usenet-indexer/tests/functional/release_complete.spec.ts

import { test } from '@japa/runner'
import db from '@adonisjs/lucid/services/db'
import testUtils from '@adonisjs/core/services/test_utils'
import ReleaseCompleteness from '#services/ReleaseCompleteness'

/** A release with `n` files (each with a binary of size 100000 + i), named f<i>.bin. */
async function release(n: number) {
  const now = new Date()
  const [row] = await db
    .table('releases')
    .insert({ release_key: `complete-${n}`.padEnd(40, '0'), name: `Complete.Test.${n}`, source: 'stem', groups: '[]', first_seen_at: now, last_seen_at: now })
    .returning('id')
  const releaseId = Number(row.id)
  await db.rawQuery(
    `WITH f AS (
       INSERT INTO files (filename, poster, date, parts, groups, segments, release_id, file_key, created_at, updated_at)
       SELECT 'f' || i || '.bin', 'x', 0, 1, '[]', '\\x'::bytea, ?::bigint, 'complete-' || ?::int || '-' || i, now(), now()
       FROM generate_series(1, ?::int) i
       RETURNING id, filename)
     INSERT INTO binaries (file_key, poster, subject, name, total_parts, groups, first_seen_at, last_seen_at, size, file_id, release_id)
     SELECT 'complete-b-' || ?::int || '-' || f.id, 'x', 'x', f.filename, 1, '[]', now(), now(), 100000 + substr(f.filename, 2, length(f.filename) - 5)::int, f.id, ?::bigint
     FROM f`,
    [releaseId, n, n, n, releaseId]
  )
  return releaseId
}

/** The par2 listing of `count` files f1..f<count> (sizes as in release()), in one INSERT, as Par2Service.attach does. */
async function listing(releaseId: number, count: number, offset = 0) {
  await db.rawQuery(
    `INSERT INTO release_files (release_id, par2_file_id, name, size, md5_16k, set_id)
     SELECT ?::bigint, md5(i::text), 'f' || i || '.bin', 100000 + i, md5(i::text), 'set' FROM generate_series(1 + ?::int, ?::int + ?::int) i`,
    [releaseId, offset, count, offset]
  )
  // Triggers only note the release; gc:scheduler's sweep recomputes it.
  await ReleaseCompleteness.sweep()
  return Boolean((await db.from('releases').where('id', releaseId).first()).complete)
}

test.group('release completeness', (group) => {
  group.each.setup(() => testUtils.db().withGlobalTransaction())

  test('file inserts keep counts, size, extensions and search text current, once per statement', async ({ assert }) => {
    const id = await release(3)
    const row = await db.from('releases').where('id', id).first()
    assert.equal(Number(row.file_count), 3)
    assert.deepEqual(row.extensions, ['bin'])
    // No par2 listing, no file total, not complete: its completeness can't change, so it isn't noted.
    const [{ count }] = await db.from('release_dirty').where('release_id', id).count('* as count')
    assert.equal(Number(count), 0)
  })

  test('a release with a par2 listing is noted once per insert, and complete by file names alone', async ({ assert }) => {
    const now = new Date()
    const [row] = await db
      .table('releases')
      .insert({ release_key: 'names-only'.padEnd(40, '0'), name: 'Names.Only', source: 'stem', groups: '[]', first_seen_at: now, last_seen_at: now })
      .returning('id')
    const id = Number(row.id)
    await db.rawQuery(
      `INSERT INTO release_files (release_id, par2_file_id, name, size, md5_16k, set_id)
       SELECT ?::bigint, md5(i::text), 'n' || i || '.rar', 100000 + i, md5(i::text), 'set' FROM generate_series(1, 3) i`,
      [id]
    )
    await ReleaseCompleteness.sweep()
    // Seeded files: no binaries, matched by name only.
    await db.rawQuery(
      `INSERT INTO files (filename, poster, date, parts, groups, segments, release_id, file_key, created_at, updated_at)
       SELECT 'n' || i || '.rar', 'x', 0, 1, '[]', '\\x'::bytea, ?::bigint, 'names-only-' || i, now(), now() FROM generate_series(1, 3) i`,
      [id]
    )
    const [{ count }] = await db.from('release_dirty').where('release_id', id).count('* as count')
    assert.equal(Number(count), 1)
    await ReleaseCompleteness.sweep()
    assert.isTrue(Boolean((await db.from('releases').where('id', id).first()).complete))
  })

  test('complete when every listed file of 64 KiB or more is there, by size or name', async ({ assert }) => {
    const id = await release(3)
    assert.isTrue(await listing(id, 3))
    const other = await release(2)
    // f3 is listed but missing.
    assert.isFalse(await listing(other, 3))
  })

  test('a par2 listing of thousands of files is checked in one go', async ({ assert }) => {
    const id = await release(3000)
    await ReleaseCompleteness.sweep()
    const started = Date.now()
    assert.isTrue(await listing(id, 3000))
    // Was cubic: minutes for a few thousand rows.
    assert.isBelow(Date.now() - started, 5000)
  }).timeout(60_000)
})