vantage6-algorithm-tools 4.3.4__tar.gz → 4.4.0__tar.gz

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vantage6-algorithm-tools
-Version: 4.3.4
+Version: 4.4.0
 Summary: Vantage6 algorithm tools
 Home-page: https://github.com/vantage6/vantage6
 Requires-Python: >=3.6

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6/algorithm/client/__init__.py

@@ -258,9 +258,13 @@ class AlgorithmClient(ClientBase):
             self.parent.log.info("--> Attempting to decode results!")
             result = None
             if response.get("result"):
-                result = json_lib.loads(
-                    base64s_to_bytes(response.get("result")).decode()
-                )
+                try:
+                    result = json_lib.loads(
+                        base64s_to_bytes(response.get("result")).decode()
+                    )
+                except Exception as e:
+                    self.parent.log.error("Unable to load results")
+                    self.parent.log.debug(e)
             return result
 
         def from_task(self, task_id: int) -> list[Any]:

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6/algorithm/client/_version.py

@@ -7,7 +7,7 @@ with open(os.path.join(here, "__build__")) as fp:
     __build__ = json.load(fp)
 
 # Module version
-version_info = (4, 3, 4, "final", __build__, 0)
+version_info = (4, 4, 0, "final", __build__, 0)
 
 # Module version stage suffix map
 _specifier_ = {"alpha": "a", "beta": "b", "candidate": "rc", "final": ""}

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6/algorithm/tools/decorators.py

@@ -10,7 +10,7 @@ import pandas as pd
 
 from vantage6.algorithm.client import AlgorithmClient
 from vantage6.algorithm.tools.mock_client import MockAlgorithmClient
-from vantage6.algorithm.tools.util import info, error, warn, get_env_var
+from vantage6.algorithm.tools.util import info, error, warn
 from vantage6.algorithm.tools.wrappers import load_data
 from vantage6.algorithm.tools.preprocessing import preprocess_data
 
@@ -93,12 +93,12 @@ def _algorithm_client() -> callable:
             if mock_client is not None:
                 return func(mock_client, *args, **kwargs)
             # read server address from the environment
-            host = get_env_var("HOST")
-            port = get_env_var("PORT")
-            api_path = get_env_var("API_PATH")
+            host = os.environ["HOST"]
+            port = os.environ["PORT"]
+            api_path = os.environ["API_PATH"]
 
             # read token from the environment
-            token_file = get_env_var("TOKEN_FILE")
+            token_file = os.environ["TOKEN_FILE"]
             info("Reading token")
             with open(token_file) as fp:
                 token = fp.read().strip()
@@ -195,7 +195,7 @@ def data(number_of_databases: int = 1) -> callable:
 
                 # do any data preprocessing here
                 info(f"Applying preprocessing for database '{label}'")
-                env_prepro = get_env_var(f"{label.upper()}_PREPROCESSING")
+                env_prepro = os.environ.get(f"{label.upper()}_PREPROCESSING")
                 if env_prepro is not None:
                     preprocess = json.loads(env_prepro)
                     data_ = preprocess_data(data_, preprocess)
@@ -309,7 +309,7 @@ def metadata(func: callable) -> callable:
         >>> def my_algorithm(metadata: RunMetaData, <other arguments>):
         >>>     pass
         """
-        token_file = get_env_var("TOKEN_FILE")
+        token_file = os.environ["TOKEN_FILE"]
         info("Reading token")
         with open(token_file) as fp:
             token = fp.read().strip()
@@ -322,10 +322,10 @@ def metadata(func: callable) -> callable:
             node_id=payload["node_id"],
             collaboration_id=payload["collaboration_id"],
             organization_id=payload["organization_id"],
-            temporary_directory=Path(get_env_var("TEMPORARY_FOLDER")),
-            output_file=Path(get_env_var("OUTPUT_FILE")),
-            input_file=Path(get_env_var("INPUT_FILE")),
-            token_file=Path(get_env_var("TOKEN_FILE")),
+            temporary_directory=Path(os.environ["TEMPORARY_FOLDER"]),
+            output_file=Path(os.environ["OUTPUT_FILE"]),
+            input_file=Path(os.environ["INPUT_FILE"]),
+            token_file=Path(os.environ["TOKEN_FILE"]),
         )
         return func(metadata, *args, **kwargs)
 
