Merge pull request #24 from proxymesh/cursor/vulnerability-findings-memory-4cec

Harden CONNECT header handling against injection and spoofing

Author: proxymesh

.github/dependabot.yml

@@ -11,3 +11,7 @@ updates:
     directory: "/"
     schedule:
       interval: "weekly"
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"

.github/workflows/github_release_on_release_branch_merge.yml

@@ -26,12 +26,12 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout merge commit
-        uses: actions/checkout@v6
+        uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803
         with:
           ref: ${{ github.event.pull_request.merge_commit_sha }}
 
       - name: Set up Python
-        uses: actions/setup-python@v6
+        uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
         with:
           python-version: "3.12"
 

.github/workflows/proxy_integration_tests.yml

@@ -12,16 +12,40 @@ permissions:
   contents: read
 
 jobs:
+  unit:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803
+        with:
+          persist-credentials: false
+
+      - name: Set up Python
+        uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
+        with:
+          python-version: "3.x"
+
+      - name: Install system dependencies (pycurl)
+        run: sudo apt-get update && sudo apt-get install -y libcurl4-openssl-dev
+
+      - name: Install package and test dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -e .
+          pip install urllib3 requests aiohttp httpx pycurl
+
+      - name: Run header security unit tests
+        run: python test_header_security.py
+
   integration:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803
         with:
           persist-credentials: false
 
       - name: Set up Python
-        uses: actions/setup-python@v6
+        uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
         with:
           python-version: "3.x"
 

.github/workflows/publish.yml

@@ -30,7 +30,7 @@ jobs:
     outputs:
       publish: ${{ steps.decide.outputs.publish }}
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803
         if: github.event_name == 'workflow_run'
         with:
           ref: main
@@ -64,12 +64,12 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803
         with:
           ref: ${{ github.event_name == 'workflow_run' && 'main' || github.event_name == 'release' && github.ref || 'main' }}
 
       - name: Set up Python
-        uses: actions/setup-python@v6
+        uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1
         with:
           python-version: "3.x"
 
@@ -82,7 +82,7 @@ jobs:
         run: python -m build
 
       - name: Store distribution packages
-        uses: actions/upload-artifact@v7
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a
         with:
           name: python-package-distributions
           path: dist/
@@ -102,13 +102,13 @@ jobs:
 
     steps:
       - name: Download distribution packages
-        uses: actions/download-artifact@v8
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c
         with:
           name: python-package-distributions
           path: dist/
 
       - name: Publish to PyPI
-        uses: pypa/gh-action-pypi-publish@release/v1
+        uses: pypa/gh-action-pypi-publish@dc37677b2e1c63e2034f94d8a5b11f265b73ba33
 
   publish-to-testpypi:
     name: Publish to TestPyPI
@@ -122,12 +122,12 @@ jobs:
 
     steps:
       - name: Download distribution packages
-        uses: actions/download-artifact@v8
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c
         with:
           name: python-package-distributions
           path: dist/
 
       - name: Publish to TestPyPI
-        uses: pypa/gh-action-pypi-publish@release/v1
+        uses: pypa/gh-action-pypi-publish@dc37677b2e1c63e2034f94d8a5b11f265b73ba33
         with:
           repository-url: https://test.pypi.org/legacy/

docs/aiohttp.rst

@@ -284,7 +284,7 @@ When establishing an HTTPS connection through a proxy, the connector:
 * Creates a CONNECT request to the proxy server
 * Includes any custom proxy headers you've specified
 * Captures the proxy's response headers (e.g., ``X-ProxyMesh-IP``)
-* Stores them so they can be merged into the final response
+* Stores them so safe CONNECT values can be merged into the final response without overwriting origin headers. Hop-by-hop and security-sensitive names are omitted. The full CONNECT set is on ``response.proxy_headers``.
 
