Files
usenet-indexer/lib/sevenzip/list.ts
T
Daan MeijerandClaude Opus 5.5 2dbca9b152
Test and deploy / test (push) Successful in 1m49s
Test and deploy / deploy (push) Successful in 2m15s
Content inspection and archive walks on seeding nodes (relay)
Inspection is split into prepare (database), probe (lib/content/probe:
downloads, decoding and parsing, no database) and apply (database).
worker:collection runs all three; seed:node with SEED_INSPECTORS leases
jobs from the same queues through the seed API (BullMQ manual leases),
fetches articles through the seed API and posts the probes back.

A node now needs only SEED_URL and SEED_TOKEN: the settings the app
requires but a node never uses get placeholders for seed:node.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-06 11:17:02 +02:00

172 lines
6.4 KiB
TypeScript

import { execFile } from 'node:child_process'
import { mkdtemp, open, rm } from 'node:fs/promises'
import { tmpdir } from 'node:os'
import path from 'node:path'
import { promisify } from 'node:util'
import { decodeSegment, type DecodedSegment } from '../yenc/index.js'
import type { RarInfo } from '../rar/parse.js'
import type { BodyFetcher } from '../rar/segment_reader.js'
import {
checkHeader,
encodedHeaderEncrypted,
listingStart,
parseBsdtarListing,
parseSevenZipStart,
type SevenZipStart,
} from './index.js'
const run = promisify(execFile)
/** The tail read for a listing: header plus packed header; more is not a header. */
const MAX_TAIL_BYTES = 16 * 1024 * 1024
/**
* Lists a 7z archive, also a split one (.7z.001, .002, …), from the first
* segment of the first volume (`firstData`, decoded) and the last segment(s)
* of the last one (CONCEPT.md D19). Those bytes are put at their place in a
* sparse file the size of the whole archive, which bsdtar (libarchive)
* lists: it only reads the start and the header at the end. Null when it
* isn't a 7z start, or the tail doesn't hold the header its start points to.
*/
export async function listSevenZip(
firstData: Buffer,
volumeCount: number,
lastVolume: { number: number; id: string }[],
fetch: BodyFetcher,
warn: (message: string) => void = () => {}
): Promise<RarInfo | null> {
const start = parseSevenZipStart(firstData)
if (!start) return null
const split = volumeCount > 1
if (start.headerSize === 0) return sevenZipInfo([], false, false, split)
const tail = await readTail(lastVolume, start, fetch)
if (!tail) return null
// Encrypted headers (-mhe): known from the header alone, no listing possible (most 7z uploads).
if ('headersEncrypted' in tail) return sevenZipInfo([], true, true, split)
const dir = await mkdtemp(path.join(tmpdir(), '7z-'))
try {
const archive = path.join(dir, 'archive.7z')
const handle = await open(archive, 'w')
try {
await handle.write(firstData, 0, Math.min(firstData.length, start.archiveSize), 0)
await handle.write(tail.data, 0, tail.data.length, tail.from)
await handle.truncate(start.archiveSize)
} finally {
await handle.close()
}
return await bsdtar(archive, split, warn)
} finally {
await rm(dir, { recursive: true, force: true })
}
}
/** Lists the sparse archive; a trial extraction of one entry tells whether its data is encrypted. */
async function bsdtar(archive: string, split: boolean, warn: (message: string) => void): Promise<RarInfo | null> {
let stdout = ''
try {
stdout = (await run('bsdtar', ['-tvf', archive], { timeout: 60_000, maxBuffer: 32 * 1024 * 1024 })).stdout
} catch (error: any) {
if (/header is encrypted/i.test(String(error.stderr))) return sevenZipInfo([], true, true, split)
warn(`bsdtar could not list a 7z archive: ${String(error.stderr || error.message).trim()}`)
return null
}
const entries = parseBsdtarListing(stdout)
const sample = entries.filter((e) => !e.directory).sort((a, b) => a.size - b.size)[0]
let encrypted = false
if (sample) {
try {
await run('bsdtar', ['-xOf', archive, sample.name], { timeout: 30_000, maxBuffer: 4 * 1024 * 1024, encoding: 'buffer' })
} catch (error: any) {
// Without the data (it's all zeros here), anything else fails too: only this message counts.
encrypted = /content is encrypted/i.test(String(error.stderr))
}
}
return sevenZipInfo(entries, false, encrypted, split)
}
export function sevenZipInfo(
entries: { name: string; size: number; directory: boolean }[],
headersEncrypted: boolean,
encrypted: boolean,
split: boolean
): RarInfo {
return {
format: '7z',
volume: split,
firstVolume: true,
solid: false,
headersEncrypted,
truncated: false,
next: 0,
entries: entries.map((e) => ({
name: e.name,
size: e.size,
packedSize: 0,
directory: e.directory,
encrypted: encrypted && !e.directory,
splitBefore: false,
splitAfter: false,
stored: false,
dataOffset: 0,
})),
}
}
/**
* The archive's tail from the last volume: the header (checked against its
* CRC) and, for an encoded header, the packed stream before it. `from` is
* its position in the whole archive.
*/
async function readTail(
last: { number: number; id: string }[],
start: SevenZipStart,
fetch: BodyFetcher
): Promise<{ from: number; data: Buffer } | { headersEncrypted: true } | null> {
const ids = new Map(last.map((s) => [s.number, s]))
const numbers = [...ids.keys()]
if (numbers.length === 0) return null
const segments = new Map<number, DecodedSegment>()
const segment = async (n: number) => {
if (!segments.has(n)) {
const id = ids.get(n)?.id
if (!id) throw new Error(`part ${n} of the last 7z volume is missing`)
segments.set(n, decodeSegment(await fetch(id), { lenientSize: true }))
}
return segments.get(n)!
}
const lastNumber = numbers[numbers.length - 1]
const lastSegment = await segment(lastNumber)
const volumeSize = lastSegment.meta.begin.size
const volumeStart = start.archiveSize - volumeSize
const partBegin = (lastSegment.meta.part?.begin ?? 1) - 1
const partSize = lastNumber > 1 ? partBegin / (lastNumber - 1) : volumeSize
if (volumeStart < 0 || !Number.isInteger(partSize) || partSize <= 0) return null
// Bytes [from, to) of the whole archive, from the last volume's parts.
const read = async (from: number, to: number): Promise<Buffer | null> => {
if (from < volumeStart) return null
const a = from - volumeStart
const b = to - volumeStart
const pieces: Buffer[] = []
for (let n = Math.floor(a / partSize) + 1; n <= Math.floor((b - 1) / partSize) + 1; n++) {
const s = await segment(n)
const offset = (s.meta.part?.begin ?? 1) - 1
if (offset !== (n - 1) * partSize) return null
pieces.push(s.data.subarray(Math.max(0, a - offset), Math.max(0, Math.min(s.data.length, b - offset))))
}
const data = Buffer.concat(pieces)
return data.length === b - a ? data : null
}
const header = await read(start.headerStart, start.archiveSize)
if (!header || !checkHeader(start, header)) return null
if (encodedHeaderEncrypted(header)) return { headersEncrypted: true }
const from = Math.min(listingStart(start, header), start.headerStart)
if (start.archiveSize - from > MAX_TAIL_BYTES) return null
const data = await read(from, start.archiveSize)
return data ? { from, data } : null
}