import Database from "bun:sqlite";
import {
  analyzeTHWord,
  deconstructSyllable,
  segmentateThai,
  type SorSyl,
  type ThaiNLPRes,
  sorSyl,
  getThaiFreq,
  SorBSyl,
} from "../calls/nlp";
import pdb from "./prosodydb";
import { cleanIpa } from "../utils";
import { handleFile } from "./utils";
import { Tone } from "../types/phonetics";
import { AsyncRes } from "../types";

const errors: string[] = [];
async function readDump(lang: string) {
  await pdb.init();
  pdb.addLanguage("th", "thai");
  let count = 0;
  const langdb = new Database(
    `/home/y/code/prosody/resources/wiktionary/${lang}.db`,
  );
  let langrows: any = langdb.query("SELECT data FROM langs");
  // langrows = langrows.slice(10);
  const freqMap = await getFrequency();
  for (const langrow of langrows) {
    count++;
    console.log(count);
    // if (count <= 10000) continue;
    if (count > 300) break;
    const j = JSON.parse(langrow.data);
    const word = j.word.trim();
    if (!word) continue;
    const split = word.split(" ");
    const res =
      split.length > 1
        ? await handleIdiom(lang, word)
        : await handleWord(lang, word, j, freqMap);
    if ("error" in res) {
      console.error(res.error);
      break;
    }
  }
  console.dir(errors);
}

async function handleWord(
  lang: string,
  word: string,
  j: any,
  freqMap: Map<string, number>,
): AsyncRes<string> {
  // TODO add categories but add a tag to see what classifying scheme we're using
  //
  const sounds = j.sounds || [];
  const hasIpa = sounds.find((s: any) => "ipa" in s);
  const hwikiRhyme = sounds.find((s: any) => "rhymes" in s);
  const wikiRhyme = hwikiRhyme ? hwikiRhyme.rhymes : null;
  if (!hasIpa) {
    // console.error("no ipa!!", word);
    // console.dir(j, { depth: null });
    return { error: "meh no ipa" };
  }
  const freq = freqMap.get(word) || null;
  // const wordId = pdb.addWord(word, lang, freq, null);
  // WIPE
  const wordId = 0;
  // console.log(analyzed);
  for (let snd of sounds)
    if ("ipa" in snd) {
      const res = await handleIpa(wordId, word, lang, j, snd, wikiRhyme);
      if ("error" in res) return res;
    }
  return { ok: "" };
}
async function handleIpa(
  wordId: number | bigint,
  word: string,
  lang: string,
  j: any,
  snd: any,
  wikiRhyme: string | null,
) {
  const tags = JSON.stringify(snd.tags) || null;
  const ipa = snd.ipa;
  const syls = await sorSyl(word, lang, ipa);
  // console.log(syls, "sorsyl");

  console.log(word);
  console.log(ipa);
  pdb.addPronunciation(wordId, ipa, syls.syls.length, tags, null);
  // set word rhyme
  const wordRhyme = syls.syls.reduce((acc: string, itemm: SorBSyl) => {
    const item = itemm.ipa;
    if (!item.stressed && !acc) return acc;
    if (item.stressed && !acc) return `${acc}${item.rhyme}`;
    else return `${acc}${item.all}`;
  }, "");
  if (wordRhyme) pdb.addWordRhyme(wordId, wordRhyme, j.lang_code, wikiRhyme);

  for (let i = 0; i < syls.syls.length; i++) {
    const syl = syls.syls[i]!;
    const res = await handleSyllable(syl, wordId, i);
    if ("error" in res) return res;
  }
  return { ok: "" };
}
async function handleSyllable(
  syl: SorBSyl,
  wordId: number | bigint,
  idx: number,
): AsyncRes<string> {
  try {
    pdb.addSyllable(
      wordId,
      idx + 1,
      syl.ipa.stressed,
      "th",
      syl.ipa.all,
      syl.ipa.long,
      syl.spelling.all,
      { spelling: syl.spelling.onset, ipa: syl.ipa.onset },
      { spelling: syl.spelling.medial, ipa: syl.ipa.medial },
      { spelling: syl.spelling.nucleus, ipa: syl.ipa.nucleus },
      { spelling: syl.spelling.coda, ipa: syl.ipa.coda },
      { spelling: syl.spelling.rhyme, ipa: syl.ipa.rhyme },
      { letters: "", numbers: 0, name: "" },
      null,
    );
    return { ok: "" };
  } catch (e) {
    // console.log("well fuck", syl);
    // console.error(e);
    return { error: `${e}` };
  }
}
async function handleIdiom(lang: string, idiom: string): AsyncRes<string> {
  try {
    pdb.addIdiom(idiom, lang);
    // TODO later set idiom_words once all words are populated
    // console.log();
    return { ok: "" };
  } catch (e) {
    return { error: `${e}` };
  }
}
// ช้า ๆ
// งก ๆ
// หงก ๆ

async function getFrequency() {
  const freqMap = new Map<number, string>();
  await handleFile(
    "/home/y/code/prosody/hanchu/datasets/unigram_freq.csv",
    (line, idx) => {
      const [spelling, frequency] = line.split(",");
      freqMap.set(Number(frequency!), spelling!);
    },
  );
  const orderedMap = new Map<string, number>();
  const keys = Array.from(freqMap.keys()).sort();
  for (let i = 0; i < keys.length; i++) {
    const val = freqMap.get(keys[i]!)!;
    orderedMap.set(val, i + 1);
  }
  return orderedMap;
}

readDump("en");