Merge branch 'master' of github.com:mongodb/mongo-python-driver

blink1073 · blink1073 · commit e91b817ecbba · 2025-09-17T14:24:33.000-05:00
diff --git a/.evergreen/generated_configs/variants.yml b/.evergreen/generated_configs/variants.yml
@@ -79,17 +79,6 @@ buildvariants:
       TEST_NAME: atlas_connect
     tags: [pr]
 
-  # Atlas data lake tests
-  - name: atlas-data-lake-ubuntu-22
-    tasks:
-      - name: .test-no-orchestration
-    display_name: Atlas Data Lake Ubuntu-22
-    run_on:
-      - ubuntu2204-small
-    expansions:
-      TEST_NAME: data_lake
-    tags: [pr]
-
   # Aws auth tests
   - name: auth-aws-ubuntu-20
     tasks:
diff --git a/.evergreen/resync-specs.sh b/.evergreen/resync-specs.sh
@@ -76,9 +76,6 @@ do
     auth)
       cpjson auth/tests/ auth
       ;;
-    atlas-data-lake-testing|data_lake)
-      cpjson atlas-data-lake-testing/tests/ data_lake
-      ;;
     bson-binary-vector|bson_binary_vector)
       cpjson bson-binary-vector/tests/ bson_binary_vector
       ;;
diff --git a/.evergreen/run-tests.sh b/.evergreen/run-tests.sh
@@ -25,11 +25,7 @@ else
   exit 1
 fi
 
-# List the packages.
-uv sync ${UV_ARGS} --reinstall --quiet
-uv pip list
-
 # Start the test runner.
-uv run ${UV_ARGS} .evergreen/scripts/run_tests.py "$@"
+uv run ${UV_ARGS} --reinstall .evergreen/scripts/run_tests.py "$@"
 
 popd
diff --git a/.evergreen/scripts/generate_config.py b/.evergreen/scripts/generate_config.py
@@ -322,14 +322,6 @@ def create_no_c_ext_variants():
     return [create_variant(tasks, display_name, host=host)]
 
 
-def create_atlas_data_lake_variants():
-    host = HOSTS["ubuntu22"]
-    tasks = [".test-no-orchestration"]
-    expansions = dict(TEST_NAME="data_lake")
-    display_name = get_variant_name("Atlas Data Lake", host)
-    return [create_variant(tasks, display_name, tags=["pr"], host=host, expansions=expansions)]
-
-
 def create_mod_wsgi_variants():
     host = HOSTS["ubuntu22"]
     tasks = [".mod_wsgi"]
diff --git a/.evergreen/scripts/generate_config_utils.py b/.evergreen/scripts/generate_config_utils.py
@@ -273,7 +273,7 @@ def generate_yaml(tasks=None, variants=None):
     out = ShrubService.generate_yaml(project)
     # Dedent by two spaces to match what we use in config.yml
     lines = [line[2:] for line in out.splitlines()]
-    print("\n".join(lines))  # noqa: T201
+    print("\n".join(lines))
 
 
 ##################
diff --git a/.evergreen/scripts/resync-all-specs.py b/.evergreen/scripts/resync-all-specs.py
@@ -11,7 +11,7 @@
 
 def resync_specs(directory: pathlib.Path, errored: dict[str, str]) -> None:
     """Actually sync the specs"""
-    print("Beginning to sync specs")  # noqa: T201
+    print("Beginning to sync specs")
     for spec in os.scandir(directory):
         if not spec.is_dir():
             continue
@@ -27,11 +27,11 @@ def resync_specs(directory: pathlib.Path, errored: dict[str, str]) -> None:
             )
         except CalledProcessError as exc:
             errored[spec.name] = exc.stderr
-    print("Done syncing specs")  # noqa: T201
+    print("Done syncing specs")
 
