immoscout24 adress finder

[sheddixx]

Hi,

Stumbled across your app. Really cool and working great for now.
i was wondering if you could implement the adressfinder of immoscout24. Its a Browser plugin. Not sure if you can implement it somehow.

https://github.com/kidzki/immo24-address-finder

[orangecoding]

I don't know what you mean. All the addresses are also being scraped.

[sheddixx]

I mean, there is only the city and very rarely the complete address with street and number on Immoscout24.
With the plugin below it uses the speedcheck on their site to get the complete adress with the streetname.

Sorry when i didnt describe it right.

[orangecoding]

I'm still unsure, you know I have 10 different jobs running with immoscout and all of which show the address..

Which Fredy version are you using?

[sheddixx]

no problem : ) , i am on version V16.3.0.

on picture 1, thats what I see in your app. only the city is shown

on picture 2, thats the plugin and i can see the street etc.

[orangecoding]

@sheddixx you might want to upvote this.
kidzki/immo24-address-finder#1

[sheddixx]

Very cool :)

[orangecoding]

I built a very simple function, based on the repository which works fine in the browser.. HOWEVER, as Fredy is not using puppeteer to get Immoscout listings, but their Mobile Api, this doesn't really help as this approach does not work here...

async function extractImmoscoutAddress() {
  const extractEncoded = () => {
    const re = /"obj_telekomInternetUrlAddition"\s*:\s*"([^"]+)"/g;

    for (const s of Array.from(document.scripts)) {
      const txt = s.textContent || '';
      let m;
      while ((m = re.exec(txt)) !== null) {
        if (m[1]) return m[1];
      }
    }

    const html = document.documentElement.innerHTML;
    const m2 = html.match(/"obj_telekomInternetUrlAddition"\s*:\s*"([^"]+)"/);
    return m2 ? m2[1] : null;
  };

  const b64Normalize = (b64) => (b64 || '').replace(/-/g, '+').replace(/_/g, '/').replace(/\s+/g, '');

  const b64ToBytes = (b64) => {
    const norm = b64Normalize(b64);
    const pad = norm.length % 4 === 0 ? '' : '='.repeat(4 - (norm.length % 4));
    const bin = atob(norm + pad);
    return Uint8Array.from(bin, (c) => c.charCodeAt(0));
  };

  const bytesToStringSmart = (bytes) => {
    try {
      return new TextDecoder('utf-8', { fatal: true }).decode(bytes);
    } catch {}
    try {
      return new TextDecoder('windows-1252', { fatal: true }).decode(bytes);
    } catch {}
    try {
      return new TextDecoder('iso-8859-1', { fatal: true }).decode(bytes);
    } catch {}
    let s = '';
    for (let i = 0; i < bytes.length; i++) s += String.fromCharCode(bytes[i] & 0xff);
    try {
      return decodeURIComponent(escape(s));
    } catch {
      return s;
    }
  };

  const fixDoubleUtf8 = (s) => {
    if (!s) return s;
    if (!/[Ã][\x80-\xBF]/.test(s)) return s;
    const bytes = Uint8Array.from([...s].map((ch) => ch.charCodeAt(0) & 0xff));
    try {
      return new TextDecoder('utf-8', { fatal: true }).decode(bytes);
    } catch {
      return s;
    }
  };

  const decodeTelekomAddition = (encoded) => {
    if (!encoded) return null;

    let urlDecoded;
    try {
      urlDecoded = decodeURIComponent(String(encoded).replace(/\+/g, '%20'));
    } catch {
      urlDecoded = encoded;
    }

    // Try Base64 first
    let jsonCandidate = null;
    try {
      const bytes = b64ToBytes(urlDecoded);
      jsonCandidate = bytesToStringSmart(bytes);
    } catch {}

    if (jsonCandidate) {
      jsonCandidate = fixDoubleUtf8(jsonCandidate);
      try {
        return JSON.parse(jsonCandidate);
      } catch {}
      try {
        return JSON.parse(decodeURIComponent(jsonCandidate));
      } catch {}
    }

    // Fallback: direct JSON
    try {
      return JSON.parse(urlDecoded);
    } catch {}
    try {
      return JSON.parse(decodeURIComponent(urlDecoded));
    } catch {}

    return null;
  };

  const enc = extractEncoded();
  if (!enc) return null;

  const obj = decodeTelekomAddition(enc) || {};
  const address = {
    street: obj.strasse || obj.street || '',
    no: obj.hausnummer || obj.houseNumber || obj.housenumber || '',
    plz: obj.plz || obj.zip || obj.postalCode || '',
    city: obj.ort || obj.city || '',
    district: obj.ortsteil || obj.district || ''
  };

  return address;
}