Use Funda mobile app fingerprint and add open-source rationale

Author: 0xMH

.github/workflows/publish.yml

@@ -1,7 +1,9 @@
 name: Publish to PyPI
 
 on:
-  workflow_dispatch:
+  push:
+    branches: [main]
+    paths: [pyproject.toml]
 
 jobs:
   publish:
@@ -18,13 +20,27 @@ jobs:
         id: version
         run: echo "version=$(grep '^version' pyproject.toml | cut -d'"' -f2)" >> $GITHUB_OUTPUT
 
+      - name: Check if release exists
+        id: check
+        run: |
+          if gh release view "v${{ steps.version.outputs.version }}" &>/dev/null; then
+            echo "exists=true" >> $GITHUB_OUTPUT
+          else
+            echo "exists=false" >> $GITHUB_OUTPUT
+          fi
+        env:
+          GH_TOKEN: ${{ github.token }}
+
       - name: Build
+        if: steps.check.outputs.exists == 'false'
         run: uv build
 
       - name: Publish to PyPI
+        if: steps.check.outputs.exists == 'false'
         run: uv publish --token ${{ secrets.PYPI_TOKEN }}
 
       - name: Create GitHub Release
+        if: steps.check.outputs.exists == 'false'
         run: gh release create "v${{ steps.version.outputs.version }}" --title "v${{ steps.version.outputs.version }}" --generate-notes
         env:
           GH_TOKEN: ${{ github.token }}

README.md

@@ -10,11 +10,17 @@ The only working real Python API for Funda ([funda.nl](https://www.funda.nl)) 
 
 [![Star History Chart](https://api.star-history.com/svg?repos=0xMH/pyfunda&type=Date)](https://star-history.com/#0xMH/pyfunda&Date)
 
-## Installation
+## Why I'm open-sourcing this?
 
-```bash
-pip install pyfunda
-```
+After pyfunda, I got messages asking why I'd give this away when aggregators will just take it and sell it. They're right, every week there's a new "revolutionary AI-powered housing finder" charging €40/month or a €250 "success fee.". They all pull from the same one or two sources and wrap it in a fancy UI completely built with AI.
+
+That's exactly why I'm open-sourcing it.
+
+These services are selling air to people who are looking for any kind of hope. The data is public. The APIs aren't hard to figure out. You shouldn't have to pay someone to refresh a webpage for you. Funda could kill this entire market overnight by offering a public API. They don't, so here we are.
+
+Here's the code, do it yourself. Send my library link to any AI service you use and ask it to build whatever tool you think will make your life easier while searching for your next home.
+
+With pyfunda, I've already done all the heavy lifting for you.
 
 ## Why pyfunda?
 
@@ -36,6 +42,12 @@ Funda has no public API. If you want Dutch real estate data programmatically, yo
 - 70+ fields including photos, floorplans, coordinates, and listing dates
 - Stable mobile API that doesn't break when the website changes
 