 
 def apply_patches():
-    print("Beginning to apply patches")  # noqa: T201
+    print("Beginning to apply patches")
     subprocess.run(["bash", "./.evergreen/remove-unimplemented-tests.sh"], check=True)  # noqa: S603, S607
     subprocess.run(
         ["git apply -R --allow-empty --whitespace=fix ./.evergreen/spec-patch/*"],  # noqa: S607
@@ -56,7 +56,6 @@ def check_new_spec_directories(directory: pathlib.Path) -> list[str]:
         "client_side_operations_timeout": "csot",
         "mongodb_handshake": "handshake",
         "load_balancers": "load_balancer",
-        "atlas_data_lake_testing": "atlas",
         "connection_monitoring_and_pooling": "connection_monitoring",
         "command_logging_and_monitoring": "command_logging",
         "initial_dns_seedlist_discovery": "srv_seedlist",
@@ -96,7 +95,7 @@ def write_summary(errored: dict[str, str], new: list[str], filename: Optional[st
         pr_body += "\n"
     if pr_body != "":
         if filename is None:
-            print(f"\n{pr_body}")  # noqa: T201
+            print(f"\n{pr_body}")
         else:
             with open(filename, "w") as f:
                 # replacements made for proper json
diff --git a/.evergreen/scripts/run_tests.py b/.evergreen/scripts/run_tests.py
@@ -10,6 +10,12 @@
 from pathlib import Path
 from shutil import which
 
+try:
+    import importlib_metadata
+except ImportError:
+    from importlib import metadata as importlib_metadata
+
+
 import pytest
 from utils import DRIVERS_TOOLS, LOGGER, ROOT, run_command
 
@@ -23,6 +29,21 @@
 SUB_TEST_NAME = os.environ.get("SUB_TEST_NAME")
 
 
+def list_packages():
+    packages = dict()
+    for distribution in importlib_metadata.distributions():
+        packages[distribution.name] = distribution
+    print("Package             Version     URL")
+    print("------------------- ----------- ----------------------------------------------------")
+    for name in sorted(packages):
+        distribution = packages[name]
+        url = ""
+        if distribution.origin is not None:
+            url = distribution.origin.url
+        print(f"{name:20s}{distribution.version:12s}{url}")
+    print("------------------- ----------- ----------------------------------------------------\n")
+
+
 def handle_perf(start_time: datetime):
     end_time = datetime.now()
     elapsed_secs = (end_time - start_time).total_seconds()
@@ -121,6 +142,9 @@ def handle_aws_lambda() -> None:
 
 
 def run() -> None:
+    # List the installed packages.
+    list_packages()
+
     # Handle green framework first so they can patch modules.
     if GREEN_FRAMEWORK:
         handle_green_framework()
diff --git a/.evergreen/scripts/setup_tests.py b/.evergreen/scripts/setup_tests.py
@@ -214,18 +214,8 @@ def handle_test_env() -> None:
                 if key in os.environ:
                     write_env(key, os.environ[key])
 
-    if test_name == "data_lake":
-        # Stop any running mongo-orchestration which might be using the port.
-        run_command(f"bash {DRIVERS_TOOLS}/.evergreen/stop-orchestration.sh")
-        run_command(f"bash {DRIVERS_TOOLS}/.evergreen/atlas_data_lake/setup.sh")
-        AUTH = "auth"
-
     if AUTH != "noauth":
-        if test_name == "data_lake":
-            config = read_env(f"{DRIVERS_TOOLS}/.evergreen/atlas_data_lake/secrets-export.sh")
-            DB_USER = config["ADL_USERNAME"]
-            DB_PASSWORD = config["ADL_PASSWORD"]
-        elif test_name == "auth_oidc":
+        if test_name == "auth_oidc":
             DB_USER = config["OIDC_ADMIN_USER"]
             DB_PASSWORD = config["OIDC_ADMIN_PWD"]
         elif test_name == "search_index":
@@ -356,7 +346,9 @@ def handle_test_env() -> None:
             setup_libmongocrypt()
 
         # TODO: Test with 'pip install pymongocrypt'
-        UV_ARGS.append("--group pymongocrypt_source")
+        UV_ARGS.append(
+            "--with pymongocrypt@git+https://github.com/mongodb/libmongocrypt@master#subdirectory=bindings/python"
+        )
 
         # Use the nocrypto build to avoid dependency issues with older windows/python versions.
         BASE = ROOT / "libmongocrypt/nocrypto"
diff --git a/.evergreen/scripts/teardown_tests.py b/.evergreen/scripts/teardown_tests.py
@@ -57,10 +57,6 @@
 
     teardown_mod_wsgi()
 
