runnable 0.11.1__tar.gz → 0.11.2__tar.gz

{runnable-0.11.1 → runnable-0.11.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: runnable
-Version: 0.11.1
+Version: 0.11.2
 Summary: A Compute agnostic pipelining software
 Home-page: https://github.com/vijayvammi/runnable
 License: Apache-2.0

{runnable-0.11.1 → runnable-0.11.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "runnable"
-version = "0.11.1"
+version = "0.11.2"
 description = "A Compute agnostic pipelining software"
 authors = ["Vijay Vammi <mesanthu@gmail.com>"]
 license = "Apache-2.0"
@@ -53,6 +53,10 @@ matplotlib = "^3.8.3"
 [tool.poetry.group.release.dependencies]
 python-semantic-release = "^9.4.2"
 
+
+[tool.poetry.group.examples.dependencies]
+pandas = "^2.2.2"
+
 [tool.poetry.extras]
 docker = ['docker']
 notebook = ['ploomber-engine']

{runnable-0.11.1 → runnable-0.11.2}/runnable/__init__.py

@@ -38,3 +38,4 @@ from runnable.sdk import (  # noqa
 
 
 # TODO: Think of way of generating dag hash without executor configuration
+# TODO: Add examples of map parameters and task types

{runnable-0.11.1 → runnable-0.11.2}/runnable/datastore.py

@@ -312,8 +312,10 @@ class RunLog(BaseModel):
         summary["Catalog Location"] = _context.catalog_handler.get_summary()
         summary["Full Run log present at: "] = _context.run_log_store.get_summary()
 
-        summary["Final Parameters"] = {p: v.description for p, v in self.parameters.items()}
-        summary["Collected metrics"] = {p: v.description for p, v in self.parameters.items() if v.kind == "metric"}
+        run_log = _context.run_log_store.get_run_log_by_id(run_id=_context.run_id, full=True)
+
+        summary["Final Parameters"] = {p: v.description for p, v in run_log.parameters.items()}
+        summary["Collected metrics"] = {p: v.description for p, v in run_log.parameters.items() if v.kind == "metric"}
 
         return summary
 

{runnable-0.11.1 → runnable-0.11.2}/runnable/extensions/catalog/file_system/implementation.py

@@ -226,7 +226,7 @@ class FileSystemCatalog(BaseCatalog):
         for cataloged_file in cataloged_files:
             if str(cataloged_file).endswith("execution.log"):
                 continue
-            print(cataloged_file.name)
+
             if cataloged_file.is_file():
                 shutil.copy(cataloged_file, run_catalog / cataloged_file.name)
             else:

{runnable-0.11.1 → runnable-0.11.2}/runnable/extensions/nodes.py

@@ -505,7 +505,7 @@ class MapNode(CompositeNode):
             for _, v in map_variable.items():
                 for branch_return in self.branch_returns:
                     param_name, param_type = branch_return
-                    raw_parameters[f"{param_name}_{v}"] = param_type.copy()
+                    raw_parameters[f"{v}_{param_name}"] = param_type.copy()
         else:
             for branch_return in self.branch_returns:
                 param_name, param_type = branch_return
@@ -606,9 +606,9 @@ class MapNode(CompositeNode):
                     param_name, _ = branch_return
                     to_reduce = []
                     for iter_variable in iterate_on:
-                        to_reduce.append(params[f"{param_name}_{iter_variable}"].get_value())
+                        to_reduce.append(params[f"{iter_variable}_{param_name}"].get_value())
 
-                    param_name = f"{param_name}_{v}"
+                    param_name = f"{v}_{param_name}"
                     params[param_name].value = reducer_f(to_reduce)
                     params[param_name].reduced = True
         else:
@@ -617,7 +617,7 @@ class MapNode(CompositeNode):
 
                 to_reduce = []
                 for iter_variable in iterate_on:
-                    to_reduce.append(params[f"{param_name}_{iter_variable}"].get_value())
+                    to_reduce.append(params[f"{iter_variable}_{param_name}"].get_value())
 
                 params[param_name].value = reducer_f(*to_reduce)
                 params[param_name].reduced = True

{runnable-0.11.1 → runnable-0.11.2}/runnable/extensions/run_log_store/generic_chunked.py

@@ -7,7 +7,16 @@ from string import Template
 from typing import Any, Dict, Optional, Sequence, Union
 
 from runnable import defaults, exceptions
-from runnable.datastore import BaseRunLogStore, BranchLog, RunLog, StepLog
+from runnable.datastore import (
+    BaseRunLogStore,
+    BranchLog,
+    JsonParameter,
+    MetricParameter,
+    ObjectParameter,
+    Parameter,
+    RunLog,
+    StepLog,
+)
 
 logger = logging.getLogger(defaults.LOGGER_NAME)
 
