genesis-flow 1.0.7__py3-none-any.whl → 1.0.9__py3-none-any.whl

{genesis_flow-1.0.7.dist-info → genesis_flow-1.0.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: genesis-flow
-Version: 1.0.7
+Version: 1.0.9
 Summary: Genesis-Flow: MLflow v3.1.4 compatible fork for Genesis platform
 Maintainer-email: Databricks <mlflow-oss-maintainers@googlegroups.com>
 License: Copyright 2018 Databricks, Inc.  All rights reserved.
@@ -316,7 +316,6 @@ Genesis-Flow is a secure, lightweight, and scalable ML operations platform built
 
 ### Security-First Design
 - **Input validation** against SQL injection and path traversal attacks
-- **Secure model loading** with restricted pickle deserialization
 - **Authentication** and authorization ready for enterprise deployment
 - **Security patches** for all known vulnerabilities in dependencies
 
@@ -504,7 +503,6 @@ export MLFLOW_POSTGRES_USERNAME="user@tenant"
 export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account.json"
 
 # Security configuration
-export MLFLOW_ENABLE_SECURE_MODEL_LOADING=true
 export MLFLOW_STRICT_INPUT_VALIDATION=true
 ```
 
@@ -761,7 +759,6 @@ entry_points = {
 
 - ✅ **Input validation** against injection attacks
 - ✅ **Path traversal protection** for file operations  
-- ✅ **Secure pickle loading** with restricted unpickling
 - ✅ **Authentication hooks** for enterprise SSO integration
 - ✅ **Audit logging** for compliance requirements
 - ✅ **Encrypted communication** support

{genesis_flow-1.0.7.dist-info → genesis_flow-1.0.9.dist-info}/RECORD

@@ -1,4 +1,4 @@
-genesis_flow-1.0.7.dist-info/licenses/LICENSE.txt,sha256=Y5U1Xebzka__NZlqMPtBsYm0mRpMtUmTrONatpoL-ig,11382
+genesis_flow-1.0.9.dist-info/licenses/LICENSE.txt,sha256=Y5U1Xebzka__NZlqMPtBsYm0mRpMtUmTrONatpoL-ig,11382
 mlflow/__init__.py,sha256=-_r__N5Afed81pLVtr2wKbHQIA0aj9u9n_7kWGxLWi4,11194
 mlflow/__main__.py,sha256=_PcdoxKehR_a2MI6GqBfzYzRCXZhVyDCSdbxDWVlWd4,39
 mlflow/cli.py,sha256=f1ObrWZ03HgRiRoVEE1Gffe-dGcSY7CxJyEFgb5VUMM,26137
@@ -311,7 +311,7 @@ mlflow/pyfunc/backend.py,sha256=5-tjUnY1nPecoNCxFjhLjPbhHsDGR9_0P9ym02_J3HA,2077
 mlflow/pyfunc/context.py,sha256=nXQcP61XR7cKx6CeFsdAM4TD-NOrIEt_GQSfJ990xc4,3007
 mlflow/pyfunc/dbconnect_artifact_cache.py,sha256=-Ji_GU-NLvGyxYrOFhR81SYn4E6ykiFcZl35hRQ9c6A,5769
 mlflow/pyfunc/mlserver.py,sha256=ER7tXcOZ-tAUblMbgqPVT2E8BtBGH1HXLSSOmsQUB_c,1271
-mlflow/pyfunc/model.py,sha256=aDwfNKgoVw4sl9W0WZfOa9st5WTsSF3NWbF5TgM3bDs,64507
+mlflow/pyfunc/model.py,sha256=sOM5lgC0cvZmbv-u-NVE4xwWKv6WskP_xR8YKVNvOIg,63797
 mlflow/pyfunc/spark_model_cache.py,sha256=sFFAi-LxXEJdqk8szm32XxK5IFsISpxxVjAWjWbHZ1U,2091
 mlflow/pyfunc/stdin_server.py,sha256=gNkWxlCz3KLu6jvZNeNytmKAy4j5gIXOpIehrizm43s,1362
 mlflow/pyfunc/loaders/__init__.py,sha256=W3Bny093PoREOCqavIzKJ3lYeZUBXhxk3EDkCqc_lBc,276
@@ -364,7 +364,7 @@ mlflow/server/graphql/graphql_errors.py,sha256=B-U26g2vsJoKNg2hax-Gl3A66LB-y5adB