 You typically don't need to use this class directly - it's automatically configured when using ``ProxyClientSession``.
 

docs/httpx.rst

@@ -375,7 +375,7 @@ The extension classes work by intercepting the CONNECT request/response cycle du
 
 2. **CONNECT Response**: The proxy responds with a CONNECT response (status 200) that may include proxy information headers in the response (e.g., ``X-ProxyMesh-IP: 192.168.1.1``)
 
-3. **Header Merging**: These proxy response headers are captured during the tunnel establishment and stored. When the actual HTTP request is made through the tunnel, the proxy response headers are merged into the final HTTP response headers using ``merge_headers()``
+3. **Header Merging**: Safe CONNECT headers are stored and later merged into the origin response with ``merge_headers()``. Origin headers are never overwritten, and hop-by-hop or security-sensitive names such as ``Set-Cookie`` or ``Location`` from CONNECT are not copied.
 
 4. **Access**: Your application can then access both the target server's response headers and the proxy's response headers from the same response object
 

docs/requests.rst

@@ -46,6 +46,8 @@ This section shows you how to quickly get up and running with proxy headers in r
 
 That's it! The ``requests_adapter`` module handles sending your custom headers to the proxy and makes proxy response headers available in the response.
 
+CONNECT response headers are copied onto ``response.headers`` except hop-by-hop and security-sensitive names (``Set-Cookie``, ``Location``, ``Content-Type``, and similar). Existing origin headers are never overwritten. The full CONNECT set is on ``response.proxy_headers``.
+
 Using Proxies with requests
 ---------------------------
 

docs/urllib3.rst

@@ -45,6 +45,8 @@ This section shows you how to quickly get up and running with proxy headers in u
 
 That's it! The ``ProxyHeaderManager`` handles sending your custom headers to the proxy and makes proxy response headers available in the response.
 
+CONNECT response headers are copied onto ``response.headers`` except hop-by-hop and security-sensitive names (``Set-Cookie``, ``Location``, ``Content-Type``, and similar). Existing origin headers are never overwritten. The full CONNECT set is on ``response.proxy_headers``.
+
 Using Proxies with urllib3
 --------------------------
 

python_proxy_headers/aiohttp_proxy.py

@@ -6,6 +6,7 @@
 from aiohttp.helpers import reify
 from aiohttp import hdrs
 from multidict import CIMultiDict, CIMultiDictProxy
+from .header_utils import merge_proxy_response_headers
 
 class ProxyTCPConnector(TCPConnector):
 	async def _create_proxy_connection(self, req: ClientRequest, traces, timeout):
@@ -143,13 +144,21 @@ async def send(self, conn):
 		return resp
 
 class ProxyClientResponse(ClientResponse):
+	@reify
+	def proxy_headers(self):
+		"""Headers from the proxy CONNECT response, not merged into origin headers."""
+		raw = getattr(self, "_proxy_headers", None)
+		if raw:
+			return CIMultiDictProxy(CIMultiDict(raw))
+		return CIMultiDictProxy(CIMultiDict())
+
 	@reify
 	def headers(self):
 		proxy_headers = getattr(self, '_proxy_headers', None)
 
 		if proxy_headers:
 			headers = CIMultiDict(self._headers)
-			headers.extend(proxy_headers)
+			merge_proxy_response_headers(headers, proxy_headers)
 			return CIMultiDictProxy(headers)
 		else:
 			return self._headers

python_proxy_headers/autoscraper_proxy.py

@@ -35,6 +35,7 @@
         "Install it with: pip install autoscraper"
     )
 
+from .header_utils import validate_headers
 from .requests_adapter import ProxySession
 
 
@@ -71,7 +72,7 @@ def __init__(
         stack_list: Optional[List] = None
     ):
         super().__init__(stack_list=stack_list)
-        self._proxy_headers = proxy_headers or {}
+        self._proxy_headers = validate_headers(proxy_headers)
         self._session: Optional[ProxySession] = None
 
     def _get_session(self) -> ProxySession:
@@ -90,7 +91,7 @@ def set_proxy_headers(self, proxy_headers: Dict[str, str]):
         Args:
             proxy_headers: New proxy headers to use
         """
-        self._proxy_headers = proxy_headers
+        self._proxy_headers = validate_headers(proxy_headers)
         if self._session is not None:
             self._session.close()
             self._session = None