Move from MD5 to CRC32 for hashing test IDs

CHANGELOG.rst

@@ -14,6 +14,10 @@ Unreleased
   `PR #687 <https://github.com/pytest-dev/pytest-randomly/issues/687>`__.
   Thanks to Bryce Drennan for the suggestion in `Issue #600 <https://github.com/pytest-dev/pytest-randomly/issues/600>`__ and initial implementation in `PR #617 <https://github.com/pytest-dev/pytest-randomly/pull/617>`__.
 
+* Move from MD5 to CRC32 for hashing test IDs, as it’s 5x faster and we don’t need cryptographic security.
+
+  `Issue #686 <https://github.com/pytest-dev/pytest-randomly/issues/686>`__.
+
 3.16.0 (2024-10-25)
 -------------------
 

src/pytest_randomly/__init__.py

@@ -8,6 +8,7 @@
 from itertools import groupby
 from types import ModuleType
 from typing import Any, Callable, TypeVar
+from zlib import crc32
 
 from _pytest.config import Config
 from _pytest.config.argparsing import Parser
@@ -198,17 +199,17 @@ def pytest_report_header(config: Config) -> str:
 
 def pytest_runtest_setup(item: Item) -> None:
     if item.config.getoption("randomly_reset_seed"):
-        _reseed(item.config, int.from_bytes(_md5(item.nodeid), "big") - 1)
+        _reseed(item.config, _crc32(item.nodeid) - 1)
 
 
 def pytest_runtest_call(item: Item) -> None:
     if item.config.getoption("randomly_reset_seed"):
-        _reseed(item.config, int.from_bytes(_md5(item.nodeid), "big"))
+        _reseed(item.config, _crc32(item.nodeid))
 
 
 def pytest_runtest_teardown(item: Item) -> None:
     if item.config.getoption("randomly_reset_seed"):
-        _reseed(item.config, int.from_bytes(_md5(item.nodeid), "big") + 1)
+        _reseed(item.config, _crc32(item.nodeid) + 1)
 
 
 @hookimpl(tryfirst=True)
@@ -227,11 +228,11 @@ def pytest_collection_modifyitems(config: Config, items: list[Item]) -> None:
             )
         )
 
-    def _module_key(module_item: tuple[ModuleType | None, list[Item]]) -> bytes:
+    def _module_key(module_item: tuple[ModuleType | None, list[Item]]) -> int:
         module, _items = module_item
         if module is None:
-            return _md5(f"{seed}::None")
-        return _md5(f"{seed}::{module.__name__}")
+            return _crc32(f"{seed}::None")
+        return _crc32(f"{seed}::{module.__name__}")
 
     modules_items.sort(key=_module_key)
 
@@ -248,19 +249,19 @@ def _get_module(item: Item) -> ModuleType | None:
 def _shuffle_by_class(items: list[Item], seed: int) -> list[Item]:
     klasses_items: list[tuple[type[Any] | None, list[Item]]] = []
 
-    def _item_key(item: Item) -> bytes:
-        return _md5(f"{seed}::{item.nodeid}")
+    def _item_key(item: Item) -> int:
+        return _crc32(f"{seed}::{item.nodeid}")
 
     for klass, group in groupby(items, _get_cls):
         klass_items = list(group)
         klass_items.sort(key=_item_key)
         klasses_items.append((klass, klass_items))
 
-    def _cls_key(klass_items: tuple[type[Any] | None, list[Item]]) -> bytes:
+    def _cls_key(klass_items: tuple[type[Any] | None, list[Item]]) -> int:
         klass, items = klass_items
         if klass is None:
-            return _md5(f"{seed}::None")
-        return _md5(f"{seed}::{klass.__module__}.{klass.__qualname__}")
+            return _crc32(f"{seed}::None")
+        return _crc32(f"{seed}::{klass.__module__}.{klass.__qualname__}")
 
     klasses_items.sort(key=_cls_key)
 
