DRILL-3425 doesn't seem to be about adding filename information to the
directory name information that Drill already provides.
Also, other users have requested the addition of the file's simple name to the
pathname segments currently available through dir0, dir1, etc. See
DRILL-3474 (Filename should be an available column when querying a directory)
<https://issues.apache.org/jira/browse/DRILL-3474>.
Daniel
Christopher Matta wrote:
Seems like this was brought up in DRILL-3425
<https://issues.apache.org/jira/browse/DRILL-3425> and identified as not
something that was in the design spec.
Chris Matta
[email protected]
215-701-3146
On Thu, Sep 24, 2015 at 2:57 PM, Tim Harper <[email protected]> wrote:
So, if the files are in subdirectories, I DO get dir0, but if I query a
single directory with a handful of files, there is no way (it seems) to see
any file name information.
On Thu, Sep 24, 2015 at 12:32 PM, Tim Harper <[email protected]> wrote:
With apache drill, I'm able to query a directory of JSON files just fine,
by invoking:
select * from file.`/path/to/data` t;
All of JSON files are selected, and the data comes as I'd expect.
However,
no fields are returned describing from which file the data came. I'd like
to be able to use this signal in my query, in addition.
I apologize if this is clearly documented somewhere. I've looked and I'm
having a hard time.
This is the configuration for my file storage plugin:
{
"type": "file",
"enabled": true,
"connection": "file:///",
"workspaces": {
"scimmia": {
"location": "/home/scimmia",
"writable": false,
"defaultInputFormat": null
}
},
"formats": {
"json": {
"type": "json"
}
}
}
--
Daniel Barclay
MapR Technologies
