Optimize endpoint blacklist (#13)

* performance: re.compile endpoint blacklist
* added documentation

Author: jpcaruana

howfast_apm/flask.py

@@ -6,7 +6,7 @@
 from flask import Flask, request
 
 from .core import CoreAPM
-from .utils import is_in_blacklist
+from .utils import is_in_blacklist, compile_endpoints
 
 logger = logging.getLogger('howfast_apm')
 
@@ -42,7 +42,7 @@ def __init__(
         app.wsgi_app = self
 
         if endpoints_blacklist:
-            self.endpoints_blacklist = endpoints_blacklist
+            self.endpoints_blacklist = compile_endpoints(*endpoints_blacklist)
         else:
             self.endpoints_blacklist = []
 

howfast_apm/utils.py

@@ -1,22 +1,28 @@
 import fnmatch
+import re
 
 from typing import List
 
 
+def compile_endpoints(*blacklist):
+    """Compiles a list endpoints to a list of regexes"""
+    return [re.compile(fnmatch.translate(x)) for x in blacklist]
+
+
 def is_in_blacklist(uri: str, blacklist: List[str]) -> bool:
     """
     Return True if the URI is blacklisted.
 
-      >>> is_in_blacklist('/some-real-uri/', ['/exact-uri/'])
+      >>> is_in_blacklist('/some-real-uri/', compile_endpoints('/exact-uri/'))
       False
-      >>> is_in_blacklist('/exact-uri/', ['/exact-uri/'])
+      >>> is_in_blacklist('/exact-uri/', compile_endpoints('/exact-uri/'))
       True
-      >>> is_in_blacklist('/job/42/retry', ['/job/*'])
+      >>> is_in_blacklist('/job/42/retry', compile_endpoints('/job/*'))
       True
 
     """
     for pattern in blacklist:
-        if fnmatch.fnmatch(uri, pattern):
+        if pattern.match(uri):
             return True
     return False
 

tests/test_utils.py

@@ -1,25 +1,27 @@
-from howfast_apm import utils
+from howfast_apm.utils import compile_endpoints, is_in_blacklist
 
 
 def test_is_blacklist_exact():
     """ The blacklist util should support exact strings matching """
-    assert utils.is_in_blacklist('/some-real-uri/', ['/exact-uri/']) is False
-    assert utils.is_in_blacklist('/exact-uri/', ['/exact-uri/']) is True
-    assert utils.is_in_blacklist('/exact-uri/with-subpath', ['/exact-uri/']) is False
+    blacklist = compile_endpoints('/exact-uri/')
+    assert is_in_blacklist('/some-real-uri/', blacklist) is False
+    assert is_in_blacklist('/exact-uri/', blacklist) is True
+    assert is_in_blacklist('/exact-uri/with-subpath', blacklist) is False
 
     # With multiple patterns
-    assert utils.is_in_blacklist('/exact-uri-1/', ['/exact-uri-1/', '/exact-uri-2/']) is True
-    assert utils.is_in_blacklist('/exact-uri-2/', ['/exact-uri-1/', '/exact-uri-2/']) is True
-    assert utils.is_in_blacklist('/exact-uri-3/', ['/exact-uri-1/', '/exact-uri-2/']) is False
+    multiple_blacklist = compile_endpoints('/exact-uri-1/', '/exact-uri-2/')
+    assert is_in_blacklist('/exact-uri-1/', multiple_blacklist) is True
+    assert is_in_blacklist('/exact-uri-2/', multiple_blacklist) is True
+    assert is_in_blacklist('/exact-uri-3/', multiple_blacklist) is False
 
 
 def test_is_blacklist_glob():
     """ The blacklist util should support shell-like matching """
-    blacklist = ['/jobs/*/results', '/support/*']
-    assert utils.is_in_blacklist('/jobs/42/results', blacklist) is True
-    assert utils.is_in_blacklist('/jobs/42/', blacklist) is False
+    blacklist = compile_endpoints('/jobs/*/results', '/support/*')
+    assert is_in_blacklist('/jobs/42/results', blacklist) is True
+    assert is_in_blacklist('/jobs/42/', blacklist) is False
 
-    assert utils.is_in_blacklist('/support/', blacklist) is True
-    assert utils.is_in_blacklist('/support/tickets/23', blacklist) is True
-    assert utils.is_in_blacklist('/support/admin', blacklist) is True
-    assert utils.is_in_blacklist('/support', blacklist) is False
+    assert is_in_blacklist('/support/', blacklist) is True
+    assert is_in_blacklist('/support/tickets/23', blacklist) is True
+    assert is_in_blacklist('/support/admin', blacklist) is True
+    assert is_in_blacklist('/support', blacklist) is False