berichtsheftgenerator/web/scripts/verify-suggestions.mjs

/**
 * Qualitätstor für Suggestions (S06).
 * Failt bei Müll, ungültigem type/weight, Duplikaten oder zu wenig Originalen.
 *
 * Usage: npm run check:suggestions
 */
import { readFileSync } from "fs";
import { dirname, join } from "path";
import { fileURLToPath } from "url";

const __dirname = dirname(fileURLToPath(import.meta.url));
const ROOT = join(__dirname, "..");
const SUGGESTIONS_PATH = join(ROOT, "src", "data", "suggestions.ts");
const CLEAN_PATH = join(__dirname, "clean-phrases.json");
const MIN_ORIGINAL_SHARE = 0.7;
const VALID_TYPES = new Set(["betrieb", "schule"]);

/** Müll: Header/Signatur/Nummern-Artefakte – nicht echte Sätze mit „zur Unterschrift …“. */
const JUNK_RE =
  /Datum und Unterschrift|\bLfd\.?\s*Nr\b|……|…………|\.{4,}|\d{5,}|Ausbildungsnachweis|Für die Zeit|^Unterschrift\b|^Abteilung\b|Zuordnung zum Lernziel/i;

/**
 * @param {string} src
 * @returns {{ suggestions: any[], themes: string[] }}
 */
function parseSuggestionsTs(src) {
  const sugKey = "export const SUGGESTIONS: Suggestion[] = ";
  const themeKey = "export const WEEK_THEMES: string[] = ";
  const sugStart = src.indexOf(sugKey);
  if (sugStart < 0) throw new Error("SUGGESTIONS-Export nicht gefunden");
  const afterSug = sugStart + sugKey.length;
  const themeStart = src.indexOf(themeKey, afterSug);
  if (themeStart < 0) throw new Error("WEEK_THEMES-Export nicht gefunden");
  const sugJson = src.slice(afterSug, themeStart).trim().replace(/;$/, "");
  const afterTheme = themeStart + themeKey.length;
  const nextExport = src.indexOf("\nexport const ", afterTheme);
  if (nextExport < 0) throw new Error("Ende von WEEK_THEMES nicht gefunden");
  const themeJson = src.slice(afterTheme, nextExport).trim().replace(/;$/, "");
  return {
    suggestions: JSON.parse(sugJson),
    themes: JSON.parse(themeJson),
  };
}

function fail(msg) {
  console.error(`FAIL: ${msg}`);
}

function main() {
  const errors = [];
  const { suggestions, themes } = parseSuggestionsTs(
    readFileSync(SUGGESTIONS_PATH, "utf8"),
  );
  const clean = JSON.parse(readFileSync(CLEAN_PATH, "utf8"));
  const originalTexts = new Set(
    (clean.activities ?? []).map((a) =>
      (typeof a === "string" ? a : a.text).toLowerCase().replace(/\s+/g, " ").trim(),
    ),
  );
  const themeText = (t) => (typeof t === "string" ? t : t?.text);
  const realThemes = new Set(
    [...(clean.themes ?? []), ...(clean.themesWeighted ?? [])]
      .map(themeText)
      .map((t) =>
        String(t ?? "")
          .replace(/&gt;/g, ">")
          .replace(/,\s*$/, "")
          .trim()
          .toLowerCase(),
      )
      .filter(Boolean),
  );

  if (!Array.isArray(suggestions) || suggestions.length === 0) {
    errors.push("SUGGESTIONS leer oder ungültig");
  }

  const seen = new Map();
  let junkCount = 0;
  let badType = 0;
  let badWeight = 0;
  let dupCount = 0;

  for (const s of suggestions) {
    const text = String(s?.text ?? "").replace(/\s+/g, " ").trim();
    if (JUNK_RE.test(text)) {
      junkCount++;
      if (junkCount <= 8) fail(`Müll-Regex: id=${s.id} "${text}"`);
    }
    if (!VALID_TYPES.has(s?.type)) {
      badType++;
      if (badType <= 8) fail(`type fehlt/ungültig: id=${s?.id} type=${s?.type}`);
    }
    if (!(Number(s?.weight) >= 1)) {
      badWeight++;
      if (badWeight <= 8) fail(`weight < 1: id=${s?.id} weight=${s?.weight}`);
    }
    const key = text.toLowerCase();
    if (seen.has(key)) {
      dupCount++;
      if (dupCount <= 8) fail(`Duplikat: "${text}" (ids ${seen.get(key)}, ${s.id})`);
    } else {
      seen.set(key, s.id);
    }
  }

  if (junkCount) errors.push(`${junkCount} Phrase(n) matchen Müll-Regex`);
  if (badType) errors.push(`${badType} Phrase(n) ohne gültigen type`);
  if (badWeight) errors.push(`${badWeight} Phrase(n) mit weight < 1`);
  if (dupCount) errors.push(`${dupCount} case-insensitive Duplikat(e)`);

  const betrieb = suggestions.filter((s) => s.type === "betrieb");
  const schule = suggestions.filter((s) => s.type === "schule");
  const exactBetrieb = betrieb.filter((s) =>
    originalTexts.has(String(s.text).toLowerCase().replace(/\s+/g, " ").trim()),
  );
  const originalShare = betrieb.length ? exactBetrieb.length / betrieb.length : 0;
  if (originalShare < MIN_ORIGINAL_SHARE) {
    errors.push(
      `Original-Anteil Betrieb ${(originalShare * 100).toFixed(1)}% < ${MIN_ORIGINAL_SHARE * 100}% (${exactBetrieb.length}/${betrieb.length})`,
    );
  }

  const fakeThemes = themes.filter((t) => !realThemes.has(String(t).trim().toLowerCase()));
  if (fakeThemes.length) {
    errors.push(
      `${fakeThemes.length} WEEK_THEMES nicht in clean-phrases: ${fakeThemes.slice(0, 5).join(" · ")}`,
    );
  }

  const top10 = [...betrieb]
    .sort((a, b) => b.weight - a.weight || a.text.localeCompare(b.text, "de"))
    .slice(0, 10);

  console.log("=== Suggestions Qualitätstor ===");
  console.log(`Gesamt: ${suggestions.length} (Betrieb ${betrieb.length}, Schule ${schule.length})`);
  console.log(
    `Original-Anteil Betrieb: ${(originalShare * 100).toFixed(1)}% (${exactBetrieb.length}/${betrieb.length})`,
  );
  console.log(`WEEK_THEMES: ${themes.length} (echte Themen)`);
  console.log("Top-10 Betrieb:");
  for (const s of top10) console.log(`  ${s.weight}× ${s.text}`);

  if (errors.length) {
    console.error("\nQualitätstor FEHLGESCHLAGEN:");
    for (const e of errors) console.error(` - ${e}`);
    process.exit(1);
  }

  console.log("\nQualitätstor OK");
}

main();