anyscale 0.24.36__tar.gz → 0.24.37__tar.gz

{anyscale-0.24.36 → anyscale-0.24.37}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: anyscale
-Version: 0.24.36
+Version: 0.24.37
 Summary: Command Line Interface for Anyscale
 Author: Anyscale Inc.
 License: AS License

anyscale-0.24.37/anyscale/__init__.py

@@ -0,0 +1,105 @@
+import inspect
+import logging
+import os
+from sys import path
+from typing import Any, Dict, List, Optional
+
+from anyscale import version
+
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logging.StreamHandler())
+logger.setLevel(os.environ.get("ANYSCALE_LOGLEVEL", "WARN"))
+
+anyscale_dir = os.path.dirname(os.path.abspath(__file__))
+path.append(os.path.join(anyscale_dir, "client"))
+path.append(os.path.join(anyscale_dir, "sdk"))
+
+import anyscale
+from anyscale import compute_config, image, integrations, job, service
+from anyscale._private.anyscale_client import AnyscaleClient, AnyscaleClientInterface
+from anyscale._private.sdk.base_sdk import Timer
+from anyscale.authenticate import AuthenticationBlock
+from anyscale.cli_logger import BlockLogger
+from anyscale.cluster import get_job_submission_client_cluster_info
+from anyscale.cluster_compute import get_cluster_compute_from_name
+from anyscale.compute_config import ComputeConfigSDK
+from anyscale.connect import ClientBuilder
+from anyscale.job import JobSDK
+from anyscale.sdk.anyscale_client.sdk import AnyscaleSDK
+from anyscale.service import ServiceSDK
+
+
+# Note: indentation here matches that of connect.py::ClientBuilder.
+BUILDER_HELP_FOOTER = """
+        See ``anyscale.ClientBuilder`` for full documentation of
+        this experimental feature."""
+
+# Auto-add all Anyscale connect builder functions to the top-level.
+for attr, _ in inspect.getmembers(ClientBuilder, inspect.isfunction):
+    if attr.startswith("_"):
+        continue
+
+    def _new_builder(attr: str) -> Any:
+        target = getattr(ClientBuilder, attr)
+
+        def new_session_builder(*a: List[Any], **kw: Dict[str, Any]) -> Any:
+            builder = ClientBuilder()
+            return target(builder, *a, **kw)
+
+        new_session_builder.__name__ = attr
+        new_session_builder.__doc__ = target.__doc__ + BUILDER_HELP_FOOTER
+        new_session_builder.__signature__ = inspect.signature(target)  # type: ignore
+
+        return new_session_builder
+
+    globals()[attr] = _new_builder(attr)
+
+__version__ = version.__version__
+
+ANYSCALE_ENV = os.environ.copy()
+
+
+class Anyscale:
+    def __init__(
+        self, *, auth_token: Optional[str] = None, _host: Optional[str] = None,
+    ):
+        auth_block = AuthenticationBlock(
+            cli_token=auth_token, host=_host, raise_structured_exception=True,
+        )
+        self._anyscale_client = AnyscaleClient(
+            api_clients=(auth_block.anyscale_api_client, auth_block.api_client),
+            host=_host,
+        )
+        self._job_sdk = JobSDK(client=self._anyscale_client)
+        self._service_sdk = ServiceSDK(client=self._anyscale_client)
+        self._compute_config_sdk = ComputeConfigSDK(client=self._anyscale_client)
+
+    @classmethod
+    def _init_private(
+        cls, *, client: AnyscaleClientInterface, logger: BlockLogger, timer: Timer
+    ):
+        # Private constructor used to inject fakes for testing.
+        obj = cls.__new__(cls)
+        super(Anyscale, obj).__init__()
+        obj._anyscale_client = client  # noqa: SLF001
+        obj._job_sdk = JobSDK(client=client, logger=logger, timer=timer)  # noqa: SLF001
+        obj._service_sdk = ServiceSDK(  # noqa: SLF001
+            client=client, logger=logger, timer=timer
+        )
+        obj._compute_config_sdk = ComputeConfigSDK(  # noqa: SLF001
+            client=client, logger=logger, timer=timer
+        )
+        return obj
+
+    @property
+    def job(self) -> JobSDK:  # noqa: F811
+        return self._job_sdk
+
+    @property
+    def service(self) -> ServiceSDK:  # noqa: F811
+        return self._service_sdk
+
+    @property
+    def compute_config(self) -> ComputeConfigSDK:  # noqa: F811
+        return self._compute_config_sdk

