usenet-indexer/lib/headers/subject.ts

export interface ParsedSubject {
  /** Filename from the quoted part of the subject. */
  name: string
  /** Segment number `k` from `(k/N)`. */
  segment: number
  /** Segment count `N` from `(k/N)`. */
  totalSegments: number
  /** File counter `[i/M]`, when present. */
  fileIndex?: number
  fileTotal?: number
  /**
   * The subject with the segment counter removed. Identical for every segment
   * of one uploaded file, so it is part of the file identity (see fileKey).
   */
  base: string
}

const SEGMENT_COUNTER = /\((\d+)\/(\d+)\)/g
/** A file name's extension: a dot and 2 to 4 letters or digits at the end. */
const EXTENSION = /[^.\s]\.[A-Za-z0-9]{2,4}$/

/**
 * The file name of a subject without a quoted one, from the text before its
 * segment counter (LOG.md 2026-10-11 (5): most "obfuscated" subjects of older
 * groups): with one quote only, the name before it (after a "[", else after
 * the last space: `[keira_s04_0037_l.jpg" yEnc] (2/2)`); without quotes, the
 * text when it ends in a file name (after a "] " or " - ": `DVDFab
 * 8.2.0.5.rar (006/332)`, `[33/117] 674a37548b09.dat (1/1)`), else its
 * first word that is one (`Emily_18_-_Bed.part01.rar Emily 18 bed video
 * (01/24)`). Null when no file name is found: the subject stays obfuscated.
 * Words are split on spaces only (both implementations alike).
 */
/**
 * The file name of a subject whose quoted text has no extension ("My
 * \"Official\" remaining set's of Inna … 025.jpg", "\"Dr. Lust\" - … - Ed_DRL_VDR_LM.part09.rar"):
 * a name with one elsewhere, else null (random quoted names keep theirs).
 */
export function nameBesideQuotes(quoted: string, beforeCounter: string): string | null {
  if (EXTENSION.test(quoted)) return null
  // A stray quote took the last pair ("(01/26) "01 Time Machine.mp3" "Silicon Dream_BY_WaRLoRD yEnc",
  // "… "x.nfo" " yEnc", "… "x.nzb" yEnc""): the last quoted part that is a file name.
  const named = [...beforeCounter.matchAll(/"([^"]*)"/g)].map((m) => m[1]).filter((q) => EXTENSION.test(q))
  if (named.length > 0) return named.at(-1)!
  const name = unquotedName(beforeCounter.replace(/"[^"]*"/g, ' '))
  return name && EXTENSION.test(name) ? name : null
}

/** A file name that is a date alone ("15_07_2015.rar"). */
const DATE_ONLY = /^(?:\d{1,2}[-_.]\d{1,2}[-_.](?:19|20)\d{2}|(?:19|20)\d{2}[-_.]\d{1,2}[-_.]\d{1,2})\.[A-Za-z0-9]{2,4}$/
const IMAGE_NAME = /\.(jpe?g|png|gif|bmp|webp|tiff?)$/i
/** A host ("www.libranews.nl", "www.x2l.nl"): its ".nl" is no extension. */
const HOST = /www\./i

