[
https://issues.apache.org/jira/browse/PIG-4673?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14876190#comment-14876190
]
Rohini Palaniswamy commented on PIG-4673:
-----------------------------------------
Good feature [[email protected]]. Would you be interested in enhancing
this UDF for better performance in a new jira?
http://stackoverflow.com/questions/7661460/replace-multiple-substrings-at-once/7661573#7661573
- You can basically compile the Pattern once and cache it (Have a limit on the
cache if the search strings are variable and not constant) and do the multiple
replace in one go. Have seen a lot of jobs suffer in performance because of
UDFs with regex match and not reusing compiled Pattern.
> Built In UDF - REPLACE_MULTI : For a given string, search and replace all
> occurrences of search keys with replacement values.
> ------------------------------------------------------------------------------------------------------------------------------
>
> Key: PIG-4673
> URL: https://issues.apache.org/jira/browse/PIG-4673
> Project: Pig
> Issue Type: New Feature
> Components: piggybank
> Affects Versions: site
> Reporter: Murali Rao
> Assignee: Murali Rao
> Priority: Minor
> Labels: None
> Fix For: 0.16.0
>
> Attachments: PIG-4673-1.patch, replace_multi_udf.patch
>
>
> Lets say we have a string = 'A1B2C3D4'. Our objective is to replace A with 1,
> B with 2, C with 3 and D with 4 to derive 11223344 string.
> Using existing REPLACE method
> REPLACE(REPLACE(REPLACE(REPLACE('A1B2C3D4','A','1'),'B','2'),'C','3'),'D','4')
>
> With proposed UDF : REPLACE_MULTI method
> General Syntax :
> REPLACE_MULTI ( sourceString, [ search1#replacement1, ... ] )
> REPLACE_MULTI ( 'A1B2C3D4', [ 'A'#'1','B'#'2', 'C'#'3', 'D'#'4' ] )
> Advantage :
> 1. Function calls are reduced.
> 2. Ease to code and better readable.
>
> Let me know your thoughts/ inputs on having this UDF in Piggy Bank. Will take
> this up based on this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
