> You could pre-parse the input as well, create a user defined function, or > use the streaming > logic in MAP/REDUCE/TRANSFORM.
Are you referring to http://wiki.apache.org/hadoop/Hive/LanguageManual/Transform for map/reduce/transform? From that page, the following points are not very clear: 1. How do I use a custom Java function for mapping the input? 2. Can my mapper take the whole row as a single input and transform it into multiple columns? (All examples on the Wiki seem to have an equal number of input parameters and output values. Can they be different? Sorry, newbie question!) 3. Does the mapping script/function need to be present locally or on the DFS? About user defined functions, http://wiki.apache.org/hadoop/Hive/LanguageManual/UDF, does not seem to have any information on how to create them. Saurabh. -- http://nandz.blogspot.com http://foodieforlife.blogspot.com
