metaflow 2.16.6__py2.py3-none-any.whl → 2.16.8__py2.py3-none-any.whl

metaflow/packaging_sys/tar_backend.py

@@ -1,7 +1,7 @@
 import tarfile
 
 from io import BytesIO
-from typing import IO, List, Optional, Union
+from typing import Any, IO, List, Optional, Union
 
 from .backend import PackagingBackend
 
@@ -56,6 +56,13 @@ class TarPackagingBackend(PackagingBackend):
     def cls_open(cls, content: IO[bytes]) -> tarfile.TarFile:
         return tarfile.open(fileobj=content, mode="r:gz")
 
+    @classmethod
+    def cls_member_name(cls, member: Union[tarfile.TarInfo, str]) -> str:
+        """
+        Returns the name of the member as a string.
+        """
+        return member.name if isinstance(member, tarfile.TarInfo) else member
+
     @classmethod
     def cls_has_member(cls, archive: tarfile.TarFile, name: str) -> bool:
         try:
@@ -76,11 +83,17 @@ class TarPackagingBackend(PackagingBackend):
     def cls_extract_members(
         cls,
         archive: tarfile.TarFile,
-        members: Optional[List[str]] = None,
+        members: Optional[List[Any]] = None,
         dest_dir: str = ".",
     ) -> None:
         archive.extractall(path=dest_dir, members=members)
 
     @classmethod
-    def cls_list_members(cls, archive: tarfile.TarFile) -> Optional[List[str]]:
+    def cls_list_members(
+        cls, archive: tarfile.TarFile
+    ) -> Optional[List[tarfile.TarInfo]]:
+        return archive.getmembers() or None
+
+    @classmethod
+    def cls_list_names(cls, archive: tarfile.TarFile) -> Optional[List[str]]:
         return archive.getnames() or None

metaflow/packaging_sys/v1.py

@@ -61,23 +61,25 @@ class MetaflowCodeContentV1(MetaflowCodeContentV1Base):
         else:
             new_modules = []
 
-        self._modules = {
-            name: _ModuleInfo(
-                name,
-                set(
-                    Path(p).resolve().as_posix()
-                    for p in getattr(mod, "__path__", [mod.__file__])
-                ),
-                mod,
-                True,  # This is a Metaflow module (see filter below)
-            )
-            for (name, mod) in new_modules
-        }
-
-        # Filter the modules
-        self._modules = {
-            name: info for name, info in self._modules.items() if criteria(info.module)
-        }
+        self._modules = {}  # type: Dict[str, _ModuleInfo]
+        # We do this explicitly module by module to harden it against misbehaving
+        # modules like the one in:
+        # https://github.com/Netflix/metaflow/issues/2512
+        # We will silently ignore modules that are not well built.
+        for name, mod in new_modules:
+            try:
+                minfo = _ModuleInfo(
+                    name,
+                    set(
+                        Path(p).resolve().as_posix()
+                        for p in getattr(mod, "__path__", [mod.__file__])
+                    ),
+                    mod,
+                    True,  # This is a Metaflow module (see filter below)
+                )
+            except:
+                continue
+            self._modules[name] = minfo
 
         # Contain metadata information regarding the distributions packaged.
         # This allows Metaflow to "fake" distribution information when packaged
@@ -355,16 +357,14 @@ class MetaflowCodeContentV1(MetaflowCodeContentV1Base):
                 )
                 yield json.dumps(self.create_mfcontent_info()).encode(
                     "utf-8"
-                ), os.path.join(self._code_dir, MFCONTENT_MARKER)
+                ), MFCONTENT_MARKER
             else:
                 for k in self._other_content.keys():
                     yield "<generated %s content>" % (os.path.basename(k)), k
                 yield "<generated %s content>" % (
                     os.path.basename(self._dist_info_file)
                 ), os.path.join(self._other_dir, self._dist_info_file)
-                yield "<generated %s content>" % MFCONTENT_MARKER, os.path.join(
-                    self._code_dir, MFCONTENT_MARKER
-                )
+                yield "<generated %s content>" % MFCONTENT_MARKER, MFCONTENT_MARKER
 
