Update version to 0.2.1rc5 and remove hard dependency to dp_accounting

analysis/parameter_tuning.py

@@ -142,8 +142,10 @@ def _find_candidate_parameters(
     if calculate_sum_per_partition_param:
         if hist.linf_sum_contributions_histogram.bins[0].lower >= 0:
             logging.warning(
-                "max_sum_per_partition should not contain negative sums because min_sum_per_partition tuning is not supported yet and therefore tuning for max_sum_per_partition works only when linf_sum_contributions_histogram does not negative sums"
-            )
+                "max_sum_per_partition should not contain negative sums because"
+                " min_sum_per_partition tuning is not supported yet and "
+                "therefore tuning for max_sum_per_partition works only when "
+                "linf_sum_contributions_histogram does not negative sums")
 
     if calculate_l0_param and calculate_linf_count:
         l0_bounds, linf_bounds = _find_candidates_parameters_in_2d_grid(

pipeline_dp/__init__.py

@@ -38,4 +38,4 @@
 from pipeline_dp.pipeline_backend import PipelineBackend
 from pipeline_dp.pipeline_backend import SparkRDDBackend
 
-__version__ = '0.2.1rc4'
+__version__ = '0.2.1rc5'

pipeline_dp/budget_accounting.py

@@ -32,6 +32,11 @@
     pass
 
 
+def _check_pldlib_imported() -> bool:
+    import sys
+    return "dp_accounting.pld.privacy_loss_distribution" in sys.modules
+
+
 @dataclass
 class MechanismSpec:
     """Specifies the parameters for a DP mechanism.
@@ -446,6 +451,13 @@ def __init__(self,
         super().__init__(total_epsilon, total_delta, num_aggregations,
                          aggregation_weights)
 
+        if not _check_pldlib_imported():
+            raise ImportError("dp_accounting library is not imported. It is"
+                              "required for using PLD budget accounting. "
+                              "Please install dp_accounting library or use"
+                              "NaiveBudgetAccountant instead of "
+                              "PLDBudgetAccountant")
+
         self.minimum_noise_std = None
         self._pld_discretization = pld_discretization
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pipeline-dp"
-version = "0.2.1rc4"
+version = "0.2.1rc5"
 description = ""
 authors = ["Chinmay Shah <chinmayshah3899@gmail.com>", "Vadym Doroshenko <dvadym@google.com>"]
 license = "Apache-2.0"
@@ -9,7 +9,6 @@ license = "Apache-2.0"
 python = "^3.8, < 3.12"
 python-dp = "^1.1.4"
 numpy = "^1.20.1"
-dp_accounting = "^0.4.2"
 
 [tool.poetry.dev-dependencies]
 yapf = "^0.32.0"

setup.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.2.1rc4
+current_version = 0.2.1rc5
 commit = True
 tag = True
 

setup.py

@@ -11,8 +11,7 @@
 install_requires = \
 ['numpy>=1.20.1,<2.0.0',
  'python-dp>=1.1.4',
- 'scipy>=1.7.3,<2.0.0',
- 'dp_accounting>=0.4.2'
+ 'scipy>=1.7.3,<2.0.0'
 ]
 
 
@@ -22,7 +21,7 @@ def read(fname):
 
 setup_kwargs = {
     'name': 'pipeline-dp',
-    'version': '0.2.1rc4',
+    'version': '0.2.1rc5',
     'description': 'Framework for applying differential privacy to large datasets using batch processing systems',
     'author': 'Chinmay Shah',
     'author_email': 'chinmayshah3899@gmail.com',

tests/budget_accounting_test.py

@@ -196,9 +196,8 @@ def test_not_enough_aggregations(self, use_num_aggregations):
             budget_accountant.compute_budgets()
 
 
-@unittest.skipIf(
-    sys.version_info.major == 3 and sys.version_info.minor == 8,
-    "There are some problems with dp_accounting library in python 3.8")
+@unittest.skipIf(sys.version_info.major == 3 and sys.version_info.minor <= 8,
+                 "dp_accounting library only support python >=3.9")
 class PLDBudgetAccountantTest(unittest.TestCase):
 
     def test_noise_not_calculated(self):

tests/dp_engine_test.py

@@ -1094,8 +1094,8 @@ def run_e2e_private_partition_selection_large_budget(self, col, backend):
         return col
 
     @unittest.skipIf(
-        sys.version_info.major == 3 and sys.version_info.minor == 8,
-        "There are some problems with dp_accounting library in python 3.8")
+        sys.version_info.major == 3 and sys.version_info.minor <= 8,
+        "dp_accounting library only support python >=3.9")
     @parameterized.parameters(False, True)
     def test_run_e2e_count_public_partition_local(self, pld_accounting):
         Accountant = pipeline_dp.PLDBudgetAccountant if pld_accounting else pipeline_dp.NaiveBudgetAccountant
@@ -1224,6 +1224,9 @@ def test_min_max_sum_per_partition(self):
         self.assertLen(output, 1)
         self.assertAlmostEqual(output[0][1].sum, -3, delta=0.1)
 
+    @unittest.skipIf(
+        sys.version_info.major == 3 and sys.version_info.minor <= 8,
+        "dp_accounting library only support python >=3.9")
     def test_pld_not_supported_metrics(self):
         with self.assertRaisesRegex(
                 NotImplementedError,
@@ -1237,6 +1240,9 @@ def test_pld_not_supported_metrics(self):
             engine.aggregate([1], aggregate_params,
                              self._get_default_extractors(), public_partitions)
 
+    @unittest.skipIf(
+        sys.version_info.major == 3 and sys.version_info.minor <= 8,
+        "dp_accounting library only support python >=3.9")
     def test_pld_not_support_private_partition_selection(self):
         with self.assertRaisesRegex(
                 NotImplementedError,