 mlflow/server/graphql/graphql_no_batching.py,sha256=3J7h6-qn6nbonYZWWCBlOgMC37NiSnKvA6P_rp5sSxc,2972
 mlflow/server/graphql/graphql_schema_extensions.py,sha256=i6DeNbzLcvBKXc_jLG_dCdXwA2cY__XPM6bU1dGQ8bM,2329
 mlflow/shap/__init__.py,sha256=i6UAJd-XYu7Nw6rbfxaGvr5vfK-6gUfU2MXMAZvSM-c,25558
-mlflow/sklearn/__init__.py,sha256=ovyR0zHB_HKoysOlElzx2EsonkisucHR8PslBsVMtfA,86026
+mlflow/sklearn/__init__.py,sha256=unYCnl3rXRlwptsluRiLjooR_KBM6anOZrvbzgiKe1A,85957
 mlflow/sklearn/utils.py,sha256=CM9z-UD65RBN4sIdEoJWhazkdmE9Wa9InjsU61y1XfE,38912
 mlflow/smolagents/__init__.py,sha256=V2sfkuWxQ090CaIjwBh7icD0bSgYP3jKD5MRzy__uW0,2202
 mlflow/smolagents/autolog.py,sha256=5GJpqmXs8dd2tf4jkL6fk7Fxb0TsxAMykB865M-BpK8,4770
@@ -613,7 +613,6 @@ mlflow/utils/requirements_utils.py,sha256=79BL9iFqg9H4VRYiqk-gTLlrfLyUY7KVG4BDV8
 mlflow/utils/rest_utils.py,sha256=R5MZk0Ck358VeW-DKqpfSyDHbWjg_slQTpRfr4ZLDLM,26664
 mlflow/utils/search_logged_model_utils.py,sha256=JWczQxogo4u9Gr3gfoSVf7tVny1e-rxfAchh3Vqy6RU,4289
 mlflow/utils/search_utils.py,sha256=PgXtwlViHvmahY63otykh5iJVW0PR9DIwetrvCCBw2U,86142
-mlflow/utils/secure_loading.py,sha256=1d-Hf0QkU6WkGjLKvAKcXv7sLa6xl2utig1lT-Hhyp4,10262
 mlflow/utils/security_validation.py,sha256=cLBrLNOoVJuAffO677v0m_v4i-01v7aLk-sJk87mpkQ,12830
 mlflow/utils/server_cli_utils.py,sha256=gbT5CVkOLorSw-y8bnNSBEP9mClcVTagtTN5SK5Azhw,2381
 mlflow/utils/spark_utils.py,sha256=zUbQIRAtwyU3rDJK8m7v42KO2TSGn28Coe_UYBEhIWA,395
@@ -641,8 +640,8 @@ mlflow/utils/autologging_utils/metrics_queue.py,sha256=bwpMX7Go6xFxrpYROi6rDBdVt
 mlflow/utils/autologging_utils/safety.py,sha256=IwbTbusyE87Hc4qkhhvMikoaZqX9kpr972FWS2B8goc,51465
 mlflow/utils/autologging_utils/versioning.py,sha256=2hSN4KXFWEJCcopDdLG6BiPeqSoqjETeNMuUBsCCwlI,3762
 mlflow/utils/import_hooks/__init__.py,sha256=werje98Woelkbwrhtlb8wmRdt3RtiL--LqGru7Xh3YU,13589
-genesis_flow-1.0.7.dist-info/METADATA,sha256=Z4t8Zn9OIxkH7d-hTozfYP0c8_-VB-WQ0KWo8XxyDHw,33401
-genesis_flow-1.0.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-genesis_flow-1.0.7.dist-info/entry_points.txt,sha256=YZZiTHSYQpv8vou6gdqWI17piK9Pm0P3BmZ-xepUtnw,449
-genesis_flow-1.0.7.dist-info/top_level.txt,sha256=wm8UqYyUHI21EvrTDHb3eYICy0dOVDLBhAL-jp5zbuI,7
-genesis_flow-1.0.7.dist-info/RECORD,,
+genesis_flow-1.0.9.dist-info/METADATA,sha256=yiOvGLekYICCUhdHRvUEjtFrmOoFYZ2xNn2wzXkSS2w,33229
+genesis_flow-1.0.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+genesis_flow-1.0.9.dist-info/entry_points.txt,sha256=YZZiTHSYQpv8vou6gdqWI17piK9Pm0P3BmZ-xepUtnw,449
+genesis_flow-1.0.9.dist-info/top_level.txt,sha256=wm8UqYyUHI21EvrTDHb3eYICy0dOVDLBhAL-jp5zbuI,7
+genesis_flow-1.0.9.dist-info/RECORD,,

mlflow/pyfunc/model.py

@@ -818,28 +818,10 @@ def _maybe_compress_cloudpickle_dump(python_model, path, compression):
 
 
 def _maybe_decompress_cloudpickle_load(path, compression):
-    """
-    Genesis-Flow: Secure model loading with safety checks.