     def _metaflow_distribution_files(self) -> Generator[Tuple[str, str], None, None]:
         debug.package_exec("Including Metaflow from '%s'" % self._metaflow_root)

metaflow/plugins/argo/argo_workflows.py

@@ -3226,8 +3226,8 @@ class ArgoWorkflows(object):
                     Trigger().template(
                         TriggerTemplate(self.name)
                         # Trigger a deployed workflow template
-                        .argo_workflow_trigger(
-                            ArgoWorkflowTrigger()
+                        .k8s_trigger(
+                            StandardK8STrigger()
                             .source(
                                 {
                                     "resource": {
@@ -4205,6 +4205,10 @@ class TriggerTemplate(object):
         self.payload = tree()
         self.payload["name"] = name
 
+    def k8s_trigger(self, k8s_trigger):
+        self.payload["k8s"] = k8s_trigger.to_json()
+        return self
+
     def argo_workflow_trigger(self, argo_workflow_trigger):
         self.payload["argoWorkflow"] = argo_workflow_trigger.to_json()
         return self
@@ -4279,3 +4283,57 @@ class TriggerParameter(object):
 
     def __str__(self):
         return json.dumps(self.payload, indent=4)
+
+
+class StandardK8STrigger(object):
+    # https://pkg.go.dev/github.com/argoproj/argo-events/pkg/apis/sensor/v1alpha1#StandardK8STrigger
+
+    def __init__(self):
+        tree = lambda: defaultdict(tree)
+        self.payload = tree()
+        self.payload["operation"] = "create"
+
+    def operation(self, operation):
+        self.payload["operation"] = operation
+        return self
+
+    def group(self, group):
+        self.payload["group"] = group
+        return self
+
+    def version(self, version):
+        self.payload["version"] = version
+        return self
+
+    def resource(self, resource):
+        self.payload["resource"] = resource
+        return self
+
+    def namespace(self, namespace):
+        self.payload["namespace"] = namespace
+        return self
+
+    def source(self, source):
+        self.payload["source"] = source
+        return self
+
+    def parameters(self, trigger_parameters):
+        if "parameters" not in self.payload:
+            self.payload["parameters"] = []
+        for trigger_parameter in trigger_parameters:
+            self.payload["parameters"].append(trigger_parameter.to_json())
+        return self
+
+    def live_object(self, live_object=True):
+        self.payload["liveObject"] = live_object
+        return self
+
+    def patch_strategy(self, patch_strategy):
+        self.payload["patchStrategy"] = patch_strategy
+        return self
+
+    def to_json(self):
+        return self.payload
+
+    def __str__(self):
+        return json.dumps(self.payload, indent=4)

metaflow/plugins/argo/argo_workflows_deployer_objects.py

@@ -310,6 +310,43 @@ class ArgoWorkflowsDeployedFlow(DeployedFlow):
 
         return cls(deployer=d)
 
+    @classmethod
+    def get_triggered_run(
+        cls, identifier: str, run_id: str, metadata: Optional[str] = None
+    ):
+        """
+        Retrieves a `ArgoWorkflowsTriggeredRun` object from an identifier, a run id and
+        optional metadata.
+
+        Parameters
+        ----------
+        identifier : str
+            Deployer specific identifier for the workflow to retrieve
+        run_id : str
+            Run ID for the which to fetch the triggered run object
+        metadata : str, optional, default None
+            Optional deployer specific metadata.
+
+        Returns
+        -------
+        ArgoWorkflowsTriggeredRun
+            A `ArgoWorkflowsTriggeredRun` object representing the
+            triggered run on argo workflows.
+        """
+        deployed_flow_obj = cls.from_deployment(identifier, metadata)
+        return ArgoWorkflowsTriggeredRun(
+            deployer=deployed_flow_obj.deployer,
+            content=json.dumps(
+                {
+                    "metadata": deployed_flow_obj.deployer.metadata,
+                    "pathspec": "/".join(
+                        (deployed_flow_obj.deployer.flow_name, run_id)
+                    ),
+                    "name": run_id,
+                }
+            ),
+        )
+
     @property
     def production_token(self) -> Optional[str]:
         """

