Closed kaatish closed 4 years ago
Don't think the repro code is right here. The file should be something like
{"col0":10,"col1":+234}
, and read_json should set lines
parameter to True
.
Also, can you please run the same code with pandas.read_json to confirm that behavior is indeed unexpected.
Moved to P1 because Pandas does not support this case either.
Describe the bug If an integer/float is encountered in a json file with a minus sign before it, it is handled here and here. The minus sign is ignored and the remaining string is checked against the pattern of a float etc.
The same is not done for a '+' sign as it is done in the case of csv here.
Steps/Code to reproduce bug Create
temp.json
as :{"col0":10,"col1":+234}
. Attempting to read the file without = cudf.read_json('temp.json', lines=True)
parses the second column as object.Expected behavior
out = cudf.read_json('temp.json', lines=True)
should give :Environment overview
Click here to see environment details