lila/packages/db/src/generating-deck.ts

import fs from "node:fs/promises";
import { db } from "@glossa/db";
import { translations, terms, decks, deck_terms } from "@glossa/db/schema";
import { inArray, and, eq, ne, countDistinct } from "drizzle-orm";

type DbOrTx = Parameters<Parameters<typeof db.transaction>[0]>[0];

const config = {
  pathToWordlist: "./src/data/wordlists/top1000englishnouns",
  deckName: "top english nouns",
  deckDescription: "Most frequently used English nouns for vocabulary practice",
  sourceLanguage: "en",
  sourcePOS: "noun",
} as const;

const readWordList = async () => {
  const raw = await fs.readFile(config.pathToWordlist, "utf8");
  const words = [
    ...new Set(
      raw
        .split("\n")
        .map((w) => w.trim().toLowerCase())
        .filter(Boolean),
    ),
  ];
  return words;
};

const resolveSourceTerms = async (words: string[]) => {
  const rows = await db
    .select({ text: translations.text, termId: translations.term_id })
    .from(translations)
    .innerJoin(terms, eq(translations.term_id, terms.id))
    .where(
      and(
        inArray(translations.text, words),
        eq(translations.language_code, config.sourceLanguage),
        eq(terms.pos, config.sourcePOS),
      ),
    );

  const wordToTermIds = new Map<string, string[]>();
  for (const row of rows) {
    const word = row.text.toLowerCase();

    if (!wordToTermIds.has(word)) {
      wordToTermIds.set(word, []);
    }
    wordToTermIds.get(word)!.push(row.termId);
  }
  // Deduplicate: multiple words can map to the same term ID (e.g. via synonyms)
  const termIds = [...new Set(Array.from(wordToTermIds.values()).flat())];
  const missingWords = words.filter((w) => !wordToTermIds.has(w));

  return { termIds, missingWords };
};

const writeMissingWordsToFile = async (missingWords: string[]) => {
  const outputPath = `${config.pathToWordlist}-missing`;
  await fs.writeFile(outputPath, missingWords.join("\n"), "utf8");
};

const validateLanguages = async (sourceLanguage: string, termIds: string[]) => {
  const coverage = await db
    .select({
      language: translations.language_code,
      coveredCount: countDistinct(translations.term_id),
    })
    .from(translations)
    .where(
      and(
        inArray(translations.term_id, termIds),
        ne(translations.language_code, sourceLanguage),
      ),
    )
    .groupBy(translations.language_code);

  const validatedLanguages = coverage
    .filter((row) => Number(row.coveredCount) === termIds.length)
    .map((row) => row.language);

  return { coverage, validatedLanguages };
};

const findExistingDeck = async (tx: DbOrTx) => {
  const existing = await tx
    .select({ id: decks.id, validatedForLanguages: decks.validated_languages })
    .from(decks)
    .where(
      and(
        eq(decks.name, config.deckName),
        eq(decks.source_language, config.sourceLanguage),
      ),
    );
  return existing[0] ?? null;
};

const createDeck = async (tx: DbOrTx, validatedLanguages: string[]) => {
  const result = await tx
    .insert(decks)
    .values({
      name: config.deckName,
      description: config.deckDescription,
      source_language: config.sourceLanguage,
      validated_languages: validatedLanguages,
      is_public: false,
    })
    .returning({ id: decks.id });
  const created = result[0];
  if (!created) throw new Error("Failed to create deck: no row returned");
  return created.id;
};

const addTermsToDeck = async (
  tx: DbOrTx,
  deckId: string,
  termIds: string[],
): Promise<number> => {
  if (termIds.length === 0) return 0;

  await tx
    .insert(deck_terms)
    .values(termIds.map((termId) => ({ deck_id: deckId, term_id: termId })))
    .onConflictDoNothing();

  return termIds.length;
};

const updateValidatedLanguages = async (
  tx: DbOrTx,
  deckId: string,
  validatedLanguages: string[],
): Promise<void> => {
  await tx
    .update(decks)
    .set({ validated_languages: validatedLanguages })
    .where(eq(decks.id, deckId));
};

const main = async () => {
  console.log("📖 Reading word list...");
  const sourceWords = await readWordList();
  console.log(`   ${sourceWords.length} words loaded\n`);

  console.log("🔍 Checking against database...");
  const { termIds, missingWords } = await resolveSourceTerms(sourceWords);
  console.log(`   ${termIds.length} terms found`);
  console.log(`   ${missingWords.length} words not found in DB\n`);

  console.log("🖊️ Writing missing words to file...\n");
  await writeMissingWordsToFile(missingWords);

  console.log("✅ Validating languages...");
  const { coverage, validatedLanguages } = await validateLanguages(
    config.sourceLanguage,
    termIds,
  );
  console.log(
    `   Validated languages: ${JSON.stringify(validatedLanguages)}\n`,
  );

  console.log("🔬 Language coverage breakdown...");
  for (const row of coverage) {
    console.log(
      `  ${row.language}: ${row.coveredCount} / ${termIds.length} terms covered`,
    );
  }

  console.log("🃏 Looking for existing deck...");
  const addedCount = await db.transaction(async (tx) => {
    const existingDeck = await findExistingDeck(tx);
    const deckId = existingDeck
      ? existingDeck.id
      : await createDeck(tx, validatedLanguages);

    const addedCount = await addTermsToDeck(tx, deckId, termIds);

    const currentLanguages = existingDeck?.validatedForLanguages ?? [];
    const hasChanged =
      JSON.stringify([...currentLanguages].sort()) !==
      JSON.stringify([...validatedLanguages].sort());

    if (hasChanged) {
      await updateValidatedLanguages(tx, deckId, validatedLanguages);
    }

    return addedCount;
  });
  const alreadyPresentCount = termIds.length - addedCount;

  console.log("━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━");
  console.log("📊 Summary");
  console.log("━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━");
  console.log(`   Words loaded from wordlist : ${sourceWords.length}`);
  console.log(
    `   Words matched in DB        : ${sourceWords.length - missingWords.length}`,
  );
  console.log(`   Words not found in DB      : ${missingWords.length}`);
  console.log(`   Term IDs resolved          : ${termIds.length}`);
  console.log(`   Terms added to deck        : ${addedCount}`);
  console.log(`   Terms already in deck      : ${alreadyPresentCount}`);
  console.log(
    `   Validated languages        : ${validatedLanguages.length > 0 ? validatedLanguages.join(", ") : "none"}`,
  );
  console.log("━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━");
};

main().catch((error) => {
  console.error(error);
  process.exit(1);
});