usenet-indexer/tests/unit/release/identity.spec.ts

import { test } from '@japa/runner'
import { fileStem, releaseIdentity, subjectRemainder } from '#lib/release/index'
import { looksLikeFilename, weakFileKey } from '#lib/headers/index'

test.group('release / fileStem', () => {
  test('reduces all files of a set to the same stem', ({ assert }) => {
    const names = [
      '1790546808_nicovideo_jp_watch_sm26679715.tar.zst',
      '1790546808_nicovideo_jp_watch_sm26679715.tar.zst.par2',
      '1790546808_nicovideo_jp_watch_sm26679715.tar.zst.vol03+04.par2',
    ]
    assert.deepEqual(new Set(names.map(fileStem)), new Set(['1790546808_nicovideo_jp_watch_sm26679715']))
  })

  test('handles rar, split and par2 volume naming', ({ assert }) => {
    assert.equal(fileStem('lJzuoIKA3T8xjzAD3vqtw0.part036.rar'), 'lJzuoIKA3T8xjzAD3vqtw0')
    assert.equal(fileStem('Some.Movie.2024.r07'), 'Some.Movie.2024')
    assert.equal(fileStem('Some.Movie.2024.mkv.001'), 'Some.Movie.2024')
    assert.equal(fileStem('Some.Movie.2024.2160p.WEB-DL.H.265.mkv'), 'Some.Movie.2024.2160p.WEB-DL.H.265')
    assert.equal(fileStem('Some.Movie.2024.H.264.nzb'), 'Some.Movie.2024.H.264')
    assert.equal(fileStem('Some.Movie.2024.vol127-128.par2'), 'Some.Movie.2024')
    assert.equal(fileStem('Some.Movie.2024.nfo'), 'Some.Movie.2024')
    assert.equal(fileStem('00-artist-album-web-2026.m3u'), '00-artist-album-web-2026')
    assert.equal(fileStem('f6dd6b156692a52b.bin'), 'f6dd6b156692a52b')
  })

  test('never reduces a name to nothing', ({ assert }) => {
    assert.equal(fileStem('.par2'), '.par2')
  })
})

test.group('release / subjectRemainder', () => {
  test('is empty when the subject only has counters, filename and size', ({ assert }) => {
    assert.equal(subjectRemainder('[1/4] - "x.tar.zst" yEnc 15411409'), '')
    assert.equal(subjectRemainder('. - [00/88] - "60SHkTQSvn5RZGExwju2.nzb" yEnc'), '')
  })

  test('keeps a release name written outside the quotes', ({ assert }) => {
    assert.equal(
      subjectRemainder('Some.Show.S01E02.1080p [03/40] - "abc.part02.rar" yEnc'),
      'Some Show S01E02 1080p'
    )
  })
})

test.group('release / releaseIdentity', () => {
  test('groups a par2 set with its archive by stem, per poster', ({ assert }) => {
    const base = { poster: 'p', weak: false }
    const a = releaseIdentity({ ...base, subject: '[1/4] - "x.tar.zst" yEnc 1', name: 'x.tar.zst' })!
    const b = releaseIdentity({ ...base, subject: '[3/4] - "x.tar.zst.vol01+02.par2" yEnc 2', name: 'x.tar.zst.vol01+02.par2' })!
    assert.equal(a.key, b.key)
    assert.equal(a.name, 'x')
    assert.equal(a.fileTotal, 4)
    const other = releaseIdentity({ ...base, poster: 'q', subject: '[1/4] - "x.tar.zst" yEnc 1', name: 'x.tar.zst' })!
    assert.notEqual(a.key, other.key)
  })

  test('a par2 index posted without [i/M] counter joins its release', ({ assert }) => {
    const base = { poster: 'nicovideo_archiver', weak: false }
    const archive = releaseIdentity({ ...base, subject: '[1/6] - "x.tar.zst" yEnc 105054576', name: 'x.tar.zst' })!
    const index = releaseIdentity({ ...base, subject: '"x.tar.zst.par2" yEnc 3100', name: 'x.tar.zst.par2' })!
    assert.equal(index.key, archive.key)
    assert.equal(archive.fileTotal, 6)
    assert.isUndefined(index.fileTotal)
  })