metaflow/plugins/aws/step_functions/step_functions_deployer_objects.py

@@ -84,6 +84,22 @@ class StepFunctionsDeployedFlow(DeployedFlow):
             "from_deployment is not implemented for StepFunctions"
         )
 
+    @classmethod
+    def get_triggered_run(
+        cls, identifier: str, run_id: str, metadata: Optional[str] = None
+    ):
+        """
+        This method is not currently implemented for Step Functions.
+
+        Raises
+        ------
+        NotImplementedError
+            This method is not implemented for Step Functions.
+        """
+        raise NotImplementedError(
+            "get_triggered_run is not implemented for StepFunctions"
+        )
+
     @property
     def production_token(self: DeployedFlow) -> Optional[str]:
         """

metaflow/plugins/cards/card_modules/convert_to_native_type.py

@@ -146,7 +146,13 @@ class TaskToDict:
             # Python 3.13 removes the standard ``imghdr`` module. Metaflow
             # vendors a copy so we can keep using ``what`` to detect image
             # formats irrespective of the Python version.
-            from metaflow._vendor import imghdr
+            import warnings
+
+            with warnings.catch_warnings():
+                warnings.filterwarnings(
+                    "ignore", category=DeprecationWarning, module="imghdr"
+                )
+                from metaflow._vendor import imghdr
 
             resp = imghdr.what(None, h=data_object)
             # Only accept types supported on the web

metaflow/plugins/kubernetes/kubernetes_jobsets.py

@@ -6,6 +6,7 @@ from collections import namedtuple
 from metaflow.exception import MetaflowException
 from metaflow.metaflow_config import KUBERNETES_JOBSET_GROUP, KUBERNETES_JOBSET_VERSION
 from metaflow.tracing import inject_tracing_vars
+from metaflow._vendor import yaml
 
 from .kube_utils import qos_requests_and_limits
 
@@ -1025,34 +1026,32 @@ class KubernetesArgoJobSet(object):
 
     def dump(self):
         client = self._kubernetes_sdk
-
-        data = json.dumps(
-            client.ApiClient().sanitize_for_serialization(
-                dict(
-                    apiVersion=self._group + "/" + self._version,
-                    kind="JobSet",
-                    metadata=client.api_client.ApiClient().sanitize_for_serialization(
-                        client.V1ObjectMeta(
-                            name=self.name,
-                            labels=self._labels,
-                            annotations=self._annotations,
-                        )
-                    ),
-                    spec=dict(
-                        replicatedJobs=[self.control.dump(), self.worker.dump()],
-                        suspend=False,
-                        startupPolicy=None,
-                        successPolicy=None,
-                        # The Failure Policy helps setting the number of retries for the jobset.
-                        # but we don't rely on it and instead rely on either the local scheduler
-                        # or the Argo Workflows to handle retries.
-                        failurePolicy=None,
-                        network=None,
-                    ),
-                    status=None,
-                )
+        js_dict = client.ApiClient().sanitize_for_serialization(
+            dict(
+                apiVersion=self._group + "/" + self._version,
+                kind="JobSet",
+                metadata=client.api_client.ApiClient().sanitize_for_serialization(
+                    client.V1ObjectMeta(
+                        name=self.name,
+                        labels=self._labels,
+                        annotations=self._annotations,
+                    )
+                ),
+                spec=dict(
+                    replicatedJobs=[self.control.dump(), self.worker.dump()],
+                    suspend=False,
+                    startupPolicy=None,
+                    successPolicy=None,
+                    # The Failure Policy helps setting the number of retries for the jobset.
+                    # but we don't rely on it and instead rely on either the local scheduler
+                    # or the Argo Workflows to handle retries.
+                    failurePolicy=None,
+                    network=None,
+                ),
+                status=None,
             )
         )
+        data = yaml.dump(js_dict, default_flow_style=False, indent=2)
         # The values we populate in the Jobset manifest (for Argo Workflows) piggybacks on the Argo Workflow's templating engine.
         # Even though Argo Workflows's templating helps us constructing all the necessary IDs and populating the fields
         # required by Metaflow, we run into one glitch. When we construct JSON/YAML serializable objects,
@@ -1067,7 +1066,6 @@ class KubernetesArgoJobSet(object):
         # Since the value of `num_parallel` can be dynamic and can change from run to run, we need to ensure that the
         # value can be passed-down dynamically and is **explicitly set as a integer** in the Jobset Manifest submitted as a
         # part of the Argo Workflow
-
-        quoted_substring = '"{{=asInt(inputs.parameters.workerCount)}}"'
+        quoted_substring = "'{{=asInt(inputs.parameters.workerCount)}}'"
         unquoted_substring = "{{=asInt(inputs.parameters.workerCount)}}"
         return data.replace(quoted_substring, unquoted_substring)

