groovy

Add ability to 'slice' Map, querying multiple keys at once

Details

  • Type: New Feature New Feature
  • Status: Reopened Reopened
  • Priority: Major Major
  • Resolution: Unresolved
  • Affects Version/s: None
  • Fix Version/s: None
  • Component/s: groovy-jdk
  • Labels:
    None
  • Testcase included:
    yes
  • Patch Submitted:
    Yes
  • Number of attachments :
    2

Description

It was brought up on the dev mailing list that you could query multiple list values at once, but not multiple Map values:

Message from groovy-dev by Jon Cox

Groovy currently lets you say:

def    x       =  [100,200,300,400]
assert x[0]    == 100
assert x[2]    == 300
assert x[0,2]  == [100,300]

You can say:

def    y        =  [moo: 100, cow: 200, egg:300 hen:400]
assert y["moo"] == 100
assert y["cow"] == 300

But currently, you cannot slice a Map like this:

assert y["moo","egg"] == [100,300]

This patch adds that functionality, so you can do (from the unit test)

    void testMapSlice() {
        def m = [ a:1, b:2, c:3 ]
        
        assert m[ 'a', 'b' ] == [ 1, 2 ]
        assert m[ 'a', 'c' ] == [ 1, 3 ]
        assert m[ 'a', 'd', 'c' ] == [ 1, null, 3 ]
    }

Note that the resultant List contains null for keys that were not found. This (I believe) differs from the way that Perl handles this (I believe they are just skipped in perl), but I think that having the null values gives the developer more information.

And they could be filtered out by doing something like:

m[ 'a', 'd', 'c' ].findAll { it }

Hope it's ok!

  1. File
    map_slice.diff
    08/Jun/11 4:24 AM
    2 kB
    Tim Yates
  2. File
    slice_instead_of_getAt.diff
    10/Jun/11 3:18 AM
    2 kB
    Tim Yates

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Jon Cox added a comment -

Wow, that was quick!

Incidentally, hash slices in Perl work just like yours
(nulls are returned for keys that have no associated value).

The behavior you've implemented looks correct/intuitive.

Tim Yates, you rock.
Thanks!

Cheers,
-Jon

Show
Jon Cox added a comment - Wow, that was quick! Incidentally, hash slices in Perl work just like yours (nulls are returned for keys that have no associated value). The behavior you've implemented looks correct/intuitive. Tim Yates, you rock. Thanks! Cheers, -Jon
Hide
Permalink
Paul King added a comment - - edited

This looks interesting and close to ready for inclusion. The main thing that worries me is that there is a slight lack of symmetry. For example, in the examples given, we can do both of these:

assert y["moo"] == 100
assert y["moo","egg"] == [100,300]

but there is no way to for instance just return [100].

This might not seem important, but if you consider the following code:

def m = [:]
def key = [1, 2]
m[key] = 'foo'
println m[key]

It currently prints 'foo' but the above change would result in '[null, null]'. So it is a breaking change albeit one that hopefully doesn't arise too often in current code.

Show
Paul King added a comment - - edited This looks interesting and close to ready for inclusion. The main thing that worries me is that there is a slight lack of symmetry. For example, in the examples given, we can do both of these: assert y[ "moo" ] == 100 assert y[ "moo" , "egg" ] == [100,300] but there is no way to for instance just return [100] . This might not seem important, but if you consider the following code: def m = [:] def key = [1, 2] m[key] = 'foo' println m[key] It currently prints ' foo ' but the above change would result in ' [null, null] '. So it is a breaking change albeit one that hopefully doesn't arise too often in current code.
Hide
Permalink
Tim Yates added a comment - - edited

Gah, can't believe I missed that, thanks Paul!

Not sure how to get round this blocker...I think overloading getAt to give you seamless integration is a no-go... So the best I could come up with was to change the getAt method to:

public static <K,V> List<V> slice(Map<K,V> self, K... keys) {
    ArrayList<V> ret = new ArrayList<V>() ;
    for( K key : keys ) {
        ret.add( self.get( key ) ) ;
    }
    return ret ;
}

So thn you can do:

def map = [ a:10, b:20 ]
assert map.slice( 'a', 'b' ) == [ 10, 20 ]

and maps with list keys work too:

def m = [ (['a','b']):10, c:20 ]
assert m.slice( [ 'a', 'b' ] ) == [ 10 ]
assert m.slice( [ 'a', 'b' ], 'c', 'd' ) == [ 10, 20, null ]

Does this look ok? I guess even without the seamless integration, it's still a nice method to have?

Show
Tim Yates added a comment - - edited Gah, can't believe I missed that, thanks Paul! Not sure how to get round this blocker...I think overloading getAt to give you seamless integration is a no-go... So the best I could come up with was to change the getAt method to: public static <K,V> List<V> slice(Map<K,V> self, K... keys) { ArrayList<V> ret = new ArrayList<V>() ; for( K key : keys ) { ret.add( self.get( key ) ) ; } return ret ; } So thn you can do: def map = [ a:10, b:20 ] assert map.slice( 'a', 'b' ) == [ 10, 20 ] and maps with list keys work too: def m = [ (['a','b']):10, c:20 ] assert m.slice( [ 'a', 'b' ] ) == [ 10 ] assert m.slice( [ 'a', 'b' ], 'c', 'd' ) == [ 10, 20, null ] Does this look ok? I guess even without the seamless integration, it's still a nice method to have?
Hide
Permalink
Tim Yates added a comment -

Ignore this... I just discovered you can do:

[ a:10, b:20, c:30 ].subMap( [ 'a', 'X', 'c' ] ).values()

to get exactly the same functionality (the intermediate subMap call actually gives you more functionality) with existing methods...

Show
Tim Yates added a comment - Ignore this... I just discovered you can do: [ a:10, b:20, c:30 ].subMap( [ 'a', 'X', 'c' ] ).values() to get exactly the same functionality (the intermediate subMap call actually gives you more functionality) with existing methods...
Hide
Permalink
Tim Yates added a comment -

It was pointed out to me that although it duplicates the functionality, the slice method is much nicer and this issue should remain open... Sorry about the faffing...

Show
Tim Yates added a comment - It was pointed out to me that although it duplicates the functionality, the slice method is much nicer and this issue should remain open... Sorry about the faffing...

People

  • Assignee:
    Unassigned
    Reporter:
    Tim Yates
Vote (1)
Watch (1)

Dates

  • Created:
    Updated: