ruff recommendations

Author: barrust

probables/blooms/bloom.py

@@ -308,10 +308,7 @@ def export_c_header(self, filename: Union[str, Path]) -> None:
         Args:
             filename (str): The filename to which the Bloom Filter will be written."""
         data = ("  " + line for line in wrap(", ".join(f"0x{e:02x}" for e in bytearray.fromhex(self.export_hex())), 80))
-        if self._type in ["regular", "regular-on-disk"]:
-            bloom_type = "standard BloomFilter"
-        else:
-            bloom_type = "CountingBloomFilter"
+        bloom_type = "standard BloomFilter" if self._type in ["regular", "regular-on-disk"] else "CountingBloomFilter"
 
         with open(filename, "w", encoding="utf-8") as file:
             print(f"/* BloomFilter Export of a {bloom_type} */", file=file)
@@ -570,9 +567,7 @@ def _verify_bloom_similarity(self, second: SimpleBloomT) -> bool:
         hash_match = self.number_hashes != second.number_hashes
         same_bits = self.number_bits != second.number_bits
         next_hash = self.hashes("test") != second.hashes("test")
-        if hash_match or same_bits or next_hash:
-            return False
-        return True
+        return not (hash_match or same_bits or next_hash)
 
 
 class BloomFilterOnDisk(BloomFilter):
@@ -673,7 +668,7 @@ def _load(self, file: Union[str, Path], hash_function: Union[HashFuncT, None] =
             fpr, n_hashes, n_bits = self._get_optimized_params(est_els, fpr)
             self._set_values(est_els, fpr, n_hashes, n_bits, hash_function)
         # setup a few additional items
-        self.__file_pointer = open(file, "r+b")  # type: ignore
+        self.__file_pointer = open(file, "r+b")  # type: ignore  # noqa: SIM115
         self._bloom = mmap(self.__file_pointer.fileno(), 0)  # type: ignore
         self._on_disk = True
 

probables/cuckoo/countingcuckoo.py

@@ -136,9 +136,7 @@ def frombytes(
 
     def __contains__(self, val: KeyT) -> bool:
         """setup the `in` keyword"""
-        if self.check(val) > 0:
-            return True
-        return False
+        return self.check(val) > 0
 
     @property
     def unique_elements(self) -> int:

probables/cuckoo/cuckoo.py

@@ -313,9 +313,7 @@ def check(self, key: KeyT) -> bool:
             bool: True if likely present, False if definately not"""
         idx_1, idx_2, fingerprint = self._generate_fingerprint_info(key)
         is_present = self._check_if_present(idx_1, idx_2, fingerprint)
-        if is_present is not None:
-            return True
-        return False
+        return is_present is not None
 
     def remove(self, key: KeyT) -> bool:
         """Remove an element from the filter

tests/bloom_test.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-# -*- coding: utf-8 -*-
 """ Unittest class """
 
 import hashlib
@@ -13,11 +12,11 @@
 sys.path.insert(0, str(this_dir))
 sys.path.insert(0, str(this_dir.parent))
 
-from probables import BloomFilter, BloomFilterOnDisk
-from probables.constants import UINT64_T_MAX
-from probables.exceptions import InitializationError, NotSupportedError
-from probables.hashes import hash_with_depth_int
-from tests.utilities import calc_file_md5, different_hash
+from probables import BloomFilter, BloomFilterOnDisk  # noqa: E402
+from probables.constants import UINT64_T_MAX  # noqa: E402
+from probables.exceptions import InitializationError, NotSupportedError  # noqa: E402
+from probables.hashes import hash_with_depth_int  # noqa: E402
+from tests.utilities import calc_file_md5, different_hash  # noqa: E402
 
 DELETE_TEMP_FILES = True
 
@@ -252,7 +251,7 @@ def test_bf_stats(self):
         )
         blm = BloomFilter(est_elements=10, false_positive_rate=0.05)
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
         stats = str(blm)
         self.assertEqual(stats, msg)
@@ -262,7 +261,7 @@ def test_bf_export_hex(self):
         hex_val = "6da491461a6bba4d000000000000000a000000000000000a3d4ccccd"
         blm = BloomFilter(est_elements=10, false_positive_rate=0.05)
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
         hex_out = blm.export_hex()
 
@@ -295,23 +294,23 @@ def test_bf_export_c_header(self):
         hex_val = "6da491461a6bba4d000000000000000a000000000000000a3d4ccccd"
         blm = BloomFilter(est_elements=10, false_positive_rate=0.05)
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
         with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:
             blm.export_c_header(fobj.name)
 
             # now load the file, parse it and do some tests!
-            with open(fobj.name, "r") as fobj:
+            with open(fobj.name) as fobj:
                 data = fobj.readlines()
         data = [x.strip() for x in data]
 
         self.assertEqual("/* BloomFilter Export of a standard BloomFilter */", data[0])
         self.assertEqual("#include <inttypes.h>", data[1])
-        self.assertEqual("const uint64_t estimated_elements = {};".format(blm.estimated_elements), data[2])
-        self.assertEqual("const uint64_t elements_added = {};".format(blm.elements_added), data[3])
-        self.assertEqual("const float false_positive_rate = {};".format(blm.false_positive_rate), data[4])
-        self.assertEqual("const uint64_t number_bits = {};".format(blm.number_bits), data[5])
-        self.assertEqual("const unsigned int number_hashes = {};".format(blm.number_hashes), data[6])
+        self.assertEqual(f"const uint64_t estimated_elements = {blm.estimated_elements};", data[2])
+        self.assertEqual(f"const uint64_t elements_added = {blm.elements_added};", data[3])
+        self.assertEqual(f"const float false_positive_rate = {blm.false_positive_rate};", data[4])
+        self.assertEqual(f"const uint64_t number_bits = {blm.number_bits};", data[5])
+        self.assertEqual(f"const unsigned int number_hashes = {blm.number_hashes};", data[6])
         self.assertEqual("const unsigned char bloom[] = {", data[7])
         self.assertEqual("};", data[-1])
 
@@ -481,7 +480,7 @@ def test_bf_clear(self):
         blm = BloomFilter(est_elements=10, false_positive_rate=0.05)
         self.assertEqual(blm.elements_added, 0)
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
         self.assertEqual(blm.elements_added, 10)
 
@@ -518,13 +517,13 @@ def my_hash(key, depth=1, encoding="utf-8"):
         self.assertNotEqual(md5_out, md5_val)
 
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
 
         self.assertEqual(blm.elements_added, 11)
 
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             self.assertTrue(blm.check(tmp))
 
         self.assertEqual(blm.hashes("this is a test", 5), results)
@@ -563,13 +562,13 @@ def my_hash(key, depth, encoding="utf-8"):
         self.assertNotEqual(md5_out, md5_val)
 
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
 
         self.assertEqual(blm.elements_added, 11)
 
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             self.assertTrue(blm.check(tmp))
 
         self.assertEqual(blm.hashes("this is a test", 5), results)
@@ -714,7 +713,7 @@ def test_bfod_close_del(self):
             blm.add("this is a test")
             del blm
             try:
-                self.assertEqual(True, blm)
+                self.assertEqual(True, blm)  # noqa: F821
             except UnboundLocalError as ex:
                 msg1 = "local variable 'blm' referenced before assignment"
                 msg2 = "cannot access local variable 'blm' where it is not associated with a value"
@@ -731,7 +730,7 @@ def test_bfod_close_del(self):
     # export to new file
     def test_bfod_export(self):
         """export to on disk to new file"""
-        with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:
+        with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:  # noqa: SIM117
             with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj1:
                 blm = BloomFilterOnDisk(fobj.name, 10, 0.05)
                 blm.add("this is a test")
@@ -781,7 +780,7 @@ def test_bfod_export_hex(self):
         with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:
             blm = BloomFilterOnDisk(fobj.name, est_elements=10, false_positive_rate=0.05)
             for i in range(0, 10):
-                tmp = "this is a test {0}".format(i)
+                tmp = f"this is a test {i}"
                 blm.add(tmp)
             hex_out = blm.export_hex()
             self.assertEqual(hex_out, hex_val)
@@ -813,24 +812,24 @@ def test_bfod_export_c_header(self):
         with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:
             blm = BloomFilterOnDisk(fobj.name, est_elements=10, false_positive_rate=0.05)
             for i in range(0, 10):
-                tmp = "this is a test {0}".format(i)
+                tmp = f"this is a test {i}"
                 blm.add(tmp)
             with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:
                 blm.export_c_header(fobj.name)
 