@@ -164,7 +173,9 @@ class ChunkedRunLogStore(BaseRunLogStore):
             raise Exception(f"Name is required during retrieval for {log_type}")
 
         naming_pattern = self.naming_pattern(log_type=log_type, name=name)
+
         matches = self.get_matches(run_id=run_id, name=naming_pattern, multiple_allowed=multiple_allowed)
+
         if matches:
             if not multiple_allowed:
                 contents = self._retrieve(name=matches)  # type: ignore
@@ -370,10 +381,17 @@ class ChunkedRunLogStore(BaseRunLogStore):
         Raises:
             RunLogNotFoundError: If the run log for run_id is not found in the datastore
         """
-        parameters = {}
+        parameters: Dict[str, Parameter] = {}
         try:
             parameters_list = self.retrieve(run_id=run_id, log_type=self.LogTypes.PARAMETER, multiple_allowed=True)
-            parameters = {key: value for param in parameters_list for key, value in param.items()}
+            for param in parameters_list:
+                for key, value in param.items():
+                    if value["kind"] == "json":
+                        parameters[key] = JsonParameter(**value)
+                    if value["kind"] == "metric":
+                        parameters[key] = MetricParameter(**value)
+                    if value["kind"] == "object":
+                        parameters[key] = ObjectParameter(**value)
         except EntityNotFoundError:
             # No parameters are set
             pass
@@ -401,7 +419,7 @@ class ChunkedRunLogStore(BaseRunLogStore):
             self.store(
                 run_id=run_id,
                 log_type=self.LogTypes.PARAMETER,
-                contents={key: value},
+                contents={key: value.model_dump(by_alias=True)},
                 name=key,
             )
 

{runnable-0.11.1 → runnable-0.11.2}/runnable/tasks.py

@@ -1,4 +1,5 @@
 import contextlib
+import copy
 import importlib
 import io
 import json
@@ -99,6 +100,20 @@ class BaseTaskType(BaseModel):
         """
         raise NotImplementedError()
 
+    def _diff_parameters(
+        self, parameters_in: Dict[str, Parameter], context_params: Dict[str, Parameter]
+    ) -> Dict[str, Parameter]:
+        diff: Dict[str, Parameter] = {}
+        for param_name, param in context_params.items():
+            if param_name in parameters_in:
+                if parameters_in[param_name] != param:
+                    diff[param_name] = param
+                continue
+
+            diff[param_name] = param
+
+        return diff
+
     @contextlib.contextmanager
     def expose_secrets(self):
         """Context manager to expose secrets to the execution.
@@ -128,7 +143,7 @@ class BaseTaskType(BaseModel):
             if param.reduced is False:
                 context_param = param_name
                 for _, v in map_variable.items():  # type: ignore
-                    context_param = f"{context_param}_{v}"
+                    context_param = f"{v}_{context_param}"
 
                 if context_param in params:
                     params[param_name].value = params[context_param].value
@@ -147,6 +162,8 @@ class BaseTaskType(BaseModel):
 
         log_file = open(log_file_name, "w")
 
+        parameters_in = copy.deepcopy(params)
+
         f = io.StringIO()
         try:
             with contextlib.redirect_stdout(f):
@@ -168,7 +185,8 @@ class BaseTaskType(BaseModel):
 
             # Update parameters
             # This should only update the parameters that are changed at the root level.
-            self._context.run_log_store.set_parameters(parameters=params, run_id=self._context.run_id)
+            diff_parameters = self._diff_parameters(parameters_in=parameters_in, context_params=params)
+            self._context.run_log_store.set_parameters(parameters=diff_parameters, run_id=self._context.run_id)
 
 
 def task_return_to_parameter(task_return: TaskReturns, value: Any) -> Parameter:
@@ -259,7 +277,7 @@ class PythonTaskType(BaseTaskType):  # pylint: disable=too-few-public-methods
                         param_name = task_return.name
                         if map_variable:
                             for _, v in map_variable.items():
-                                param_name = f"{param_name}_{v}"
+                                param_name = f"{v}_{param_name}"
 
                         output_parameters[param_name] = output_parameter
 

{runnable-0.11.1 → runnable-0.11.2}/runnable/extensions/run_log_store/chunked_file_system/implementation.py

@@ -35,10 +35,10 @@ class ChunkedFileSystemRunLogStore(ChunkedRunLogStore):
             name (str): The suffix of the file name to check in the run log store.
         """
         log_folder = self.log_folder_with_run_id(run_id=run_id)
-
         sub_name = Template(name).safe_substitute({"creation_time": ""})
 
         matches = list(log_folder.glob(f"{sub_name}*"))
+
         if matches:
             if not multiple_allowed:
                 if len(matches) > 1: