Search code examples
regexhadoophiveudfmapr

Hive -e and regex replace udf issue

I am working on Hive 0.13 in MapR distro. I am seeing a weird issue when i run the following query in hive using the hive -e option.

Below is my query :

select regexp_replace('$60,825.48','\\$|\,','');

The above query works fine when I run from the hive shell directly and gives desired the output as 60825.48.

Now, When I run the same using the hive -e as follows :

hive -e "select regexp_replace('$60,825.48','\\$|\,','');"
Total MapReduce CPU Time Spent: 890 msec
OK
0825.48
Time taken: 7.813 seconds, Fetched: 1 row(s)

The "6" gets trimmed off from the desired output.

Also, The same works fine with Hive -f option too.

Could you please help me identify the issue?

Solution

  • Try this

    hive -e 'select regexp_replace("$60,825.48","\\$|\,","");'

    If your objective is to remove everything except numbers and '.', then you can try this as well

    hive -e 'select regexp_replace("$60,825.48","[^0-9.]","");'