anyscale 0.24.13__tar.gz → 0.24.15__tar.gz

{anyscale-0.24.13 → anyscale-0.24.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: anyscale
-Version: 0.24.13
+Version: 0.24.15
 Summary: Command Line Interface for Anyscale
 Author: Anyscale Inc.
 License: AS License

{anyscale-0.24.13 → anyscale-0.24.15}/anyscale/__init__.py

@@ -18,8 +18,11 @@ path.append(os.path.join(anyscale_dir, "sdk"))
 from anyscale import integrations
 from anyscale.cluster import get_job_submission_client_cluster_info
 from anyscale.cluster_compute import get_cluster_compute_from_name
+import anyscale.compute_config
 from anyscale.connect import ClientBuilder
+import anyscale.job
 from anyscale.sdk.anyscale_client.sdk import AnyscaleSDK
+import anyscale.service
 
 
 # Note: indentation here matches that of connect.py::ClientBuilder.

{anyscale-0.24.13 → anyscale-0.24.15}/anyscale/_private/anyscale_client/anyscale_client.py

@@ -1,8 +1,11 @@
+from abc import ABC, abstractmethod
+import io
 import logging
 import os
 import pathlib
 import time
-from typing import Callable, Dict, List, Optional, Tuple
+from typing import Any, Callable, Dict, IO, List, Optional, Tuple
+from urllib.parse import urlparse
 
 import requests
 
@@ -60,6 +63,7 @@ from anyscale.util import (
 )
 from anyscale.utils.runtime_env import (
     is_workspace_dependency_tracking_disabled,
+    parse_dot_env_file,
     WORKSPACE_REQUIREMENTS_FILE_PATH,
     zip_local_dir,
 )
@@ -78,6 +82,39 @@ OVERWRITE_EXISTING_CLOUD_STORAGE_FILES = (
 internal_logger = logging.getLogger(__name__)
 
 
+class AWSS3ClientInterface(ABC):
+    @abstractmethod
+    def download_fileobj(self, Bucket: str, Key: str, Fileobj: IO[Any],) -> None:
+        """Download a file from an S3 bucket to a file-like object."""
+        raise NotImplementedError
+
+
+class GCSBlobInterface(ABC):
+    @abstractmethod
+    def download_to_file(self, fileobj: IO[Any]) -> None:
+        """Download the blob to a file-like object."""
+        raise NotImplementedError
+
+    @abstractmethod
+    def exists(self) -> bool:
+        """Check if the blob exists."""
+        raise NotImplementedError
+
+
+class GCSBucketInterface(ABC):
+    @abstractmethod
+    def blob(self, object_name: str) -> GCSBlobInterface:
+        """Get a blob object for the given object name."""
+        raise NotImplementedError
+
+
+class GCPGCSClientInterface(ABC):
+    @abstractmethod
+    def bucket(self, bucket: str) -> GCSBucketInterface:
+        """Get a bucket object for the given bucket name."""
+        raise NotImplementedError
+
+
 class AnyscaleClient(AnyscaleClientInterface):
     # Number of entries to fetch per request for list endpoints.
     LIST_ENDPOINT_COUNT = 50
@@ -90,6 +127,8 @@ class AnyscaleClient(AnyscaleClientInterface):
         workspace_requirements_file_path: str = WORKSPACE_REQUIREMENTS_FILE_PATH,
         logger: Optional[BlockLogger] = None,
         host: Optional[str] = None,
+        s3_client: Optional[AWSS3ClientInterface] = None,
+        gcs_client: Optional[GCPGCSClientInterface] = None,
     ):
         if api_clients is None:
             auth_block: AuthenticationBlock = get_auth_api_client(
@@ -100,6 +139,8 @@ class AnyscaleClient(AnyscaleClientInterface):
         self._external_api_client, self._internal_api_client = api_clients
         self._workspace_requirements_file_path = workspace_requirements_file_path
         self._sleep = sleep or time.sleep
+        self._s3_client = s3_client
+        self._gcs_client = gcs_client
 
