dont run xfail + add test deps

Author: aktech

.github/reusable-build/action.yml

@@ -16,11 +16,6 @@ runs:
     with:
       python-version: ${{ inputs.python-version }}
 
-  - name: Upgrade pip
-    shell: bash
-    run: |
-      python -m pip install --upgrade pip pytest
-
   - name: Build the package for Python ${{ inputs.python-version }}
     shell: bash
     run: |

pyproject.toml

@@ -19,3 +19,13 @@ requires = [
   # Required for using org_tensorflow bazel repository.
   "numpy~=1.22.0",
 ]
+
+[project]
+name = "tensorflow-data-validation"
+
+[project.optional-dependencies]
+test = [
+  "pytest",
+  "scikit-learn",
+  "scipy",
+]

tensorflow_data_validation/api/stats_api_test.py

@@ -44,7 +44,7 @@ class StatsAPITest(absltest.TestCase):
   def _get_temp_dir(self):
     return tempfile.mkdtemp()
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_stats_pipeline(self):
     record_batches = [
         pa.RecordBatch.from_arrays([
@@ -203,7 +203,7 @@ def test_stats_pipeline(self):
     }
     """, statistics_pb2.DatasetFeatureStatisticsList())
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_stats_pipeline_with_examples_with_no_values(self):
     record_batches = [
         pa.RecordBatch.from_arrays([
@@ -321,7 +321,7 @@ def test_stats_pipeline_with_examples_with_no_values(self):
           test_util.make_dataset_feature_stats_list_proto_equal_fn(
               self, expected_result, check_histograms=False))
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_stats_pipeline_with_zero_examples(self):
     expected_result = text_format.Parse(
         """
@@ -343,7 +343,7 @@ def test_stats_pipeline_with_zero_examples(self):
           test_util.make_dataset_feature_stats_list_proto_equal_fn(
               self, expected_result, check_histograms=False))
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_stats_pipeline_with_sample_rate(self):
     record_batches = [
         pa.RecordBatch.from_arrays(
@@ -493,7 +493,7 @@ def test_write_stats_to_tfrecord_and_binary(self):
 
 class MergeDatasetFeatureStatisticsListTest(absltest.TestCase):
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_merges_two_shards(self):
     stats1 = text_format.Parse(
         """

tensorflow_data_validation/api/validation_api_test.py

@@ -3241,7 +3241,7 @@ def _assert_skew_pairs_equal(self, actual, expected) -> None:
     for each in actual:
       self.assertIn(each, expected)
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_detect_feature_skew(self):
     training_data = [
         text_format.Parse("""

tensorflow_data_validation/coders/csv_decoder_test.py

@@ -366,7 +366,7 @@
 ]
 
 
-@pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed. ")
+@pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed. ")
 class CSVDecoderTest(parameterized.TestCase):
   """Tests for CSV decoder."""
 

tensorflow_data_validation/integration_tests/sequence_example_e2e_test.py

@@ -1738,7 +1738,7 @@
 ]
 
 
-@pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed. ")
+@pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed. ")
 class SequenceExampleStatsTest(parameterized.TestCase):
 
   @classmethod

tensorflow_data_validation/skew/feature_skew_detector_test.py

@@ -142,7 +142,7 @@ def _make_ex(identifier: str,
 
 class FeatureSkewDetectorTest(parameterized.TestCase):
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_detect_feature_skew(self):
     baseline_examples, test_examples, _ = get_test_input(
         include_skewed_features=True, include_close_floats=True)
@@ -194,7 +194,7 @@ def test_detect_feature_skew(self):
           skew_result,
           test_util.make_skew_result_equal_fn(self, expected_result))
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_detect_no_skew(self):
     baseline_examples, test_examples, _ = get_test_input(
         include_skewed_features=False, include_close_floats=False)
@@ -224,7 +224,7 @@ def test_detect_no_skew(self):
       util.assert_that(skew_sample, make_sample_equal_fn(self, 0, []),
                        'CheckSkewSample')
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_obtain_skew_sample(self):
     baseline_examples, test_examples, skew_pairs = get_test_input(
         include_skewed_features=True, include_close_floats=False)
@@ -248,7 +248,7 @@ def test_obtain_skew_sample(self):
           skew_sample, make_sample_equal_fn(self, sample_size,
                                             potential_samples))
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_empty_inputs(self):
     baseline_examples, test_examples, _ = get_test_input(
         include_skewed_features=True, include_close_floats=True)
@@ -304,7 +304,7 @@ def test_empty_inputs(self):
                        make_sample_equal_fn(self, 0, expected_result),
                        'CheckSkewSample')
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_float_precision_configuration(self):
     baseline_examples, test_examples, _ = get_test_input(
         include_skewed_features=True, include_close_floats=True)
@@ -395,7 +395,7 @@ def test_no_identifier_features(self):
         _ = ((baseline_examples, test_examples)
              | feature_skew_detector.DetectFeatureSkewImpl([]))
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_duplicate_identifiers_allowed_with_duplicates(self):
     base_example_1 = text_format.Parse(
         """
@@ -469,7 +469,7 @@ def test_duplicate_identifiers_allowed_with_duplicates(self):
           skew_result,
           test_util.make_skew_result_equal_fn(self, expected_result))
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_duplicate_identifiers_not_allowed_with_duplicates(self):
     base_example_1 = text_format.Parse(
         """
@@ -535,7 +535,7 @@ def test_duplicate_identifiers_not_allowed_with_duplicates(self):
     self.assertLen(actual_counter, 1)
     self.assertEqual(actual_counter[0].committed, 1)
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_skips_missing_identifier_example(self):
     base_example_1 = text_format.Parse(
         """
@@ -576,7 +576,7 @@ def test_skips_missing_identifier_example(self):
     runner = p.run()
     runner.wait_until_finish()
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_empty_features_equivalent(self):
     base_example_1 = text_format.Parse(
         """
@@ -626,7 +626,7 @@ def test_empty_features_equivalent(self):
     runner = p.run()
     runner.wait_until_finish()
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_empty_features_not_equivalent_to_missing(self):
     base_example_1 = text_format.Parse(
         """
@@ -699,7 +699,7 @@ def test_telemetry(self):
     self.assertLen(actual_counter, 1)
     self.assertEqual(actual_counter[0].committed, 1)
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_confusion_analysis(self):
 
     baseline_examples = [
@@ -834,7 +834,7 @@ def test_confusion_analysis_errors(self, input_example, expected_error_regex):
                     feature_skew_detector.ConfusionConfig(name='val'),
                 ]))[feature_skew_detector.CONFUSION_KEY]
 
-  @pytest.mark.xfail(run=True, reason="PR 260 This test fails and needs to be fixed.")
+  @pytest.mark.xfail(run=False, reason="PR 260 This test fails and needs to be fixed.")
   def test_match_stats(self):
     baseline_examples = [
         _make_ex('id0'),