@@ -355,11 +355,11 @@ def get_ohdsi_metadata(label: str) -> OHDSIMetaData:
     for var in expected_env_vars:
         _check_environment_var_exists_or_exit(f"{label_}_DB_PARAM_{var}")
 
-    tmp = Path(get_env_var("TEMPORARY_FOLDER"))
+    tmp = Path(os.environ["TEMPORARY_FOLDER"])
     metadata = OHDSIMetaData(
-        database=get_env_var(f"{label_}_DB_PARAM_CDM_DATABASE"),
-        cdm_schema=get_env_var(f"{label_}_DB_PARAM_CDM_SCHEMA"),
-        results_schema=get_env_var(f"{label_}_DB_PARAM_RESULTS_SCHEMA"),
+        database=os.environ[f"{label_}_DB_PARAM_CDM_DATABASE"],
+        cdm_schema=os.environ[f"{label_}_DB_PARAM_CDM_SCHEMA"],
+        results_schema=os.environ[f"{label_}_DB_PARAM_RESULTS_SCHEMA"],
         incremental_folder=tmp / "incremental",
         cohort_statistics_folder=tmp / "cohort_statistics",
         export_folder=tmp / "export",
@@ -417,10 +417,10 @@ def _create_omop_database_connection(label: str) -> callable:
         _check_environment_var_exists_or_exit(f"{label_}_DB_PARAM_{var}")
 
     info("Reading OHDSI environment variables")
-    dbms = get_env_var(f"{label_}_DB_PARAM_DBMS")
-    uri = get_env_var(f"{label_}_DATABASE_URI")
-    user = get_env_var(f"{label_}_DB_PARAM_USER")
-    password = get_env_var(f"{label_}_DB_PARAM_PASSWORD")
+    dbms = os.environ[f"{label_}_DB_PARAM_DBMS"]
+    uri = os.environ[f"{label_}_DATABASE_URI"]
+    user = os.environ[f"{label_}_DB_PARAM_USER"]
+    password = os.environ[f"{label_}_DB_PARAM_PASSWORD"]
     info(f" - dbms: {dbms}")
     info(f" - uri: {uri}")
     info(f" - user: {user}")
@@ -460,20 +460,20 @@ def _get_data_from_label(label: str) -> pd.DataFrame:
         Data from the database
     """
     # Load the input data from the input file - this may e.g. include the
-    database_uri = get_env_var(f"{label.upper()}_DATABASE_URI")
+    database_uri = os.environ[f"{label.upper()}_DATABASE_URI"]
     info(f"Using '{database_uri}' with label '{label}' as database")
 
     # Get the database type from the environment variable, this variable is
     # set by the vantage6 node based on its configuration file.
-    database_type = get_env_var(f"{label.upper()}_DATABASE_TYPE", "csv").lower()
+    database_type = os.environ.get(f"{label.upper()}_DATABASE_TYPE", "csv").lower()
 
     # Load the data based on the database type. Try to provide environment
     # variables that should be available for some data types.
     return load_data(
         database_uri,
         database_type,
-        query=get_env_var(f"{label.upper()}_QUERY"),
-        sheet_name=get_env_var(f"{label.upper()}_SHEET_NAME"),
+        query=os.environ.get(f"{label.upper()}_QUERY"),
+        sheet_name=os.environ.get(f"{label.upper()}_SHEET_NAME"),
     )
 
 
@@ -488,7 +488,7 @@ def _get_user_database_labels() -> list[str]:
     """
     # read the labels that the user requested, which is a comma
     # separated list of labels.
-    labels = get_env_var("USER_REQUESTED_DATABASE_LABELS")
+    labels = os.environ["USER_REQUESTED_DATABASE_LABELS"]
     return labels.split(",")
 