         # Cached IDs and models to avoid duplicate lookups.
         self._default_project_id: Optional[str] = None
@@ -108,6 +149,35 @@ class AnyscaleClient(AnyscaleClientInterface):
         self._logger = logger or BlockLogger()
         self._host = host or ANYSCALE_HOST
 
+    @property
+    def s3_client(self) -> AWSS3ClientInterface:
+        if self._s3_client is None:
+            # initialize the s3 client lazily so that we import the boto3 library only when needed.
+            try:
+                import boto3
+                import botocore.config
+            except ImportError:
+                raise RuntimeError(
+                    "Could not import the Amazon S3 Python API via `import boto3`.  Please check your installation or try running `pip install boto3`."
+                )
+            self._s3_client = boto3.client(
+                "s3", config=botocore.config.Config(signature_version="s3v4")
+            )  # type: ignore
+        return self._s3_client  # type: ignore
+
+    @property
+    def gcs_client(self) -> GCPGCSClientInterface:
+        if self._gcs_client is None:
+            # initialize the gcs client lazily so that we import the google cloud storage library only when needed.
+            try:
+                from google.cloud import storage
+            except ImportError:
+                raise RuntimeError(
+                    "Could not import the Google Storage Python API via `from google.cloud import storage`.  Please check your installation or try running `pip install --upgrade google-cloud-storage`."
+                )
+            self._gcs_client = storage.Client()
+        return self._gcs_client  # type: ignore
+
     @property
     def host(self) -> str:
         return self._host
@@ -143,6 +213,84 @@ class AnyscaleClient(AnyscaleClientInterface):
 
         return self._workspace_requirements_file_path
 
+    def _download_file_from_google_cloud_storage(
+        self, bucket: str, object_name: str
+    ) -> Optional[bytes]:
+        try:
+            bucket_obj = self.gcs_client.bucket(bucket)
+            blob = bucket_obj.blob(object_name)
+            fileobj = io.BytesIO()
+            if blob.exists():
+                blob.download_to_file(fileobj)
+                return fileobj.getvalue()
+            return None
+        except Exception as e:  # noqa: BLE001
+            raise RuntimeError(
+                f"Failed to download the working directory from Google Cloud Storage. Error {repr(e)}"
+                "Please validate you have exported cloud credentials with the correct read permissions and the intended bucket exists in your Cloud Storage account. "
+            ) from e
+
+    def _download_file_from_s3(self, bucket: str, object_key: str) -> Optional[bytes]:
+        try:
+            from botocore.exceptions import ClientError
+        except Exception:  # noqa: BLE001
+            raise RuntimeError(
+                "Could not download file from S3: Could not import the Amazon S3 Python API via `import boto3`.  Please check your installation or try running `pip install boto3`."
+            )
+        try:
+            fileobj = io.BytesIO()
+            self.s3_client.download_fileobj(bucket, object_key, fileobj)
+            return fileobj.getvalue()
+        except ClientError as e:
+            if e.response["Error"]["Code"] == "404":
+                return None
+            raise
+        except Exception as e:  # noqa: BLE001
+            raise RuntimeError(
+                f"Failed to download the working directory from S3. Error {repr(e)}"
+                "Please validate you have exported cloud credentials with the correct read permissions and the intended bucket exists in your S3 account. "
+            ) from e
+
+    def _download_file_from_remote_storage(self, remote_uri: str) -> Optional[bytes]:
+        parsed_uri = urlparse(remote_uri)
+        service = parsed_uri.scheme
+        bucket = parsed_uri.netloc
+        object_name = parsed_uri.path.lstrip("/")
+        if service == "s3":
+            return self._download_file_from_s3(bucket, object_name)
+        if service == "gs":
+            return self._download_file_from_google_cloud_storage(bucket, object_name)
+        return None
+
+    def get_workspace_env_vars(self) -> Optional[Dict[str, str]]:
+        system_storage_path = os.environ.get("ANYSCALE_INTERNAL_SYSTEM_STORAGE", "")
+        workspace_id = os.environ.get("ANYSCALE_EXPERIMENTAL_WORKSPACE_ID", "")
+        workspace_artifact_path = (
+            os.path.join(
+                system_storage_path, "workspace_tracking_dependencies", workspace_id,
+            )
+            if workspace_id and system_storage_path
+            else ""
+        )
+        workspace_dot_env_path = (
+            os.path.join(workspace_artifact_path, ".env")
+            if workspace_artifact_path
+            else ""
+        )
+
+        if not self.inside_workspace() or not workspace_dot_env_path:
+            return None
+
+        dot_env = self._download_file_from_remote_storage(workspace_dot_env_path)
+        if dot_env:
+            parsed_dot_env = parse_dot_env_file(dot_env)
+            if parsed_dot_env:
+                self.logger.info(
+                    f"Using workspace runtime dependencies env vars: {parsed_dot_env}."
+                )
+            return parsed_dot_env
+        return None
+
     def get_current_workspace_cluster(self) -> Optional[Cluster]:
         # Checks for the existence of the ANYSCALE_EXPERIMENTAL_WORKSPACE_ID env var.
         if not is_anyscale_workspace():
@@ -156,13 +304,20 @@ class AnyscaleClient(AnyscaleClientInterface):
 
