Fix test hypothesis (#4927)

* fix:test-hypothesis

---------

Co-authored-by: Roman Gershman <[email protected]>

Author: cunla

.github/workflows/test-fakeredis.yml

@@ -24,16 +24,6 @@ jobs:
     permissions:
       pull-requests: write
       checks: read
-    # services:
-    #   redis:
-    #     image: docker.dragonflydb.io/dragonflydb/dragonfly:${{ matrix.DRAGONFLY_VERSION }}
-    #     ports:
-    #       - 6380:6379
-    #     options: >-
-    #       --health-cmd "redis-cli ping"
-    #       --health-interval 10s
-    #       --health-timeout 5s
-    #       --health-retries 5
 
     steps:
       - uses: actions/checkout@v4
@@ -69,14 +59,8 @@ jobs:
       - name: Run tests
         working-directory: tests/fakeredis
         run: |
-          poetry run pytest test/ \
-            --ignore test/test_hypothesis/test_hash.py  \
-            --ignore test/test_hypothesis/test_set.py  \
-            --ignore test/test_hypothesis/test_zset.py  \
-            --ignore test/test_hypothesis/test_joint.py  \
-            --ignore test/test_hypothesis/test_transaction.py  \
-            --ignore test/test_mixins/test_bitmap_commands.py \
-            --junit-xml=results-tests.xml  --html=report-tests.html -v
+          poetry run pytest -s test/ \
+          --junit-xml=results-tests.xml  --html=report-tests.html -v
         continue-on-error: true  # For now to mark the flow as successful
 
       - name: Show Dragonfly stats

tests/fakeredis/poetry.lock

@@ -25,7 +25,7 @@ pytest = "^8.3"
 pytest-timeout = "^2.3.1"
 pytest-asyncio = "^0.24"
 pytest-cov = "^5.0"
-pytest-mock =  "^3.14"
+pytest-mock = "^3.14"
 pytest-html = "^4.1"
 
 [tool.pytest.ini_options]
@@ -45,4 +45,5 @@ generate_report_on_test = true
 render_collapsed = "failed,error"
 addopts = [
     "--self-contained-html",
+    "--import-mode=importlib",
 ]

tests/fakeredis/pyproject.toml

@@ -0,0 +1,33 @@
+import hypothesis.strategies as st
+
+from .. import test_hypothesis as tests
+from ..test_hypothesis.base import BaseTest, common_commands, commands
+from ..test_hypothesis.test_string import string_commands
+
+bad_commands = (
+    # redis-py splits the command on spaces, and hangs if that ends up being an empty list
+    commands(
+        st.text().filter(lambda x: bool(x.split())), st.lists(st.binary() | st.text())
+    )
+)
+
+
+class TestJoint(BaseTest):
+    create_command_strategy = (
+        tests.TestString.create_command_strategy
+        | tests.TestHash.create_command_strategy
+        | tests.TestList.create_command_strategy
+        | tests.TestSet.create_command_strategy
+        | tests.TestZSet.create_command_strategy
+    )
+    command_strategy = (
+        tests.TestServer.server_commands
+        | tests.TestConnection.connection_commands
+        | string_commands
+        | tests.TestHash.hash_commands
+        | tests.TestList.list_commands
+        | tests.TestSet.set_commands
+        | tests.TestZSet.zset_commands
+        | common_commands
+        | bad_commands
+    )

tests/fakeredis/test/test_hypotesis_joint/__init__.py

@@ -0,0 +1,19 @@
+__all__ = [
+    "TestConnection",
+    "TestHash",
+    "TestList",
+    "TestServer",
+    "TestSet",
+    "TestString",
+    "TestTransaction",
+    "TestZSet",
+]
+
+from .test_connection import TestConnection
+from .test_hash import TestHash
+from .test_list import TestList
+from .test_server import TestServer
+from .test_set import TestSet
+from .test_string import TestString
+from .test_transaction import TestTransaction
+from .test_zset import TestZSet

tests/fakeredis/test/test_hypotesis_joint/test_joint.py

@@ -1,5 +1,6 @@
 import functools
 import math
+import string
 import sys
 from typing import Any, List, Tuple, Type, Optional
 
@@ -34,11 +35,15 @@ def sample_attr(draw, name):
 values = sample_attr("values")
 scores = sample_attr("scores")
 
+eng_text = st.builds(
+    lambda x: x.encode(), st.text(alphabet=string.ascii_letters, min_size=1)
+)
 ints = st.integers(min_value=MIN_INT, max_value=MAX_INT)
 int_as_bytes = st.builds(lambda x: str(_default_normalize(x)).encode(), ints)
-float_as_bytes = st.builds(
-    lambda x: repr(_default_normalize(x)).encode(), st.floats(width=32)
+floats = st.floats(
+    width=32, allow_nan=False, allow_subnormal=False, allow_infinity=False
 )
+float_as_bytes = st.builds(lambda x: repr(_default_normalize(x)).encode(), floats)
 counts = st.integers(min_value=-3, max_value=3) | ints
 # Redis has an integer overflow bug in swapdb, so we confine the numbers to
 # a limited range (https://github.com/antirez/redis/issues/5737).
@@ -51,8 +56,8 @@ def sample_attr(draw, name):
 ) | st.binary().filter(lambda x: b"\0" not in x)
 
