BUG: Reading large CSV files with `pyarrow` when values contain newline character.

[matteosantama]

Pandas version checks

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

• [X] 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

rows = []
for i in range(1_000_000):
    rows.append({"text": "ab\ncd", "i": i})

df = pd.DataFrame(rows)
df.to_csv("./example.csv", index=False)
pd.read_csv("./example.csv", engine="pyarrow")

Issue Description

pd.read_csv fails when reading large CSV files with engine="payarrow" if values contain newline characters. The error is

ParserError: CSV parser got out of sync with chunker. This can mean the 
data file contains cell values spanning multiple lines; please consider 
enabling the option 'newlines_in_values'.

Note the file must be large to trigger the error. Either pandas should enable this flag internally, or expose the option to the user.

Expected Behavior

Reading the file succeeds with engine="python" and I would expect consistency between the two options.

Installed Versions

In [7]: pd.show_versions() INSTALLED VERSIONS ------------------ commit : d9cdd2ee5a58015ef6f4d15c7226110c9aab8140 python : 3.11.9.final.0 python-bits : 64 OS : Darwin OS-release : 23.5.0 Version : Darwin Kernel Version 23.5.0: Wed May 1 20:13:18 PDT 2024; root:xnu-10063.121.3~5/RELEASE_ARM64_T6030 machine : arm64 processor : arm byteorder : little LC_ALL : None LANG : en_US.UTF-8 LOCALE : en_US.UTF-8 pandas : 2.2.2 numpy : 1.26.4 pytz : 2024.1 dateutil : 2.9.0.post0 setuptools : None pip : 24.0 Cython : None pytest : None hypothesis : None sphinx : None blosc : None feather : None xlsxwriter : None lxml.etree : None html5lib : None pymysql : None psycopg2 : None jinja2 : None IPython : 8.25.0 pandas_datareader : None adbc-driver-postgresql: None adbc-driver-sqlite : None bs4 : None bottleneck : None dataframe-api-compat : None fastparquet : None fsspec : None gcsfs : None matplotlib : None numba : None numexpr : None odfpy : None openpyxl : None pandas_gbq : None pyarrow : 16.1.0 pyreadstat : None python-calamine : None pyxlsb : None s3fs : None scipy : None sqlalchemy : None tables : None tabulate : None xarray : None xlrd : None zstandard : None tzdata : 2024.1 qtpy : None pyqt5 : None

[matteosantama]

From the latest pyarrow documentation

newlines_in_values, optional (default False) Whether newline characters are allowed in CSV values. Setting this to True reduces the performance of multi-threaded CSV reading.

Enabling it by default would probably be a mistake. The pyarrow engine (with its multi-threaded capabilities) is the preferred option for large CSV files, though, so it'd be a shame for it to fail in this scenario.

If the pyarrow engine is here to stay, I'd recommend exposing newlines_in_values to the user.

[tilovashahrin]

To keep the pyarrow engine, you'll need to use the pyarrow library directly to handle CSV files that contain newline characters. This involves using the ParseOptions class from pyarrow.csv to set the newlines_in_values option to True.

Example

import pyarrow as pa
import pandas as pd

rows = []
for i in range(1_000_000):
    rows.append({"text": "ab\ncd", "i": i})

df = pd.DataFrame(rows)
# Define parse options to allow newlines in values
parse_options = pv.ParseOptions(newlines_in_values=True)

# Read the CSV file using pyarrow
table = pv.read_csv("example.csv", parse_options=parse_options)

# Convert the Arrow Table to a Pandas DataFrame
df = table.to_pandas()
df

[gosuchoi]

take

[wooseogchoi]

take

[wooseogchoi]

take

[wooseogchoi]

@WillAyd I would like to introduce a new argument in order to expose pyarrow's 'newlines_in_values' to the user because I cannot find any suitable in the current parameters. Could you please suggest new parametrer name for this, 'newlines_in_values' which might be used by another engines in the future.

[WillAyd]

Reading through the issue I don't think we actually want to change anything here - the solution from @tilovashahrin should work.

Can you check if that works for you? If so, we should add a test for it to pandas (if one doesn't already exist) and maybe update the documentation to show how to do it

[wooseogchoi]

@WillAyd With some modification, the codes above are working. I will add it as example in the read_csv doc. Also I will check the test cases. If it is not there, I will add one. Thx