anonymize

// Live demo — runs the same French paragraph through the anonymizer under
// four configurations, from regex-only to regex + full ML coverage. Prints
// each result with ANSI highlights so the contribution of each layer is
// obvious.
//
// Run: `npm run demo`
//   First call downloads ~280 MB of model weights to the HF cache.
//   Subsequent runs reuse the cache (sub-second model load).

import { anonymize } from "../src/anonymizer.ts";
import { detectWithModel, type PiiModelOptions } from "../src/piiModel.ts";

// ANSI helpers — colorize pseudonyms so the changes between scenarios pop.
const C = {
  reset: "\x1b[0m",
  bold: "\x1b[1m",
  dim: "\x1b[2m",
  green: "\x1b[32m",
  cyan: "\x1b[36m",
  yellow: "\x1b[33m",
  magenta: "\x1b[35m",
  red: "\x1b[31m",
  blue: "\x1b[34m",
};

const COLOR_BY_TYPE: Record<string, string> = {
  EMAIL: C.cyan,
  TEL: C.cyan,
  CB: C.yellow,
  NIR: C.yellow,
  IBAN: C.yellow,
  NOM: C.green,
  CUSTOM: C.magenta,
  ADDRESS: C.blue,
  DATE: C.red,
  IDDOC: C.red,
};

function colorize(text: string): string {
  return text.replace(/\[(\w+)_\d+\]/g, (m, type) => {
    const c = COLOR_BY_TYPE[type] ?? C.bold;
    return `${c}${C.bold}${m}${C.reset}`;
  });
}

// A deliberately diverse French paragraph: it mixes the formats our regex
// already covers (email/phone/IBAN/NIR/CB) with the unstructured PII the
// model is supposed to add (names, addresses, dates, ID document numbers).
const SAMPLE = [
  "Compte-rendu — Marie Dupont, née le 14/03/1985 à Lille.",
  "Adresse : 27 rue de la République, 69002 Lyon.",
  "Contact : marie.dupont@example.fr, +33 6 12 34 56 78.",
  "Passeport n° 19FH84235, NIR 2 89 04 78 342 163 49.",
  "Virement reçu sur FR14 2004 1010 0505 0001 3M02 606 ; carte 4111 1111 1111 1111.",
  "Le dossier a été confirmé par Pierre Martin lors de la réunion du 12/05/2025.",
].join(" ");

type Scenario = {
  name: string;
  description: string;
  modelOptions: PiiModelOptions;
};

const SCENARIOS: Scenario[] = [
  {
    name: "1. Regex only (baseline)",
    description: "Plugin in its default config — no ML model loaded.",
    modelOptions: {},
  },
  {
    name: "2. Regex + names",
    description: "detectPiiWithModel ON, modelDetectNames ON. Pseudonyms for names appear.",
    modelOptions: { detectNames: true },
  },
  {
    name: "3. Regex + names + addresses",
    description: "Adds modelDetectAddresses. Streets, cities, building numbers, zipcodes get [ADDRESS_N].",
    modelOptions: { detectNames: true, detectAddresses: true },
  },
  {
    name: "4. Regex + all ML categories",
    description: "All four ML sub-flags ON. Dates and ID documents now redacted too.",
    modelOptions: {
      detectNames: true,
      detectAddresses: true,
      detectDates: true,
      detectIdDocs: true,
    },
  },
];

function fmtCounts(counts: Record<string, number>): string {
  const entries = Object.entries(counts);
  if (entries.length === 0) return "(nothing redacted)";
  return entries
    .sort(([a], [b]) => a.localeCompare(b))
    .map(([k, v]) => `${k}=${v}`)
    .join("  ");
}

async function main() {
  console.log(`${C.bold}== anonymize plugin — live demo ==${C.reset}`);
  console.log(`${C.dim}Sample text (${SAMPLE.length} chars):${C.reset}`);
  console.log(SAMPLE);
  console.log("");

  for (const scenario of SCENARIOS) {
    console.log(`${C.bold}${scenario.name}${C.reset}`);
    console.log(`${C.dim}${scenario.description}${C.reset}`);

    const t0 = Date.now();
    const modelSpans = await detectWithModel(SAMPLE, scenario.modelOptions);
    const tModel = Date.now() - t0;

    const t1 = Date.now();
    const result = anonymize(SAMPLE, { modelSpans });
    const tRedact = Date.now() - t1;

    console.log(`${C.dim}Timing: model=${tModel} ms, redact=${tRedact} ms${C.reset}`);
    console.log(`${C.dim}Counts: ${fmtCounts(result.counts)}${C.reset}`);
    console.log("");
    console.log(colorize(result.anonymized));
    console.log("");

    if (Object.keys(result.mapping).length > 0) {
      console.log(`${C.dim}Mapping:${C.reset}`);
      for (const [pseudo, original] of Object.entries(result.mapping)) {
        const c = COLOR_BY_TYPE[pseudo.replace(/[\[\]_\d]/g, "")] ?? "";
        console.log(`  ${c}${pseudo}${C.reset} → ${original}`);
      }
    }
    console.log(`${C.dim}${"─".repeat(72)}${C.reset}\n`);
  }

  console.log(`${C.dim}Tip: scenarios 2-4 reuse the model from scenario 2 — note the timing drop.${C.reset}`);
}

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