fix: Expose URL in wsgi middleware

Author: untitaker

sentry_sdk/integrations/_wsgi_common.py

@@ -16,12 +16,6 @@ def extract_into_event(self, event):
 
         content_length = self.content_length()
         request_info = event.setdefault("request", {})
-        try:
-            request_info["url"] = self.url()
-        except Exception:
-            # Django sometimes crashes with KeyError trying to build an
-            # absolute URL
-            pass
 
         if _should_send_default_pii():
             request_info["cookies"] = dict(self.cookies())
@@ -56,9 +50,6 @@ def content_length(self):
         except ValueError:
             return 0
 
-    def url(self):
-        raise NotImplementedError()
-
     def cookies(self):
         raise NotImplementedError()
 

sentry_sdk/integrations/django/__init__.py

@@ -159,9 +159,6 @@ def _got_request_exception(request=None, **kwargs):
 
 
 class DjangoRequestExtractor(RequestExtractor):
-    def url(self):
-        return self.request.build_absolute_uri(self.request.path)
-
     def env(self):
         return self.request.META
 

sentry_sdk/integrations/flask.py

@@ -85,9 +85,6 @@ def _request_started(sender, **kwargs):
 
 
 class FlaskRequestExtractor(RequestExtractor):
-    def url(self):
-        return "%s://%s%s" % (self.request.scheme, self.request.host, self.request.path)
-
     def env(self):
         return self.request.environ
 

sentry_sdk/integrations/wsgi.py

@@ -2,10 +2,51 @@
 
 from sentry_sdk.hub import Hub, _should_send_default_pii
 from sentry_sdk.utils import capture_internal_exceptions, event_from_exception
-from sentry_sdk._compat import reraise, implements_iterator
+from sentry_sdk._compat import PY2, reraise, implements_iterator
 from sentry_sdk.integrations._wsgi_common import _filter_headers
 
 
+if PY2:
+
+    def wsgi_decoding_dance(s, charset="utf-8", errors="replace"):
+        return s.decode(charset, errors)
+
+
+else:
+
+    def wsgi_decoding_dance(s, charset="utf-8", errors="replace"):
+        return s.encode("latin1").decode(charset, errors)
+
+
+def get_host(environ):
+    """Return the host for the given WSGI environment. Yanked from Werkzeug."""
+    if "HTTP_HOST" in environ:
+        rv = environ["HTTP_HOST"]
+        if environ["wsgi.url_scheme"] == "http" and rv.endswith(":80"):
+            rv = rv[:-3]
+        elif environ["wsgi.url_scheme"] == "https" and rv.endswith(":443"):
+            rv = rv[:-4]
+    else:
+        rv = environ["SERVER_NAME"]
+        if (environ["wsgi.url_scheme"], environ["SERVER_PORT"]) not in (
+            ("https", "443"),
+            ("http", "80"),
+        ):
+            rv += ":" + environ["SERVER_PORT"]
+
+    return rv
+
+
+def get_request_url(environ):
+    """Return the absolute URL without query string for the given WSGI
+    environment."""
+    return "%s://%s/%s" % (
+        environ.get("wsgi.url_scheme"),
+        get_host(environ),
+        wsgi_decoding_dance(environ.get("PATH_INFO") or "").lstrip("/"),
+    )
+
+
 class SentryWsgiMiddleware(object):
     __slots__ = ("app",)
 
@@ -139,6 +180,9 @@ def event_processor(event, hint):
                 if "ip_address" not in user_info:
                     user_info["ip_address"] = get_client_ip(environ)
 
+            if "url" not in request_info:
+                request_info["url"] = get_request_url(environ)
+
             if "query_string" not in request_info:
                 request_info["query_string"] = environ.get("QUERY_STRING")
 

tests/integrations/wsgi/test_wsgi.py

@@ -30,6 +30,7 @@ def test_basic(sentry_init, crashing_app, capture_events):
         "headers": {"Content-Length": "0", "Content-Type": "", "Host": "localhost"},
         "method": "GET",
         "query_string": "",
+        "url": "http://localhost/",
     }