arraylake 1.0.0__tar.gz → 1.0.1__tar.gz

{arraylake-1.0.0 → arraylake-1.0.1}/.gitignore

@@ -146,3 +146,6 @@ client/arraylake/_version.py
 
 # claude
 **/.claude/
+!/.claude/
+/.claude/*
+!/.claude/skills/

{arraylake-1.0.0 → arraylake-1.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arraylake
-Version: 1.0.0
+Version: 1.0.1
 Summary: Python client for ArrayLake
 Project-URL: Documentation, https://docs.earthmover.io/
 Project-URL: Changelog, https://docs.earthmover.io/changelog
@@ -16,10 +16,10 @@ Requires-Dist: donfig<1.0,>=0.7
 Requires-Dist: httpx<0.29,>=0.23
 Requires-Dist: icechunk<3.0.0,>=2.0.1
 Requires-Dist: numpy<3.0,>=1.23
-Requires-Dist: packaging<26.0,>=23.0
+Requires-Dist: packaging<27.0,>=23.0
 Requires-Dist: pydantic[email]<3.0,>=2.9
 Requires-Dist: python-dateutil<3.0,>=2.8
-Requires-Dist: rich<15.0,>=12.6
+Requires-Dist: rich<16.0,>=12.6
 Requires-Dist: ruamel-yaml<1.0,>=0.17
 Requires-Dist: structlog<26.0,>=24.1
 Requires-Dist: typer<1.0,>=0.12.0

{arraylake-1.0.0 → arraylake-1.0.1}/arraylake/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '1.0.0'
-__version_tuple__ = version_tuple = (1, 0, 0)
+__version__ = version = '1.0.1'
+__version_tuple__ = version_tuple = (1, 0, 1)
 
 __commit_id__ = commit_id = None

{arraylake-1.0.0 → arraylake-1.0.1}/arraylake/api_utils.py

@@ -1,9 +1,11 @@
 from __future__ import annotations
 
 import asyncio
+import functools
 import json
 import os
 import random
+import uuid
 import warnings
 import weakref
 from collections.abc import AsyncGenerator, Callable, Generator, Mapping
@@ -82,6 +84,19 @@ def _get_proxy_from_env() -> str | None:
     return None
 
 
+@functools.cache
+def _get_session_id() -> str:
+    """Session ID for this Python process.
+
+    Propagated as the ``arraylake-session-id`` header on every outbound request and
+    baked into ``ArraylakeHttpClient._default_headers`` at construction time, so it
+    rides along to dask workers via the existing pickle path on the client.
+    """
+    session_id = str(uuid.uuid4())
+    logger.info("arraylake session id: %s", session_id)
+    return session_id
+
+
 # this is a hashable key to use for the global client cache
 # we only need one client for each key
 @dataclass(eq=True, frozen=True)
@@ -302,6 +317,7 @@ class ArraylakeHttpClient:
             # technically we don't enforce this, for now it's for debugging purposes
             "client-name": "arraylake-python-client",
             "client-version": arraylake.__version__,
+            "arraylake-session-id": _get_session_id(),
         }
 
         self._async_lock = asyncio.Lock()

{arraylake-1.0.0 → arraylake-1.0.1}/arraylake/cli/auth.py

@@ -101,14 +101,7 @@ def token():
         raise typer.Exit(code=1)
     try:
         token_value = handler.tokens.id_token.get_secret_value()
-        rich_console.print(
-            Panel(
-                f"[dim]{token_value}[/dim]",
-                title="id_token",
-                subtitle=f"[link=https://jwt.io/#id_token={token_value}][blue]:link: Inspect with JWT.io[/blue][/link]",
-                subtitle_align="right",
-            )
-        )
+        typer.echo(token_value)
     except (AuthException, AttributeError):
         rich_console.print(Align(f"[yellow]{NOT_LOGGED_IN_MESSAGE}[/yellow]", align="center"))
         raise typer.Exit(code=1)

