ob-metaflow 2.16.8.2rc0__py2.py3-none-any.whl → 2.16.8.2rc2__py2.py3-none-any.whl

metaflow/metaflow_environment.py

@@ -203,19 +203,6 @@ class MetaflowEnvironment(object):
                     "mfcontent_version": 1,
                 }
             )
-
-        extra_exports = []
-        for k, v in MetaflowPackage.get_post_extract_env_vars(
-            code_package_metadata, dest_dir="$(pwd)"
-        ).items():
-            if k.endswith(":"):
-                # If the value ends with a colon, we override the existing value
-                extra_exports.append("export %s=%s" % (k[:-1], v))
-            else:
-                extra_exports.append(
-                    "export %s=%s:$(printenv %s)" % (k, v.replace('"', '\\"'), k)
-                )
-
         cmds = (
             [
                 BASH_MFLOG,
@@ -239,7 +226,12 @@ class MetaflowEnvironment(object):
             + MetaflowPackage.get_extract_commands(
                 code_package_metadata, "job.tar", dest_dir="."
             )
-            + extra_exports
+            + [
+                "export %s=%s:$(printenv %s)" % (k, v.replace('"', '\\"'), k)
+                for k, v in MetaflowPackage.get_post_extract_env_vars(
+                    code_package_metadata, dest_dir="."
+                ).items()
+            ]
             + [
                 "mflog 'Task is starting.'",
                 "flush_mflogs",

metaflow/package/__init__.py

@@ -17,6 +17,7 @@ from ..packaging_sys.utils import suffix_filter, walk
 from ..metaflow_config import DEFAULT_PACKAGE_SUFFIXES
 from ..exception import MetaflowException
 from ..user_configs.config_parameters import dump_config_values
+from ..util import get_metaflow_root
 from .. import R
 
 DEFAULT_SUFFIXES_LIST = DEFAULT_PACKAGE_SUFFIXES.split(",")
@@ -75,22 +76,12 @@ class MetaflowPackage(object):
             from ..user_decorators.user_flow_decorator import FlowMutatorMeta
             from ..user_decorators.user_step_decorator import UserStepDecoratorMeta
 
-            # Be very defensive here to filter modules in case there are
-            # some badly behaved modules that have weird values for
-            # METAFLOW_PACKAGE_POLICY for example.
-            try:
-                if (
-                    m.__name__ in FlowMutatorMeta._import_modules
-                    or m.__name__ in UserStepDecoratorMeta._import_modules
-                    or (
-                        hasattr(m, "METAFLOW_PACKAGE_POLICY")
-                        and m.METAFLOW_PACKAGE_POLICY == "include"
-                    )
-                ):
-                    return True
-                return False
-            except:
-                return False
+            if (
+                m.__name__ in FlowMutatorMeta._import_modules
+                or m.__name__ in UserStepDecoratorMeta._import_modules
+                or hasattr(m, "METAFLOW_PACKAGE")
+            ):
+                return True
 
         if mfcontent is None:
             self._mfcontent = MetaflowCodeContentV1(criteria=_module_selector)
@@ -359,10 +350,10 @@ class MetaflowPackage(object):
         """
         backend = cls.get_backend(pkg_metadata)
         with backend.cls_open(archive) as opened_archive:
-            include_members = MetaflowCodeContent.get_archive_content_members(
+            include_names = MetaflowCodeContent.get_archive_content_names(
                 opened_archive, content_types, backend
             )
-            backend.cls_extract_members(opened_archive, include_members, dest_dir)
+            backend.extract_members(include_names, dest_dir)
 
     def user_tuples(self, timeout: Optional[float] = None):
         # Wait for at least the blob to be formed

metaflow/packaging_sys/__init__.py

@@ -118,7 +118,9 @@ class MetaflowCodeContent:
         return handling_cls.get_filename_impl(mfcontent_info, filename, content_type)
 
     @classmethod
-    def get_env_vars_for_packaged_metaflow(cls, dest_dir: str) -> Dict[str, str]:
+    def get_env_vars_for_packaged_metaflow(
+        cls, dest_dir: str
+    ) -> Optional[Dict[str, str]]:
         """
         Get the environment variables that are needed to run Metaflow when it is
         packaged. This is typically used to set the PYTHONPATH to include the
@@ -126,19 +128,17 @@ class MetaflowCodeContent:
 