metaflow/plugins/pypi/conda_decorator.py

@@ -243,9 +243,11 @@ class CondaStepDecorator(StepDecorator):
         # Ensure local installation of Metaflow is visible to user code
         python_path = self.__class__._metaflow_home.name
         addl_env_vars = {}
-        if self.__class__._addl_env_vars is not None:
+        if self.__class__._addl_env_vars:
             for key, value in self.__class__._addl_env_vars.items():
-                if key == "PYTHONPATH":
+                if key.endswith(":"):
+                    addl_env_vars[key[:-1]] = value
+                elif key == "PYTHONPATH":
                     addl_env_vars[key] = os.pathsep.join([value, python_path])
                 else:
                     addl_env_vars[key] = value

metaflow/runner/click_api.py

@@ -43,6 +43,7 @@ from metaflow._vendor.click.types import (
 )
 from metaflow.decorators import add_decorator_options
 from metaflow.exception import MetaflowException
+from metaflow.flowspec import _FlowState
 from metaflow.includefile import FilePathClass
 from metaflow.metaflow_config import CLICK_API_PROCESS_CONFIG
 from metaflow.parameters import JSONTypeClass, flow_context
@@ -171,7 +172,6 @@ def _lazy_load_command(
     _self,
     name: str,
 ):
-
     # Context is not used in get_command so we can pass None. Since we pin click,
     # this won't change from under us.
 
@@ -516,6 +516,11 @@ class MetaflowAPI(object):
         # Note that if CLICK_API_PROCESS_CONFIG is False, we still do this because
         # it will init all parameters (config_options will be None)
         # We ignore any errors if we don't check the configs in the click API.
+
+        # Init all values in the flow mutators and then process them
+        for decorator in self._flow_cls._flow_state.get(_FlowState.FLOW_MUTATORS, []):
+            decorator.external_init()
+
         new_cls = self._flow_cls._process_config_decorators(
             config_options, process_configs=CLICK_API_PROCESS_CONFIG
         )
@@ -541,14 +546,16 @@ def extract_all_params(cmd_obj: Union[click.Command, click.Group]):
 
     for each_param in cmd_obj.params:
         if isinstance(each_param, click.Argument):
-            arg_params_sigs[each_param.name], annotations[each_param.name] = (
-                get_inspect_param_obj(each_param, inspect.Parameter.POSITIONAL_ONLY)
-            )
+            (
+                arg_params_sigs[each_param.name],
+                annotations[each_param.name],
+            ) = get_inspect_param_obj(each_param, inspect.Parameter.POSITIONAL_ONLY)
             arg_parameters[each_param.name] = each_param
         elif isinstance(each_param, click.Option):
-            opt_params_sigs[each_param.name], annotations[each_param.name] = (
-                get_inspect_param_obj(each_param, inspect.Parameter.KEYWORD_ONLY)
-            )
+            (
+                opt_params_sigs[each_param.name],
+                annotations[each_param.name],
+            ) = get_inspect_param_obj(each_param, inspect.Parameter.KEYWORD_ONLY)
             opt_parameters[each_param.name] = each_param
 
         defaults[each_param.name] = each_param.default

metaflow/runner/deployer.py

@@ -231,6 +231,67 @@ class DeployedFlowMeta(type):
                 }
             )
 
