doctolib-finder — find a doctor + earliest slots on Doctolib

Goal: given a specialty, a location and a conventionnement preference, return a

ranked list of practitioners with their sector and next real availability, so the

user can book the soonest secteur-1 (or any) appointment.

Doctolib blocks headless bots and shows a DataDome captcha. The trick that works: drive a

visible Google Chrome and let the user solve the captcha by hand in the window. A

persistent profile keeps that trust across runs.

This skill is a single file. The entire program lives in the code block in §2 below.

There is no separate .js file and no npm install. To run it, write that block to a temp

file and execute it with Node.

0. Preflight

The script drives the installed Google Chrome over the DevTools Protocol (CDP) using only

Node built-ins (child_process + global fetch + global WebSocket).

node -v                         # need Node 22+ (global WebSocket)
ls "/Applications/Google Chrome.app" >/dev/null && echo "chrome ok"

1. Materialize the program, then run it

Write the §2 code block to a temp file and launch it with the user's flags:

mkdir -p /tmp/doctolib-finder
# (write the §2 block to /tmp/doctolib-finder/find-doctors.js — see below)
node /tmp/doctolib-finder/find-doctors.js \
  --specialty dermatologue \
  --where paris-75011 --where paris-75020 \
  --pages 3 --sector 1 --new-patients

The cleanest way to materialize it is a heredoc that copies the block verbatim:

cat > /tmp/doctolib-finder/find-doctors.js <<'DOCTOLIB_EOF' … DOCTOLIB_EOF.

Flags:

• --specialty Doctolib URL slug (dermatologue, medecin-generaliste, ophtalmologue, dentiste, gynecologue-medical, …). Default dermatologue.
• --where Doctolib location slug used in the URL: paris-75011, paris-75020, paris-20e-arrondissement, lyon, or a metro paris-metro-voltaire. Repeatable to sweep several areas.
• --pages Result pages per location (20 cards/page). Default 3.
• --sector 1, 2, or any. Default any.
• --new-patients Drop practitioners whose online booking is reserved to existing patients.
• --headless No window. Only works once a captcha-trusted profile exists, and availability is less reliable (slot calendar often doesn't load → "dispo non lisible"). Use visible mode for accurate slots.

A visible Chrome opens on the first search page. The user accepts cookies + solves the

captcha once; the script waits up to 3 min for results, then scrapes silently across all

pages/locations. Output: a ranked table in the terminal + full JSON at

/tmp/doctolib-finder/results.json.

2. The program (self-contained, zero deps)

#!/usr/bin/env node
/**
 * doctolib-finder — find a doctor on Doctolib and rank by sector + soonest availability.
 * Self-contained: drives the installed Google Chrome via the DevTools Protocol (CDP) using
 * only Node built-ins (child_process + global fetch + global WebSocket). Requires Node 22+.
 */
const fs = require('fs');
const os = require('os');
const path = require('path');
const { spawn, execSync } = require('child_process');

const CHROME = '/Applications/Google Chrome.app/Contents/MacOS/Google Chrome';
const PORT = 9333;
const PROFILE = path.join(os.homedir(), '.cache', 'doctolib-finder', 'profile');
const OUTDIR = '/tmp/doctolib-finder';

function parseArgs(argv) {
  const a = { specialty: 'dermatologue', where: [], pages: 3, sector: 'any', newPatients: false, headless: false };
  for (let i = 2; i < argv.length; i++) {
    const k = argv[i];
    if (k === '--specialty') a.specialty = argv[++i];
    else if (k === '--where') a.where.push(argv[++i]);
    else if (k === '--pages') a.pages = parseInt(argv[++i], 10) || 3;
    else if (k === '--sector') a.sector = String(argv[++i]);
    else if (k === '--new-patients') a.newPatients = true;
    else if (k === '--headless') a.headless = true;
  }
  if (a.where.length === 0) a.where = ['paris-75011'];
  return a;
}
const ARGS = parseArgs(process.argv);
const sleep = (ms) => new Promise((r) => setTimeout(r, ms));