         Returns
         -------
-        Dict[str, str]
+        Optional[Dict[str, str]]
             The environment variables that are needed to run Metaflow when it is
-            packaged it present.
+            packaged -- None if there are no such variables (not packaged for example)
         """
-        mfcontent_info = cls._extract_mfcontent_info(dest_dir)
+        mfcontent_info = cls._extract_mfcontent_info()
         if mfcontent_info is None:
             # No MFCONTENT_MARKER file found -- this is not a packaged Metaflow code
             # package so no environment variables to set.
-            return {}
+            return None
         handling_cls = cls._get_mfcontent_class(mfcontent_info)
-        v = handling_cls.get_post_extract_env_vars_impl(dest_dir)
-        v["METAFLOW_EXTRACTED_ROOT:"] = dest_dir
-        return v
+        return handling_cls.get_post_extract_env_vars_impl(dest_dir)
 
     @classmethod
     def get_archive_info(
@@ -216,15 +216,15 @@ class MetaflowCodeContent:
         )
 
     @classmethod
-    def get_archive_content_members(
+    def get_archive_content_names(
         cls,
         archive: Any,
         content_types: Optional[int] = None,
         packaging_backend: Type[PackagingBackend] = TarPackagingBackend,
-    ) -> List[Any]:
+    ) -> List[str]:
         mfcontent_info = cls._extract_archive_mfcontent_info(archive, packaging_backend)
         handling_cls = cls._get_mfcontent_class(mfcontent_info)
-        return handling_cls.get_archive_content_members_impl(
+        return handling_cls.get_archive_content_names_impl(
             mfcontent_info, archive, content_types, packaging_backend
         )
 
@@ -276,9 +276,7 @@ class MetaflowCodeContent:
                 "Invalid package -- unknown version %s in info: %s"
                 % (version_id, cls._mappings)
             )
-        v = cls._mappings[version_id].get_post_extract_env_vars_impl(dest_dir)
-        v["METAFLOW_EXTRACTED_ROOT:"] = dest_dir
-        return v
+        return cls._mappings[version_id].get_post_extract_env_vars_impl(dest_dir)
 
     # Implement the _impl methods in the base subclass (in this file). These need to
     # happen with as few imports as possible to prevent circular dependencies.
@@ -339,14 +337,14 @@ class MetaflowCodeContent:
         raise NotImplementedError("get_archive_filename_impl not implemented")
 
     @classmethod
-    def get_archive_content_members_impl(
+    def get_archive_content_names_impl(
         cls,
         mfcontent_info: Optional[Dict[str, Any]],
         archive: Any,
         content_types: Optional[int] = None,
         packaging_backend: Type[PackagingBackend] = TarPackagingBackend,
-    ) -> List[Any]:
-        raise NotImplementedError("get_archive_content_members_impl not implemented")
+    ) -> List[str]:
+        raise NotImplementedError("get_archive_content_names_impl not implemented")
 
     @classmethod
     def get_post_extract_env_vars_impl(cls, dest_dir: str) -> Dict[str, str]:
@@ -525,22 +523,19 @@ class MetaflowCodeContent:
         return mfcontent_info
 
     @classmethod
-    def _extract_mfcontent_info(
-        cls, target_dir: Optional[str] = None
-    ) -> Optional[Dict[str, Any]]:
-        target_dir = target_dir or "_local"
-        if target_dir in cls._cached_mfcontent_info:
-            return cls._cached_mfcontent_info[target_dir]
+    def _extract_mfcontent_info(cls) -> Optional[Dict[str, Any]]:
+        if "_local" in cls._cached_mfcontent_info:
+            return cls._cached_mfcontent_info["_local"]
 
