Merge pull request #288 from cavokz/speed-up-high-cardinality-generation

Speed up generation of (some) high cardinality data

geneve/utils/__init__.py

@@ -19,6 +19,7 @@
 
 import functools
 import json
+import re
 import shutil
 import sys
 from contextlib import contextmanager
@@ -32,12 +33,12 @@
 from . import dirs, epr
 
 random = Random()
+wc_re = re.compile(r"\*|\?")
 
 
 def has_wildcards(s):
-    if not isinstance(s, str):
-        return False
-    return s.find("?") + s.find("*") > -2
+    if isinstance(s, str):
+        return bool(wc_re.search(s))
 
 
 def expand_wildcards(s, alphabet, min_star_len, max_star_len):

geneve/utils/solution_space.py

@@ -77,11 +77,16 @@ def __bool__(self):
     def __contains__(self, item):
         if self.__exclude and item in self.__exclude:
             return False
-        return (
-            self.__set == everything()
-            or any(item == value for value, _ in self.__set)
-            or any(fnmatchcase(str(item).lower(), str(value).lower()) for value, _ in self.__set)
-        )
+        if self.__set == everything():
+            return True
+        item = str(item).lower()
+        for value, wc in self.__set:
+            value = str(value).lower()
+            if item == value:
+                return True
+            if wc and fnmatchcase(item, value):
+                return True
+        return False
 
     def __and__(self, other):
         ss = self.__copy__()

tests/test_solution_space.py

@@ -30,6 +30,7 @@ class TestStrings(tu.SeededTestCase, unittest.TestCase):
     def test_contains(self):
         testcases = [
             ("a", "a", True),
+            ("a", "A", True),
             ("aa", "a", False),
             ("a", "aa", False),
             ("a?", "a?", True),