-# Tear down data_lake if applicable.
-elif TEST_NAME == "data_lake":
-    run_command(f"{DRIVERS_TOOLS}/.evergreen/atlas_data_lake/teardown.sh")
-
 # Tear down coverage if applicable.
 if os.environ.get("COVERAGE"):
     shutil.rmtree(".pytest_cache", ignore_errors=True)
diff --git a/.evergreen/scripts/utils.py b/.evergreen/scripts/utils.py
@@ -33,7 +33,6 @@ class Distro:
     "atlas_connect": "atlas_connect",
     "auth_aws": "auth_aws",
     "auth_oidc": "auth_oidc",
-    "data_lake": "data_lake",
     "default": "",
     "default_async": "default_async",
     "default_sync": "default",
@@ -57,7 +56,6 @@ class Distro:
     "auth_oidc",
     "atlas_connect",
     "aws_lambda",
-    "data_lake",
     "mockupdb",
     "ocsp",
 ]
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -355,13 +355,6 @@ Note: these tests can only be run from an Evergreen Linux host that has the Pyth
 The `mode` can be `standalone` or `embedded`.  For the `replica_set` version of the tests, use
 `TOPOLOGY=replica_set just run-server`.
 
-### Atlas Data Lake tests.
-
-You must have `docker` or `podman` installed locally.
-
-- Run `just setup-tests data_lake`.
-- Run `just run-tests`.
-
 ### OCSP tests
 
 - Export the orchestration file, e.g. `export ORCHESTRATION_FILE=rsa-basic-tls-ocsp-disableStapling.json`.
diff --git a/pyproject.toml b/pyproject.toml
@@ -60,9 +60,6 @@ coverage = [
 mockupdb = [
     "mockupdb@git+https://github.com/mongodb-labs/mongo-mockup-db@master"
 ]
-pymongocrypt_source = [
-    "pymongocrypt@git+https://github.com/mongodb/libmongocrypt@master#subdirectory=bindings/python"
-]
 perf = ["simplejson"]
 typing = [
     "mypy==1.18.1",
@@ -134,7 +131,6 @@ markers = [
     "auth: tests that rely on authentication",
     "ocsp: tests that rely on ocsp",
     "atlas_connect: tests that rely on an atlas connection",
-    "data_lake: tests that rely on atlas data lake",
     "perf: benchmark tests",
     "search_index: search index helper tests",
     "kms: client-side field-level encryption tests using kms",
@@ -239,6 +235,7 @@ dummy-variable-rgx = "^(_+|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?)|dummy.*)$"
 
 [tool.ruff.lint.per-file-ignores]
 "pymongo/__init__.py" = ["E402"]
+".evergreen/scripts/*.py" = ["T201"]
 "test/*.py" = ["PT", "E402", "PLW", "SIM", "E741", "PTH", "S", "B904", "E722", "T201",
     "RET", "ARG", "F405", "B028", "PGH001", "B018", "F403", "RUF015", "E731", "B007",
     "UP031", "F401", "B023", "F811"]
diff --git a/requirements/test.txt b/requirements/test.txt
@@ -1,2 +1,3 @@
 pytest>=8.2
 pytest-asyncio>=0.24.0
+importlib_metadata>=7.0;python_version < "3.13"
diff --git a/test/__init__.py b/test/__init__.py
@@ -121,7 +121,6 @@ def __init__(self):
         self.sessions_enabled = False
         self.client = None  # type: ignore
         self.conn_lock = threading.Lock()
-        self.is_data_lake = False
         self.load_balancer = TEST_LOADBALANCER
         self._fips_enabled = None
         if self.load_balancer:
@@ -199,16 +198,6 @@ def _init_client(self):
         self.mongoses = []
         self.connection_attempts = []
         self.client = self._connect(host, port)
-        if self.client is not None:
-            # Return early when connected to dataLake as mongohoused does not
-            # support the getCmdLineOpts command and is tested without TLS.
-            if os.environ.get("TEST_DATA_LAKE"):
-                self.is_data_lake = True
-                self.auth_enabled = True
-                self.client.close()
-                self.client = self._connect(host, port, username=db_user, password=db_pwd)
-                self.connected = True
-                return
 
         if HAVE_SSL and not self.client:
             # Is MongoDB configured for SSL?
@@ -501,14 +490,6 @@ def require_connection(self, func):
             func=func,
         )
 
