Class ByteReader

constructor

• new ByteReader(bytes: Uint8Array, pos?: number): ByteReader

  Parameters

  • bytes: Uint8Array
  • pos: number = 0

  Returns ByteReader

Readonlybytes

pos

remaining

• get remaining(): number

  Returns number

peekU8

• peekU8(offset?: number): number

  Parameters

  • offset: number = 0

  Returns number

readBytes

• readBytes(n: number): Uint8Array

  Parameters

  • n: number

  Returns Uint8Array

readI32LE

• readI32LE(): number

  Returns number

readU16LE

• readU16LE(): number

  Returns number

readU32LE

• readU32LE(): number

  Returns number

readU64LE

• readU64LE(): bigint

  Returns bigint

readU8

• readU8(): number

  Returns number

readVarInt

• readVarInt(): bigint

  Bitcoin CompactSize varint. A wider encoding than the value needs is accepted here, where Bitcoin Core's ReadCompactSize refuses it, so no such transaction is in any block. The consequence is recorded in parsers.props.test.ts under "varint: canonicality" and is not what a reader expects: parseTx derives the txid from serializeStripped, which writes every count canonically, so a padded copy of a real transaction reproduces the real txid and its txid merkle proof passes rather than failing. Nothing forged rides along, because a wider encoding states the same count and every field parses identically, the padding being dropped. The wtxid is the hash of the bytes as received, so the same copy fails an L3 wtxid proof, and the 64-byte leaf/node rule reads the stripped serialization, so padding cannot carry a transaction around it. What stays open is stated in the legacy-transaction test: tx.raw is not pinned by either identifier, so future code must not treat it as though it were.

  Returns bigint

readVarIntNum

• readVarIntNum(): number

  Returns number