{anyscale-0.24.36 → anyscale-0.24.37}/anyscale/_private/anyscale_client/anyscale_client.py

@@ -1,10 +1,11 @@
 from abc import ABC, abstractmethod
+import asyncio
 import io
 import logging
 import os
 import pathlib
 import time
-from typing import Any, Callable, Dict, IO, List, Optional, Tuple
+from typing import Any, Callable, cast, Dict, IO, List, Optional, Tuple
 from urllib.parse import urlparse
 
 import requests
@@ -16,8 +17,9 @@ from anyscale._private.anyscale_client.common import (
     RUNTIME_ENV_PACKAGE_FORMAT,
 )
 from anyscale._private.models.image_uri import ImageURI
+from anyscale.api_utils.job_logs_util import _get_job_logs_from_storage_bucket_streaming
 from anyscale.authenticate import AuthenticationBlock, get_auth_api_client
-from anyscale.cli_logger import BlockLogger
+from anyscale.cli_logger import BlockLogger, LogsLogger
 from anyscale.client.openapi_client.api.default_api import DefaultApi as InternalApi
 from anyscale.client.openapi_client.models import (
     ArchiveStatus,
@@ -154,7 +156,7 @@ class AnyscaleClient(AnyscaleClientInterface):
         self._gcs_client = gcs_client
 
         # Cached IDs and models to avoid duplicate lookups.
-        self._default_project_id_from_cloud_id: Dict[str, str] = {}
+        self._default_project_id_from_cloud_id: Dict[Optional[str], str] = {}
         self._cloud_id_cache: Dict[Optional[str], str] = {}
         self._current_workspace_cluster: Optional[Cluster] = None
         self._logger = logger or BlockLogger()
@@ -322,10 +324,43 @@ class AnyscaleClient(AnyscaleClientInterface):
 
         return self._current_workspace_cluster
 
-    def get_project_id(self, parent_cloud_id: str) -> str:
+    def _get_project_id_by_name(
+        self, *, parent_cloud_id: Optional[str] = None, name: Optional[str] = None
+    ) -> str:
+        # Find if project with name already exists
+        existing_projects = self._internal_api_client.find_project_by_project_name_api_v2_projects_find_by_name_get(
+            name
+        ).results
+        if len(existing_projects) == 0:
+            raise ValueError(f"Project '{name}' was not found.")
+        else:
+            for project in existing_projects:
+                if project.parent_cloud_id == parent_cloud_id:
+                    return project.id
+            raise ValueError(
+                f"{len(existing_projects)} project(s) found with name '{name}' and none matched cloud_id '{parent_cloud_id}'"
+            )
+
+    def _get_project_id_by_cloud_id(
+        self, *, parent_cloud_id: Optional[str] = None,
+    ) -> str:
         workspace_cluster = self.get_current_workspace_cluster()
         if workspace_cluster is not None:
-            return workspace_cluster.project_id
+            if (
+                workspace_cluster.cluster_compute_config is not None
+                and workspace_cluster.cluster_compute_config.cloud_id == parent_cloud_id
+            ):
+                return workspace_cluster.project_id
+            elif workspace_cluster.cluster_compute_id is not None:
+                workspace_cluster_compute = self.get_compute_config(
+                    workspace_cluster.cluster_compute_id
+                )
+                if (
+                    workspace_cluster_compute is not None
+                    and workspace_cluster_compute.config is not None
+                    and workspace_cluster_compute.config.cloud_id == parent_cloud_id
+                ):
+                    return workspace_cluster.project_id
 
         if self._default_project_id_from_cloud_id.get(parent_cloud_id) is None:
             # Cloud isolation organizations follow the permissions model in https://docs.anyscale.com/organization-and-user-account/access-controls
@@ -340,6 +375,16 @@ class AnyscaleClient(AnyscaleClientInterface):
 
         return self._default_project_id_from_cloud_id[parent_cloud_id]
 
