ARROW-143 Switch to Pytest as Test Runner (#126)

Author: blink1073

.github/workflows/test-python.yml

@@ -12,7 +12,7 @@ concurrency:
 defaults:
   run:
     working-directory: ./bindings/python
-    shell: bash
+    shell: bash -eux {0}
 
 jobs:
 
@@ -44,6 +44,10 @@ jobs:
           python-version: ${{ matrix.python-version }}
           cache: 'pip'
           cache-dependency-path: '**/setup.cfg'
+      - name: Set up env
+        run: |
+           echo "LIBBSON_INSTALL_DIR=$PWD/libbson" >> $GITHUB_ENV
+           echo "LD_LIBRARY_PATH=$PWD/libbson/lib" >> $GITHUB_ENV
       - name: Start MongoDB on Linux
         if: ${{ startsWith(runner.os, 'Linux') }}
         uses: supercharge/[email protected]
@@ -65,22 +69,23 @@ jobs:
           net start MongoDB
       - name: Install libbson
         run: |
-          LIBBSON_INSTALL_DIR=$(pwd)/libbson ./build-libbson.sh
+          ./build-libbson.sh
       - name: Install Python dependencies
         run: |
           python -m pip install -U pip
       - name: Install pymongoarrow
         run: |
           # Install the library with no deps
-          LIBBSON_INSTALL_DIR=$(pwd)/libbson python -m pip install -vv -e "."
+          python -m pip install -v -e "."
       - name: Ensure imports with no test deps
         run: |
           python -c "from pymongoarrow.monkey import patch_all; patch_all()"
-      - name: Run tests
+      - name: Install the test dependencies
+        run: |
+          python -m pip install -v -e ".[test]"
+      - name: Run the tests
         run: |
-          # Install the test deps
-          LIBBSON_INSTALL_DIR=$(pwd)/libbson python -m pip install -vv -e ".[test]"
-          PYTHONWARNINGS=error LD_LIBRARY_PATH=$(pwd)/libbson/lib python -m unittest discover test -v
+          PYTHONWARNINGS=error python -m pytest -vv
       - name: Check the manifest
         run: |
           pip install check-manifest

bindings/python/docs/source/developer/installation.rst

@@ -72,7 +72,7 @@ Test
 To run the test suite, you will need a MongoDB instance running on
 ``localhost`` using port ``27017``. To run the entire test suite, do::
 
-  (pymongoarrow) $ python -m unittest discover test
+  (pymongoarrow) $ python -m pytest
 
 Running Linters
 ---------------

bindings/python/setup.cfg

@@ -47,4 +47,4 @@ exclude =
     docs
 
 [options.extras_require]
-test = pandas;pytz
+test = pandas;pytz;pytest

bindings/python/test/test_arrow.py

@@ -17,7 +17,7 @@
 import unittest.mock as mock
 from datetime import datetime
 from test import client_context
-from test.utils import AllowListEventListener, TestNullsBase
+from test.utils import AllowListEventListener, NullsTestMixin
 
 import pyarrow
 import pymongo
@@ -39,7 +39,7 @@
 from pytz import timezone
 
 
-class TestArrowApiMixin:
+class ArrowApiTestMixin:
     @classmethod
     def setUpClass(cls):
         if not client_context.connected:
@@ -468,15 +468,15 @@ def test_empty_nested_objects(self):
         self.round_trip(data, Schema(schema))
 
 
-class TestArrowExplicitApi(TestArrowApiMixin, unittest.TestCase):
+class TestArrowExplicitApi(ArrowApiTestMixin, unittest.TestCase):
     def run_find(self, *args, **kwargs):
         return find_arrow_all(self.coll, *args, **kwargs)
 
     def run_aggregate(self, *args, **kwargs):
         return aggregate_arrow_all(self.coll, *args, **kwargs)
 
 
-class TestArrowPatchedApi(TestArrowApiMixin, unittest.TestCase):
+class TestArrowPatchedApi(ArrowApiTestMixin, unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         patch_all()
@@ -528,7 +528,7 @@ def test_find_decimal128(self):
         self.assertEqual(table, expected)
 