 

vantage6-algorithm-tools-4.4.0/vantage6/algorithm/tools/exceptions.py

@@ -0,0 +1,138 @@
+class AlgorithmError(Exception):
+    """Generic exception raised when an algorithm fails."""
+
+
+# ---------------- Privacy exceptions ----------------
+class PrivacyViolation(AlgorithmError):
+    """Generic exception raised for data privacy concerns."""
+
+
+class PrivacyThresholdViolation(PrivacyViolation):
+    """
+    Raised when privacy threshold is violated.
+
+    Example usage:
+    - The number of rows in the data is too low.
+    - Returning the results of the algorithm would violate privacy.
+    """
+
+
+# ---------------- Data exceptions ----------------
+
+
+class DataError(AlgorithmError):
+    """Generic error raised with data handling."""
+
+
+class DataReadError(DataError):
+    """Raised when data reading fails.
+
+    Example usage:
+    - File not found.
+    - File is not in the right format.
+    - File is not readable.
+    - File is empty.
+    """
+
+
+class DataTypeError(DataError):
+    """Raised when data type is invalid.
+
+    Example usage:
+    - String column is selected by user for numeric operation.
+    """
+
+
+# ---------------- Runtime exceptions ----------------
+class AlgorithmRuntimeError(AlgorithmError):
+    """Generic error raised when an algorithm fails at runtime."""
+
+
+class AlgorithmExecutionError(AlgorithmRuntimeError):
+    """Raised when algorithm function fails.
+
+    Use when the algorithm function raises an exception.
+    """
+
+
+class MaxIterationsReached(AlgorithmRuntimeError):
+    """Raised when the maximum number of iterations is reached."""
+
+
+class ConvergenceError(AlgorithmRuntimeError):
+    """Raised when the algorithm fails to converge."""
+
+
+# ---------------- Client exceptions ----------------
+
+
+class ClientError(AlgorithmError):
+    """Generic error raised when call to the algorithm client fails."""
+
+
+# the most common client errors are defined separately for clarity
+class SubtakCreationError(ClientError):
+    """Raised when subtask creation fails."""
+
+
+class CollectOrganizationError(ClientError):
+    """Raised when organization collection fails."""
+
+
+class CollectResultsError(ClientError):
+    """Raised when result collection fails."""
+
+
+# ---------------- Input exceptions ----------------
+
+
+class InputError(AlgorithmError):
+    """Generic error raised with algorithm input handling.
+    Example usage:
+    - User input is invalid.
+    - Subtask fails due to invalid input received from the parent task.
+    """
+
+
+class UserInputError(InputError):
+    """Raised when user input is invalid.
+
+    Example usage:
+    - User input is not in the expected format.
+    """
+
+
+class DeserializationError(InputError):
+    """Raised when result deserialization fails."""
+
+
+# TODO v5+ remove this alias, which is there for backwards compatibility
+DeserializationException = DeserializationError
+
+
+class EnvironmentVariableError(InputError):
+    """Raised when environment variable is not found."""
+
+
+# ---------------- Initialization exceptions ----------------
+
+
+class AlgorithmInitializationError(AlgorithmError):
+    """Generic error raised when algorithm initialization fails."""
+
+
+class AlgorithmModuleNotFoundError(AlgorithmInitializationError):
+    """
+    Raised when the algorithm module is not found.
+
+    Note that if this error is raised, the algorithm image is not built correctly.
+    """
+
+
+class MethodNotFoundError(AlgorithmInitializationError):
+    """
+    Raised when the algorithm method is not found.
+
+    This error may be raised if the user calls a non-existing method, or if the
+    algorithm image is not built correctly.
+    """

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6/algorithm/tools/util.py

@@ -1,6 +1,8 @@
 import sys
 import os
 import base64
+import binascii
+
 from vantage6.common.globals import STRING_ENCODING, ENV_VAR_EQUALS_REPLACEMENT
 
 
@@ -40,6 +42,9 @@ def error(msg: str) -> None:
     sys.stdout.write(f"error > {msg}\n")
 