+    def get_project_id(
+        self, *, parent_cloud_id: Optional[str] = None, name: Optional[str] = None
+    ) -> str:
+        if name is not None:
+            return self._get_project_id_by_name(
+                parent_cloud_id=parent_cloud_id, name=name
+            )
+        else:
+            return self._get_project_id_by_cloud_id(parent_cloud_id=parent_cloud_id)
+
     def _get_cloud_id_for_compute_config_id(self, compute_config_id: str) -> str:
         cluster_compute: ClusterCompute = self._external_api_client.get_cluster_compute(
             compute_config_id
@@ -733,8 +778,18 @@ class AnyscaleClient(AnyscaleClientInterface):
 
         return service
 
+    def get_project(self, project_id: str) -> Optional[Project]:
+        return self._internal_api_client.get_project_api_v2_projects_project_id_get(
+            project_id
+        ).result
+
     def get_job(
-        self, name: Optional[str], job_id: Optional[str]
+        self,
+        *,
+        name: Optional[str],
+        job_id: Optional[str],
+        cloud: Optional[str],
+        project: Optional[str],
     ) -> Optional[ProductionJob]:
         if job_id is not None:
             try:
@@ -745,8 +800,8 @@ class AnyscaleClient(AnyscaleClientInterface):
                 raise e from None
         else:
             paging_token = None
-            cloud_id = self.get_cloud_id()
-            project_id = self.get_project_id(parent_cloud_id=cloud_id)
+            cloud_id = self.get_cloud_id(cloud_name=cloud)
+            project_id = self.get_project_id(parent_cloud_id=cloud_id, name=project)
             result: Optional[ProductionJob] = None
             while True:
                 resp = self._external_api_client.list_production_jobs(
@@ -853,3 +908,25 @@ class AnyscaleClient(AnyscaleClientInterface):
                 )
 
         return info.file_uri
+
+    def logs_for_job_run(self, job_run_id: str) -> str:
+        """Retrieves logs from the streaming job logs folder in S3/GCS"""
+        logs = ""
+        next_page_token = None
+        cluster_journal_events_start_line = 0
+
+        (logs, next_page_token, cluster_journal_events_start_line, _,) = asyncio.run(
+            _get_job_logs_from_storage_bucket_streaming(
+                self._internal_api_client,
+                cast(
+                    LogsLogger, self.logger
+                ),  # self.logger is a BlockLogger, cast to LogsLogger subclass
+                job_run_id=job_run_id,
+                remove_escape_chars=False,
+                next_page_token=next_page_token,
+                cluster_journal_events_start_line=cluster_journal_events_start_line,
+                no_cluster_journal_events=True,
+            )
+        )
+
+        return logs

{anyscale-0.24.36 → anyscale-0.24.37}/anyscale/_private/anyscale_client/common.py

@@ -8,6 +8,7 @@ from anyscale.client.openapi_client.models import (
     CreateInternalProductionJob,
     DecoratedComputeTemplate,
     InternalProductionJob,
+    Project,
 )
 from anyscale.client.openapi_client.models.production_job import ProductionJob
 from anyscale.sdk.anyscale_client.models import (
@@ -92,8 +93,17 @@ class AnyscaleClientInterface(ABC):
         raise NotImplementedError
 
     @abstractmethod
-    def get_project_id(self, parent_cloud_id: str) -> str:
-        """Get the default project ID for this user.
+    def get_project_id(
+        self, *, parent_cloud_id: Optional[str] = None, name: Optional[str] = None
+    ) -> str:
+        """Get the project ID.
+
+        If name is passed, projects with that name will be checked for a project with
+        a cloud matching parent_cloud_id. If a match is found the cloud id is returned,
+        otherwise a ValueError is raised.
+
+        If name is not passed, but parent_cloud_id is, the default project for that cloud
+        will be used.
 
