I tried out using the example you shared. Saw the same failures. However,
this worked for me:
> select `hello.yoko`, yousayyes from dfs.`/Users/agirish/abc.json`;
+-------------+------------+
| hello.yoko | yousayyes |
+-------------+------------+
| goodbye | isayno |
+-------------+------------+
1 row selected (0.092 seconds)
Basically, instead of a select * I used select <columnName> and escaped the
column using back quotes. See if this can work for you. I'm not sure if
there is an alternate way - someone else can comment.
-Abhishek
On Mon, Oct 12, 2015 at 9:45 AM, John Omernik <j...@omernik.com> wrote:
> I am getting some data that seems well formed, except for there are dots in
> some of the key names.
>
> Basically, something like ip.src seems to be making it fail with a "Field
> References must be singular names"
>
> I am using the MapR Packaged Drill 1.2.
>
>
> I can prove this out with
>
> {"hello":"goodbye", "yousayyes":"isayno"}
>
> works fine but
>
> {"hello.yoko":"goodbye", "yousayyes":"isayno"}
>
> Screams loudly that there is a failure.
>
> Is there a way we could handle dots in the key name on JSON?
>
> John
>
