Using your sed approach, this should replace all NULL with empty character sed 's/[\t]/,/g; s/NULL//g' > myfile.csv If there is a chance that NULL is a substring of a value you will need to do the following where ^ is beginning of line and $ is end of line and , is your field delimiter sed 's/[\t]/,/g; s/^NULL,/,/g; s/,NULL,/,,/g; s/,NULL$/,/g;' > myfile.csv Note that if your resultset is large, it is probably best to use Pig on HDFS and not sed (to leverage the parallel processing of hadoop and save yourself a lot of time. Note also: To use empty character as nulls in the actual hive table, use the following in the DDL TBLPROPERTIES('serialization.null.format'='');
... View more
