Inconsistent behaviour on sep/delimiter for pandas.read_csv

[dahlbaek]

Related: #7662

Code Sample, a copy-pastable example if possible

from io import StringIO

import pandas as pd


CSV = "a|b\n1|2"
print(pd.read_csv(StringIO(CSV), sep=None, engine='python'))
print('is not the same as')
print(pd.read_csv(StringIO(CSV), delimiter=None, engine='python'))
print('\nand no warning is emitted by')
print(pd.read_csv(StringIO(CSV), sep='|', delimiter=' '))

Problem description

According to the documentation,

delimiter : str, default None
    Alternative argument name for sep.

Thus, I would expect sep and delimiter to be interchangable.

Expected Output

Specifying delimiter=None is equivalent to specifying sep=None, and specifying both sep and delimiter emits a warning or causes an error. Alternatively, either sep or delimiter should be deprecated.

Output of pd.show_versions()

No module named 'dask'

INSTALLED VERSIONS
------------------
commit: None
python: 3.6.6.final.0
python-bits: 64
OS: Linux
OS-release: 4.13.0-45-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL:
LANG: en_DK.UTF-8
LOCALE: en_DK.UTF-8

pandas: 0.24.0.dev0+332.g1f6ddc4
pytest: 3.6.3
pip: 10.0.1
setuptools: 39.0.1
Cython: 0.28.4
numpy: 1.14.5
scipy: 1.1.0
pyarrow: 0.9.0
xarray: 0.10.7
IPython: 6.4.0
sphinx: 1.7.5
patsy: None
dateutil: 2.7.3
pytz: 2018.5
blosc: 1.5.1
bottleneck: 1.2.1
tables: 3.4.4
numexpr: 2.6.5
feather: 0.4.0
matplotlib: 2.2.2
openpyxl: 2.5.4
xlrd: 1.1.0
xlwt: 1.3.0
xlsxwriter: 1.0.5
lxml: 4.2.3
bs4: 4.6.0
html5lib: 1.0.1
sqlalchemy: 1.2.9
pymysql: 0.9.2
psycopg2: None
jinja2: 2.10
s3fs: 0.1.5
fastparquet: 0.1.5
pandas_gbq: None
pandas_datareader: None
gcsfs: 0.1.1

[WillAyd]

I'd be +1 to deprecate delimiter and just keep sep - let's see what others think.

[jreback]

i believe we have this discussion and kept this because it’s compatible with other languages

there might be an open issue about this

but sure would be ok to deprecate

[dahlbaek]

The only reason I might not be in favor of deprecating delimiter is that it is supported by the Frictionless Data CSV Dialect, as well as csv.Dialect. On the other hand sep seems to be the favored name in pandas, so deprecating sep might also not be a good idea…

[WillAyd]

There's some overlap with the dialect but obviously we don't align on all the keywords and naming conventions thereof. I'd still stick with sep given it's usage in pandas and there fact that there's no delimiter in to_csv

[minggli]

happy to work on this.

[jorisvandenbossche]

I think we should decide on deprecating delimeter or not together we the other discussion on aliases with _, as one of the arguments there to keep old names is the consistency with csv module. If we do that there, I think it also applies on the delimiter keyword.

We can also simply add to the docstring that delimiter does not support the "sniffing feature".

(which is a bit an esoteric feature anyhow IMO)

[jorisvandenbossche]

The other issue I mean is: #22639