export function unquotedName(beforeCounter: string): string | null {
  let text = beforeCounter.trimEnd()
  if (/yenc$/i.test(text)) text = text.slice(0, -4).trimEnd()
  const quote = text.indexOf('"')
  if (quote !== -1) {
    if (text.indexOf('"', quote + 1) !== -1) return null
    const before = text.slice(0, quote)
    const bracket = before.lastIndexOf('[')
    if (bracket !== -1) {
      // A file counter that lost its closing bracket ("[03/21 Name.part01.rar") isn't part of the name.
      const inBracket = before.slice(bracket + 1).trim().replace(/^\d+\/\d+ +/, '')
      if (EXTENSION.test(inBracket)) return inBracket
    }
    const word = before.slice(before.lastIndexOf(' ') + 1)
    return EXTENSION.test(word) ? word : null
  }
  // A single-quoted name at the end ("xlady post You, Me And Dupree www.x2l.nl [001/106] 'You, Me And Dupree.nfo'").
  const single = /'([^']*[^.\s']\.[A-Za-z0-9]{2,4})'$/.exec(text)
  if (single) return single[1]
  if (EXTENSION.test(text)) {
    const bracket = text.lastIndexOf('] ')
    // A track or part number ("Album - CD2 - 07 - Artist - Title.mp3"): the name starts there, so the rest
    // is the album, the same for every track (LOG.md 2026-10-11 (10)); else after the last " - ".
    const track = / - \d{1,3} - /.exec(text)
    const dash = track ? track.index : text.lastIndexOf(' - ')
    // Or after an arrow ("… -> [0001 of 0076] -> Exclusive-tgirls_000000-lg.jpg").
    const arrow = text.lastIndexOf('-> ')
    const cut = Math.max(bracket === -1 ? 0 : bracket + 2, dash === -1 ? 0 : dash + 3, arrow === -1 ? 0 : arrow + 3)
    // A date alone after the cut ("Ballers.2014.S01E01.HDTV.x264-KILLERS[ettv] 15_07_2015.rar", "Computer Bild - Ausgabe
    // 152015 - 15_07_2015.rar": a dump's every item was "15_07_2015.rar"; LOG.md 2026-10-13 (42)): the whole text is the name.
    const name = DATE_ONLY.test(text.slice(cut).trim()) ? text.trim() : text.slice(cut).trim()
    // An image after numbered text without a separator: its last word ("<abm> … AB: part2 0573-1002 043.jpg"; the whole
    // subject was the name, so each image a release). Not an index alone ("Teen Tease Self Shooter sad1 (10).jpg").
    if (cut === 0 && IMAGE_NAME.test(name) && name.includes(' ')) {
      const space = name.lastIndexOf(' ')
      const last = name.slice(space + 1)
      if (/\d$/.test(name.slice(0, space).trimEnd()) && !last.startsWith('(')) return last
    }
    // Ending in a host is no file name: the one in brackets or a word follows.
    if (!HOST.test(name)) return name && EXTENSION.test(name) ? name : null
  }
  // A name in brackets ("[01/77][LIBRATEAM_M2219009.par2] … www.libranews.nl", "-^-[Mnozil Brass - Seven.nzb]-^-").
  const bracketed = [...text.matchAll(/\[([^[\]]*[^.\s[\]]\.[A-Za-z0-9]{2,4})\]/g)].at(-1)
  if (bracketed) return bracketed[1]
  for (const word of text.split(' ')) if (EXTENSION.test(word) && !HOST.test(word)) return word
  return null
}
const FILE_COUNTER = /\[(\d+)\/(\d+)\]/

/**
 * Parses a binary post subject such as
 * `[07/45] - "Some.Release.part06.rar" yEnc (12/137) 51315352`.
 *
 * Returns null for subjects without a quoted filename and segment counter
 * (obfuscated posts), which need their yEnc header read from the body instead.
 */
export function parseSubject(subject: string): ParsedSubject | null {
  const nameEnd = subject.lastIndexOf('"')
  const nameStart = nameEnd > 0 ? subject.lastIndexOf('"', nameEnd - 1) : -1

  // The segment counter is the last "(k/N)" after the filename (anywhere without a quote); a file size may follow it.
  let counter: RegExpExecArray | undefined
  for (const match of subject.matchAll(SEGMENT_COUNTER)) {
    if (match.index > nameEnd) counter = match
  }
  if (!counter) return null
  const quoted = nameStart !== -1 ? subject.slice(nameStart + 1, nameEnd) : null
  const name = quoted !== null ? (nameBesideQuotes(quoted, subject.slice(0, counter.index)) ?? quoted) : unquotedName(subject.slice(0, counter.index))
  if (!name) return null

  const segment = Number.parseInt(counter[1], 10)
  const totalSegments = Number.parseInt(counter[2], 10)
  if (totalSegments < 1 || segment < 1 || segment > totalSegments) return null

  const base = (subject.slice(0, counter.index) + subject.slice(counter.index + counter[0].length))
    .replace(/\s+/g, ' ')
    .trim()

  const fileCounter = FILE_COUNTER.exec(subject.slice(0, nameStart !== -1 ? nameStart : counter.index))
  return {
    name,
    segment,
    totalSegments,
    fileIndex: fileCounter ? Number.parseInt(fileCounter[1], 10) : undefined,
    fileTotal: fileCounter ? Number.parseInt(fileCounter[2], 10) : undefined,
    base,
  }
}