                 # now load the file, parse it and do some tests!
-                with open(fobj.name, "r") as fobj:
+                with open(fobj.name) as fobj:
                     data = fobj.readlines()
 
         data = [x.strip() for x in data]
 
         self.assertEqual("/* BloomFilter Export of a standard BloomFilter */", data[0])
         self.assertEqual("#include <inttypes.h>", data[1])
-        self.assertEqual("const uint64_t estimated_elements = {};".format(blm.estimated_elements), data[2])
-        self.assertEqual("const uint64_t elements_added = {};".format(blm.elements_added), data[3])
-        self.assertEqual("const float false_positive_rate = {};".format(blm.false_positive_rate), data[4])
-        self.assertEqual("const uint64_t number_bits = {};".format(blm.number_bits), data[5])
-        self.assertEqual("const unsigned int number_hashes = {};".format(blm.number_hashes), data[6])
+        self.assertEqual(f"const uint64_t estimated_elements = {blm.estimated_elements};", data[2])
+        self.assertEqual(f"const uint64_t elements_added = {blm.elements_added};", data[3])
+        self.assertEqual(f"const float false_positive_rate = {blm.false_positive_rate};", data[4])
+        self.assertEqual(f"const uint64_t number_bits = {blm.number_bits};", data[5])
+        self.assertEqual(f"const unsigned int number_hashes = {blm.number_hashes};", data[6])
         self.assertEqual("const unsigned char bloom[] = {", data[7])
         self.assertEqual("};", data[-1])
 
@@ -844,7 +843,7 @@ def test_bfod_clear(self):
             blm = BloomFilterOnDisk(filepath=fobj.name, est_elements=10, false_positive_rate=0.05)
             self.assertEqual(blm.elements_added, 0)
             for i in range(0, 10):
-                tmp = "this is a test {0}".format(i)
+                tmp = f"this is a test {i}"
                 blm.add(tmp)
             self.assertEqual(blm.elements_added, 10)
 

tests/countingbloom_test.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-# -*- coding: utf-8 -*-
 """ Unittest class """
 
 import hashlib
@@ -13,9 +12,9 @@
 sys.path.insert(0, str(this_dir))
 sys.path.insert(0, str(this_dir.parent))
 
-from probables import CountingBloomFilter
-from probables.exceptions import InitializationError
-from tests.utilities import calc_file_md5, different_hash
+from probables import CountingBloomFilter  # noqa: E402
+from probables.exceptions import InitializationError  # noqa: E402
+from tests.utilities import calc_file_md5, different_hash  # noqa: E402
 
 DELETE_TEMP_FILES = True
 
@@ -87,7 +86,7 @@ def test_cbf_stats(self):
         )
         blm = CountingBloomFilter(est_elements=10, false_positive_rate=0.05)
         for i in range(0, 10):
-            blm.add("this is a test {0}".format(i))
+            blm.add(f"this is a test {i}")
         stats = str(blm)
         self.assertEqual(stats, msg)
 
@@ -96,7 +95,7 @@ def test_cbf_clear(self):
         blm = CountingBloomFilter(est_elements=10, false_positive_rate=0.05)
         self.assertEqual(blm.elements_added, 0)
         for i in range(0, 10):
-            blm.add("this is a test {0}".format(i))
+            blm.add(f"this is a test {i}")
         self.assertEqual(blm.elements_added, 10)
 
         blm.clear()
@@ -215,7 +214,7 @@ def test_cbf_export_hex(self):
 
         blm = CountingBloomFilter(est_elements=10, false_positive_rate=0.05)
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
         hex_out = blm.export_hex()
 
@@ -277,25 +276,25 @@ def test_cbf_export_c_header(self):
         )
         blm = CountingBloomFilter(est_elements=10, false_positive_rate=0.05)
         for i in range(0, 10):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
 
         with NamedTemporaryFile(dir=os.getcwd(), suffix=".blm", delete=DELETE_TEMP_FILES) as fobj:
             blm.export_c_header(fobj.name)
 