+            def _get_triggered_run_injected_method():
+                def f(
+                    cls,
+                    identifier: str,
+                    run_id: str,
+                    metadata: Optional[str] = None,
+                    impl: str = DEFAULT_FROM_DEPLOYMENT_IMPL.replace("-", "_"),
+                ) -> "TriggeredRun":
+                    """
+                    Retrieves a `TriggeredRun` object from an identifier, a run id and optional
+                    metadata. The `impl` parameter specifies the deployer implementation
+                    to use (like `argo-workflows`).
+
+                    Parameters
+                    ----------
+                    identifier : str
+                        Deployer specific identifier for the workflow to retrieve
+                    run_id : str
+                        Run ID for the which to fetch the triggered run object
+                    metadata : str, optional, default None
+                        Optional deployer specific metadata.
+                    impl : str, optional, default given by METAFLOW_DEFAULT_FROM_DEPLOYMENT_IMPL
+                        The default implementation to use if not specified
+
+                    Returns
+                    -------
+                    TriggeredRun
+                        A `TriggeredRun` object representing the triggered run corresponding
+                        to the identifier and the run id.
+                    """
+                    if impl in allowed_providers:
+                        return (
+                            allowed_providers[impl]
+                            .deployed_flow_type()
+                            .get_triggered_run(identifier, run_id, metadata)
+                        )
+                    else:
+                        raise ValueError(
+                            f"No deployer '{impl}' exists; valid deployers are: "
+                            f"{list(allowed_providers.keys())}"
+                        )
+
+                f.__name__ = "get_triggered_run"
+                return f
+
+            def _per_type_get_triggered_run_injected_method(method_name, impl):
+                def f(
+                    cls,
+                    identifier: str,
+                    run_id: str,
+                    metadata: Optional[str] = None,
+                ):
+                    return (
+                        allowed_providers[impl]
+                        .deployed_flow_type()
+                        .get_triggered_run(identifier, run_id, metadata)
+                    )
+
+                f.__name__ = method_name
+                return f
+
             def _from_deployment_injected_method():
                 def f(
                     cls,
@@ -347,10 +408,16 @@ class DeployedFlowMeta(type):
                 "list_deployed_flows",
                 classmethod(_list_deployed_flows_injected_method()),
             )
+            setattr(
+                cls,
+                "get_triggered_run",
+                classmethod(_get_triggered_run_injected_method()),
+            )
 
             for impl in allowed_providers:
                 from_deployment_method_name = f"from_{impl}"
                 list_deployed_flows_method_name = f"list_{impl}"
+                get_triggered_run_method_name = f"get_triggered_{impl}_run"
 
                 setattr(
                     cls,
@@ -372,6 +439,16 @@ class DeployedFlowMeta(type):
                     ),
                 )
 
+                setattr(
+                    cls,
+                    get_triggered_run_method_name,
+                    classmethod(
+                        _per_type_get_triggered_run_injected_method(
+                            get_triggered_run_method_name, impl
+                        )
+                    ),
+                )
+
         return cls
 
