Reputation: 759
I want to use a better constructor for HashMap so I wrote one like this:
public static Map<S, T> newMap(S obj1, T obj2, Object... objects) {
h = new HashMap<S, T> ();
h.put(obj1, obj2);
... // parse object and put it in h.
}
My problem is I have few methods that take variadic arguments themselves and transfer the arguments to the newMap method.
For example,
public static void foo(S arg0, T arg1, Object... kv) {
...
newMap(tmp0, tmp1, kv); // WRONG: kv is an array and not a variadic transfer.
}
and called as:
foo(arg0, arg1, key1, val1, key2, val2);
This works fine but, when the call to newMap is:
newMap(tmp0, tmp1, tmp2, tmp3, kv);
and call to foo is
foo(arg0, arg1)
then the empty array [] sits in kv. Which creates problems with parsing, because we now have uneven number of elements in the "objects" array.
objects = { tmp2, tmp3, [] }
Can this be done without walking through the arguments in the kv object array and reconstructing an array with all objects [tmp0, tmp1, tmp2, tmp3] for the newMap call.
Refer: Better Map Constructor
Upvotes: 0
Views: 455
Reputation: 533680
You can do this instead.
public static <K, V> Map<K, V> newMap(K key1, V value1, Object... kvs) {
Map<K, V> m = new HashMap<>();
m.put(key1, value1);
addToMap(map, kvs);
return m;
}
private static <K, V> void addToMap(Map<K,V> map, Object[] kvs) {
for(int i = 0; i < kvs.length; i++) {
Object key = kvs[i];
if (key instanceof Object[]) {
addToMap(map, (Object[]) key);
} else if (++i < kvs.length) {
map.put((K) key, (V) kvs[i]);
}
}
}
This will unpack nested Object[] if they are "keys", if they are "values" they will remain as Object[].
Upvotes: 2