// ---------- minimal CDP client (built-ins only) ----------
class CDP {
  constructor(wsUrl) {
    this.ws = new WebSocket(wsUrl);
    this.id = 0;
    this.pending = new Map();
    this.ready = new Promise((res, rej) => {
      this.ws.addEventListener('open', () => res());
      this.ws.addEventListener('error', (e) => rej(new Error('ws error: ' + (e.message || 'unknown'))));
    });
    this.ws.addEventListener('message', (ev) => {
      let msg; try { msg = JSON.parse(ev.data); } catch { return; }
      if (msg.id && this.pending.has(msg.id)) {
        const { resolve, reject } = this.pending.get(msg.id);
        this.pending.delete(msg.id);
        if (msg.error) reject(new Error(msg.error.message)); else resolve(msg.result);
      }
    });
  }
  send(method, params = {}) {
    const id = ++this.id;
    return new Promise((resolve, reject) => {
      this.pending.set(id, { resolve, reject });
      this.ws.send(JSON.stringify({ id, method, params }));
      setTimeout(() => { if (this.pending.has(id)) { this.pending.delete(id); reject(new Error('CDP timeout: ' + method)); } }, 60000);
    });
  }
  async navigate(url) { await this.send('Page.navigate', { url }); }
  async eval(expression) {
    const r = await this.send('Runtime.evaluate', { expression, returnByValue: true, awaitPromise: true });
    return r && r.result ? r.result.value : undefined;
  }
  close() { try { this.ws.close(); } catch {} }
}

async function getJSON(url) { const res = await fetch(url); return res.json(); }

async function launchChrome() {
  fs.mkdirSync(PROFILE, { recursive: true });
  try { execSync(`pkill -f "doctolib-finder/profile" 2>/dev/null`); } catch {}
  for (const f of ['SingletonLock', 'SingletonCookie', 'SingletonSocket']) {
    try { fs.rmSync(path.join(PROFILE, f), { force: true }); } catch {}
  }
  await sleep(500);
  const args = [
    `--remote-debugging-port=${PORT}`,
    `--user-data-dir=${PROFILE}`,
    '--no-first-run', '--no-default-browser-check', '--disable-sync',
    '--disable-features=Translate,DialMediaRouteProvider',
    'about:blank',
  ];
  if (ARGS.headless) args.unshift('--headless=new');
  const child = spawn(CHROME, args, { stdio: 'ignore', detached: false });
  let wsUrl = null;
  for (let i = 0; i < 60; i++) {
    try {
      const list = await getJSON(`http://127.0.0.1:${PORT}/json`);
      const pageT = list.find((t) => t.type === 'page' && t.webSocketDebuggerUrl);
      if (pageT) { wsUrl = pageT.webSocketDebuggerUrl; break; }
    } catch {}
    await sleep(500);
  }
  if (!wsUrl) { try { child.kill(); } catch {} throw new Error('Chrome DevTools endpoint introuvable sur le port ' + PORT); }
  return { child, wsUrl };
}