         mfcontent_info = None  # type: Optional[Dict[str, Any]]
-        if target_dir == "_local":
-            root = os.environ.get("METAFLOW_EXTRACTED_ROOT", get_metaflow_root())
-        else:
-            root = target_dir
-        if os.path.exists(os.path.join(root, MFCONTENT_MARKER)):
-            with open(os.path.join(root, MFCONTENT_MARKER), "r", encoding="utf-8") as f:
+        if os.path.exists(os.path.join(get_metaflow_root(), MFCONTENT_MARKER)):
+            with open(
+                os.path.join(get_metaflow_root(), MFCONTENT_MARKER),
+                "r",
+                encoding="utf-8",
+            ) as f:
                 mfcontent_info = json.load(f)
-        cls._cached_mfcontent_info[target_dir] = mfcontent_info
+        cls._cached_mfcontent_info["_local"] = mfcontent_info
         return mfcontent_info
 
     def get_package_version(self) -> int:
@@ -632,13 +627,13 @@ class MetaflowCodeContentV0(MetaflowCodeContent, version_id=0):
         return None
 
     @classmethod
-    def get_archive_content_members_impl(
+    def get_archive_content_names_impl(
         cls,
         mfcontent_info: Optional[Dict[str, Any]],
         archive: Any,
         content_types: Optional[int] = None,
         packaging_backend: Type[PackagingBackend] = TarPackagingBackend,
-    ) -> List[Any]:
+    ) -> List[str]:
         """
         For V0, we use a static list of known files to classify the content
         """
@@ -654,20 +649,16 @@ class MetaflowCodeContentV0(MetaflowCodeContent, version_id=0):
             "condav2-1.cnd": ContentType.OTHER_CONTENT.value,
         }
         to_return = []
-        for member in packaging_backend.cls_list_members(archive):
-            filename = packaging_backend.cls_member_name(member)
-            added = False
+        for filename in packaging_backend.cls_list_members(archive):
             for prefix, classification in known_prefixes.items():
                 if (
                     prefix[-1] == "/" and filename.startswith(prefix)
                 ) or prefix == filename:
                     if content_types & classification:
-                        to_return.append(member)
-                        added = True
-                        break
-            if not added and content_types & ContentType.USER_CONTENT.value:
-                # Everything else is user content
-                to_return.append(member)
+                        to_return.append(filename)
+                elif content_types & ContentType.USER_CONTENT.value:
+                    # Everything else is user content
+                    to_return.append(filename)
         return to_return
 
     @classmethod
@@ -714,7 +705,7 @@ class MetaflowCodeContentV1Base(MetaflowCodeContent, version_id=1):
         cls, mfcontent_info: Optional[Dict[str, Any]], filename: str, in_archive: bool
     ) -> str:
         if in_archive:
-            return os.path.join(cls._other_dir, filename)
+            return filename
         return os.path.join(get_metaflow_root(), "..", cls._other_dir, filename)
 
     @classmethod
@@ -722,7 +713,7 @@ class MetaflowCodeContentV1Base(MetaflowCodeContent, version_id=1):
         cls, mfcontent_info: Optional[Dict[str, Any]], filename: str, in_archive: bool
     ) -> str:
         if in_archive:
-            return os.path.join(cls._code_dir, filename)
+            return filename
         return os.path.join(get_metaflow_root(), filename)
 
     @classmethod
@@ -841,38 +832,37 @@ class MetaflowCodeContentV1Base(MetaflowCodeContent, version_id=1):
         return None
 
     @classmethod
-    def get_archive_content_members_impl(
+    def get_archive_content_names_impl(
         cls,
         mfcontent_info: Optional[Dict[str, Any]],
         archive: Any,
         content_types: Optional[int] = None,
         packaging_backend: Type[PackagingBackend] = TarPackagingBackend,
-    ) -> List[Any]:
+    ) -> List[str]:
         to_return = []
         module_content = set(mfcontent_info.get("module_files", []))
-        for member in packaging_backend.cls_list_members(archive):
-            filename = packaging_backend.cls_member_name(member)
+        for filename in packaging_backend.cls_list_members(archive):
             if filename.startswith(cls._other_dir) and (
                 content_types & ContentType.OTHER_CONTENT.value
             ):
-                to_return.append(member)
+                to_return.append(filename)
             elif filename.startswith(cls._code_dir):
                 # Special case for marker which is a other content even if in code.
-                if filename == MFCONTENT_MARKER:
+                if filename == f"{cls._code_dir}/{MFCONTENT_MARKER}":
                     if content_types & ContentType.OTHER_CONTENT.value:
-                        to_return.append(member)
+                        to_return.append(filename)
                     else:
                         continue
                 # Here it is either module or code
                 if os.path.join(cls._code_dir, filename) in module_content:
                     if content_types & ContentType.MODULE_CONTENT.value:
-                        to_return.append(member)
+                        to_return.append(filename)
                 elif content_types & ContentType.CODE_CONTENT.value:
-                    to_return.append(member)
+                    to_return.append(filename)
             else:
                 if content_types & ContentType.USER_CONTENT.value:
                     # Everything else is user content
-                    to_return.append(member)
+                    to_return.append(filename)
         return to_return
 
