`CIDv1.buffer` property recomputes on every access

[sumanjeet0012]

CIDv1.buffer recomputes the byte representation on every access by calling multicodec.add_prefix() and concatenating bytes. Go stores the bytes once in the Cid struct. For CIDs accessed frequently (e.g., in sets, dicts, or comparisons), this causes unnecessary overhead.

Problem

In cid/cid.py:

class CIDv1(BaseCID):
    @property
    def buffer(self) -> bytes:
        return b"".join([
            bytes([self.version]),
            multicodec.add_prefix(self.codec, self.multihash)
        ])
        # ← Recomputed every time .buffer is accessed

This is called by:

• encode() — every time the CID is serialized
• to_bytes() — alias for buffer
• key_string() — calls buffer
• __hash__() — via BaseCID, uses multihash not buffer (OK)
• prefix() — calls mh.decode(self.multihash) (separate issue)

Proposed Solution

Cache the buffer on first computation:

class CIDv1(BaseCID):
    def __init__(self, codec, multihash):
        super().__init__(1, codec, multihash)
        self._buffer = None

    @property
    def buffer(self) -> bytes:
        if self._buffer is None:
            self._buffer = b"".join([
                bytes([self.version]),
                multicodec.add_prefix(self.codec, self.multihash)
            ])
        return self._buffer

Related

• File: cid/cid.py