-    def require_data_lake(self, func):
-        """Run a test only if we are connected to Atlas Data Lake."""
-        return self._require(
-            lambda: self.is_data_lake,
-            "Not connected to Atlas Data Lake on self.pair",
-            func=func,
-        )
-
     def require_version_min(self, *ver):
         """Run a test only if the server version is at least ``version``."""
         other_version = Version(*ver)
@@ -1230,21 +1211,6 @@ def teardown():
         garbage.append(f"  gc.get_referrers: {gc.get_referrers(g)!r}")
     if garbage:
         raise AssertionError("\n".join(garbage))
-    c = client_context.client
-    if c:
-        if not client_context.is_data_lake:
-            try:
-                c.drop_database("pymongo-pooling-tests")
-                c.drop_database("pymongo_test")
-                c.drop_database("pymongo_test1")
-                c.drop_database("pymongo_test2")
-                c.drop_database("pymongo_test_mike")
-                c.drop_database("pymongo_test_bernie")
-            except AutoReconnect:
-                # PYTHON-4982
-                if sys.implementation.name.lower() != "pypy":
-                    raise
-        c.close()
     print_running_clients()
 
 
diff --git a/test/asynchronous/__init__.py b/test/asynchronous/__init__.py
@@ -121,7 +121,6 @@ def __init__(self):
         self.sessions_enabled = False
         self.client = None  # type: ignore
         self.conn_lock = threading.Lock()
-        self.is_data_lake = False
         self.load_balancer = TEST_LOADBALANCER
         self._fips_enabled = None
         if self.load_balancer:
@@ -199,16 +198,6 @@ async def _init_client(self):
         self.mongoses = []
         self.connection_attempts = []
         self.client = await self._connect(host, port)
-        if self.client is not None:
-            # Return early when connected to dataLake as mongohoused does not
-            # support the getCmdLineOpts command and is tested without TLS.
-            if os.environ.get("TEST_DATA_LAKE"):
-                self.is_data_lake = True
-                self.auth_enabled = True
-                await self.client.close()
-                self.client = await self._connect(host, port, username=db_user, password=db_pwd)
-                self.connected = True
-                return
 
         if HAVE_SSL and not self.client:
             # Is MongoDB configured for SSL?
@@ -501,14 +490,6 @@ def require_connection(self, func):
             func=func,
         )
 
-    def require_data_lake(self, func):
-        """Run a test only if we are connected to Atlas Data Lake."""
-        return self._require(
-            lambda: self.is_data_lake,
-            "Not connected to Atlas Data Lake on self.pair",
-            func=func,
-        )
-
     def require_version_min(self, *ver):
         """Run a test only if the server version is at least ``version``."""
         other_version = Version(*ver)
@@ -1246,21 +1227,6 @@ async def async_teardown():
         garbage.append(f"  gc.get_referrers: {gc.get_referrers(g)!r}")
     if garbage:
         raise AssertionError("\n".join(garbage))
-    c = async_client_context.client
-    if c:
-        if not async_client_context.is_data_lake:
-            try:
-                await c.drop_database("pymongo-pooling-tests")
-                await c.drop_database("pymongo_test")
-                await c.drop_database("pymongo_test1")
-                await c.drop_database("pymongo_test2")
-                await c.drop_database("pymongo_test_mike")
-                await c.drop_database("pymongo_test_bernie")
-            except AutoReconnect:
-                # PYTHON-4982
-                if sys.implementation.name.lower() != "pypy":
-                    raise
-        await c.close()
     print_running_clients()
 
 
diff --git a/test/asynchronous/test_data_lake.py b/test/asynchronous/test_data_lake.py
diff --git a/test/test_data_lake.py b/test/test_data_lake.py
diff --git a/tools/synchro.py b/tools/synchro.py
diff --git a/uv.lock b/uv.lock