{arraylake-1.0.0 → arraylake-1.0.1}/arraylake/client.py

@@ -558,16 +558,17 @@ class AsyncClient:
             if response.virtual_chunk_credentials:
                 credentials_map: dict[str, icechunk.AnyS3Credential | icechunk.AnyGcsCredential | icechunk.AnyAzureCredential | None] = {}
                 for vcc_prefix, vcc_creds in response.virtual_chunk_credentials.items():
-                    # Handle runtime VCCs (e.g., filtered subscriptions) - create VCC on the fly
+                    # Create VCC objects for all server-provided VCCs, not just runtime ones,
+                    # because icechunk's config merge replaces virtual_chunk_containers
+                    # rather than extending them.
+                    store_config = vcc_creds.to_vcc_store_config(user_id=UUID(response.principal_id))
+                    container = icechunk.VirtualChunkContainer(
+                        url_prefix=vcc_prefix,
+                        store=store_config,
+                    )
                     if vcc_creds.runtime_vcc_name is not None:
-                        bucket_config = await self.get_bucket_config(org=vcc_creds.org, nickname=vcc_creds.bucket_nickname)
-                        store_config = bucket_config.to_vcc_store_config(user_id=UUID(response.principal_id))
-                        container = icechunk.VirtualChunkContainer(
-                            url_prefix=vcc_prefix,
-                            store=store_config,
-                        )
                         container.name = vcc_creds.runtime_vcc_name
-                        config.set_virtual_chunk_container(container)
+                    config.set_virtual_chunk_container(container)
 
                     # TODO: build refresh closures here using vcc_creds.org, vcc_creds.bucket_nickname,
                     # and vcc_creds.platform so that credentials can be refreshed when they expire. (See https://github.com/earth-mover/arraylake/issues/5122)

{arraylake-1.0.0 → arraylake-1.0.1}/arraylake/types.py

@@ -1232,6 +1232,10 @@ class VirtualChunkCredentials(BaseModel):
 
     The org, bucket_nickname, and platform fields are needed so the client can
     construct credential refresh closures for delegated-creds buckets.
+
+    The store config fields (region, endpoint_url, allow_http, force_path_style)
+    allow the client to construct VirtualChunkContainer objects without extra
+    API calls to fetch the full bucket config.
     """
 
     # TODO: consider a dedicated server-side refresh endpoint to avoid exposing cross-org metadata to the client?
@@ -1243,6 +1247,55 @@ class VirtualChunkCredentials(BaseModel):
     # where subsync writes placeholder refs like vcc://__al_source/CHUNK_ID)
     # If None, assume the VCC already exists
     runtime_vcc_name: str | None = None
+    # Store config fields for constructing VCCs without extra get_bucket_config() calls.
+    # Populated from the bucket's extra_config via from_bucket().
+    region: str | None = None
+    endpoint_url: str | None = None
+    allow_http: bool = False
+    force_path_style: bool = False
+
+    @classmethod
+    def from_bucket(
+        cls,
+        org: OrgName,
+        bucket: BucketModel,
+        credentials: TempCredentials | None,
+    ) -> VirtualChunkCredentials:
+        """Build VirtualChunkCredentials from a bucket model, extracting store config fields."""
+        extra = bucket.extra_config
+        use_ssl = extra.get("use_ssl", True)
+        region = extra.get("region_name")
+        endpoint_url = extra.get("endpoint_url")
+        return cls(
+            credentials=credentials,
+            org=org,
+            bucket_nickname=bucket.nickname,
+            platform=bucket.platform,
+            region=str(region) if region is not None else None,
+            endpoint_url=str(endpoint_url) if endpoint_url is not None else None,
+            allow_http=not use_ssl,
+            force_path_style=bool(extra.get("force_path_style", False if use_ssl else True)),
+        )
+
+    def to_vcc_store_config(self, user_id: UUID | None = None) -> icechunk.AnyObjectStoreConfig:
+        """Create an Icechunk ObjectStoreConfig from the embedded store config fields."""
+        if self.platform in ("s3", "s3-compatible", "minio"):
+            options = icechunk.S3Options(
+                region=self.region,
+                endpoint_url=self.endpoint_url,
+                allow_http=self.allow_http,
+                force_path_style=self.force_path_style,
+            )
+            return icechunk.ObjectStoreConfig.S3(options)
+        elif self.platform == "gs":
+            from arraylake import __version__ as arraylake_version
+
+            config = {"user_agent": f"arraylake/{arraylake_version} (uuid={str(user_id)})" if user_id else f"arraylake/{arraylake_version}"}
+            return icechunk.ObjectStoreConfig.Gcs(config)
+        elif self.platform == "azure":
+            return icechunk.ObjectStoreConfig.Azure({})
+        else:
+            raise ValueError(f"Unsupported platform for VCC store config: {self.platform}")
 
 
 class OpenRepoResponse(BaseModel):