-    """
-    from mlflow.utils.secure_loading import SecureModelLoader, SecurityError
-    
     _check_compression_supported(compression)
-    
-    # For compressed files, we need to decompress first then load securely
-    if compression and compression != "none":
-        import tempfile
-        file_open = _COMPRESSION_INFO.get(compression, {}).get("open", open)
-        with file_open(path, "rb") as compressed_f:
-            with tempfile.NamedTemporaryFile(delete=False) as temp_f:
-                temp_f.write(compressed_f.read())
-                temp_path = temp_f.name
-        try:
-            return SecureModelLoader.safe_cloudpickle_load(temp_path)
-        finally:
-            os.unlink(temp_path)
-    else:
-        # Direct secure loading for uncompressed files
-        return SecureModelLoader.safe_cloudpickle_load(path)
+    file_open = _COMPRESSION_INFO.get(compression, {}).get("open", open)
+    with file_open(path, "rb") as f:
+        return cloudpickle.load(f)
 
 
 if IS_PYDANTIC_V2_OR_NEWER:

mlflow/sklearn/__init__.py

@@ -465,13 +465,14 @@ def _load_model_from_local_file(path, serialization_format):
             ),
             error_code=INVALID_PARAMETER_VALUE,
         )
-    # Genesis-Flow: Use secure model loading to prevent code execution attacks
-    from mlflow.utils.secure_loading import SecureModelLoader
-    
     if serialization_format == SERIALIZATION_FORMAT_PICKLE:
-        return SecureModelLoader.safe_pickle_load(path)
+        import pickle
+        with open(path, "rb") as f:
+            return pickle.load(f)
     elif serialization_format == SERIALIZATION_FORMAT_CLOUDPICKLE:
-        return SecureModelLoader.safe_cloudpickle_load(path)
+        import cloudpickle
+        with open(path, "rb") as f:
+            return cloudpickle.load(f)
 
 
 def _load_pyfunc(path):

mlflow/utils/secure_loading.py

@@ -1,285 +0,0 @@
-"""
-Genesis-Flow Secure Model Loading
-
-This module provides secure alternatives to pickle-based model loading,
-addressing security vulnerabilities in model deserialization.
-"""
-
-import io
-import pickle
-import cloudpickle
-import logging
-import hashlib
-from typing import Any, Set, Optional, Union, Type
-from pathlib import Path
-
-logger = logging.getLogger(__name__)
-
-# Allowlist of safe classes that can be unpickled
-SAFE_PICKLE_CLASSES = {
-
-    # STD types
-    'time.time',
-
-    # NumPy types
-    'numpy.ndarray',
-    'numpy.dtype',
-    'numpy.int32', 'numpy.int64', 'numpy.float32', 'numpy.float64',
-    'numpy.bool_', 'numpy.str_',
-
-    # Pandas types
-    'pandas.core.frame.DataFrame',
-    'pandas.core.series.Series',
-    'pandas.core.index.Index',
-    'pandas.core.dtypes.dtypes.CategoricalDtype',
-
-    # Scikit-learn estimators
-    'sklearn.linear_model._base.LinearRegression',
-    'sklearn.linear_model._logistic.LogisticRegression',
-    'sklearn.ensemble._forest.RandomForestClassifier',
-    'sklearn.ensemble._forest.RandomForestRegressor',
-    'sklearn.tree._classes.DecisionTreeClassifier',
-    'sklearn.tree._classes.DecisionTreeRegressor',
-    'sklearn.svm._classes.SVC',
-    'sklearn.svm._classes.SVR',
-
-    # Built-in types
-    'builtins.dict', 'builtins.list', 'builtins.tuple', 'builtins.set',
-    'builtins.str', 'builtins.int', 'builtins.float', 'builtins.bool',
-    'builtins.type',
-
-    # Collections
-    'collections.OrderedDict',
-    'collections.defaultdict',
-
-    # MLflow types
-    'mlflow.models.signature.ModelSignature',
-    'mlflow.models.signature._TypeHints',
-    'mlflow.types.schema.Schema',
-    'mlflow.pyfunc.model.PythonModel',
-
-    # Cloudpickle internals
-    'cloudpickle.cloudpickle._make_skeleton_class',
-    'cloudpickle.cloudpickle._class_setstate',
-    'cloudpickle.cloudpickle._make_function',
-    'cloudpickle.cloudpickle._builtin_type',
-    'cloudpickle.cloudpickle._function_setstate',