 # Redis has integer overflow bugs in time computations, which is why we set a maximum.
-expires_seconds = st.integers(min_value=100000, max_value=MAX_INT)
-expires_ms = st.integers(min_value=100000000, max_value=MAX_INT)
+expires_seconds = st.integers(min_value=5, max_value=1_000)
+expires_ms = st.integers(min_value=5_000, max_value=50_000)
 
 
 class WrappedException:
@@ -98,6 +103,8 @@ def _sort_list(lst):
 
 
 def _normalize_if_number(x):
+    if isinstance(x, list):
+        return [_normalize_if_number(i) for i in x]
     try:
         res = float(x)
         return x if math.isnan(res) else res
@@ -149,6 +156,7 @@ def normalize(self):
             b"sinter",
             b"sunion",
             b"smembers",
+            b"hexpire",
         }
         if command in unordered:
             return _sort_list
@@ -159,7 +167,7 @@ def normalize(self):
     def testable(self) -> bool:
         """Whether this command is suitable for a test.
 
-        The fuzzer can create commands with behaviour that is non-deterministic, not supported, or which hits redis bugs.
+        The fuzzer can create commands with behavior that is non-deterministic, not supported, or which hits redis bugs.
         """
         N = len(self.args)
         if N == 0:
@@ -201,20 +209,6 @@ def commands(*args, **kwargs):
     | commands(st.just("sort"), keys, *zero_or_more("asc", "desc", "alpha"))
 )
 
-attrs = st.fixed_dictionaries(
-    {
-        "keys": st.lists(st.binary(), min_size=2, max_size=5, unique=True),
-        "fields": st.lists(st.binary(), min_size=2, max_size=5, unique=True),
-        "values": st.lists(
-            st.binary() | int_as_bytes | float_as_bytes,
-            min_size=2,
-            max_size=5,
-            unique=True,
-        ),
-        "scores": st.lists(st.floats(width=32), min_size=2, max_size=5, unique=True),
-    }
-)
-
 
 @hypothesis.settings(max_examples=1000)
 class CommonMachine(hypothesis.stateful.RuleBasedStateMachine):
@@ -291,6 +285,16 @@ def _compare(self, command: Command) -> None:
             for n, r, f in zip(self.transaction_normalize, real_result, fake_result):
                 assert n(f) == n(r)
             self.transaction_normalize = []
+        elif isinstance(fake_result, list):
+            assert len(fake_result) == len(real_result), (
+                f"Discrepancy when running command {command}, fake({fake_result}) != real({real_result})",
+            )
+            for i in range(len(fake_result)):
+                assert fake_result[i] == real_result[i] or (
+                    type(fake_result[i]) is float
+                    and fake_result[i] == pytest.approx(real_result[i])
+                ), f"Discrepancy when running command {command}, fake({fake_result}) != real({real_result})"
+
         else:
             assert fake_result == real_result or (
                 type(fake_result) is float and fake_result == pytest.approx(real_result)
@@ -307,7 +311,26 @@ def _compare(self, command: Command) -> None:
         ):
             self.transaction_normalize = []
 
-    @initialize(attrs=attrs)
+    @initialize(
+        attrs=st.fixed_dictionaries(
+            dict(
+                keys=st.lists(eng_text, min_size=2, max_size=5, unique=True),
+                fields=st.lists(eng_text, min_size=2, max_size=5, unique=True),
+                values=st.lists(
+                    eng_text | int_as_bytes | float_as_bytes,
+                    min_size=2,
+                    max_size=5,
+                    unique=True,
+                ),
+                scores=st.lists(
+                    floats,
+                    min_size=2,
+                    max_size=5,
+                    unique=True,
+                ),
+            )
+        )
+    )
     def init_attrs(self, attrs):
         for key, value in attrs.items():
             setattr(self, key, value)

tests/fakeredis/test/test_hypothesis/__init__.py

@@ -1,6 +1,6 @@
 import hypothesis.strategies as st
 
-from test.test_hypothesis.base import BaseTest, commands, values, common_commands
+from .base import BaseTest, commands, values, common_commands
 
 
 class TestConnection(BaseTest):

tests/fakeredis/test/test_hypothesis/base.py

@@ -1,6 +1,6 @@
 import hypothesis.strategies as st
 
-from test.test_hypothesis.base import (
+from .base import (
     BaseTest,
     commands,
     values,
@@ -37,7 +37,7 @@ class TestHash(BaseTest):
             expires_seconds,
             st.just("fields"),
             st.just(2),
-            st.lists(fields, min_size=2, max_size=2),
+            st.lists(fields, min_size=2, max_size=2, unique=True),
         )
     )
     create_command_strategy = commands(