Yeah the record delimiter currently has to be a character. I did find https://issues.apache.org/jira/browse/SQOOP-1175 upstream tracking this. Could you add a note saying you also would like this feature on the Jira? It'll give the community a sense of priority on it.
On Mon, Apr 6, 2015 at 9:48 AM, Brenden Cobb <[email protected]> wrote: > Abe- Thanks for your reply. I was hasty to refer to a limitation as a > bug. > On import "--lines-terminated-by" ignores more than one character e.g. > "||" will result in "|". > On Export "--input-lines-terminated-by" accepts more characters, as long > as each record is on a separate line. > > For example a file: > whiskers|cat#^ > fido|dog#^ > > This will export fine with "--input-lines-terminated-by '#^'", however a > file like "whiskers|cat#^fido|dog#^" won't break into records properly, > writing only one field to the database. > > I tested this on the latest version of Sqoop and there's not change. > More robust options would be nice, but upstream workarounds aren't too > difficult. > > Thanks, > BC > > > > From: Abraham Elmahrek <[email protected]> > Reply-To: "[email protected]" <[email protected]> > Date: Friday, April 3, 2015 3:13 PM > To: "[email protected]" <[email protected]> > Subject: Re: Record terminator > > Hey man, > > What bug is this? Also, could you provide your sqoop command? > > This probably works in later versions of Sqoop. > > -Abe > > On Wed, Apr 1, 2015 at 8:25 AM, Brenden Cobb <[email protected]> > wrote: > >> Hi- >> Running Cloudera's Sqoop 1.4.3, I'm hitting the apparent bug that >> "--lines-terminated-by" doesn't honor any value other than newline. >> >> Wondering If there's any direct Sqoop way around this. Also, have later >> Sqoop releases fixed this issue? >> >> Thanks for your help. >> >> -BC >> >> >> > >
