#!/usr/bin/env node
// One-off: recover the word labels trapped in the free-text (code 99) notes.
//
// Before the triage tool had a word selector, the only way to say what a clip
// actually contained was a note. One 284-clip pass produced "do uh", "to uh",
// "when uhh", "uhh literally", "duuhh", "uh-huh" and three separate clean
// "and"s -- all real human judgement, none of it filterable, because
// arrange-poly.mjs gates on the ASR token and never sees a note.
//
// This turns the unambiguous ones into words.json entries. It is deliberately
// conservative and it PRINTS EVERY SKIP: a wrong label is worse than a missing
// one, because the arranger would then act on it.
//
//   node notes-to-words.mjs           report only
//   node notes-to-words.mjs --write   write words.json
import { readFileSync, writeFileSync, existsSync } from "node:fs";
import path from "node:path";
import { SONG_DATA } from "./paths.mjs";

const DIR = path.resolve(path.dirname(new URL(import.meta.url).pathname));
const WRITE = process.argv.includes("--write");

const load = (f, d) => {
  try {
    return JSON.parse(readFileSync(f, "utf8"));
  } catch {
    return d;
  }
};

const reasons = load(path.join(DIR, "reasons.json"), {});
const salvaged = load(path.join(SONG_DATA, "cand2", "_salvaged.json"), { candidates: [] });

// A note about a WINDOW is not a note about a word. These are exactly the clips
// the served-audio tool can now rescue, so labelling them from a complaint
// about the old boundary would bake in the wrong answer.
const WINDOW_COMPLAINT = /cut ?off|cut uh|cut "uh"/i;

// The quoted phrase is the label. Prose around it is commentary.
const QUOTED = /[""„«]?"([^"]{1,40})"[""»]?/;

// A note with no quotes at all is only usable when it IS the word: one short
// token, no sentence.
const BARE_WORD = /^[A-Za-z'’\-.]{2,20}$/;

const rows = [];
const skips = [];

for (const [key, v] of Object.entries(reasons)) {
  if (v.code !== 99) continue;
  const text = (v.text ?? "").trim();
  if (!text) {
    skips.push([key, text, "no text"]);
    continue;
  }
  if (WINDOW_COMPLAINT.test(text)) {
    skips.push([key, text, "complains about the window, not the word"]);
    continue;
  }

  const quotes = text.match(/"/g)?.length ?? 0;
  let word = null;
  let how = "";

  if (quotes === 2) {
    const m = text.match(QUOTED);
    if (m) {
      word = m[1].trim();
      how = "quoted phrase";
    }
  } else if (quotes === 0 && BARE_WORD.test(text)) {
    word = text;
    how = "bare word";
  } else if (quotes > 2) {
    skips.push([key, text, `${quotes / 2} quoted phrases -- ambiguous`]);
    continue;
  }

  if (!word) {
    skips.push([key, text, "no unambiguous word in it"]);
    continue;
  }
  // "clean", "Isolated", "long" and friends describe the clip, not the word.
  // "literal" is deliberately NOT in this list: "Literal uh" means the word was
  // spoken AS a word rather than as filler, which is exactly the distinction the
  // arranger needs, and stripping it would flatten the label to plain "uh".
  word = word.replace(/^(clean|isolated|quick|long)\s+/i, "").trim();
  if (!word) {
    skips.push([key, text, "nothing left after dropping the qualifier"]);
    continue;
  }
  if (word.split(/\s+/).length > 4) {
    skips.push([key, text, "longer than four words"]);
    continue;
  }
  rows.push({ key, word, how, text });
}

// A re-trimmed clip lives in the palette under a NEW key, so the label has to be
// written there too or the arranger never finds it.
const alias = new Map();
for (const c of salvaged.candidates ?? []) {
  const from = c.salvagedFrom ?? c.sortedFrom;
  if (!from) continue;
  alias.set(from, `${c.video}@${(+c.start).toFixed(2)}`);
}

const words = load(path.join(DIR, "words.json"), {});
let written = 0;
for (const r of rows) {
  const entry = { word: r.word, from: "note" };
  words[r.key] = entry;
  written += 1;
  const nk = alias.get(r.key);
  if (nk && nk !== r.key) {
    words[nk] = entry;
    written += 1;
  }
}

console.log(`converted ${rows.length} notes into ${written} label entries`);
for (const r of rows) console.log(`  ${r.key.padEnd(26)} ${JSON.stringify(r.word).padEnd(22)} <- ${r.text}`);
console.log(`\nskipped ${skips.length}:`);
for (const [key, text, why] of skips) console.log(`  ${key.padEnd(26)} ${why.padEnd(42)} ${JSON.stringify(text)}`);

if (WRITE) {
  writeFileSync(path.join(DIR, "words.json"), JSON.stringify(words, null, 1));
  console.log(`\nwrote words.json (${Object.keys(words).length} labels)`);
} else {
  console.log(`\n(report only -- pass --write to save)`);
}
if (!existsSync(path.join(SONG_DATA, "cand2"))) {
  console.log("note: SONG_DATA/cand2 not found, so re-trimmed aliases were not resolved");
}