         return self._current_workspace_cluster
 
-    def get_project_id(self) -> str:
+    def get_project_id(self, parent_cloud_id: str) -> str:
         workspace_cluster = self.get_current_workspace_cluster()
         if workspace_cluster is not None:
             return workspace_cluster.project_id
 
         if self._default_project_id is None:
-            default_project: Project = self._external_api_client.get_default_project().result
+            # Cloud isolation organizations follow the permissions model in https://docs.anyscale.com/organization-and-user-account/access-controls
+            # TODO(nikita): Remove this FF check after completing the cloud isolation migration in Q2 2024
+            cloud_isolation_ff_on = self._internal_api_client.check_is_feature_flag_on_api_v2_userinfo_check_is_feature_flag_on_get(
+                "cloud-isolation-phase-1"
+            ).result.is_on
+            default_project: Project = self._external_api_client.get_default_project(
+                parent_cloud_id=(parent_cloud_id if cloud_isolation_ff_on else None)
+            ).result
             self._default_project_id = default_project.id
 
         return self._default_project_id
@@ -361,7 +516,6 @@ class AnyscaleClient(AnyscaleClientInterface):
         self, cluster_env_name: str, base_build_id: str, requirements: List[str]
     ) -> str:
         if requirements:
-            self.logger.info(f"Building image with requirements: {requirements}.")
             base_build = self._external_api_client.get_cluster_environment_build(
                 base_build_id
             ).result
@@ -369,6 +523,7 @@ class AnyscaleClient(AnyscaleClientInterface):
                 base_build.status == ClusterEnvironmentBuildStatus.SUCCEEDED
                 and base_build.docker_image_name
             ):
+                self.logger.info(f"Using tracked python packages: {requirements}")
                 lines = [
                     "# syntax=docker/dockerfile:1",
                     f"FROM {base_build.docker_image_name}",
@@ -417,6 +572,7 @@ class AnyscaleClient(AnyscaleClientInterface):
                 build_id
             ).result
             if build.status == ClusterEnvironmentBuildStatus.SUCCEEDED:
+                self.logger.info("")
                 return
             elif build.status == ClusterEnvironmentBuildStatus.FAILED:
                 raise RuntimeError(f"Image build {build_id} failed.")
@@ -425,7 +581,8 @@ class AnyscaleClient(AnyscaleClientInterface):
 
