24.0
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 5.1.0
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.1.3
IPython : 8.22.2
pandas_datareader : None
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : 4.12.3
bottleneck : 1.3.8
dataframe-api-compat : None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : 2.9.0
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 11.0.0
pyreadstat : None
python-calamine : None
pyxlsb : None
s3fs : None
scipy : 1.12.0
sqlalchemy : None
tables : None
tabulate : 0.9.0
xarray : None
xlrd : None
zstandard : 0.22.0
tzdata : 2024.1
qtpy : None
pyqt5 : None

[rohanjain101]

Likely related to #57666 (comment)

[eshaready]

take

[mroeschke]

Closing as a duplicate of #57666