         If running in a workspace, returns the workspace project ID.
         """
@@ -226,11 +236,24 @@ class AnyscaleClientInterface(ABC):
         """
         raise NotImplementedError
 
+    @abstractmethod
+    def get_project(self, project_id: str) -> Optional[Project]:
+        """Get a project by id.
+
+        Returns None if not found.
+        """
+        raise NotImplementedError
+
     @abstractmethod
     def get_job(
-        self, name: Optional[str], job_id: Optional[str]
+        self,
+        *,
+        name: Optional[str],
+        job_id: Optional[str],
+        cloud: Optional[str],
+        project: Optional[str],
     ) -> Optional[ProductionJob]:
-        """Get a job by name.
+        """Get a job by either name or id. Filter by cloud and project.
 
         Returns None if not found.
         """
@@ -289,7 +312,7 @@ class AnyscaleClientInterface(ABC):
         *,
         cloud_id: str,
         excludes: Optional[List[str]] = None,
-        overwrite_existing_file: bool = False
+        overwrite_existing_file: bool = False,
     ) -> str:
         """Upload the provided directory to cloud storage and return a URI for it.
 
@@ -302,3 +325,11 @@ class AnyscaleClientInterface(ABC):
         default if the target file URI already exists it will not be overwritten.
         """
         raise NotImplementedError
+
+    @abstractmethod
+    def logs_for_job_run(self, job_run_id: str) -> str:
+        """Returns the logs associated with a particular job run.
+
+        The job_run_id is the backend id for the job run.
+        """
+        raise NotImplementedError

{anyscale-0.24.36 → anyscale-0.24.37}/anyscale/_private/anyscale_client/fake_anyscale_client.py

@@ -18,9 +18,10 @@ from anyscale.client.openapi_client.models import (
     HaJobGoalStates,
     HaJobStates,
     InternalProductionJob,
+    ProductionJob,
     ProductionJobStateTransition,
+    Project,
 )
-from anyscale.client.openapi_client.models.production_job import ProductionJob
 from anyscale.cluster_compute import parse_cluster_compute_name_version
 from anyscale.sdk.anyscale_client.configuration import Configuration
 from anyscale.sdk.anyscale_client.models import (
@@ -82,6 +83,7 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
         self._services: Dict[str, ServiceModel] = {}
         self._jobs: Dict[str, ProductionJob] = {}
         self._job_runs: Dict[str, List[APIJobRun]] = defaultdict(list)
+        self._project_to_id: Dict[Optional[str] : Dict[Optional[str], str]] = {}
         self._rolled_out_model: Optional[ApplyServiceModel] = None
         self._sent_workspace_notifications: List[WorkspaceNotification] = []
         self._rolled_back_service: Optional[Tuple[str, Optional[int]]] = None
@@ -91,6 +93,7 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
         self._upload_uri_mapping: Dict[str, str] = {}
         self._submitted_job: Optional[CreateInternalProductionJob] = None
         self._env_vars: Optional[Dict[str, str]] = None
+        self._job_run_logs: Dict[str, str] = {}
 
         # Cloud ID -> Cloud.
         self._clouds: Dict[str, Cloud] = {
@@ -184,13 +187,66 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
         if self.inside_workspace():
             self._sent_workspace_notifications.append(notification)
 
-    def get_project_id(self, parent_cloud_id: str) -> str:  # noqa: ARG002
+    def _find_project_cloud_id_tuples_by_name(self, name):
+        """Returns list of (cloud_id, project_id) tuples with name == name."""
+        project_id_cloud_id_pairs = []
+        for cloud_id, cloud_project_dict in self._project_to_id.items():
+            for p_name, p_id in cloud_project_dict.items():
+                if name == p_name:
+                    project_id_cloud_id_pairs.append((cloud_id, p_id))
+        return project_id_cloud_id_pairs
+
+    def _get_project_id_by_name(
+        self, *, parent_cloud_id: Optional[str] = None, name: Optional[str] = None
+    ) -> str:
+        # items of existing_projects are (cloud_id, project_id)
+        existing_projects = self._find_project_cloud_id_tuples_by_name(name)
+        if len(existing_projects) == 0:
+            raise ValueError(f"Project '{name}' was not found.")
+        else:
+            for cloud_id, project_id in existing_projects:
+                if cloud_id == parent_cloud_id:
+                    return project_id
+            raise ValueError(
+                f"{len(existing_projects)} project(s) found with name '{name}' and none matched cloud_id '{parent_cloud_id}'"
+            )
+
+    def _get_project_id_by_cloud_id(
+        self, *, parent_cloud_id: Optional[str] = None,
+    ) -> str:
         workspace_cluster = self.get_current_workspace_cluster()
         if workspace_cluster is not None:
-            return workspace_cluster.project_id
+            if (
+                workspace_cluster.cluster_compute_config is not None
+                and workspace_cluster.cluster_compute_config.cloud_id == parent_cloud_id
+            ):
+                return workspace_cluster.project_id
+            elif workspace_cluster.cluster_compute_id is not None:
+                workspace_cluster_compute = self.get_compute_config(
+                    workspace_cluster.cluster_compute_id
+                )
+                if (
+                    workspace_cluster_compute is not None
+                    and workspace_cluster_compute.config is not None
+                    and workspace_cluster_compute.config.cloud_id == parent_cloud_id
+                ):
+                    return workspace_cluster.project_id
 