             # now load the file, parse it and do some tests!
-            with open(fobj.name, "r") as fobj:
+            with open(fobj.name) as fobj:
                 data = fobj.readlines()
 
         data = [x.strip() for x in data]
 
         self.assertEqual("/* BloomFilter Export of a CountingBloomFilter */", data[0])
         self.assertEqual("#include <inttypes.h>", data[1])
-        self.assertEqual("const uint64_t estimated_elements = {};".format(blm.estimated_elements), data[2])
-        self.assertEqual("const uint64_t elements_added = {};".format(blm.elements_added), data[3])
-        self.assertEqual("const float false_positive_rate = {};".format(blm.false_positive_rate), data[4])
-        self.assertEqual("const uint64_t number_bits = {};".format(blm.number_bits), data[5])
-        self.assertEqual("const unsigned int number_hashes = {};".format(blm.number_hashes), data[6])
+        self.assertEqual(f"const uint64_t estimated_elements = {blm.estimated_elements};", data[2])
+        self.assertEqual(f"const uint64_t elements_added = {blm.elements_added};", data[3])
+        self.assertEqual(f"const float false_positive_rate = {blm.false_positive_rate};", data[4])
+        self.assertEqual(f"const uint64_t number_bits = {blm.number_bits};", data[5])
+        self.assertEqual(f"const unsigned int number_hashes = {blm.number_hashes};", data[6])
         self.assertEqual("const unsigned char bloom[] = {", data[7])
         self.assertEqual("};", data[-1])
 
@@ -410,7 +409,7 @@ def test_cbf_remove(self):
         blm = CountingBloomFilter(est_elements=10, false_positive_rate=0.05)
         self.assertEqual(blm.elements_added, 0)
         for i in range(0, 5):
-            tmp = "this is a test {0}".format(i)
+            tmp = f"this is a test {i}"
             blm.add(tmp)
         self.assertEqual(blm.elements_added, 5)
         res = blm.remove("this is a test 0")

tests/countingcuckoo_test.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-# -*- coding: utf-8 -*-
 """ Unittest class """
 
 import hashlib
@@ -13,8 +12,8 @@
 sys.path.insert(0, str(this_dir))
 sys.path.insert(0, str(this_dir.parent))
 
-from probables import CountingCuckooFilter, CuckooFilterFullError
-from tests.utilities import calc_file_md5
+from probables import CountingCuckooFilter, CuckooFilterFullError  # noqa: E402
+from tests.utilities import calc_file_md5  # noqa: E402
 
 DELETE_TEMP_FILES = True
 

tests/countminsketch_test.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-# -*- coding: utf-8 -*-
 """ Unittest class """
 
 import hashlib
@@ -13,10 +12,10 @@
 sys.path.insert(0, str(this_dir))
 sys.path.insert(0, str(this_dir.parent))
 
-from probables import CountMeanMinSketch, CountMeanSketch, CountMinSketch, HeavyHitters, StreamThreshold
-from probables.constants import INT32_T_MAX, INT32_T_MIN, INT64_T_MAX, INT64_T_MIN
-from probables.exceptions import CountMinSketchError, InitializationError, NotSupportedError
-from tests.utilities import calc_file_md5, different_hash
+from probables import CountMeanMinSketch, CountMeanSketch, CountMinSketch, HeavyHitters, StreamThreshold  # noqa: E402
+from probables.constants import INT32_T_MAX, INT32_T_MIN, INT64_T_MAX, INT64_T_MIN  # noqa: E402
+from probables.exceptions import CountMinSketchError, InitializationError, NotSupportedError  # noqa: E402
+from tests.utilities import calc_file_md5, different_hash  # noqa: E402
 
 DELETE_TEMP_FILES = True
 

tests/cuckoo_test.py

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-# -*- coding: utf-8 -*-
 """ Unittest class """
 
 import hashlib
@@ -13,8 +12,8 @@
 sys.path.insert(0, str(this_dir))
 sys.path.insert(0, str(this_dir.parent))
 
-from probables import CuckooFilter, CuckooFilterFullError, InitializationError
-from tests.utilities import calc_file_md5
+from probables import CuckooFilter, CuckooFilterFullError, InitializationError  # noqa: E402
+from tests.utilities import calc_file_md5  # noqa: E402
 
 DELETE_TEMP_FILES = True
 
@@ -77,10 +76,10 @@ def my_hash(key):
             hash_function=my_hash,
         )
         for i in range(50):
-            cko.add("this is a test - {}".format(i))
+            cko.add(f"this is a test - {i}")
 
         for i in range(50):
-            self.assertTrue("this is a test - {}".format(i) in cko)
+            self.assertTrue(f"this is a test - {i}" in cko)
 
     def test_cuckoo_filter_remove(self):
         """test removing from the cuckoo filter"""