-    'cloudpickle.cloudpickle._make_empty_cell',
-    'cloudpickle.cloudpickle._make_cell',
-    'cloudpickle.cloudpickle.subimport',
-
-    # Sentence Transformers
-    'sentence_transformers.SentenceTransformer.SentenceTransformer',
-    'sentence_transformers.model_card.SentenceTransformerModelCardData',
-    'sentence_transformers.models.Transformer.Transformer',
-    'sentence_transformers.models.Pooling.Pooling',
-    'sentence_transformers.models.Normalize.Normalize',
-
-    # Torch
-    'torch.torch_version.TorchVersion',
-    'torch._utils._rebuild_tensor_v2',
-    'torch.storage._load_from_bytes',
-    'torch.nn.modules.sparse.Embedding',
-    'torch._utils._rebuild_parameter',
-    'torch.nn.modules.normalization.LayerNorm',
-    'torch.nn.modules.dropout.Dropout',
-    'torch.nn.modules.container.ModuleList',
-    'torch.nn.modules.linear.Linear',
-    'torch.nn.modules.activation.Tanh',
-    'torch.float32',
-    'torch._C._nn.gelu',
-
-    # Transformers
-    'transformers.models.bert.modeling_bert.BertModel',
-    'transformers.models.bert.modeling_bert.BertEmbeddings',
-    'transformers.models.bert.modeling_bert.BertEncoder',
-    'transformers.models.bert.modeling_bert.BertLayer',
-    'transformers.models.bert.modeling_bert.BertAttention',
-    'transformers.models.bert.modeling_bert.BertSdpaSelfAttention',
-    'transformers.models.bert.modeling_bert.BertSelfOutput',
-    'transformers.models.bert.modeling_bert.BertIntermediate',
-    'transformers.models.bert.modeling_bert.BertOutput',
-    'transformers.models.bert.modeling_bert.BertPooler',
-    'transformers.models.bert.configuration_bert.BertConfig',
-    'transformers.models.bert.tokenization_bert_fast.BertTokenizerFast',
-    'transformers.activations.GELUActivation',
-
-    # Tokenizers
-    'tokenizers.Tokenizer',
-    'tokenizers.models.Model',
-    'tokenizers.AddedToken',
-
-    # PyTorch
-    'model.load_checkpoint',
-    'torch.device',
-    'model.SepClassifier',
-}
-
-
-class RestrictedUnpickler(pickle.Unpickler):
-    """
-    Secure unpickler that only allows safe, whitelisted classes.
-    
-    This prevents arbitrary code execution during model deserialization
-    by restricting which classes can be instantiated.
-    """
-    
-    def __init__(self, file, *, safe_classes: Optional[Set[str]] = None):
-        super().__init__(file)
-        self.safe_classes = safe_classes or SAFE_PICKLE_CLASSES
-        
-    def find_class(self, module: str, name: str) -> Type:
-        """
-        Override to restrict class loading to safe classes only.
-        
-        Args:
-            module: Module name
-            name: Class name
-            
-        Returns:
-            Class object if safe
-            
-        Raises:
-            SecurityError: If class is not in allowlist
-        """
-        full_name = f"{module}.{name}"
-        
-        # Check if the class is in our safe list
-        if full_name in self.safe_classes:
-            logger.debug(f"Loading safe class: {full_name}")
-            return super().find_class(module, name)
-        
-        # Additional check for known safe modules
-        safe_modules = {
-            'numpy': ['ndarray', 'dtype', 'int32', 'int64', 'float32', 'float64', 'bool_'],
-            'pandas': ['DataFrame', 'Series', 'Index'],
-            'builtins': ['dict', 'list', 'tuple', 'set', 'str', 'int', 'float', 'bool'],
-        }
-        
-        if module in safe_modules and name in safe_modules[module]:
-            logger.debug(f"Loading safe module class: {full_name}")
-            return super().find_class(module, name)
-        
-        # Log and block unsafe class
-        logger.warning(f"Blocked potentially unsafe class: {full_name}")
-        raise pickle.UnpicklingError(
-            f"Security: Class '{full_name}' is not in the allowlist. "
-            f"If this is a legitimate model class, add it to SAFE_PICKLE_CLASSES."
