feat: new UDFs for working with Maps #5536
Conversation
|
|
||
| @UdfDescription( | ||
| name = "map_union", | ||
| description = "Returns a new map containing the union of all entries from both input maps. " |
There was a problem hiding this comment.
What if the maps have the same key with different values? Currently, the latter will be in the result. Maybe point this out in the description
There was a problem hiding this comment.
yeah - this is described in the docs file but not here, will add something to call it out here also
| public class MapUnion { | ||
|
|
||
| @Udf | ||
| public <T> Map<String, T> union(final Map<String, T> input1, final Map<String, T> input2) { |
There was a problem hiding this comment.
Add @UdfParameter just so that it can compile without the -parameters in the pom file?
There was a problem hiding this comment.
Thank you @blueedgenick ! LGTM! Maybe just add the @UdfParameter to the udfs so that they can compile without problems if the -parameter is missing from the pom file
|
Thanks @vpapavas ! ok to merge with these changes ? |
Jim's doc feedback Co-authored-by: Jim Galasyn <[email protected]>
Jim's doc feedback Co-authored-by: Jim Galasyn <[email protected]>
Jim's doc feedback Co-authored-by: Jim Galasyn <[email protected]>
|
Can you add the |
Three new UDFs:
map_keys(some_map)- get array of all keys from a mapmap_values(some_map)- get array of all values from a mapmap_union(map1, map2)- add all entries from 2 maps into a new mapThis is an updated version of functions and their tests originally proposed waaaay back in PR #1611 and blocked at that time by limited scope for handling generics in the UDF invocation framework.
Ideally I'd like to also make
map_unionvariadic but the invocation framework will choke on that until #5511 is resolved.