@@ -282,19 +283,15 @@ def reduce_list_of_lists(lists: list[list[T]]) -> list[T]:
 
 
 @lru_cache
-def _md5(string: str) -> bytes:
-    hasher = hashlib.md5(usedforsecurity=False)
-    hasher.update(string.encode())
-    return hasher.digest()
+def _crc32(string: str) -> int:
+    return crc32(string.encode())
 
 
 if have_faker:  # pragma: no branch
 
     @fixture(autouse=True)
     def faker_seed(pytestconfig: Config, request: SubRequest) -> int:
-        print(type(request))
-        result: int = pytestconfig.getoption("randomly_seed") + int.from_bytes(
-            _md5(request.node.nodeid),
-            "big",
+        result: int = pytestconfig.getoption("randomly_seed") + _crc32(
+            request.node.nodeid
         )
         return result

tests/test_pytest_randomly.py

@@ -248,10 +248,10 @@ def test_it():
 
     out.assert_outcomes(passed=4, failed=0)
     assert out.outlines[9:13] == [
+        "test_c.py::test_it PASSED",
         "test_b.py::test_it PASSED",
-        "test_a.py::test_it PASSED",
         "test_d.py::test_it PASSED",
-        "test_c.py::test_it PASSED",
+        "test_a.py::test_it PASSED",
     ]
 
 
@@ -268,10 +268,10 @@ def test_it():
 
     out.assert_outcomes(passed=4, failed=0)
     assert out.outlines[9:13] == [
+        "test_c.py::test_it PASSED",
         "test_b.py::test_it PASSED",
-        "test_a.py::test_it PASSED",
         "test_d.py::test_it PASSED",
-        "test_c.py::test_it PASSED",
+        "test_a.py::test_it PASSED",
     ]
 
 
@@ -308,9 +308,9 @@ def test_d(self):
     out.assert_outcomes(passed=4, failed=0)
     assert out.outlines[9:13] == [
         "test_one.py::D::test_d PASSED",
-        "test_one.py::B::test_b PASSED",
-        "test_one.py::C::test_c PASSED",
         "test_one.py::A::test_a PASSED",
+        "test_one.py::C::test_c PASSED",
+        "test_one.py::B::test_b PASSED",
     ]
 
 
@@ -341,8 +341,8 @@ def test_d(self):
     assert out.outlines[9:13] == [
         "test_one.py::T::test_c PASSED",
         "test_one.py::T::test_b PASSED",
-        "test_one.py::T::test_a PASSED",
         "test_one.py::T::test_d PASSED",
+        "test_one.py::T::test_a PASSED",
     ]
 
 
@@ -368,10 +368,10 @@ def test_d():
 
     out.assert_outcomes(passed=4, failed=0)
     assert out.outlines[9:13] == [
-        "test_one.py::test_c PASSED",
+        "test_one.py::test_d PASSED",
         "test_one.py::test_a PASSED",
+        "test_one.py::test_c PASSED",
         "test_one.py::test_b PASSED",
-        "test_one.py::test_d PASSED",
     ]
 
 
@@ -402,10 +402,10 @@ def test_d():
 
     out.assert_outcomes(passed=4, failed=0)
     assert out.outlines[9:13] == [
-        "test_one.py::test_c PASSED",
+        "test_one.py::test_d PASSED",
         "test_one.py::test_a PASSED",
+        "test_one.py::test_c PASSED",
         "test_one.py::test_b PASSED",
-        "test_one.py::test_d PASSED",
     ]
 
 
@@ -528,7 +528,15 @@ def test_b():
             assert 0
         """
     )
-    out = ourtester.runpytest("-v", "--randomly-seed=1", "--stepwise")
+    out = ourtester.runpytest("-v", "--randomly-seed=8")
+    out.assert_outcomes(failed=2)
+    # Ensure test_b runs first
+    assert out.outlines[9:11] == [
+        "test_one.py::test_b FAILED",
+        "test_one.py::test_a FAILED",
+    ]
+
+    out = ourtester.runpytest("--randomly-seed=8", "--stepwise")
     out.assert_outcomes(failed=1)
 