             elapsed_secs += poll_interval_seconds
             self.logger.info(
-                f"Waiting for image build to complete. Elapsed time: {elapsed_secs} seconds",
+                f"Waiting for image build to complete. Elapsed time: {elapsed_secs} seconds.",
+                end="\r",
             )
             self._sleep(poll_interval_seconds)
         raise TimeoutError(
@@ -544,7 +701,8 @@ class AnyscaleClient(AnyscaleClientInterface):
         # TODO(edoakes): this endpoint is very slow and there's no reason we should need
         # to use this complex list endpoint just to fetch a service by name.
         paging_token = None
-        project_id = self.get_project_id()
+        cloud_id = self.get_cloud_id()
+        project_id = self.get_project_id(parent_cloud_id=cloud_id)
         service: Optional[ServiceModel] = None
         while True:
             resp = self._external_api_client.list_services(

{anyscale-0.24.13 → anyscale-0.24.15}/anyscale/_private/anyscale_client/common.py

@@ -1,5 +1,5 @@
 from abc import ABC, abstractmethod
-from typing import List, Optional
+from typing import Dict, List, Optional
 
 from anyscale._private.models.image_uri import ImageURI
 from anyscale.client.openapi_client.models import (
@@ -59,6 +59,11 @@ class AnyscaleClientInterface(ABC):
         """
         raise NotImplementedError
 
+    @abstractmethod
+    def get_workspace_env_vars(self) -> Optional[Dict[str, str]]:
+        """Returns the environment variables specified in workspace runtime dependencies."""
+        raise NotImplementedError
+
     @abstractmethod
     def get_current_workspace_cluster(self) -> Optional[Cluster]:
         """Get the cluster model for the workspace this code is running in.
@@ -75,7 +80,7 @@ class AnyscaleClientInterface(ABC):
         raise NotImplementedError
 
     @abstractmethod
-    def get_project_id(self) -> str:
+    def get_project_id(self, parent_cloud_id: str) -> str:
         """Get the default project ID for this user.
 
         If running in a workspace, returns the workspace project ID.

{anyscale-0.24.13 → anyscale-0.24.15}/anyscale/_private/anyscale_client/fake_anyscale_client.py

@@ -71,6 +71,7 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
         self._requirements_path: Optional[str] = None
         self._upload_uri_mapping: Dict[str, str] = {}
         self._submitted_job: Optional[CreateInternalProductionJob] = None
+        self._env_vars: Optional[Dict[str, str]] = None
 
         # Cloud ID -> Cloud.
         self._clouds: Dict[str, Cloud] = {
@@ -110,8 +111,10 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
         *,
         requirements_path: Optional[str] = None,
         cluster_name: Optional[str] = None,
+        env_vars: Optional[Dict[str, str]] = None,
     ):
         self._requirements_path = requirements_path
+        self._env_vars = env_vars
         if inside_workspace:
             self._workspace_cluster = Cluster(
                 id=self.WORKSPACE_CLUSTER_ID,
@@ -136,6 +139,9 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
     def inside_workspace(self) -> bool:
         return self.get_current_workspace_cluster() is not None
 
+    def get_workspace_env_vars(self) -> Optional[Dict[str, str]]:
+        return self._env_vars
+
     def get_workspace_requirements_path(self) -> Optional[str]:
         if self.inside_workspace():
             return self._requirements_path
@@ -152,7 +158,7 @@ class FakeAnyscaleClient(AnyscaleClientInterface):
         if self.inside_workspace():
             self._sent_workspace_notifications.append(notification)
 
