usenet-indexer/lib/sevenzip/index.ts

import { crc32 } from 'node:zlib'

/**
 * 7-Zip archives, independent of AdonisJS. A 7z archive starts with a
 * 32-byte signature header that points to the archive's header (the file
 * list) at its very end, often LZMA-compressed ("encoded") with its packed
 * data just before it. So the list needs the first 32 bytes and the tail:
 * for a split archive (.7z.001, .002, …), the start of the first volume and
 * the end of the last. Listing itself is left to libarchive (bsdtar).
 */

const SIGNATURE = Buffer.from([0x37, 0x7a, 0xbc, 0xaf, 0x27, 0x1c])
const SIGNATURE_HEADER_SIZE = 32
/** Header property ids (7zFormat.txt). */
const K_END = 0x00
const K_PACK_INFO = 0x06
const K_UNPACK_INFO = 0x07
const K_SIZE = 0x09
const K_CRC = 0x0a
const K_FOLDER = 0x0b
const K_ENCODED_HEADER = 0x17

export interface SevenZipStart {
  /** Where the archive header starts, from the start of the archive. */
  headerStart: number
  headerSize: number
  headerCrc: number
  /** The archive's total size (the header is its last part). */
  archiveSize: number
}

/** Reads the signature header; null when `data` isn't a 7z start or its CRC fails. */
export function parseSevenZipStart(data: Buffer): SevenZipStart | null {
  if (data.length < SIGNATURE_HEADER_SIZE || !data.subarray(0, 6).equals(SIGNATURE)) return null
  if ((crc32(data.subarray(12, 32)) >>> 0) !== data.readUInt32LE(8)) return null
  const offset = Number(data.readBigUInt64LE(12))
  const size = Number(data.readBigUInt64LE(20))
  const headerStart = SIGNATURE_HEADER_SIZE + offset
  return { headerStart, headerSize: size, headerCrc: data.readUInt32LE(28), archiveSize: headerStart + size }
}

/** Whether `header` (the bytes at headerStart) is the archive header its signature header describes. */
export function checkHeader(start: SevenZipStart, header: Buffer): boolean {
  return header.length === start.headerSize && (crc32(header) >>> 0) === start.headerCrc
}

/**
 * Where the bytes needed for listing start: the header itself, or for an
 * encoded header the packed (compressed) header stream it points to.
 */
export function listingStart(start: SevenZipStart, header: Buffer): number {
  if (header[0] !== K_ENCODED_HEADER || header[1] !== K_PACK_INFO) return start.headerStart
  const [packPos] = readNumber(header, 2)
  return SIGNATURE_HEADER_SIZE + packPos
}

/** 7zAES, the coder of password-protected streams. */
const AES_CODER = Buffer.from([0x06, 0xf1, 0x07, 0x01])

/**
 * Whether an encoded header (`header`, the bytes at headerStart) is
 * encrypted (7z -mhe): its coders include 7zAES. Then nothing can be listed
 * without the password, and neither the packed header nor bsdtar is needed.
 * Null when the header isn't encoded or doesn't read as expected.
 */
export function encodedHeaderEncrypted(header: Buffer): boolean | null {
  try {
    if (header[0] !== K_ENCODED_HEADER || header[1] !== K_PACK_INFO) return null
    let pos = 2
    let numPackStreams: number
    ;[, pos] = readNumber(header, pos) // packPos
    ;[numPackStreams, pos] = readNumber(header, pos)
    for (let id = header[pos++]; id !== K_END; id = header[pos++]) {
      if (id === K_SIZE) {
        for (let i = 0; i < numPackStreams; i++) [, pos] = readNumber(header, pos)
      } else if (id === K_CRC) {
        const allDefined = header[pos++]
        let defined = numPackStreams
        if (!allDefined) {
          const bits = header.subarray(pos, pos + Math.ceil(numPackStreams / 8))
          pos += bits.length
          defined = [...bits].reduce((sum, byte) => sum + byte.toString(2).replaceAll('0', '').length, 0)
        }
        pos += 4 * defined
      } else {
        return null
      }
      if (pos >= header.length) return null
    }
    if (header[pos++] !== K_UNPACK_INFO || header[pos++] !== K_FOLDER) return null
    let numFolders: number
    ;[numFolders, pos] = readNumber(header, pos)
    if (numFolders < 1 || header[pos++] !== 0) return null // folders stored elsewhere
    let numCoders: number
    ;[numCoders, pos] = readNumber(header, pos)
    if (!(numCoders >= 1) || pos >= header.length) return null
    for (let c = 0; c < numCoders; c++) {
      const flags = header[pos++]
      const idSize = flags & 0x0f
      if (header.subarray(pos, pos + idSize).equals(AES_CODER)) return true
      pos += idSize
      if (flags & 0x10) {
        ;[, pos] = readNumber(header, pos)
        ;[, pos] = readNumber(header, pos)
      }
      if (flags & 0x20) {
        let size: number
        ;[size, pos] = readNumber(header, pos)
        pos += size
      }
      if (pos > header.length) return null
    }
    return false
  } catch {
    return null
  }
}

/**
 * 7z's variable-length number: the leading 1-bits of the first byte count
 * the little-endian bytes that follow; its remaining bits are the high part.
 */
export function readNumber(data: Buffer, pos: number): [value: number, next: number] {
  const first = data[pos]
  let mask = 0x80
  let value = 0
  for (let i = 0; i < 8; i++) {
    if ((first & mask) === 0) {
      value += (first & (mask - 1)) * 2 ** (8 * i)
      return [value, pos + 1 + i]
    }
    value += data[pos + 1 + i] * 2 ** (8 * i)
    mask >>= 1
  }
  return [value, pos + 9]
}

export interface SevenZipEntry {
  name: string
  size: number
  directory: boolean
}

/**
 * Entries from `bsdtar -tv` output, e.g.
 * `-rw-rw-r--  0 0      0     3000000 Sep 30 19:13 Some.Release/movie.mkv`.
 */
export function parseBsdtarListing(stdout: string): SevenZipEntry[] {
  const entries: SevenZipEntry[] = []
  for (const line of stdout.split('\n')) {
    const match = /^([-dlbcps])\S*\s+\d+\s+\S+\s+\S+\s+(\d+)\s+\w{3}\s+\d{1,2}\s+(?:\d{1,2}:\d{2}|\d{4})\s(.+)$/.exec(line)
    if (!match) continue
    const directory = match[1] === 'd'
    entries.push({ name: directory ? match[3].replace(/\/$/, '') : match[3], size: Number(match[2]), directory })
  }
  return entries
}

/** Volume number of a split 7z name (`x.7z.001` → 1), 0 for an unsplit `.7z`, null otherwise. */
export function sevenZipVolume(name: string): number | null {
  const split = /\.7z\.(\d{1,4})$/i.exec(name)
  if (split) return Number(split[1])
  return /\.7z$/i.test(name) ? 0 : null
}