 

metaflow/runner/subprocess_manager.py

@@ -152,12 +152,20 @@ class SubprocessManager(object):
         int
             The process ID of the subprocess.
         """
-        updated_env = MetaflowCodeContent.get_env_vars_for_packaged_metaflow(
-            get_metaflow_root()
-        )
-        if updated_env:
-            env = env or {}
-            env.update(updated_env)
+        env = env or {}
+        installed_root = os.environ.get("METAFLOW_EXTRACTED_ROOT", get_metaflow_root())
+
+        for k, v in MetaflowCodeContent.get_env_vars_for_packaged_metaflow(
+            installed_root
+        ).items():
+            if k.endswith(":"):
+                # Override
+                env[k[:-1]] = v
+            elif k in env:
+                env[k] = "%s:%s" % (v, env[k])
+            else:
+                env[k] = v
+
         command_obj = CommandManager(command, env, cwd)
         pid = command_obj.run(show_output=show_output)
         self.commands[pid] = command_obj
@@ -188,12 +196,12 @@ class SubprocessManager(object):
         int
             The process ID of the subprocess.
         """
-        updated_env = MetaflowCodeContent.get_env_vars_for_packaged_metaflow(
-            get_metaflow_root()
-        )
-        if updated_env:
-            env = env or {}
-            env.update(updated_env)
+        env = env or {}
+        if "PYTHONPATH" in env:
+            env["PYTHONPATH"] = "%s:%s" % (get_metaflow_root(), env["PYTHONPATH"])
+        else:
+            env["PYTHONPATH"] = get_metaflow_root()
+
         command_obj = CommandManager(command, env, cwd)
         pid = await command_obj.async_run()
         self.commands[pid] = command_obj

metaflow/vendor.py

@@ -63,14 +63,29 @@ def find_vendored_libs(vendor_dir, whitelist, whitelist_dirs):
     return vendored_libs, paths
 
 
-def fetch_licenses(*info_dir, vendor_dir):
-    for file in chain.from_iterable(map(iter_subtree, info_dir)):
-        if "LICENSE" in file.name:
-            library = file.parent.name.split("-")[0]
-            shutil.copy(file, vendor_dir / ("%s.LICENSE" % library))
-        else:
+def fetch_licenses(*info_dirs, vendor_dir):
+    for dist_info in info_dirs:
+        metadata_file = dist_info / "METADATA"
+        if not metadata_file.exists():
+            continue
+
+        project_name = None
+        for line in metadata_file.read_text("utf-8").splitlines():
+            if line.startswith("Name: "):
+                project_name = line.split("Name: ", 1)[1].strip()
+                break
+        if not project_name:
             continue
 
+        for item in dist_info.iterdir():
+            if item.is_file() and re.search(r"(LICENSE|COPYING)", item.name, re.I):
+                shutil.copy(item, vendor_dir / f"{project_name}.LICENSE")
+            elif item.is_dir() and item.name.lower() == "licenses":
+                for license_file in item.iterdir():
+                    if license_file.is_file():
+                        dest_name = f"{project_name}.{license_file.name}"
+                        shutil.copy(license_file, vendor_dir / dest_name)
+
 
 def vendor(vendor_dir):
     # remove everything
@@ -108,6 +123,8 @@ def vendor(vendor_dir):
                 "-r",
                 "_vendor/vendor_%s.txt" % subdir,
                 "--no-compile",
+                "--no-binary",
+                ":all:",
             ]
         )
 

metaflow/version.py

@@ -1 +1 @@
-metaflow_version = "2.16.6"
+metaflow_version = "2.16.8"

{metaflow-2.16.6.dist-info → metaflow-2.16.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: metaflow
-Version: 2.16.6
+Version: 2.16.8
 Summary: Metaflow: More AI and ML, Less Engineering
 Author: Metaflow Developers
 Author-email: help@metaflow.org
@@ -26,7 +26,7 @@ License-File: LICENSE
 Requires-Dist: requests
 Requires-Dist: boto3
 Provides-Extra: stubs
-Requires-Dist: metaflow-stubs==2.16.6; extra == "stubs"
+Requires-Dist: metaflow-stubs==2.16.8; extra == "stubs"
 Dynamic: author
 Dynamic: author-email
 Dynamic: classifier