  test('uses the subject text when it names the release', ({ assert }) => {
    const base = { poster: 'p', weak: false }
    const a = releaseIdentity({ ...base, subject: 'My.Release [01/40] - "abc.part01.rar" yEnc', name: 'abc.part01.rar' })!
    const b = releaseIdentity({ ...base, subject: 'My.Release [40/40] - "abc.par2" yEnc', name: 'abc.par2' })!
    assert.equal(a.key, b.key)
    assert.equal(a.name, 'My Release')
  })

  test('obfuscated volumes group by stem regardless of the random poster', ({ assert }) => {
    const a = releaseIdentity({ poster: '3Uzx7TTSFpLzz@ngPost.com', subject: '"lJzu.part033.rar" yEnc', name: 'lJzu.part033.rar', weak: true })!
    const b = releaseIdentity({ poster: 'bGLd1oaXRdrK1@ngPost.com', subject: '"lJzu.part034.rar" yEnc', name: 'lJzu.part034.rar', weak: true })!
    assert.equal(a.key, b.key)
  })

  test('random extensionless names cannot be grouped by name', ({ assert }) => {
    const name = '204b914a977e5f891590ff1099c7a8caf1a0be537bfa5bf7f3a8b1213231'
    assert.isNull(releaseIdentity({ poster: 'x', subject: `"${name}" yEnc`, name, weak: true }))
  })
})

test.group('subject / weakFileKey', () => {
  test('ignores random per-article names, keeps real filenames', ({ assert }) => {
    assert.isFalse(looksLikeFilename('204b914a977e5f891590ff1099c7a8caf1a0be537bfa5bf7f3a8b1213231'))
    assert.isTrue(looksLikeFilename('lJzuoIKA3T8xjzAD3vqtw0.part033.rar'))
    // Same file, different random names per article: same key.
    assert.equal(weakFileKey('aaaa1111', 5000, 7), weakFileKey('bbbb2222', 5000, 7))
    // Equal-sized volumes with real names: different keys.
    assert.notEqual(weakFileKey('x.part01.rar', 5000, 7), weakFileKey('x.part02.rar', 5000, 7))
    assert.notEqual(weakFileKey('aaaa1111', 5000, 7), weakFileKey('aaaa1111', 5001, 7))
  })
})

test.group('release / rarReleaseIdentity', () => {
  test('groups volumes by the shared top-level directory', async ({ assert }) => {
    const { rarReleaseIdentity } = await import('#lib/release/index')
    const first = rarReleaseIdentity([
      { name: 'LASZR_-_DYING_ALIVE-(PRO094)-WEB-2026-DIM/00-cover.jpg', size: 589382 },
      { name: 'LASZR_-_DYING_ALIVE-(PRO094)-WEB-2026-DIM/01-track.flac', size: 30_000_000 },
    ])!
    const later = rarReleaseIdentity([{ name: 'LASZR_-_DYING_ALIVE-(PRO094)-WEB-2026-DIM/03-track.flac', size: 29_000_000 }])!
    assert.equal(first.key, later.key)
    assert.equal(first.name, 'LASZR_-_DYING_ALIVE-(PRO094)-WEB-2026-DIM')
    assert.equal(first.source, 'rar')
  })

  test('without a directory, volumes of one file share its first entry', async ({ assert }) => {
    const { rarReleaseIdentity } = await import('#lib/release/index')
    const a = rarReleaseIdentity([{ name: 'Some.Show.S01E01.1080p.mkv', size: 3_000_000_000 }])!
    const b = rarReleaseIdentity([{ name: 'Some.Show.S01E01.1080p.mkv', size: 3_000_000_000 }])!
    assert.equal(a.key, b.key)
    assert.equal(a.name, 'Some.Show.S01E01.1080p')
    assert.isNull(rarReleaseIdentity([]))
  })
})