// ---------- parsing ----------
function sectorOf(txt) {
  if (/Conventionné secteur 1 avec droit permanent à dépassement/i.test(txt)) return 'secteur 1 (DP)';
  if (/Conventionné secteur 1/i.test(txt)) return 'secteur 1';
  if (/Conventionné secteur 2/i.test(txt)) return 'secteur 2';
  if (/Non conventionné/i.test(txt)) return 'non conventionné';
  if (/Hôpital public|centre de santé|Institut /i.test(txt)) return 'secteur 1';
  return '?';
}
function availabilityOf(txt) {
  const times = txt.match(/\b([01]?\d|2[0-3]):[0-5]\d\b/g);
  if (times && times.length) return { kind: 'slots', label: 'CRÉNEAUX: ' + Array.from(new Set(times)).slice(0, 6).join(', '), rank: 0 };
  const prochain = txt.match(/Prochain RDV le\s+([^\n]+)/i);
  if (prochain) return { kind: 'next', label: 'Prochain RDV le ' + prochain[1].trim(), rank: 1 };
  if (/réserve la prise de rendez-vous en ligne aux patients déjà suivis/i.test(txt))
    return { kind: 'existing-only', label: 'en ligne réservé aux patients suivis → appeler', rank: 3 };
  if (/Aucune disponibilité en ligne/i.test(txt)) return { kind: 'none', label: 'aucune dispo en ligne → appeler', rank: 4 };
  return { kind: 'unknown', label: 'dispo non lisible', rank: 5 };
}
function parseCard(txt) {
  const lines = txt.split('\n').map((s) => s.trim()).filter(Boolean);
  const name = (lines[0] || '').replace(/,\s*$/, '');
  const distM = txt.match(/(\d+[\.,]\d+)\s*km/);
  const dist = distM ? parseFloat(distM[1].replace(',', '.')) : 999;
  const addr = lines.filter((l) =>
    /\b\d{5}\b/.test(l) || /\b(Rue|Avenue|Av\.|Bd|Boulevard|Place|Pl\.|Passage|Impasse|Quai|Allée|Cours|Chemin)\b/i.test(l)
  ).slice(0, 2).join(', ');
  const video = /Consultation vidéo/i.test(txt);
  const isCenter = /Hôpital public|centre de santé|Institut |dermatologues et vénérologues/i.test(txt);
  return { name, sector: sectorOf(txt), addr, dist, video, isCenter, avail: availabilityOf(txt) };
}

const EXTRACT_JS = `(async () => {
  for (let i = 0; i < 5; i++) { window.scrollBy(0, 1400); await new Promise(r => setTimeout(r, 900)); }
  const cards = Array.from(document.querySelectorAll('.dl-card')).map(e => (e.innerText || '').trim());
  return JSON.stringify(cards);
})()`;

(async () => {
  fs.mkdirSync(OUTDIR, { recursive: true });
  const { child, wsUrl } = await launchChrome();
  const cdp = new CDP(wsUrl);
  await cdp.ready;
  await cdp.send('Page.enable').catch(() => {});

  const all = [];
  let firstLoad = true;
  for (const where of ARGS.where) {
    for (let p = 1; p <= ARGS.pages; p++) {
      const url = `https://www.doctolib.fr/${ARGS.specialty}/${where}${p > 1 ? '?page=' + p : ''}`;
      await cdp.navigate(url);
      await sleep(2500);
      if (firstLoad) {
        let n = 0;
        try { n = await cdp.eval(`document.querySelectorAll('.dl-card').length`); } catch {}
        if (!n) {
          console.log('\n================ ACTION REQUISE ================');
          console.log('Une fenêtre Chrome est ouverte. Accepte les cookies et résous le captcha.');
          console.log('Le scraping démarre dès que les résultats s\'affichent (max 3 min).');
          console.log('================================================\n');
          for (let i = 0; i < 180 && !n; i++) { await sleep(1000); try { n = await cdp.eval(`document.querySelectorAll('.dl-card').length`); } catch {} }
        }
        firstLoad = false;
      }
      let cards = [];
      try { cards = JSON.parse(await cdp.eval(EXTRACT_JS) || '[]'); } catch {}
      let n = 0;
      for (const c of cards) {
        if (!c || c.length < 30) continue;
        if (!/secteur|disponibilité|Hôpital|centre de santé|RDV|vénérologue|généraliste|Dermatolog/i.test(c)) continue;
        const parsed = parseCard(c);
        if (!parsed.name || /^Centre d'aide|^Doctolib$/i.test(parsed.name)) continue;
        all.push({ where, page: p, ...parsed });
        n++;
      }
      console.log(`${ARGS.specialty} / ${where} p${p}: ${n} fiches`);
    }
  }

