Reading a SAS file with 0 rows gives None

[AbdealiLoKo]

Code Sample, a copy-pastable example if possible

In [2]: import pandas as pd

In [3]: pd.read_csv("zero_observations.csv")
Out[3]: 
Empty DataFrame
Columns: [a, b, c, d]
Index: []

In [4]: pd.read_sas("zero_observations.sas7bdat")

In [5]: 

This is a.csv:

$ cat zero_observations.csv
a,b,c,d

Problem description

When reading a SAS file with 0 records, it gives None. But when reading a CSV file with 0 records it gives an empty dataframe.

In SAS, we can atleast identify the datatypes correctly and make an empty dataframe with this.

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.5.2.final.0 python-bits: 64 OS: Darwin OS-release: 16.7.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: en_US.UTF-8 LANG: en_US.UTF-8 LOCALE: en_US.UTF-8

pandas: 0.21.0
pytest: 3.2.1
pip: 9.0.1
setuptools: 23.0.0
Cython: 0.27.3
numpy: 1.13.3
scipy: 0.18.1
pyarrow: None
xarray: None
IPython: 6.1.0
sphinx: 1.4.9
patsy: None
dateutil: 2.6.1
pytz: 2017.3
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: 2.0.0
openpyxl: None
xlrd: 1.1.0
xlwt: None
xlsxwriter: 0.9.8
lxml: None
bs4: None
html5lib: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.8.1
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[jreback]

is this just an issue for #18184 ?

normally we do open issues first but since u opened the pr we actually don’t need an issue (fyi for the future)

[AbdealiLoKo]

This is actually an issue for 0 rows and not the 0 variables case.
The PR #18184 handles the case for 0 variables. Im not very sure how to fix the one for 0 rows hence created a issue for that

[AbdealiLoKo]

When going through this, I realized there may be some confusion in expected behavior.

In read_csv() if I give a file with 0 rows and iterator, it gives:

In [30]: for i, idf in enumerate(pd.read_csv("/tmp/a.csv", iterator=True)):
    ...:     print("IterCount:", i)
    ...:     print(idf)
    ...:     
IterCount: 0
Empty DataFrame
Columns: [a, b, c, d]
Index: []

This seems non-intuitive because even though there are 0 rows, the iterator runs atleast once. While in read_sas() the equivalent iterator runs 0 times.
In read_csv() when I give a file with 2 records, and chunksize=2 it gives an iterator with 1 item. Not two items (First one with 2 rows, Second with 0 rows)

In [32]: for i, idf in enumerate(pd.read_csv("/tmp/b.csv", iterator=True, chunksize=2)):
    ...:     print("IterCount:", i)
    ...:     print(idf)
    ...:     
IterCount: 0
   a  b  c  d
0  1  2  3  4
1  5  6  7  8

This is definitely a bit of a corner case but I'd like to know what is the expected behavior.

I believe the following makes sense:

• If it is an iterator, give None so that the iterator can stop correctly
• If it it not an iterator and the whole dataframe is being read, identify EmptyDataFrames with header names and column types and return the empty data frame

[AbdealiLoKo]

Any thoughts on this one ?

[mukesh5]

Is this issue still open? Can I pick it up?