fix(receiver): prevent requests to non-public addresses

Resolve source hostnames before fetching and reject addresses that are
not globally routable to prevent SSRF against local or private services.

Repeat the check for every HTTP request so redirects cannot bypass the
initial source validation. Treat DNS resolution failures during
verification as temporary fetch errors.

4 files changed, 82 insertions, 1 deletions

diff --git a/webmentions_ssg/forms/__init__.py b/webmentions_ssg/forms/__init__.py
index 98962ad..0c54999 100644
--- a/webmentions_ssg/forms/__init__.py
+++ b/webmentions_ssg/forms/__init__.py
@@ -4,7 +4,7 @@ from flask_wtf import FlaskForm
 from wtforms import PasswordField, StringField, SubmitField
 from wtforms.validators import URL, InputRequired, Regexp
 
-from .validators import AllowedHostname, NotEqualTo
+from .validators import AllowedHostname, NotEqualTo, PublicURL
 
 
 class LoginForm(FlaskForm):
@@ -25,6 +25,7 @@ class EndpointForm(FlaskForm):
                 message="source must begin with http or https",
             ),
             NotEqualTo("target"),
+            PublicURL(),
         ],
     )
     target = StringField(
diff --git a/webmentions_ssg/forms/validators.py b/webmentions_ssg/forms/validators.py
index 20d9ebf..e979644 100644
--- a/webmentions_ssg/forms/validators.py
+++ b/webmentions_ssg/forms/validators.py
@@ -3,6 +3,12 @@ from urllib.parse import urlsplit
 from flask import current_app
 from wtforms import ValidationError
 
+from ..url_security import (
+    AddressResolutionError,
+    NonPublicAddressError,
+    ensure_public_url,
+)
+
 
 class NotEqualTo:
     """
@@ -60,3 +66,22 @@ class AllowedHostname:
             message = field.gettext("Invalid input.")
 
         raise ValidationError(message)
+
+
+class PublicURL:
+    def __init__(self, message=None):
+        self.message = message
+
+    def __call__(self, form, field):
+        try:
+            ensure_public_url(field.data)
+        except (
+            AddressResolutionError,
+            NonPublicAddressError,
+        ) as exc:
+            message = self.message
+
+            if message is None:
+                message = field.gettext("URL must resolve to a public address.")
+
+            raise ValidationError(message) from exc
diff --git a/webmentions_ssg/tasks/receiver.py b/webmentions_ssg/tasks/receiver.py
index ea48299..9475866 100644
--- a/webmentions_ssg/tasks/receiver.py
+++ b/webmentions_ssg/tasks/receiver.py
@@ -10,6 +10,11 @@ from .. import APP_NAME, VERSION
 from .. import DATABASE as db
 from .. import HUEY as huey
 from ..models import ReceivedWebmention
+from ..url_security import (
+    AddressResolutionError,
+    NonPublicAddressError,
+    ensure_public_url,
+)
 
 
 class VerificationError(Exception):
@@ -103,6 +108,16 @@ def text_mentions_target(body: str, target_url: str) -> bool:
     return any(match.group() == target_url for match in IRI_PATTERN.finditer(body))
 
 
+def ensure_public_request(request: httpx.Request) -> None:
+    """Prevent requests to non-public network addresses."""
+    try:
+        ensure_public_url(str(request.url))
+    except NonPublicAddressError as exc:
+        raise VerificationError("Source resolves to a non-public address") from exc
+    except AddressResolutionError as exc:
+        raise TemporaryFetchError("Source hostname could not be resolved") from exc
+
+
 def fetch_source(source_url: str) -> tuple[httpx.Response, bytes]:
     """Fetch a source with limits on redirects, time, and response size."""
 
@@ -117,6 +132,9 @@ def fetch_source(source_url: str) -> tuple[httpx.Response, bytes]:
         follow_redirects=True,
         max_redirects=current_app.config.get("WEBMENTIONS_SSG_MAX_REDIRECTS", 20),
         trust_env=False,
+        event_hooks={
+            "request": [ensure_public_request],
+        },
     ) as client:
         with client.stream("GET", source_url) as response:
             match response.status_code:
diff --git a/webmentions_ssg/url_security.py b/webmentions_ssg/url_security.py
new file mode 100644
index 0000000..ce5d700
--- /dev/null
+++ b/webmentions_ssg/url_security.py
@@ -0,0 +1,37 @@
+import ipaddress
+from urllib.parse import urlsplit
+
+import dns.exception
+import dns.resolver
+
+
+class AddressResolutionError(Exception):
+    pass
+
+
+class NonPublicAddressError(Exception):
+    pass
+
+
+def ensure_public_url(url: str) -> None:
+    hostname = urlsplit(url).hostname
+
+    if hostname is None:
+        raise NonPublicAddressError("URL has no hostname")
+
+    try:
+        answers = dns.resolver.resolve_name(hostname)
+    except dns.exception.DNSException as exc:
+        raise AddressResolutionError(
+            f"Could not resolve hostname {hostname!r}"
+        ) from exc
+
+    addresses = {ipaddress.ip_address(address) for address in answers.addresses()}
+
+    if not addresses:
+        raise AddressResolutionError(f"Hostname {hostname!r} did not resolve")
+
+    if any(not address.is_global for address in addresses):
+        raise NonPublicAddressError(
+            f"Hostname {hostname!r} resolves to a non-public address"
+        )