Discussion:
Querying Data with period in name
(too old to reply)
John Omernik
2017-08-11 18:07:04 UTC
Permalink
Raw Message
Hey all,

I am querying some json and parquet data that has dots in the name. Not all
the data I may be querying will come from Drill, thus dot is a valid
character... when I go to initially explore my data, Drill throws the error
below when I run a select * query.

I understand the error, and I can create a view, selecting each column out
and renaming it for easier select * in the future. However, as a user, if
I get a new data set, this could (unless I am informed of another way here)
force me to leave drill to explore my data.

I get how using periods as field qualifiers causes issues... but if we had
had a way to read a file to get the schema, to either produce the all the
fields in a select query for easy view creation or a way to query with
periods in the name that would be awesome! It would keep users IN drill
instead of going elsewhere to explore their data.

I am open to ideas!










Error Returned - Code: 500
Error Text:
SYSTEM ERROR: UnsupportedOperationException: Unhandled field reference
"id.orig_h"; a field reference identifier must not have the form of a
qualified name (i.e., with ".").

Fragment 0:0

[Error Id: 88acd3d8-4e44-49f6-b587-24bf26f89a3b on zeta4.brewingintel.com:20005]
Vova Vysotskyi
2017-08-11 18:33:15 UTC
Permalink
Raw Message
Hi John,

Fix for the DRILL-4264
<https://issues.apache.org/jira/browse/DRILL-4264> should
solve this issue. This error appears when you try to do *select **. But
while DRILL-4264 <https://issues.apache.org/jira/browse/DRILL-4264> is not
fixed, you can try to do *select `**id.orig_h`*. It should not throw the
error.

Kind regards,
Volodymyr Vysotskyi
Post by John Omernik
Hey all,
I am querying some json and parquet data that has dots in the name. Not all
the data I may be querying will come from Drill, thus dot is a valid
character... when I go to initially explore my data, Drill throws the error
below when I run a select * query.
I understand the error, and I can create a view, selecting each column out
and renaming it for easier select * in the future. However, as a user, if
I get a new data set, this could (unless I am informed of another way here)
force me to leave drill to explore my data.
I get how using periods as field qualifiers causes issues... but if we had
had a way to read a file to get the schema, to either produce the all the
fields in a select query for easy view creation or a way to query with
periods in the name that would be awesome! It would keep users IN drill
instead of going elsewhere to explore their data.
I am open to ideas!
Error Returned - Code: 500
SYSTEM ERROR: UnsupportedOperationException: Unhandled field reference
"id.orig_h"; a field reference identifier must not have the form of a
qualified name (i.e., with ".").
Fragment 0:0
[Error Id: 88acd3d8-4e44-49f6-b587-24bf26f89a3b on
zeta4.brewingintel.com:20005]
John Omernik
2017-08-11 18:48:50 UTC
Permalink
Raw Message
So, I added a comment on that JIRA... I have a parquet file, I need to
explore it... it comes from another user, I want to be able to explore with
select * but I don't know the field names... how can I do this? I know it
seems like we should always know the column names, but I don't :)
Post by Vova Vysotskyi
Hi John,
Fix for the DRILL-4264
<https://issues.apache.org/jira/browse/DRILL-4264> should
solve this issue. This error appears when you try to do *select **. But
while DRILL-4264 <https://issues.apache.org/jira/browse/DRILL-4264> is not
fixed, you can try to do *select `**id.orig_h`*. It should not throw the
error.
Kind regards,
Volodymyr Vysotskyi
Post by John Omernik
Hey all,
I am querying some json and parquet data that has dots in the name. Not
all
Post by John Omernik
the data I may be querying will come from Drill, thus dot is a valid
character... when I go to initially explore my data, Drill throws the
error
Post by John Omernik
below when I run a select * query.
I understand the error, and I can create a view, selecting each column
out
Post by John Omernik
and renaming it for easier select * in the future. However, as a user,
if
Post by John Omernik
I get a new data set, this could (unless I am informed of another way
here)
Post by John Omernik
force me to leave drill to explore my data.
I get how using periods as field qualifiers causes issues... but if we
had
Post by John Omernik
had a way to read a file to get the schema, to either produce the all the
fields in a select query for easy view creation or a way to query with
periods in the name that would be awesome! It would keep users IN drill
instead of going elsewhere to explore their data.
I am open to ideas!
Error Returned - Code: 500
SYSTEM ERROR: UnsupportedOperationException: Unhandled field reference
"id.orig_h"; a field reference identifier must not have the form of a
qualified name (i.e., with ".").
Fragment 0:0
[Error Id: 88acd3d8-4e44-49f6-b587-24bf26f89a3b on
zeta4.brewingintel.com:20005]
Loading...