body.ts

/**
 * Body canonicalization and content_hash — Verifiable OKF SPEC §5.
 *
 * The canonical body is defined deterministically so that signer and verifier
 * always agree byte-for-byte. No markdown-semantic normalization is performed
 * in v0 (no link/heading/whitespace collapsing inside lines).
 */
import { createHash } from "node:crypto";

const utf8Encoder = new TextEncoder();

/** Decode bytes as UTF-8, rejecting invalid UTF-8 (SPEC §5 step 2). */
export function decodeUtf8(bytes: Uint8Array): string {
  return new TextDecoder("utf-8", { fatal: true }).decode(bytes);
}

/**
 * Canonicalize an (already UTF-8-decoded) body string into canonical bytes.
 * Steps 3–6 of SPEC §5:
 *   3. normalize CRLF / CR → LF
 *   4. strip trailing spaces/tabs at the end of each line
 *   5. ensure exactly one trailing LF
 *   6. re-encode UTF-8
 */
export function canonicalizeBody(body: string): Uint8Array {
  const lf = body.replace(/\r\n/g, "\n").replace(/\r/g, "\n"); // step 3
  const stripped = lf
    .split("\n")
    .map((line) => line.replace(/[ \t]+$/u, "")) // step 4
    .join("\n");
  const oneTrailingLf = stripped.replace(/\n+$/u, "") + "\n"; // step 5
  return utf8Encoder.encode(oneTrailingLf); // step 6
}

/**
 * Canonicalize body bytes directly: validates UTF-8 (§5 step 2) then applies
 * steps 3–6. Throws `TypeError` on invalid UTF-8.
 */
export function canonicalizeBodyBytes(bodyBytes: Uint8Array): Uint8Array {
  return canonicalizeBody(decodeUtf8(bodyBytes));
}

/** content_hash over canonical body bytes: "sha256:" + lowercase hex SHA-256 (§5). */
export function contentHash(canonicalBody: Uint8Array): string {
  return "sha256:" + createHash("sha256").update(canonicalBody).digest("hex");
}