 
-class TestNulls(TestNullsBase):
+class TestNulls(NullsTestMixin, unittest.TestCase):
     def find_fn(self, coll, query, schema):
         return find_arrow_all(coll, query, schema=schema)
 

bindings/python/test/test_builders.py

@@ -30,7 +30,7 @@
 from pymongoarrow.types import ObjectIdType
 
 
-class TestIntBuildersMixin:
+class IntBuildersTestMixin:
     def test_simple(self):
         builder = self.builder_cls()
         builder.append(0)
@@ -45,13 +45,13 @@ def test_simple(self):
         self.assertEqual(arr.type, self.data_type)
 
 
-class TestInt32Builder(TestCase, TestIntBuildersMixin):
+class TestInt32Builder(TestCase, IntBuildersTestMixin):
     def setUp(self):
         self.builder_cls = Int32Builder
         self.data_type = int32()
 
 
-class TestInt64Builder(TestCase, TestIntBuildersMixin):
+class TestInt64Builder(TestCase, IntBuildersTestMixin):
     def setUp(self):
         self.builder_cls = Int64Builder
         self.data_type = int64()
@@ -186,7 +186,7 @@ def test_nested(self):
         self.assertEqual(len(arr), 9)
 
 
-class TestBoolBuilderMixin:
+class BoolBuilderTestMixin:
     def test_simple(self):
         builder = BoolBuilder()
         builder.append(False)
@@ -201,7 +201,7 @@ def test_simple(self):
         self.assertEqual(arr.type, self.data_type)
 
 
-class TestBoolBuilder(TestCase, TestBoolBuilderMixin):
+class TestBoolBuilder(TestCase, BoolBuilderTestMixin):
     def setUp(self):
         self.builder_cls = BoolBuilder
         self.data_type = bool_()

bindings/python/test/test_numpy.py

@@ -15,7 +15,7 @@
 import datetime
 import unittest
 from test import client_context
-from test.utils import AllowListEventListener, TestNullsBase
+from test.utils import AllowListEventListener, NullsTestMixin
 from unittest import mock
 
 import numpy as np
@@ -311,7 +311,7 @@ def test_find_decimal128(self):
 
 # The spec for pyarrow says to_numpy is experimental, so we should expect
 # this to change in the future.
-class TestNulls(TestNullsBase, NumpyTestBase):
+class TestNulls(NullsTestMixin, NumpyTestBase):
     def table_from_dict(self, d, schema=None):
         out = {}
         for k, v in d.items():

bindings/python/test/test_pandas.py

@@ -18,7 +18,7 @@
 import unittest.mock as mock
 import warnings
 from test import client_context
-from test.utils import AllowListEventListener, TestNullsBase
+from test.utils import AllowListEventListener, NullsTestMixin
 
 import numpy as np
 import pandas as pd
@@ -347,7 +347,7 @@ def test_find_decimal128(self):
         pd.testing.assert_frame_equal(expected, table)
 
 
-class TestNulls(TestNullsBase):
+class TestNulls(NullsTestMixin, unittest.TestCase):
     def find_fn(self, coll, query, schema):
         return find_pandas_all(coll, query, schema=schema)
 

bindings/python/test/utils.py

@@ -72,7 +72,7 @@ def failed(self, event):
             super(AllowListEventListener, self).failed(event)
 
 
-class TestNullsBase(unittest.TestCase):
+class NullsTestMixin:
     def find_fn(self, coll, query, schema):
         raise NotImplementedError
 
@@ -121,7 +121,7 @@ def na_safe(self, atype):
 
     @classmethod
     def setUpClass(cls):
-        if cls is TestNullsBase:
+        if cls is NullsTestMixin:
             raise unittest.SkipTest("Base class")
 
         if not client_context.connected:
@@ -193,7 +193,7 @@ def test_other_handling(self):
         # arrow/pandas/numpy.
         for gen in [str, float, datetime.datetime, ObjectId, Decimal128]:
             con_type = self.pytype_tab_map[gen]  # Arrow/Pandas/Numpy
-            pytype = TestNullsBase.pytype_tab_map[gen]  # Arrow type specifically
+            pytype = NullsTestMixin.pytype_tab_map[gen]  # Arrow type specifically
 
             other_schema = Schema({"_id": ObjectIdType(), "other": pytype})
             others = [