performance - Is using java Map.containsKey() redundant when using map.get()

Question

I have been wondering for some time whether it is allowable within best practice to refrain from using the containsKey() method on java.util.Map and instead do a null check on the result from get().

My rationale is that it seems redundant to do the lookup of the value twice - first for the containsKey() and then again for get().

On the other hand it may be that most standard implementations of Map cache the last lookup or that the compiler can otherwise do away with the redundancy, and that for readability of the code it is preferable to maintain the containsKey() part.

I would much appreciate your comments.

Answer 1

Some Map implementations are allowed to have null values, eg HashMap, in this case if get(key) returns null it does not guarantee that there is no entry in the map associated with this key.

So if you want to know if a map contains a key use Map.containsKey. If you simply need a value mapped to a key use Map.get(key). If this map permits null values, then a return value of null does not necessarily indicate that the map contains no mapping for the key; In such case Map.containsKey is useless and will affect performance. Moreover, in case of concurrent access to a map (eg ConcurrentHashMap), after you tested Map.containsKey(key) there is a chance that the entry will be removed by another thread before you call Map.get(key).