 
+# TODO v5+ move this function to wrap.py and no longer expose it to be used by
+# algorithms but as part of _decode_env_vars. It is kept here for backwards
+# compatibility with 4.2/4.3 algorithms
 def get_env_var(var_name: str, default: str | None = None) -> str:
     """
     Get the value of an environment variable. Environment variables are encoded
@@ -69,3 +74,6 @@ def get_env_var(var_name: str, default: str | None = None) -> str:
         return base64.b32decode(encoded_env_var_value).decode(STRING_ENCODING)
     except KeyError:
         return default
+    except binascii.Error:
+        # If the decoding fails, return the original value
+        return os.environ[var_name]

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6/algorithm/tools/wrap.py

@@ -7,7 +7,7 @@ from typing import Any
 from vantage6.common.client import deserialization
 from vantage6.common import serialization
 from vantage6.algorithm.tools.util import info, error, get_env_var
-from vantage6.algorithm.tools.exceptions import DeserializationException
+from vantage6.algorithm.tools.exceptions import DeserializationError
 
 
 def wrap_algorithm(log_traceback: bool = True) -> None:
@@ -52,8 +52,11 @@ def wrap_algorithm(log_traceback: bool = True) -> None:
         exit(1)
     info(f"wrapper for {module}")
 
+    # Decode environment variables that are encoded by the node.
+    _decode_env_vars()
+
     # read input from the mounted input file.
-    input_file = get_env_var("INPUT_FILE")
+    input_file = os.environ["INPUT_FILE"]
     info(f"Reading input file {input_file}")
     input_data = load_input(input_file)
 
@@ -63,7 +66,7 @@ def wrap_algorithm(log_traceback: bool = True) -> None:
 
     # write output from the method to mounted output file. Which will be
     # transferred back to the server by the node-instance.
-    output_file = get_env_var("OUTPUT_FILE")
+    output_file = os.environ["OUTPUT_FILE"]
     info(f"Writing output to {output_file}")
 
     _write_output(output, output_file)
@@ -145,14 +148,14 @@ def load_input(input_file: str) -> Any:
 
     Raises
     ------
-    DeserializationException
+    DeserializationError
         Failed to deserialize input data
     """
     with open(input_file, "rb") as fp:
         try:
             input_data = deserialization.deserialize(fp)
-        except DeserializationException:
-            raise DeserializationException("Could not deserialize input")
+        except DeserializationError:
+            raise DeserializationError("Could not deserialize input")
     return input_data
 
 
@@ -170,3 +173,16 @@ def _write_output(output: Any, output_file: str) -> None:
     with open(output_file, "wb") as fp:
         serialized = serialization.serialize(output)
         fp.write(serialized)
+
+
+def _decode_env_vars() -> None:
+    """
+    Decode environment variables that are encoded by the node
+
+    Note that environment variables may be present that are not specific to vantage6,
+    such as HOME, PATH, etc. These are not encoded by the node and should not be
+    decoded here. The `get_env_var` function handles these properly so that the
+    original value is returned if the environment variable is not encoded.
+    """
+    for env_var in os.environ:
+        os.environ[env_var] = get_env_var(env_var)

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6_algorithm_tools.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vantage6-algorithm-tools
-Version: 4.3.4
+Version: 4.4.0
 Summary: Vantage6 algorithm tools
 Home-page: https://github.com/vantage6/vantage6
 Requires-Python: >=3.6

{vantage6-algorithm-tools-4.3.4 → vantage6-algorithm-tools-4.4.0}/vantage6_algorithm_tools.egg-info/requires.txt

@@ -3,7 +3,7 @@ pandas>=1.5.3
 PyJWT==2.6.0
 pyfiglet==0.8.post1
 SPARQLWrapper>=2.0.0
-vantage6-common==4.3.4
+vantage6-common==4.4.0
 
 [dev]
 black

vantage6-algorithm-tools-4.3.4/vantage6/algorithm/tools/exceptions.py

@@ -1,4 +0,0 @@
-class DeserializationException(Exception):
-    """Raised when result deserialization fails."""
-
-    pass