// What an answer's citations owe: every sentence that states something cites the passage it rests on,
// and every ref it cites is one a search gave. Pure, so the reading of an answer is tested.
// Headings, table rows, code, and short fragments state nothing to cite; a sentence saying the passages
// do not hold something is what the prompt asks for when they do not, and cites nothing.

const CITE = /\[(p\d+)\]/g
const SAYS_NOT_FOUND =
  /\b(no passage|not (?:find|found)|couldn['’]?t find|could not find|did not find|didn['’]?t find|do(?:es)? not (?:say|hold|mention|contain)|not in the (?:passages|filings|store))\b/i
const CLAIM_WORDS = 6

/** The sentences of an answer that state something and cite nothing. */
export function uncited(answer: string): string[] {
  const out: string[] = []
  let fenced = false
  for (const raw of answer.split('\n')) {
    const line = raw.trim()
    if (line.startsWith('```')) {
      fenced = !fenced
      continue
    }
    if (fenced || !line || line.startsWith('#') || line.startsWith('|') || line.startsWith('>')) continue
    const body = line.replace(/^(?:[-*+]|\d+[.)])\s+/, '')
    for (const sentence of body.split(/(?<=[.!?])\s+(?=["“(*\p{Lu}\d])/u)) {
      const plain = sentence.replace(/\*\*?|__?|`/g, '').trim()
      if (CITE.test(sentence) || SAYS_NOT_FOUND.test(plain)) {
        CITE.lastIndex = 0
        continue
      }
      CITE.lastIndex = 0
      const words = plain.split(/\s+/).filter(Boolean).length
      if (words >= CLAIM_WORDS || (/\d/.test(plain) && words >= 3)) out.push(plain)
    }
  }
  return out
}

/** The refs an answer cites that no search gave. */
export function unknownRefs(answer: string, known: Set<string>): string[] {
  const out = new Set<string>()
  for (const match of answer.matchAll(CITE)) if (!known.has(match[1]!)) out.add(match[1]!)
  return [...out]
}

/** What the model is told when its answer owes citations. */
export function repairText(missing: string[], unknown: string[], refs: string[]): string {
  const lines = [
    'Every sentence that states a fact must cite the passage it rests on, like [p2]. This answer does not yet:',
    ...missing.map((s) => `- Not cited: "${s}"`),
    ...unknown.map((r) => `- [${r}] is not a passage you were given.`),
    `Write the whole answer again. Cite each of these from the passages you have (${refs.join(', ') || 'none yet'}), search for what they rest on, or leave them out.`,
    'Reply with the answer alone, as the user will read it: say nothing about this rewrite, what you changed, or what you left out.',
  ]
  return lines.join('\n')
}

/**
 * An answer as it is copied: its words, then the passages it cites in the order it first cites them,
 * each with where it is from and its address, so a ref still says something once pasted elsewhere.
 */
export function withSources(answer: string, passages: { ref: string; label: string; url: string | null }[]): string {
  const byRef = new Map(passages.map((p) => [p.ref, p]))
  const cited = [...new Set([...answer.matchAll(CITE)].map((m) => m[1]!))].flatMap((ref) => byRef.get(ref) ?? [])
  if (cited.length === 0) return answer.trim()
  const lines = cited.map((p) => `[${p.ref}] ${p.label}${p.url ? ` ${p.url}` : ''}`)
  return `${answer.trim()}\n\nSources\n${lines.join('\n')}`
}