     @classmethod

metaflow/packaging_sys/backend.py

@@ -57,15 +57,6 @@ class PackagingBackend(ABC):
         """Open the archive from the given content."""
         pass
 
-    @classmethod
-    @abstractmethod
-    def cls_member_name(cls, member: Union[Any, str]) -> str:
-        """
-        Returns the name of the member as a string.
-        This is used to ensure consistent naming across different archive formats.
-        """
-        pass
-
     @classmethod
     @abstractmethod
     def cls_has_member(cls, archive: Any, name: str) -> bool:
@@ -81,20 +72,14 @@ class PackagingBackend(ABC):
     def cls_extract_members(
         cls,
         archive: Any,
-        members: Optional[List[Any]] = None,
+        members: Optional[List[str]] = None,
         dest_dir: str = ".",
     ) -> None:
         pass
 
     @classmethod
     @abstractmethod
-    def cls_list_names(cls, archive: Any) -> Optional[List[str]]:
-        pass
-
-    @classmethod
-    @abstractmethod
-    def cls_list_members(cls, archive: Any) -> Optional[List[Any]]:
-        """List all members in the archive."""
+    def cls_list_members(cls, archive: Any) -> Optional[List[str]]:
         pass
 
     def has_member(self, name: str) -> bool:
@@ -108,17 +93,17 @@ class PackagingBackend(ABC):
         raise ValueError("Cannot get member from an uncreated archive")
 
     def extract_members(
-        self, members: Optional[List[Any]] = None, dest_dir: str = "."
+        self, members: Optional[List[str]] = None, dest_dir: str = "."
     ) -> None:
         if self._archive:
             self.cls_extract_members(self._archive, members, dest_dir)
         else:
             raise ValueError("Cannot extract from an uncreated archive")
 
-    def list_names(self) -> Optional[List[str]]:
+    def list_members(self) -> Optional[List[str]]:
         if self._archive:
-            return self.cls_list_names(self._archive)
-        raise ValueError("Cannot list names from an uncreated archive")
+            return self.cls_list_members(self._archive)
+        raise ValueError("Cannot list members from an uncreated archive")
 
     def __enter__(self):
         self.create()

metaflow/packaging_sys/tar_backend.py

@@ -1,7 +1,7 @@
 import tarfile
 
 from io import BytesIO
-from typing import Any, IO, List, Optional, Union
+from typing import IO, List, Optional, Union
 
 from .backend import PackagingBackend
 
@@ -56,13 +56,6 @@ 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:
@@ -83,17 +76,11 @@ class TarPackagingBackend(PackagingBackend):
     def cls_extract_members(
         cls,
         archive: tarfile.TarFile,
-        members: Optional[List[Any]] = None,
+        members: Optional[List[str]] = None,
         dest_dir: str = ".",
     ) -> None:
         archive.extractall(path=dest_dir, members=members)
 
     @classmethod
-    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]]:
+    def cls_list_members(cls, archive: tarfile.TarFile) -> Optional[List[str]]:
         return archive.getnames() or None

metaflow/packaging_sys/v1.py

@@ -61,25 +61,23 @@ class MetaflowCodeContentV1(MetaflowCodeContentV1Base):
         else:
             new_modules = []
 
-        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
+        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)
+        }
 
         # Contain metadata information regarding the distributions packaged.
         # This allows Metaflow to "fake" distribution information when packaged
@@ -357,14 +355,16 @@ class MetaflowCodeContentV1(MetaflowCodeContentV1Base):
                 )
                 yield json.dumps(self.create_mfcontent_info()).encode(
                     "utf-8"
-                ), MFCONTENT_MARKER
+                ), os.path.join(self._code_dir, 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, MFCONTENT_MARKER
+                yield "<generated %s content>" % MFCONTENT_MARKER, os.path.join(
+                    self._code_dir, 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_client.py