         return self.DEFAULT_PROJECT_ID
 
+    def get_project_id(
+        self,
+        *,
+        parent_cloud_id: Optional[str] = None,  # noqa: ARG002
+        name: Optional[str] = None,  # noqa: ARG002
+    ) -> str:
+        if name is not None:
+            return self._get_project_id_by_name(
+                parent_cloud_id=parent_cloud_id, name=name
+            )
+        else:
+            return self._get_project_id_by_cloud_id(parent_cloud_id=parent_cloud_id)
+
     def get_cloud_id(
         self,
         *,
@@ -355,17 +411,40 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
 
         return None
 
+    def get_project(self, project_id: str) -> Optional[Project]:
+        for cloud_project_dict in self._project_to_id.values():
+            for p_name, p_id in cloud_project_dict.items():
+                if p_id == project_id:
+                    # return stub project
+                    return Project(
+                        name=p_name,
+                        id=p_id,
+                        local_vars_configuration=OPENAPI_NO_VALIDATION,
+                    )
+        return None
+
     def get_job(
-        self, name: Optional[str], job_id: Optional[str]
+        self,
+        *,
+        name: Optional[str],
+        job_id: Optional[str],
+        cloud: Optional[str],
+        project: Optional[str],
     ) -> Optional[ProductionJob]:
         if job_id is not None:
             return self._jobs.get(job_id, None)
         else:
+            cloud_id = self.get_cloud_id(cloud_name=cloud)
+            cloud_project_dict = self._project_to_id.get(cloud_id, None)
+            project_id = (
+                cloud_project_dict.get(project, None) if cloud_project_dict else None
+            )
             result: ProductionJob = None
             for job in self._jobs.values():
                 if (
                     job is not None
                     and job.name == name
+                    and (project_id is None or job.project_id == project_id)
                     and (result is None or job.created_at > result.created_at)
                 ):
                     result = job
@@ -381,6 +460,21 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
     def update_job_run(self, prod_job_id: str, model: APIJobRun):
         self._job_runs[prod_job_id].append(model)
 
+    def register_project_by_name(
+        self, name: str, cloud: str = DEFAULT_CLOUD_NAME
+    ) -> str:
+        """Helper method to create project name to project id mapping."""
+        cloud_id = self.get_cloud_id(cloud_name=cloud)
+        if cloud_id not in self._project_to_id:
+            self._project_to_id[cloud_id] = {}
+        cloud_project_dict = self._project_to_id[cloud_id]
+        if name in cloud_project_dict:
+            return cloud_project_dict[name]
+        else:
+            project_id = f"project-id-{uuid.uuid4()!s}"
+            cloud_project_dict[name] = project_id
+            return project_id
+
     @property
     def rolled_out_model(self) -> Optional[ApplyServiceModel]:
         return self._rolled_out_model
@@ -458,6 +552,7 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
                 goal_state=HaJobGoalStates.SUCCESS,
                 local_vars_configuration=OPENAPI_NO_VALIDATION,
             ),
+            project_id=model.project_id,
             local_vars_configuration=OPENAPI_NO_VALIDATION,
         )
 
