T-HUB/packages/egress/src/hls.ts

import { EgressError } from "./errors.js";
import { EgressPolicy, type ValidatedEndpoint } from "./policy.js";

interface UriOccurrence { start: number; end: number; value: string }

/** Parse attribute boundaries rather than finding URI= inside other quoted values. */
function attributeUris(attributes: string, offset: number): UriOccurrence[] {
  const uris: UriOccurrence[] = [];
  if (!/^\s*[A-Z0-9-]+\s*=/.test(attributes)) return uris;
  let cursor = 0;
  const seen = new Set<string>();
  while (cursor < attributes.length) {
    const key = /^\s*([A-Z0-9-]+)\s*=\s*/.exec(attributes.slice(cursor));
    if (!key) throw new EgressError("INVALID_PLAYLIST", "Malformed HLS attributes");
    cursor += key[0].length;
    const quoted = attributes[cursor] === '"';
    const start = cursor + (quoted ? 1 : 0);
    const end = quoted ? attributes.indexOf('"', start) : attributes.indexOf(",", start);
    const valueEnd = end < 0 && !quoted ? attributes.length : end;
    if (valueEnd < 0) throw new EgressError("INVALID_PLAYLIST", "Unterminated HLS attribute");
    if (key[1] === "URI" || key[1] === "SERVER-URI") {
      if (!quoted || start === valueEnd || seen.has(key[1])) throw new EgressError("INVALID_PLAYLIST", "Invalid HLS URI attribute");
      seen.add(key[1]);
      uris.push({ start: offset + start, end: offset + valueEnd, value: attributes.slice(start, valueEnd) });
    }
    cursor = valueEnd + (quoted ? 1 : 0);
    while (/\s/.test(attributes[cursor] ?? "") && cursor < attributes.length) cursor++;
    if (cursor === attributes.length) break;
    if (attributes[cursor] !== "," || cursor + 1 === attributes.length) throw new EgressError("INVALID_PLAYLIST", "Invalid HLS attribute suffix");
    cursor++;
  }
  return uris;
}

function playlistUris(playlist: string): { uris: UriOccurrence[]; master: boolean } {
  if (!/^#EXTM3U(?:\r?\n|$)/.test(playlist) || playlist.includes("\u0000")) throw new EgressError("INVALID_PLAYLIST", "Missing HLS header");
  const uris: UriOccurrence[] = [];
  let offset = 0;
  let awaitingVariant = false;
  let variants = 0;
  let media = false;
  for (const line of playlist.split(/(?<=\n)/)) {
    const text = line.replace(/\r?\n$/, "").trim();
    if (text && !text.startsWith("#")) {
      const start = offset + line.indexOf(text);
      uris.push({ start, end: start + text.length, value: text });
      if (awaitingVariant) { variants++; awaitingVariant = false; }
    } else if (text.startsWith("#EXT")) {
      if (text.startsWith("#EXTINF:")) media = true;
      if (text.startsWith("#EXT-X-STREAM-INF:")) {
        if (awaitingVariant) throw new EgressError("INVALID_PLAYLIST", "Variant URI is missing");
        awaitingVariant = true;
      }
      const colon = line.indexOf(":");
      if (colon >= 0) {
        const attributes = line.slice(colon + 1).replace(/\r?\n$/, "");
        const attributesFound = attributeUris(attributes, offset + colon + 1);
        uris.push(...attributesFound);
        if (text.startsWith("#EXT-X-I-FRAME-STREAM-INF:")) {
          if (!attributesFound.length) throw new EgressError("INVALID_PLAYLIST", "I-frame URI is missing");
          variants++;
        }
      }
    }
    offset += line.length;
  }
  if (awaitingVariant) throw new EgressError("INVALID_PLAYLIST", "Variant URI is missing");
  return { uris, master: variants > 0 && !media };
}

async function resolveUris(uris: UriOccurrence[], base: string | URL, policy: EgressPolicy): Promise<ValidatedEndpoint[]> {
  const baseUrl = (await policy.validate(base)).url;
  const endpoints: ValidatedEndpoint[] = [];
  // Sequential validation bounds DNS work even for large playlists; shared URI pins within this parse.
  const cache = new Map<string, ValidatedEndpoint>();
  for (const occurrence of uris) {
    if (!occurrence.value || occurrence.value.includes("{$") || /[\s\\\u0000-\u001f\u007f]/u.test(occurrence.value)) throw new EgressError("INVALID_PLAYLIST", "Invalid or unexpanded HLS URI");
    let absolute: URL;
    try { absolute = new URL(occurrence.value, baseUrl); } catch { throw new EgressError("INVALID_PLAYLIST", "Invalid HLS URI"); }
    let endpoint = cache.get(absolute.href);
    if (!endpoint) { endpoint = await policy.validate(absolute); cache.set(absolute.href, endpoint); }
    endpoints.push(endpoint);
  }
  return endpoints;
}

/** Validates master variants and URI-bearing tags (renditions, keys, maps, etc.) with DNS pinning. */
export async function validateMasterPlaylist(playlist: string, base: string | URL, policy = new EgressPolicy()): Promise<ValidatedEndpoint[]> {
  const { uris, master } = playlistUris(playlist);
  if (!master) throw new EgressError("INVALID_PLAYLIST", "Expected an HLS master playlist");
  return resolveUris(uris, base, policy);
}

/** Produces absolute, validated child URLs, preserving comments, attributes and newline style. */
export async function rewriteSegmentUris(playlist: string, base: string | URL, policy = new EgressPolicy()): Promise<string> {
  const { uris } = playlistUris(playlist);
  const endpoints = await resolveUris(uris, base, policy);
  let rewritten = playlist;
  for (let index = uris.length - 1; index >= 0; index--) {
    const uri = uris[index]!;
    rewritten = rewritten.slice(0, uri.start) + endpoints[index]!.url.href + rewritten.slice(uri.end);
  }
  return rewritten;
}