@@ -58,38 +58,21 @@ class ArgoClient(object):
                 json.loads(e.body)["message"] if e.body is not None else e.reason
             )
 
-    def get_workflow_templates(self, page_size=100):
+    def get_workflow_templates(self):
         client = self._client.get()
-        continue_token = None
-
-        while True:
-            try:
-                params = {"limit": page_size}
-                if continue_token:
-                    params["_continue"] = continue_token
-
-                response = client.CustomObjectsApi().list_namespaced_custom_object(
-                    group=self._group,
-                    version=self._version,
-                    namespace=self._namespace,
-                    plural="workflowtemplates",
-                    **params,
-                )
-
-                for item in response.get("items", []):
-                    yield item
-
-                metadata = response.get("metadata", {})
-                continue_token = metadata.get("continue")
-
-                if not continue_token:
-                    break
-            except client.rest.ApiException as e:
-                if e.status == 404:
-                    return None
-                raise ArgoClientException(
-                    json.loads(e.body)["message"] if e.body is not None else e.reason
-                )
+        try:
+            return client.CustomObjectsApi().list_namespaced_custom_object(
+                group=self._group,
+                version=self._version,
+                namespace=self._namespace,
+                plural="workflowtemplates",
+            )["items"]
+        except client.rest.ApiException as e:
+            if e.status == 404:
+                return None
+            raise ArgoClientException(
+                json.loads(e.body)["message"] if e.body is not None else e.reason
+            )
 
     def register_workflow_template(self, name, workflow_template):
         # Unfortunately, Kubernetes client does not handle optimistic

metaflow/plugins/argo/argo_workflows.py

@@ -216,14 +216,23 @@ class ArgoWorkflows(object):
         return name.replace(".", "-")
 
     @staticmethod
-    def list_templates(flow_name, all=False, page_size=100):
+    def list_templates(flow_name, all=False):
         client = ArgoClient(namespace=KUBERNETES_NAMESPACE)
 
-        for template in client.get_workflow_templates(page_size=page_size):
-            if all or flow_name == template["metadata"].get("annotations", {}).get(
-                "metaflow/flow_name", None
-            ):
-                yield template["metadata"]["name"]
+        templates = client.get_workflow_templates()
+        if templates is None:
+            return []
+
+        template_names = [
+            template["metadata"]["name"]
+            for template in templates
+            if all
+            or flow_name
+            == template["metadata"]
+            .get("annotations", {})
+            .get("metaflow/flow_name", None)
+        ]
+        return template_names
 
     @staticmethod
     def delete(name):
@@ -956,6 +965,11 @@ class ArgoWorkflows(object):
                 dag_task = DAGTask(self._sanitize(node.name)).template(
                     self._sanitize(node.name)
                 )
+            if node.type == "split-switch":
+                raise ArgoWorkflowsException(
+                    "Deploying flows with switch statement "
+                    "to Argo Workflows is not supported currently."
+                )
             elif (
                 node.is_inside_foreach
                 and self.graph[node.in_funcs[0]].type == "foreach"
@@ -3277,8 +3291,8 @@ class ArgoWorkflows(object):
                     Trigger().template(
                         TriggerTemplate(self.name)
                         # Trigger a deployed workflow template
-                        .k8s_trigger(
-                            StandardK8STrigger()
+                        .argo_workflow_trigger(
+                            ArgoWorkflowTrigger()
                             .source(
                                 {
                                     "resource": {
@@ -4256,10 +4270,6 @@ 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
@@ -4334,57 +4344,3 @@ 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_cli.py

@@ -1011,7 +1011,8 @@ def terminate(obj, run_id, authorize=None):
 )
 @click.pass_obj
 def list_workflow_templates(obj, all=None):
-    for template_name in ArgoWorkflows.list_templates(obj.flow.name, all):
+    templates = ArgoWorkflows.list_templates(obj.flow.name, all)
+    for template_name in templates:
         obj.echo_always(template_name)