@@ -489,3 +584,9 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
             ] = f"{bucket}/fake_pkg_{str(uuid.uuid4())}.zip"
 
         return self._upload_uri_mapping[local_dir]
+
+    def add_job_run_logs(self, job_run_id: str, logs: str):
+        self._job_run_logs[job_run_id] = logs
+
+    def logs_for_job_run(self, job_run_id: str) -> str:
+        return self._job_run_logs[job_run_id]

{anyscale-0.24.36 → anyscale-0.24.37}/anyscale/_private/docgen/__main__.py

@@ -51,6 +51,7 @@ ALL_MODULES = [
             anyscale.job.status,
             anyscale.job.terminate,
             anyscale.job.archive,
+            anyscale.job.get_logs,
         ],
         models=[JobConfig, JobState, JobStatus, JobRunStatus, JobRunState],
     ),

{anyscale-0.24.36 → anyscale-0.24.37}/anyscale/_private/sdk/__init__.py

@@ -33,12 +33,13 @@ def sdk_command(
 
         @wraps(f)
         def wrapper(*args: P.args, **kwargs: P.kwargs) -> T:
-            if key not in _LAZY_SDK_SINGLETONS:
-                _LAZY_SDK_SINGLETONS[key] = sdk_cls()
-
             # We disable the mypy linter here because it treats kwargs as a
             # P.kwargs object. mypy wrongly thinks kwargs can't be indexed.
-            kwargs["_sdk"] = _LAZY_SDK_SINGLETONS[key]  # type: ignore
+            if "_sdk" not in kwargs:  # type: ignore
+                if key not in _LAZY_SDK_SINGLETONS:
+                    _LAZY_SDK_SINGLETONS[key] = sdk_cls()
+
+                kwargs["_sdk"] = _LAZY_SDK_SINGLETONS[key]  # type: ignore
 
             return f(*args, **kwargs)
 
@@ -49,3 +50,27 @@ def sdk_command(
         return wrapper
 
     return _inject_typed_sdk_singleton
+
+
+def sdk_docs(
+    *, doc_py_example: str, arg_docstrings: Dict[str, str],
+) -> Callable[[Callable[P, T]], Callable[P, T]]:
+    """Decorator to add documentation for an SDK command."""
+
+    # The P and T type hints allow f's type hints to pass through this decorator.
+    # Without them, f's type hints would not be visible to the developer.
+    # See https://github.com/anyscale/product/pull/27738.
+    def _add_doc_magic_attrs(f: Callable[P, T]) -> Callable[P, T]:
+        if not doc_py_example:
+            raise ValueError(
+                f"SDK command '{f.__name__}' must provide a non-empty 'doc_py_example'."
+            )
+
+        # TODO(edoakes): validate docstrings.
+        # TODO(edoakes): move to parsing docstrings instead.
+        f.__doc_py_example__ = doc_py_example  # type: ignore
+        f.__arg_docstrings__ = arg_docstrings  # type: ignore
+
+        return f
+
+    return _add_doc_magic_attrs

{anyscale-0.24.36 → anyscale-0.24.37}/anyscale/_private/workload/workload_config.py

@@ -139,3 +139,25 @@ class WorkloadConfig(ModelBase):
             or not all(isinstance(m, str) for m in py_modules)
         ):
             raise TypeError("'py_modules' must be a list of strings.")
+
+    cloud: Optional[str] = field(
+        default=None,
+        metadata={
+            "docstring": "The Anyscale Cloud to run this workload on. If not provided, the organization default will be used (or, if running in a workspace, the cloud of the workspace)."
+        },
+    )
+
+    def _validate_cloud(self, cloud: Optional[str]):
+        if cloud is not None and not isinstance(cloud, str):
+            raise TypeError("'cloud' must be a string.")
+
+    project: Optional[str] = field(
+        default=None,
+        metadata={
+            "docstring": "The project for the workload. If not provided, the default project for the cloud will be used (or, if running in a workspace, the project of the workspace)."
+        },
+    )
+
+    def _validate_project(self, project: Optional[str]):
+        if project is not None and not isinstance(project, str):
+            raise TypeError("'project' must be a string.")