BUG: read_json converts Numeric Strings to Numbers

[WillAyd]

>>> pd.read_json("""{"a": ["01", "02", "03"]}""")
   a
0  1
1  2
2  3

These are well defined by JSON to be alphanumeric strings and not integers, so its a bit unfortunate that these get cast.

Expected Output:

>>> pd.read_json("""{"a": ["01", "02", "03"]}""")
   a
0  01
1  02
2  03

[asishm]

unsure if related, these issues are specifically dealing with numbers that have '_' like '23_42'

• BUG: read_json changes dtype (object => int)  #41408 read_json() removes underscore and converts string to integer when only numbers and underscore present #17869

[Gdango]

I think you have to define dtype=False, otherwise it will default to True.

>>> pd.read_json("""{"a": ["01", "02", "03"]}""", dtype=False)

    a
0  01
1  02
2  03

Reference: