flask: Add exclude lists for flask integration (#630)

Leverage global configurations to allow users to specify paths and hosts that they do not want to trace within their Flask applications.

OPENTELEMETRY_PYTHON_FLASK_EXCLUDED_HOSTS and OPENTELEMETRY_PYTHON_FLASK_EXCLUDED_PATHS are the env variables used. Use a comma delimited string to represent seperate hosts/paths to blacklist.

Author: lzchen

ext/opentelemetry-ext-flask/CHANGELOG.md

@@ -2,6 +2,9 @@
 
 ## Unreleased
 
+- Add exclude list for paths and hosts
+  ([#630](https://github.com/open-telemetry/opentelemetry-python/pull/630))
+
 ## 0.6b0
 
 Released 2020-03-30

ext/opentelemetry-ext-flask/README.rst

@@ -16,6 +16,17 @@ Installation
 
     pip install opentelemetry-ext-flask
 
+Configuration
+-------------
+
+Exclude lists
+*************
+Excludes certain hosts and paths from being tracked. Pass in comma delimited string into environment variables.
+Host refers to the entire url and path refers to the part of the url after the domain. Host matches the exact string that is given, where as path matches if the url starts with the given excluded path.
+
+Excluded hosts: OPENTELEMETRY_PYTHON_FLASK_EXCLUDED_HOSTS
+Excluded paths: OPENTELEMETRY_PYTHON_FLASK_EXCLUDED_PATHS
+
 
 References
 ----------

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

@@ -51,10 +51,14 @@ def hello():
 import flask
 
 import opentelemetry.ext.wsgi as otel_wsgi
-from opentelemetry import context, propagators, trace
+from opentelemetry import configuration, context, propagators, trace
 from opentelemetry.auto_instrumentation.instrumentor import BaseInstrumentor
 from opentelemetry.ext.flask.version import __version__
-from opentelemetry.util import time_ns
+from opentelemetry.util import (
+    disable_tracing_hostname,
+    disable_tracing_path,
+    time_ns,
+)
 
 logger = logging.getLogger(__name__)
 
@@ -80,17 +84,18 @@ def wrapped_app(environ, start_response):
             environ[_ENVIRON_STARTTIME_KEY] = time_ns()
 
             def _start_response(status, response_headers, *args, **kwargs):
-                span = flask.request.environ.get(_ENVIRON_SPAN_KEY)
-                if span:
-                    otel_wsgi.add_response_attributes(
-                        span, status, response_headers
-                    )
-                else:
-                    logger.warning(
-                        "Flask environ's OpenTelemetry span "
-                        "missing at _start_response(%s)",
-                        status,
-                    )
+                if not _disable_trace(flask.request.url):
+                    span = flask.request.environ.get(_ENVIRON_SPAN_KEY)
+                    if span:
+                        otel_wsgi.add_response_attributes(
+                            span, status, response_headers
+                        )
+                    else:
+                        logger.warning(
+                            "Flask environ's OpenTelemetry span "
+                            "missing at _start_response(%s)",
+                            status,
+                        )
 
                 return start_response(
                     status, response_headers, *args, **kwargs
@@ -102,6 +107,9 @@ def _start_response(status, response_headers, *args, **kwargs):
 
         @self.before_request
         def _before_flask_request():
+            # Do not trace if the url is excluded
+            if _disable_trace(flask.request.url):
+                return
             environ = flask.request.environ
             span_name = (
                 flask.request.endpoint
@@ -132,6 +140,9 @@ def _before_flask_request():
 
         @self.teardown_request
         def _teardown_flask_request(exc):
+            # Not traced if the url is excluded
+            if _disable_trace(flask.request.url):
+                return
             activation = flask.request.environ.get(_ENVIRON_ACTIVATION_KEY)
             if not activation:
                 logger.warning(
@@ -150,6 +161,20 @@ def _teardown_flask_request(exc):
             context.detach(flask.request.environ.get(_ENVIRON_TOKEN))
 
 
+def _disable_trace(url):
+    excluded_hosts = configuration.Configuration().FLASK_EXCLUDED_HOSTS
+    excluded_paths = configuration.Configuration().FLASK_EXCLUDED_PATHS
+    if excluded_hosts:
+        excluded_hosts = str.split(excluded_hosts, ",")
+        if disable_tracing_hostname(url, excluded_hosts):
+            return True
+    if excluded_paths:
+        excluded_paths = str.split(excluded_paths, ",")
+        if disable_tracing_path(url, excluded_paths):
+            return True
+    return False
+
+
 class FlaskInstrumentor(BaseInstrumentor):
     """A instrumentor for flask.Flask
 

ext/opentelemetry-ext-flask/tests/test_flask_integration.py

@@ -13,12 +13,14 @@
 # limitations under the License.
 
 import unittest
+from unittest.mock import patch
 
 from flask import Flask, request
 from werkzeug.test import Client
 from werkzeug.wrappers import BaseResponse
 
 from opentelemetry import trace as trace_api
+from opentelemetry.configuration import Configuration
 from opentelemetry.test.wsgitestutil import WsgiTestBase
 
 
@@ -45,18 +47,31 @@ def setUp(self):
         # No instrumentation code is here because it is present in the
         # conftest.py file next to this file.
         super().setUp()
-
+        Configuration._instance = None  # pylint:disable=protected-access
+        Configuration.__slots__ = []
         self.app = Flask(__name__)
 
         def hello_endpoint(helloid):
             if helloid == 500:
                 raise ValueError(":-(")
             return "Hello: " + str(helloid)
 
+        def excluded_endpoint():
+            return "excluded"
+
+        def excluded2_endpoint():
+            return "excluded2"
+
         self.app.route("/hello/<int:helloid>")(hello_endpoint)
+        self.app.route("/excluded")(excluded_endpoint)
+        self.app.route("/excluded2")(excluded2_endpoint)
 
         self.client = Client(self.app, BaseResponse)
 
+    def tearDown(self):
+        Configuration._instance = None  # pylint:disable=protected-access
+        Configuration.__slots__ = []
+
     def test_only_strings_in_environ(self):
         """
         Some WSGI servers (such as Gunicorn) expect keys in the environ object
@@ -80,9 +95,8 @@ def test_simple(self):
         expected_attrs = expected_attributes(
             {"http.target": "/hello/123", "http.route": "/hello/<int:helloid>"}
         )
-        resp = self.client.get("/hello/123")
-        self.assertEqual(200, resp.status_code)
-        self.assertEqual([b"Hello: 123"], list(resp.response))
+        self.client.get("/hello/123")
+
         span_list = self.memory_exporter.get_finished_spans()
         self.assertEqual(len(span_list), 1)
         self.assertEqual(span_list[0].name, "hello_endpoint")
@@ -126,6 +140,21 @@ def test_internal_error(self):
         self.assertEqual(span_list[0].kind, trace_api.SpanKind.SERVER)
         self.assertEqual(span_list[0].attributes, expected_attrs)
 
+    @patch.dict(
+        "os.environ",  # type: ignore
+        {
+            "OPENTELEMETRY_PYTHON_FLASK_EXCLUDED_HOSTS": "http://localhost/excluded",
+            "OPENTELEMETRY_PYTHON_FLASK_EXCLUDED_PATHS": "excluded2",
+        },
+    )
+    def test_excluded_path(self):
+        self.client.get("/hello/123")
+        self.client.get("/excluded")
+        self.client.get("/excluded2")
+        span_list = self.memory_exporter.get_finished_spans()
+        self.assertEqual(len(span_list), 1)
+        self.assertEqual(span_list[0].name, "hello_endpoint")
+
 
 if __name__ == "__main__":
     unittest.main()

opentelemetry-api/src/opentelemetry/util/__init__.py

@@ -11,9 +11,10 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+import re
 import time
 from logging import getLogger
-from typing import Union
+from typing import Sequence, Union
 
 from pkg_resources import iter_entry_points
 
@@ -33,18 +34,39 @@ def time_ns() -> int:
         return int(time.time() * 1e9)
 
 
-def _load_provider(provider: str) -> Union["TracerProvider", "MeterProvider"]:  # type: ignore
+def _load_provider(
+    provider: str,
+) -> Union["TracerProvider", "MeterProvider"]:  # type: ignore
     try:
         return next(  # type: ignore
             iter_entry_points(
                 "opentelemetry_{}".format(provider),
-                name=getattr(  # type: ignore
-                    Configuration(), provider, "default_{}".format(provider),  # type: ignore
+                name=getattr(
+                    Configuration(),  # type: ignore
+                    provider,
+                    "default_{}".format(provider),
                 ),
             )
         ).load()()
     except Exception:  # pylint: disable=broad-except
-        logger.error(
-            "Failed to load configured provider %s", provider,
-        )
+        logger.error("Failed to load configured provider %s", provider)
         raise
+
+
+# Pattern for matching up until the first '/' after the 'https://' part.
+_URL_PATTERN = r"(https?|ftp)://.*?/"
+
+
+def disable_tracing_path(url: str, excluded_paths: Sequence[str]) -> bool:
+    if excluded_paths:
+        # Match only the part after the first '/' that is not in _URL_PATTERN
+        regex = "{}({})".format(_URL_PATTERN, "|".join(excluded_paths))
+        if re.match(regex, url):
+            return True
+    return False
+
+
+def disable_tracing_hostname(
+    url: str, excluded_hostnames: Sequence[str]
+) -> bool:
+    return url in excluded_hostnames