Interface Hasher

  • All Superinterfaces:
    PrimitiveSink

    @Beta
    public interface Hasher
    extends PrimitiveSink
    A PrimitiveSink that can compute a hash code after reading the input. Each hasher should translate all multibyte values (putInt(int), putLong(long), etc) to bytes in little-endian order.

    Warning: The result of calling any methods after calling hash() is undefined.

    Warning: Using a specific character encoding when hashing a CharSequence with putString(CharSequence, Charset) is generally only useful for cross-language compatibility (otherwise prefer putUnencodedChars(java.lang.CharSequence)). However, the character encodings must be identical across languages. Also beware that Charset definitions may occasionally change between Java releases.

    Warning: Chunks of data that are put into the Hasher are not delimited. The resulting HashCode is dependent only on the bytes inserted, and the order in which they were inserted, not how those bytes were chunked into discrete put() operations. For example, the following three expressions all generate colliding hash codes:

    
     newHasher().putByte(b1).putByte(b2).putByte(b3).hash()
     newHasher().putByte(b1).putBytes(new byte[] { b2, b3 }).hash()
     newHasher().putBytes(new byte[] { b1, b2, b3 }).hash()
     

    If you wish to avoid this, you should either prepend or append the size of each chunk. Keep in mind that when dealing with char sequences, the encoded form of two concatenated char sequences is not equivalent to the concatenation of their encoded form. Therefore, putString(CharSequence, Charset) should only be used consistently with complete sequences and not broken into chunks.

    Since:
    11.0
    Author:
    Kevin Bourrillion