+## Installation
+
+```bash
+pip install pyfunda
+```
+
 ## Quick Start
 
 ```python

funda/__init__.py

@@ -13,8 +13,10 @@
     ...     print(r['title'], r['city'])
 """
 
+from importlib.metadata import version
+
 from funda.funda import Funda, FundaAPI
 from funda.listing import Listing
 
-__version__ = "2.2.0"
+__version__ = version("pyfunda")
 __all__ = ["Funda", "FundaAPI", "Listing", "__version__"]

funda/funda.py

@@ -1,10 +1,12 @@
 """Main Funda API class."""
 
+import random
 import re
 import time
 from typing import Any
 
 from curl_cffi import requests
+from curl_cffi.const import CurlHttpVersion
 
 from funda.listing import Listing
 
@@ -16,17 +18,47 @@
 API_SEARCH = "https://listing-search-wonen.funda.io/_msearch/template"
 API_WALTER = "https://api.walterliving.com/hunter/lookup"
 
-# Headers for mobile API
-HEADERS = {
-    "x-funda-app-platform": "android",
-    "content-type": "application/json",
-}
-
-SEARCH_HEADERS = {
-    "content-type": "application/json",
-    "accept": "application/json",
-    "referer": "https://www.funda.nl/",
-}
+FUNDA_JA3 = "771,4867-4865-4866-52393-52392-49195-49199-49196-49200-49161-49171-49162-49172-156-157-47-53,0-23-65281-10-11-35-13-51-45-43-21,29-23-24,0"
+
+
+
+def _make_headers(host: str, for_search: bool = False) -> list[tuple[str, str]]:
+    """Generate headers matching the Funda Android app."""
+    trace_id = str(random.randint(10**18, 10**19))
+    parent_id = hex(random.randint(10**15, 10**16))[2:]
+    tid = hex(int(time.time()))[2:] + "00000000"
+
+    headers = [
+        ("user-agent", "Dart/3.9 (dart:io)"),
+        ("x-datadog-sampling-priority", "0"),
+        ("x-datadog-origin", "rum"),
+        ("tracestate", f"dd=s:0;o:rum;p:{parent_id}"),
+        ("accept-encoding", "gzip"),
+        ("x-datadog-parent-id", trace_id),
+    ]
+
+    if for_search:
+        # Search endpoint uses referer and accept instead of x-funda-app-platform
+        headers.extend([
+            ("content-type", "application/json"),
+            ("referer", "https://www.funda.nl/"),
+            ("accept", "application/json"),
+        ])
+    else:
+        # Listing endpoint uses x-funda-app-platform
+        headers.extend([
+            ("x-funda-app-platform", "android"),
+            ("content-type", "application/json"),
+        ])
+
+    headers.extend([
+        ("traceparent", f"00-{tid}{trace_id[:16]}-{parent_id}-00"),
+        ("host", host),
+        ("x-datadog-tags", f"_dd.p.tid={tid}"),
+        ("x-datadog-trace-id", trace_id),
+    ])
+
+    return headers
 
 
 def _parse_area(value: str | None) -> int | None:
@@ -69,8 +101,7 @@ def __init__(self, timeout: int = 30):
     def session(self) -> requests.Session:
         """Lazily create HTTP session."""
         if self._session is None:
-            self._session = requests.Session(impersonate="chrome")
-            self._session.headers.update(HEADERS)
+            self._session = requests.Session()
         return self._session
 
     def close(self) -> None:
@@ -111,17 +142,26 @@ def get_listing(self, listing_id: int | str) -> Listing:
 
         # Try tinyId endpoint first (8-9 digits), then globalId (7 digits)
         listing_id_str = str(listing_id)
+        host = "listing-detail-page.funda.io"
         if len(listing_id_str) >= 8:
             url = API_LISTING_TINY.format(tiny_id=listing_id_str)
         else:
             url = API_LISTING.format(listing_id=listing_id_str)
 
-        response = self.session.get(url, timeout=self.timeout)
+        headers = _make_headers(host)
+        response = self.session.get(
+            url, headers=headers, ja3=FUNDA_JA3,
+            http_version=CurlHttpVersion.V1_1, timeout=self.timeout
+        )
 
         # If tinyId fails, try as globalId
         if response.status_code == 404 and len(listing_id_str) >= 8:
             url = API_LISTING.format(listing_id=listing_id_str)
-            response = self.session.get(url, timeout=self.timeout)
+            headers = _make_headers(host)
+            response = self.session.get(
+                url, headers=headers, ja3=FUNDA_JA3,
+                http_version=CurlHttpVersion.V1_1, timeout=self.timeout
+            )
 
         if response.status_code != 200:
             raise LookupError(f"Listing {listing_id} not found")
@@ -259,11 +299,15 @@ def search_listing(
         query = f"{index_line}\n{query_line}\n"
 
         # Retry on intermittent 400 errors from API
+        host = "listing-search-wonen.funda.io"
         for attempt in range(3):
+            headers = _make_headers(host, for_search=True)
             response = self.session.post(
                 API_SEARCH,
-                headers=SEARCH_HEADERS,
+                headers=headers,
                 data=query,
+                ja3=FUNDA_JA3,
+                http_version=CurlHttpVersion.V1_1,
                 timeout=self.timeout,
             )
             if response.status_code == 200:
@@ -458,11 +502,16 @@ def poll_new_listings(
         """
         consecutive_404s = 0
         current_id = since_id + 1
+        host = "listing-detail-page.funda.io"
 
         while consecutive_404s < max_consecutive_404s:
             url = API_LISTING.format(listing_id=current_id)
             try:
-                response = self.session.get(url, timeout=self.timeout)
+                headers = _make_headers(host)
+                response = self.session.get(
+                    url, headers=headers, ja3=FUNDA_JA3,
+                    http_version=CurlHttpVersion.V1_1, timeout=self.timeout
+                )
 
                 if response.status_code == 200:
                     consecutive_404s = 0
@@ -536,6 +585,7 @@ def get_price_history(self, listing: Listing | str) -> list[dict]:
             json=payload,
             headers={"Accept": "application/json", "Content-Type": "application/json"},
             timeout=self.timeout,
+            http_version=CurlHttpVersion.V1_1,
         )
 
         if response.status_code != 200:

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "pyfunda"
-version = "2.2.0"
+version = "2.2.1"
 description = "Python API for Funda.nl real estate listings"
 readme = "README.md"
 license = "AGPL-3.0-or-later"

test_all_flows.py

@@ -626,18 +626,18 @@ def test_session_lazy_loading():
     print("  Session lazy loading working correctly")
 
 
-@test("Session headers are set correctly")
+@test("Session is created correctly")
 def test_session_headers():
     from funda import Funda
 
     f = Funda()
     session = f.session
 
-    assert 'x-funda-app-platform' in session.headers, "Should have x-funda-app-platform header"
-    assert session.headers['x-funda-app-platform'] == 'android', "Platform should be android"
+    # Session exists (headers are generated per-request with Dart fingerprint)
+    assert session is not None, "Session should exist"
 
     f.close()
-    print("  Session headers set correctly")
+    print("  Session created correctly")
 
 
 @test("Custom timeout is respected")