-        )
-
-class SecureModelLoader:
-    """
-    Secure model loading with multiple safety mechanisms.
-    """
-    
-    @staticmethod
-    def calculate_file_hash(file_path: Union[str, Path]) -> str:
-        """Calculate SHA256 hash of a file for integrity checking."""
-        hasher = hashlib.sha256()
-        with open(file_path, 'rb') as f:
-            for chunk in iter(lambda: f.read(4096), b""):
-                hasher.update(chunk)
-        return hasher.hexdigest()
-    
-    @staticmethod
-    def safe_pickle_load(file_path: Union[str, Path], *, safe_classes: Optional[Set[str]] = None) -> Any:
-        """
-        Safely load a pickle file using restricted unpickler.
-        
-        Args:
-            file_path: Path to pickle file
-            safe_classes: Optional custom set of safe classes
-            
-        Returns:
-            Unpickled object
-            
-        Raises:
-            SecurityError: If file contains unsafe classes
-            FileNotFoundError: If file doesn't exist
-        """
-        file_path = Path(file_path)
-        
-        if not file_path.exists():
-            raise FileNotFoundError(f"Model file not found: {file_path}")
-        
-        # Log file hash for audit trail
-        file_hash = SecureModelLoader.calculate_file_hash(file_path)
-        logger.info(f"Loading model file {file_path.name} (SHA256: {file_hash[:16]}...)")
-        
-        try:
-            with open(file_path, 'rb') as f:
-                unpickler = RestrictedUnpickler(f, safe_classes=safe_classes)
-                model = unpickler.load()
-                logger.info(f"Successfully loaded model of type: {type(model).__name__}")
-                return model
-        except pickle.UnpicklingError as e:
-            logger.error(f"Security: Unsafe model file rejected: {e}")
-            raise SecurityError(f"Model file contains unsafe content: {e}") from e
-        except Exception as e:
-            logger.error(f"Error loading model: {e}")
-            raise
-    
-    @staticmethod
-    def safe_cloudpickle_load(file_path: Union[str, Path], *, safe_classes: Optional[Set[str]] = None) -> Any:
-        """
-        Safely load a cloudpickle file using restricted unpickler.
-        
-        Args:
-            file_path: Path to cloudpickle file
-            safe_classes: Optional custom set of safe classes
-            
-        Returns:
-            Unpickled object
-        """
-        file_path = Path(file_path)
-        
-        if not file_path.exists():
-            raise FileNotFoundError(f"Model file not found: {file_path}")
-        
-        # Calculate file hash for audit trail
-        file_hash = SecureModelLoader.calculate_file_hash(file_path)
-        logger.info(f"Loading cloudpickle file {file_path.name} (SHA256: {file_hash[:16]}...)")
-        
-        try:
-            with open(file_path, 'rb') as f:
-                # Use cloudpickle's load with our custom unpickler
-                # Note: This is a simplified approach - in practice, cloudpickle's
-                # load function would need to be modified to accept a custom unpickler
-                unpickler = RestrictedUnpickler(f, safe_classes=safe_classes)
-                model = unpickler.load()
-                logger.info(f"Successfully loaded cloudpickle model of type: {type(model).__name__}")
-                return model
-        except pickle.UnpicklingError as e:
-            logger.error(f"Security: Unsafe cloudpickle file rejected: {e}")
-            raise SecurityError(f"CloudPickle file contains unsafe content: {e}") from e
-        except Exception as e:
-            logger.error(f"Error loading cloudpickle model: {e}")
-            raise
-
-class SecurityError(Exception):
-    """Exception raised for security-related model loading issues."""
-    pass
-
-def add_safe_class(class_name: str) -> None:
-    """
-    Add a class to the safe loading allowlist.
-    
-    Args:
-        class_name: Full class name (e.g., 'mymodule.MyClass')
-    """
-    SAFE_PICKLE_CLASSES.add(class_name)
-    logger.info(f"Added {class_name} to safe loading allowlist")
-
-def remove_safe_class(class_name: str) -> None:
-    """
-    Remove a class from the safe loading allowlist.
-    
-    Args:
-        class_name: Full class name to remove
-    """
-    SAFE_PICKLE_CLASSES.discard(class_name)
-    logger.info(f"Removed {class_name} from safe loading allowlist")
-
-def get_safe_classes() -> Set[str]:
-    """Return a copy of the current safe classes set."""
-    return SAFE_PICKLE_CLASSES.copy()