BUG: pd.read_excel error when a cell contains string "INF"

[josephernest]

Here is how to reproduce the bug: run this script:

import pandas as pd
pd.read_excel('test.xlsx')
# idem with pd.read_excel('test.xlsx', engine='openpyxl')
# so it's not specific to xlrd or openpyxl

with this test.xlsx file:

a	b
12	1
14	1
15	1
16	INF

Error:

OverflowError: cannot convert float infinity to integer

Why does this happen / why is there a integer conversion? No automatic integer parsing has been asked.

---

PS: Congrats for this great project!

[asishm]

editing since openpyxl is the default engine now.

openpyxl

Note that this would still fail when using engine='openpyxl'. It seems like openpyxl does its own internal int conversion when parsing the file as referenced in the traceback:

  File "pandas-dev/lib/python3.8/site-packages/openpyxl/worksheet/_reader.py", line 199, in parse_cell
    value = _cast_number(value)
  File "pandas-dev/lib/python3.8/site-packages/openpyxl/worksheet/_reader.py", line 82, in _cast_number
    return int(value)
ValueError: invalid literal for int() with base 10: 'INF'

That is something that would need to be fixed upstream.

[mroeschke]

Since this is an issue with openpyxl, I think it would be better for the issue to be fixed upstream. If you could file an issue in their repo that would be great. Closing.