[httplib, requests] Sanitize urls in span metadata

ddtrace/compat.py

@@ -1,5 +1,5 @@
-import sys
 import platform
+import sys
 
 PYTHON_VERSION_INFO = sys.version_info
 PY2 = sys.version_info[0] == 2

ddtrace/contrib/httplib/patch.py

@@ -5,12 +5,11 @@
 import wrapt
 
 # Project
-from ...compat import httplib, PY2
+from ...compat import PY2, httplib, parse
 from ...ext import http as ext_http
 from ...pin import Pin
 from ...utils.wrappers import unwrap as _u
 
-
 span_name = 'httplib.request' if PY2 else 'http.client.request'
 
 log = logging.getLogger(__name__)
@@ -60,9 +59,14 @@ def _wrap_putrequest(func, instance, args, kwargs):
         method, path = args[:2]
         scheme = 'https' if isinstance(instance, httplib.HTTPSConnection) else 'http'
         port = ':{port}'.format(port=instance.port)
+
+        # strip path of non-path data
+        path = parse.urlparse(path).path
+
         if (scheme == 'http' and instance.port == 80) or (scheme == 'https' and instance.port == 443):
             port = ''
         url = '{scheme}://{host}{port}{path}'.format(scheme=scheme, host=instance.host, port=port, path=path)
+
         span.set_tag(ext_http.URL, url)
         span.set_tag(ext_http.METHOD, method)
     except Exception:

tests/contrib/httplib/test_httplib.py

@@ -7,16 +7,15 @@
 import wrapt
 
 # Project
-from ddtrace.compat import httplib, PY2
+from ddtrace.compat import PY2, httplib
 from ddtrace.contrib.httplib import patch, unpatch
 from ddtrace.contrib.httplib.patch import should_skip_request
 from ddtrace.pin import Pin
-
 from tests.opentracer.utils import init_tracer
+
 from ...test_tracer import get_dummy_tracer
 from ...util import assert_dict_issuperset, override_global_tracer
 
-
 if PY2:
     from urllib2 import urlopen, build_opener, Request
 else:
@@ -242,7 +241,7 @@ def test_httplib_request_get_request_query_string(self):
             {
                 'http.method': 'GET',
                 'http.status_code': '200',
-                'http.url': '{}?key=value&key2=value2'.format(URL_200),
+                'http.url': '{}'.format(URL_200),
             }
         )