     # Now make test_b pass
@@ -543,9 +551,9 @@ def test_b():
         """
     )
     shutil.rmtree(ourtester.path / "__pycache__")
-    out = ourtester.runpytest("-v", "--randomly-seed=1", "--stepwise")
+    out = ourtester.runpytest("--randomly-seed=8", "--stepwise")
     out.assert_outcomes(passed=1, failed=1)
-    out = ourtester.runpytest("-v", "--randomly-seed=1", "--stepwise")
+    out = ourtester.runpytest("--randomly-seed=8", "--stepwise")
     out.assert_outcomes(failed=1)
 
 
@@ -579,11 +587,11 @@ def test_factory_boy(ourtester):
         from factory.random import randgen
 
         def test_a():
-            assert randgen.random() == 0.9988532989147809
+            assert randgen.random() == 0.17867277194477893
 
 
         def test_b():
-            assert randgen.random() == 0.18032546798434612
+            assert randgen.random() == 0.8026272812225962
         """
     )
 
@@ -599,10 +607,10 @@ def test_faker(ourtester):
         fake = Faker()
 
         def test_one():
-            assert fake.name() == 'Mrs. Lisa Ryan'
+            assert fake.name() == 'Kimberly Powell'
 
         def test_two():
-            assert fake.name() == 'Kaitlyn Mitchell'
+            assert fake.name() == 'Thomas Moyer PhD'
         """
     )
 
@@ -614,10 +622,10 @@ def test_faker_fixture(ourtester):
     ourtester.makepyfile(
         test_one="""
         def test_one(faker):
-            assert faker.name() == 'Mrs. Lisa Ryan'
+            assert faker.name() == 'Kimberly Powell'
 
         def test_two(faker):
-            assert faker.name() == 'Kaitlyn Mitchell'
+            assert faker.name() == 'Thomas Moyer PhD'
         """
     )
 
@@ -634,10 +642,10 @@ def test_model_bakery(ourtester):
         from model_bakery.random_gen import gen_slug
 
         def test_a():
-            assert gen_slug(10) == 'XjpU5br7ej'
+            assert gen_slug(10) == 'whwhAKeQYE'
 
         def test_b():
-            assert gen_slug(10) == 'xJHS-PD_WT'
+            assert gen_slug(10) == 'o2N4p5UAXd'
         """
     )
 
@@ -651,10 +659,10 @@ def test_numpy(ourtester):
         import numpy as np
 
         def test_one():
-            assert np.random.rand() == 0.36687834264514585
+            assert np.random.rand() == 0.1610140063074521
 
         def test_two():
-            assert np.random.rand() == 0.7050715833365834
+            assert np.random.rand() == 0.6896867238957805
         """
     )
 
@@ -718,19 +726,19 @@ def fake_entry_points(*, group):
     assert reseed.mock_calls == [
         mock.call(1),
         mock.call(1),
-        mock.call(116362448262735926321257785636175308268),
-        mock.call(116362448262735926321257785636175308269),
-        mock.call(116362448262735926321257785636175308270),
+        mock.call(2964001072),
+        mock.call(2964001073),
+        mock.call(2964001074),
     ]
 
     reseed.mock_calls[:] = []
     pytester.runpytest_inprocess("--randomly-seed=424242")
     assert reseed.mock_calls == [
         mock.call(424242),
         mock.call(424242),
-        mock.call(116362448262735926321257785636175732509),
-        mock.call(116362448262735926321257785636175732510),
-        mock.call(116362448262735926321257785636175732511),
+        mock.call(2964425313),
+        mock.call(2964425314),
+        mock.call(2964425315),
     ]