{arraylake-1.0.0 → arraylake-1.0.1}/pyproject.toml

@@ -37,10 +37,10 @@ dependencies = [
   "httpx >= 0.23, < 0.29",
   "icechunk >= 2.0.1, < 3.0.0",
   "numpy >= 1.23, < 3.0",
-  "packaging >= 23.0, < 26.0",
+  "packaging>=23.0,<27.0",
   "pydantic[email] >= 2.9, < 3.0",
   "python-dateutil >= 2.8, < 3.0",
-  "rich >= 12.6, < 15.0",
+  "rich>=12.6,<16.0",
   "ruamel-yaml >= 0.17, < 1.0",
   "structlog >= 24.1, < 26.0",
   "typer >= 0.12.0, < 1.0",
@@ -59,25 +59,24 @@ maximal = [
 
 [dependency-groups]
 dev = [
-  "cloudpickle~=3.0.0",
+  "cloudpickle>=3.0,<3.2",
   "hatch>=1.14.2",
   "hypothesis>=6.88.1",
   "icechunk >= 2.0.1, < 3.0.0",
   "mypy~=1.11",
-  "pytest-asyncio~=0.21.1",
+  "pytest-asyncio>=0.21.1,<1.4.0",
   "pytest-click~=1.1.0",
-  "pytest-cov~=4.0.0",
+  "pytest-cov>=4.0,<7.2",
   "pytest-incremental~=0.6.0",
-  "pytest-lazy-fixture~=0.6.3",
-  "pytest-reportlog~=0.4.0",
-  "pytest-retry~=1.6.2",
-  "pytest-timeout~=2.1.0",
-  "pytest-xdist~=3.5.0",
-  "pytest~=7.3",
+  "pytest-reportlog>=0.4,<1.1",
+  "pytest-retry>=1.6.2,<1.8.0",
+  "pytest-timeout>=2.1,<2.5",
+  "pytest-xdist>=3.5,<3.9",
+  "pytest>=7.3,<10.0",
   "pyyaml~=6.0.1",
-  "respx~=0.22.0",
-  "types-cachetools~=5.3",
-  "types-python-dateutil~=2.8.19",
+  "respx>=0.22,<0.24",
+  "types-cachetools>=5.3,<7.0",
+  "types-python-dateutil>=2.8.19,<2.10.0",
   "types-pyyaml~=6.0.12",
   "types-urllib3~=1.26.25",
 ]

{arraylake-1.0.0 → arraylake-1.0.1}/tests/test_api_utils.py

@@ -18,6 +18,7 @@ from arraylake.api_utils import (
     UserAuth,
     _emitted_deprecation_warnings,
     _get_proxy_from_env,
+    _get_session_id,
     calc_backoff,
     handle_response,
     retry_on_exception,
@@ -81,6 +82,24 @@ async def test_http_request_headers(respx_mock, token, test_token_file) -> None:
     assert route.calls.last.request.headers["special"] == "header"
 
 
+def test_session_id_stable_within_process() -> None:
+    """Two clients in the same process share one session id (the point of the feature)."""
+    c1 = ArraylakeHttpClient("https://foo.com")
+    c2 = ArraylakeHttpClient("https://foo.com")
+    sid = _get_session_id()
+    assert c1._default_headers["arraylake-session-id"] == sid
+    assert c2._default_headers["arraylake-session-id"] == sid
+
+
+def test_session_id_survives_pickling() -> None:
+    """Session id rides along when the client is pickled → one dask cluster, one session."""
+    client = ArraylakeHttpClient("https://foo.com")
+    sid = client._default_headers["arraylake-session-id"]
+    assert sid
+    restored = pickle.loads(pickle.dumps(client))
+    assert restored._default_headers["arraylake-session-id"] == sid
+
+
 def test_calc_backoff() -> None:
     assert calc_backoff(0, backoff_factor=0.5, jitter_ratio=0.1, max_backoff_wait=10) == 0
     assert calc_backoff(1, backoff_factor=0.5, jitter_ratio=0.0, max_backoff_wait=10) == 0.5