test.group('release / namesRelease', () => {
  test('random per-file tags and poster ids are not a release name', async ({ assert }) => {
    const { namesRelease, subjectRemainder, releaseIdentity } = await import('#lib/release/index')
    const subject = 'elys-de81bcabef5faf6a - [Nyh5Q07Y] "JkpNIFg7En50zU1U" yEnc'
    assert.isFalse(namesRelease(subjectRemainder(subject)))
    const a = releaseIdentity({ poster: 'elys', subject, name: 'JkpNIFg7En50zU1U', weak: false })!
    assert.equal(a.source, 'stem')
    assert.isTrue(namesRelease(subjectRemainder('Some.Show.S01E02.1080p [03/40] - "abc.part02.rar" yEnc')))
    assert.isTrue(namesRelease('My Release'))
  })
})

test.group('release / post id', () => {
  test('groups random-named files of one post by poster, post id and file count', async ({ assert }) => {
    const { releaseIdentity, isPostToken } = await import('#lib/release/index')
    const base = { poster: 'jW344UtC8R2bZEegrVLz@x', weak: false }
    const a = releaseIdentity({ ...base, subject: '83110629-n [10/18] - "eiae6fulpolas2adfts6o8643n3e6j6ct8d3219aau0jc33ol9r50tohncgr46evpl0dqoqktfb7k" yEnc', name: 'eiae6fulpolas2adfts6o8643n3e6j6ct8d3219aau0jc33ol9r50tohncgr46evpl0dqoqktfb7k' })!
    const b = releaseIdentity({ ...base, subject: '83110629-n [01/18] - "2ef0vm4kb618nv17kn0qsb0j1ofh7pgqtk5a3v5lkqqf1f7g6ed0" yEnc', name: '2ef0vm4kb618nv17kn0qsb0j1ofh7pgqtk5a3v5lkqqf1f7g6ed0' })!
    assert.equal(a.key, b.key)
    assert.equal(a.source, 'post')
    assert.equal(a.fileTotal, 18)
    const other = releaseIdentity({ ...base, subject: '83110630-n [01/18] - "x" yEnc', name: 'x' })!
    assert.notEqual(a.key, other.key)

    assert.isTrue(isPostToken('83110629 n'))
    assert.isFalse(isPostToken('PRiVATE'))
    assert.isFalse(isPostToken(''))
    // Without a counter there is nothing to bound the post: fall back to the stem.
    assert.equal(releaseIdentity({ ...base, subject: '83110629-n - "x.bin" yEnc', name: 'x.bin' })!.source, 'stem')
  })
})

test.group('release / par2 names', () => {
  test('names a season pack after what its episodes share', async ({ assert }) => {
    const { par2ReleaseName } = await import('#lib/release/index')
    const eps = Array.from({ length: 20 }, (_, i) =>
      `SEAL.Team.S03E${String(i + 1).padStart(2, '0')}.1080p.AMZN.WEB-DL.AAC2.0.H.265.DUAL-nitrato.mkv`)
    assert.equal(par2ReleaseName(eps), 'SEAL.Team.S03.1080p.AMZN.WEB-DL.AAC2.0.H.265.DUAL-nitrato')
    // Movie plus sample and nfo: the movie's (longest) name.
    assert.equal(
      par2ReleaseName(['Some.Movie.2024.1080p.BluRay.x264-GRP.mkv', 'Some.Movie.2024.1080p.BluRay.x264-GRP.nfo', 'sample.mkv']),
      'Some.Movie.2024.1080p.BluRay.x264-GRP'
    )
    // Numbered parts: the shared name without the part number.
    assert.equal(par2ReleaseName(['Album.Name-2024-GRP.part01.rar', 'Album.Name-2024-GRP.part02.rar']), 'Album.Name-2024-GRP')
    assert.equal(par2ReleaseName(['only.one.file.2024.mkv']), 'only.one.file.2024')
  })
})