wsgi: Fix non-absolute http.url (fixes open-telemetry#143).

ext/opentelemetry-ext-wsgi/src/opentelemetry/ext/wsgi/__init__.py

@@ -20,6 +20,7 @@
 
 import functools
 import wsgiref.util as wsgiref_util
+from urllib.parse import urlparse
 
 from opentelemetry import trace
 from opentelemetry.ext.wsgi.version import __version__  # noqa
@@ -47,13 +48,33 @@ def _add_request_attributes(span, environ):
         span.set_attribute("http.method", environ["REQUEST_METHOD"])
 
         host = environ.get("HTTP_HOST") or environ["SERVER_NAME"]
-        span.set_attribute("http.host", host)
+        span.set_attribute("http.host", host)  # NOTE: Nonstandard.
+
+        url = environ.get("REQUEST_URI") or environ.get("RAW_URI")
+
+        if url:  # We got something, but is absolute?
+            # The simplistic ``"://" in url` is not sufficient,
+            # as that could be contained in the query string.
+            try:
+                urlparts = urlparse(url)
+            except Exception:  # pylint:disable=broad-except
+                url = wsgiref_util.request_uri(environ)
+            else:
+                if not urlparts.netloc:
+                    scheme = environ["wsgi.url_scheme"]
+                    portstr = ""
+                    if scheme == "https":
+                        if environ.get("SERVER_PORT", "443") != "443":
+                            portstr = ":" + environ["SERVER_PORT"]
+                    elif environ.get("SERVER_PORT", "80") != "80":
+                        portstr = ":" + environ["SERVER_PORT"]
+
+                    url = (
+                        scheme + "://" + (host or "localhost") + portstr + url
+                    )
+        else:
+            url = wsgiref_util.request_uri(environ)
 
-        url = (
-            environ.get("REQUEST_URI")
-            or environ.get("RAW_URI")
-            or wsgiref_util.request_uri(environ, include_query=False)
-        )
         span.set_attribute("http.url", url)
 
     @staticmethod

ext/opentelemetry-ext-wsgi/tests/test_wsgi_middleware.py

@@ -171,6 +171,36 @@ def test_request_attributes(self):
         self.assertEqual(self.span.set_attribute.call_count, len(expected))
         self.span.set_attribute.assert_has_calls(expected, any_order=True)
 
+    def test_request_attributes_with_partial_raw_uri(self):
+        self.environ["RAW_URI"] = "/#top"
+        OpenTelemetryMiddleware._add_request_attributes(  # noqa pylint: disable=protected-access
+            self.span, self.environ
+        )
+        self.span.set_attribute.assert_any_call(
+            "http.url", "http://127.0.0.1/#top"
+        )
+
+    def test_request_attributes_with_partial_raw_uri_and_nonstandard_port(
+        self
+    ):
+        self.environ["RAW_URI"] = "/#top"
+        self.environ["SERVER_PORT"] = "8080"
+        OpenTelemetryMiddleware._add_request_attributes(  # noqa pylint: disable=protected-access
+            self.span, self.environ
+        )
+        self.span.set_attribute.assert_any_call(
+            "http.url", "http://127.0.0.1:8080/#top"
+        )
+
+    def test_request_attributes_with_full_request_uri(self):
+        self.environ["REQUEST_URI"] = "http://foobar.com:8080/?foo=bar#top"
+        OpenTelemetryMiddleware._add_request_attributes(  # noqa pylint: disable=protected-access
+            self.span, self.environ
+        )
+        self.span.set_attribute.assert_any_call(
+            "http.url", "http://foobar.com:8080/?foo=bar#top"
+        )
+
     def test_response_attributes(self):
         OpenTelemetryMiddleware._add_response_attributes(  # noqa pylint: disable=protected-access
             self.span, "404 Not Found"