-    def get_project_id(self) -> str:
+    def get_project_id(self, parent_cloud_id: str) -> str:  # noqa: ARG002
         workspace_cluster = self.get_current_workspace_cluster()
         if workspace_cluster is not None:
             return workspace_cluster.project_id

anyscale-0.24.15/anyscale/_private/docgen/README.md

@@ -0,0 +1,15 @@
+Tool to autogenerate documentation for the Anyscale SDK/CLI that can be embedded in a [Docusarus](https://docusaurus.io/) site.
+
+This is used by: https://github.com/anyscale/endpoint-docs.
+
+Usage:
+```bash
+Usage: python -m anyscale._private.docgen [OPTIONS] OUTPUT_DIR
+
+  Generate markdown docs for the Anyscale CLI & SDK.
+
+Options:
+  -r, --remove-existing  If set, all files in the 'output_dir' that were not
+                         generated will be removed.
+  --help                 Show this message and exit.
+```

anyscale-0.24.15/anyscale/_private/docgen/__main__.py

@@ -0,0 +1,114 @@
+"""Entrypoint to generate reference docs for the Anyscale CLI/SDK.
+
+Usage: python -m anyscale._private.docgen --help
+"""
+import os
+
+import click
+
+import anyscale
+from anyscale._private.docgen.generator import MarkdownGenerator, Module
+from anyscale.commands import job_commands, service_commands
+from anyscale.compute_config.models import (
+    ComputeConfig,
+    HeadNodeConfig,
+    MarketType,
+    WorkerNodeGroupConfig,
+)
+from anyscale.job.models import JobConfig
+from anyscale.service.models import (
+    RayGCSExternalStorageConfig,
+    ServiceConfig,
+    ServiceState,
+    ServiceStatus,
+    ServiceVersionState,
+    ServiceVersionStatus,
+)
+
+
+# Defines all modules to be documented.
+ALL_MODULES = [
+    Module(
+        title="Job",
+        filename="job-api.md",
+        cli_prefix="anyscale job",
+        cli_commands=[job_commands.submit],
+        sdk_prefix="anyscale.job",
+        sdk_commands=[anyscale.job.submit],
+        models=[JobConfig],
+    ),
+    Module(
+        title="Service",
+        filename="service-api.md",
+        cli_prefix="anyscale service",
+        cli_commands=[
+            service_commands.deploy,
+            service_commands.status,
+            service_commands.rollback,
+            service_commands.terminate,
+        ],
+        sdk_prefix="anyscale.service",
+        sdk_commands=[
+            anyscale.service.deploy,
+            anyscale.service.status,
+            anyscale.service.rollback,
+            anyscale.service.terminate,
+        ],
+        models=[
+            ServiceConfig,
+            RayGCSExternalStorageConfig,
+            ServiceStatus,
+            ServiceState,
+            ServiceVersionStatus,
+            ServiceVersionState,
+        ],
+    ),
+    Module(
+        title="Compute Config",
+        filename="compute-config-api.md",
+        cli_prefix="anyscale compute-config",
+        cli_commands=[],
+        sdk_prefix="anyscale.compute_config",
+        sdk_commands=[],
+        models=[ComputeConfig, HeadNodeConfig, WorkerNodeGroupConfig, MarketType],
+    ),
+]
+
+
+@click.command(help="Generate markdown docs for the Anyscale CLI & SDK.")
+@click.argument("output_dir")
+@click.option(
+    "-r",
+    "--remove-existing",
+    is_flag=True,
+    default=False,
+    help="If set, all files in the 'output_dir' that were not generated will be removed.",
+)
+def generate(
+    output_dir: str, *, remove_existing: bool = False,
+):
+    if not os.path.isdir(output_dir):
+        raise RuntimeError(f"output_dir '{output_dir}' does not exist.")
+
+    gen = MarkdownGenerator(ALL_MODULES)
+    gen.generate()
+
+    generated_files = set()
+    os.makedirs(output_dir, exist_ok=True)
+    for filename, content in gen.generate().items():
+        generated_files.add(filename)
+        full_path = os.path.join(output_dir, filename)
+        print(f"Writing output file {full_path}")
+        with open(full_path, "w") as f:
+            f.write(content)
+
+    if remove_existing:
+        to_remove = set(os.listdir(output_dir)) - generated_files
+        for path in to_remove:
+            full_path = os.path.join(output_dir, path)
+            print(f"Removing existing file {full_path}")
+            os.unlink(full_path)
+
+
+if __name__ == "__main__":
+    generate()