refactor(webrtc): deduplicate _varint.py — use libp2p.utils.varint

[acul71]

Summary

PR #1309 adds libp2p/transport/webrtc/_varint.py with encode_uvarint() and decode_uvarint() for WebRTC data-channel and signaling framing. The same unsigned LEB128 logic already exists in the canonical project module libp2p/utils/varint.py, which is used by mplex, msgio, peer envelopes, identify, interop tests, and others.

This is duplicated code on the same wire format; WebRTC should reuse the shared utility rather than maintain a fourth in-tree varint implementation (alongside libp2p/utils/varint.py, the external varint PyPI package used in kad-dht/bitswap/filecoin, and _varint_size() in pubsub/rpc_queue.py).

Current duplication

WebRTC (_varint.py)	Existing (libp2p/utils/varint.py)
encode_uvarint(value)	encode_uvarint(value) — same algorithm
decode_uvarint(data, offset=0) -> (value, bytes_consumed)	decode_varint_with_size(data) -> (value, bytes_consumed) (no offset; use data[offset:])
—	decode_uvarint_from_stream(reader) — async stream read

Consumers today:

• libp2p/transport/webrtc/stream.py — encode_uvarint / decode_uvarint for SCTP message framing
• libp2p/transport/webrtc/signaling.py — encode_uvarint for writes; custom _read_uvarint() for async reads (duplicates decode_uvarint_from_stream)

The encode loops are functionally identical (while value > 0x7F vs while value >= 0x80 is equivalent). Decode-with-length maps 1:1 to decode_varint_with_size.

Suggested approach

1. Delete libp2p/transport/webrtc/_varint.py.

2. Update imports in stream.py and signaling.py:

   from libp2p.utils.varint import (
       encode_uvarint,
       decode_varint_with_size,
       decode_uvarint_from_stream,
   )

3. Replace decode_uvarint(raw) with decode_varint_with_size(raw) in stream.py (same (value, consumed) tuple).

4. Replace signaling._read_uvarint() with decode_uvarint_from_stream(stream), mapping ParseError to WebRTCSignalingError at the call site if needed.

5. Update tests in tests/core/transport/webrtc/test_signaling.py that import from _varint to use libp2p.utils.varint (or test via public signaling/stream APIs only).

No wire-format or protocol change — this is a pure refactor.

Minor API differences to handle

• Offset parameter: WebRTC decode_uvarint(data, offset) can be replaced by slicing: decode_varint_with_size(data[offset:]), then adjust consumed if callers need absolute offset (currently stream.py uses offset 0 only).
• Error types: WebRTC raises ValueError on truncated varints; utils uses ParseError in some stream paths and ValueError in others. Preserve existing WebRTC error handling at boundaries if tests depend on it.
• Empty/truncated input: Confirm edge-case behavior matches (utils decode_varint_with_size on empty bytes returns (0, 0); WebRTC raises ValueError — verify call sites never hit empty prefix, or add explicit checks).

Scope / timing

• Sub-issue of Add Support for WebRTC set-up in py-libp2p #546 (WebRTC transport).
• Safe as a post-feat(webrtc): add WebRTC transport scaffolding (spec compliance + tests) #1309 merge cleanup — does not block experimental WebRTC scaffolding landing.
• Small diff (~delete one file, update 2–3 imports, adjust tests).

Acceptance criteria

• _varint.py removed
• WebRTC stream and signaling modules import from libp2p.utils.varint
• All tests/core/transport/webrtc/ tests pass
• No change to on-wire framing behavior

Refs #546, #1309.

[yashksaini-coder]

@acul71 can I take on this ?

[acul71]

@acul71 can I take on this ?

go ahead