  const seen = new Set();
  let rows = [];
  for (const r of all) { const k = r.name + '|' + r.addr; if (seen.has(k)) continue; seen.add(k); rows.push(r); }
  if (ARGS.sector === '1') rows = rows.filter((r) => r.sector === 'secteur 1' || r.sector === 'secteur 1 (DP)');
  else if (ARGS.sector === '2') rows = rows.filter((r) => r.sector === 'secteur 2');
  if (ARGS.newPatients) rows = rows.filter((r) => r.avail.kind !== 'existing-only');
  const secScore = (s) => (s === 'secteur 1' ? 0 : s === 'secteur 1 (DP)' ? 1 : s === 'secteur 2' ? 2 : 3);
  rows.sort((a, b) => (a.avail.rank - b.avail.rank) || (secScore(a.sector) - secScore(b.sector)) || (a.dist - b.dist));

  fs.writeFileSync(path.join(OUTDIR, 'results.json'), JSON.stringify(rows, null, 2));
  console.log('\n================= RÉSULTATS (' + rows.length + ') =================');
  for (const r of rows) {
    const tags = [r.sector, r.video ? 'vidéo' : null, r.isCenter ? 'centre/hôpital' : null].filter(Boolean).join(' · ');
    const km = r.dist < 900 ? `${r.dist} km` : '';
    console.log(`\n• ${r.name}  [${tags}]  ${km}`);
    if (r.addr) console.log(`  ${r.addr}`);
    console.log(`  → ${r.avail.label}`);
  }
  console.log('\nJSON complet: ' + path.join(OUTDIR, 'results.json'));

  if (!ARGS.headless) await sleep(120000);
  cdp.close();
  try { child.kill(); } catch {}
})().catch((e) => { console.log('FATAL ' + e.message); process.exit(1); });

3. Read the ranking

Ranked by: sector match → has concrete time slots → soonest "Prochain RDV le" → distance.

Availability is one of:

• CRÉNEAUX: 10:00, 10:15 — bookable slots in the visible week (soonest, book now)
• Prochain RDV le 22 juin 2026 — next slot beyond the visible week
• réservé aux patients suivis — online booking blocked for new patients → call
• aucune dispo en ligne → call the office

4. Conventionnement (important)

• secteur 1 — no extra fees. What most users mean by "conventionné secteur 1".
• secteur 1 (DP) — secteur 1 with permanent right to dépassement: can still charge extra. Flag it; not zero-dépassement.
• secteur 2 — free fees / dépassements.
• Health centers (centre de santé, Institut …) and public hospitals (AP-HP) are secteur 1 by construction — good fallback when private secteur-1 docs have no online slots. Their agenda is often reserved to existing patients → call.

5. Book

Open the chosen practitioner's profile in the still-open Chrome (or print the Doctolib URL),

let the user select the motif and confirm. Many strong secteur-1 options (hospital services,

centres de santé, some private docs) are phone-only — surface the office phone number too.

Troubleshooting

• 0 cards / "existing browser session" — a stray Chrome holds the profile. The script

auto-kills it and clears SingletonLock; if it persists:

pkill -f "doctolib-finder/profile"; rm -f ~/.cache/doctolib-finder/profile/Singleton* then re-run.

• Captcha loops — solve it slowly in the visible window; the persistent profile remembers

the DataDome token, so later runs (even --headless) usually skip it.

• DOM changed — result cards are .dl-card; the card's innerText carries name, address,

Conventionné secteur X, time slots and Prochain RDV le …. If counts hit 0 on a loaded

page, dump a card's innerText and adjust the parser.

Notes

• Legitimate personal use: searching public Doctolib listings for one's own appointment. The

script enters no credentials; the user does any login/booking by hand.

• Profile (cookies, captcha trust) is stored at ~/.cache/doctolib-finder/profile.