ygg 0.1.51__py3-none-any.whl → 0.1.52__py3-none-any.whl

yggdrasil/databricks/workspaces/path_kind.py

@@ -8,6 +8,6 @@ __all__ = ["DatabricksPathKind"]
 
 class DatabricksPathKind(str, Enum):
     """Supported Databricks path kinds for workspace, volumes, and DBFS."""
-    WORKSPACE = "workspace"
-    VOLUME = "volume"
-    DBFS = "dbfs"
+    WORKSPACE = 1
+    VOLUME = 2
+    DBFS = 3

yggdrasil/databricks/workspaces/volumes_path.py

@@ -0,0 +1,85 @@
+import datetime as dt
+from typing import Tuple, Optional
+
+from ...libs.databrickslib import databricks_sdk, WorkspaceClient
+
+if databricks_sdk is not None:
+    from databricks.sdk.errors.platform import (
+        NotFound,
+        ResourceDoesNotExist,
+        BadRequest,
+        PermissionDenied,
+    )
+
+
+__all__ = [
+    "get_volume_status",
+    "get_volume_metadata"
+]
+
+
+def get_volume_status(
+    sdk: WorkspaceClient,
+    full_path: str,
+    check_file_first: bool = True,
+    raise_error: bool = True,
+) -> Tuple[Optional[bool], Optional[bool], Optional[int], Optional[dt.datetime]]:
+    client = sdk.files
+
+    if check_file_first:
+        try:
+            info = client.get_metadata(full_path)
+            return True, False, info.content_length, _parse_mtime(info)
+        except (NotFound, ResourceDoesNotExist, BadRequest, PermissionDenied) as e:
+            pass
+
+        try:
+            info = client.get_directory_metadata(full_path)
+            return False, True, 0, _parse_mtime(info)
+        except (NotFound, ResourceDoesNotExist, BadRequest, PermissionDenied) as e:
+            last_exception = e
+    else:
+        try:
+            info = client.get_directory_metadata(full_path)
+            return False, True, 0, _parse_mtime(info)
+        except (NotFound, ResourceDoesNotExist, BadRequest, PermissionDenied) as e:
+            pass
+
+        try:
+            info = client.get_metadata(full_path)
+            return True, False, info.content_length, _parse_mtime(info)
+        except (NotFound, ResourceDoesNotExist, BadRequest, PermissionDenied) as e:
+            last_exception = e
+
+    if raise_error and last_exception is not None:
+        raise last_exception
+
+    return None, None, None, None
+
+
+def get_volume_metadata(
+    sdk: WorkspaceClient,
+    full_name: str,
+    include_browse: bool = False,
+    raise_error: bool = True,
+):
+    client = sdk.volumes
+
+    try:
+        return client.read(
+            name=full_name,
+            include_browse=include_browse
+        )
+    except (NotFound, ResourceDoesNotExist, BadRequest, PermissionDenied):
+        if raise_error:
+            raise
+
+    return None
+
+
+
+def _parse_mtime(info):
+    if not info:
+        return dt.datetime.now(tz=dt.timezone.utc)
+
+    return dt.datetime.strptime(info.last_modified, "%a, %d %b %Y %H:%M:%S %Z").replace(tzinfo=dt.timezone.utc)

yggdrasil/libs/databrickslib.py

@@ -4,6 +4,8 @@ try:
     import databricks
     import databricks.sdk  # type: ignore
 
+    from databricks.sdk import WorkspaceClient
+
     databricks = databricks
     databricks_sdk = databricks.sdk
 except ImportError:
@@ -16,6 +18,8 @@ except ImportError:
     databricks = _DatabricksDummy
     databricks_sdk = _DatabricksDummy
 
+    WorkspaceClient = _DatabricksDummy
+
 
 def require_databricks_sdk():
     """Ensure the Databricks SDK is available before use.
@@ -34,4 +38,5 @@ __all__ = [
     "databricks",
     "databricks_sdk",
     "require_databricks_sdk",
+    "WorkspaceClient"
 ]

yggdrasil/pyutils/callable_serde.py

@@ -27,7 +27,6 @@ import dis
 import importlib
 import inspect
 import io
-import lzma
 import os
 import secrets
 import struct
@@ -66,6 +65,14 @@ def _try_import_zstd():
         return None
 
 
+def _try_import_lzma():
+    try:
+        import lzma  # type: ignore
+        return lzma
+    except Exception:
+        return None
+
+
 def _pick_zlib_level(n: int, limit: int) -> int:
     """Ramp compression level 1..9 based on how far we exceed the byte_limit."""
     ratio = n / max(1, limit)
@@ -94,13 +101,6 @@ def _encode_with_candidates(raw: bytes, *, byte_limit: int, allow_zstd: bool) ->
                 except Exception:
                     pass
 
-    for preset in (6, 9):
-        try:
-            c = lzma.compress(raw, preset=preset)
-            candidates.append(_frame(_CODEC_LZMA, len(raw), preset, c))
-        except Exception:
-            pass
-
     lvl = _pick_zlib_level(len(raw), byte_limit)
     try:
         c = zlib.compress(raw, lvl)
@@ -143,8 +143,8 @@ def _decode_result_blob(blob: bytes) -> bytes:
         raw = data
     elif codec == _CODEC_ZLIB:
         raw = zlib.decompress(data)
-    elif codec == _CODEC_LZMA:
-        raw = lzma.decompress(data)
+    # elif codec == _CODEC_LZMA:
+    #     raw = lzma.decompress(data)
     elif codec == _CODEC_ZSTD:
         zstd = _try_import_zstd()
         if zstd is None:

yggdrasil/pyutils/retry.py

@@ -82,14 +82,14 @@ def retry(
     logger:
         Optional logger for debug/info messages.
     reraise:
-        If True, re-raise the last exception after exhausting retries or timeout.
+        If True, reraise the last exception after exhausting retries or timeout.
         If False, returns None after all retries fail or timeout is hit.
     timeout:
         Optional max total time in seconds across all attempts. The timeout is
         checked after each failed attempt. In-flight attempts are not forcibly
         cancelled; we just stop scheduling new retries when the timeout is hit.
 
-    Usage
+    Example
     -----
     @retry(exceptions=(ValueError,), tries=5, delay=0.1, backoff=2)
     def flaky():

yggdrasil/types/cast/registry.py

@@ -68,20 +68,6 @@ def register_converter(
         Returns:
             Registered converter.
         """
-        sig = inspect.signature(func)
-        params = list(sig.parameters.values())
-        if any(
-            param.kind in {inspect.Parameter.VAR_POSITIONAL, inspect.Parameter.VAR_KEYWORD}
-            for param in params
-        ):
-            raise TypeError(
-                "Converters must accept exactly two positional arguments: (value, options)"
-            )
-
-        if len(params) < 2:
-            raise TypeError(
-                "Converters must accept at least two positional arguments: (value, options)"
-            )
 
         if from_hint in (Any, object):
             _any_registry[to_hint] = func

yggdrasil/types/file_format.py

@@ -0,0 +1,10 @@
+from pyarrow.dataset import FileFormat
+
+
+__all__ = [
+    "ExcelFileFormat"
+]
+
+
+class ExcelFileFormat(FileFormat):
+    pass

yggdrasil/version.py

@@ -1 +1 @@
-__version__ = "0.1.51"
+__version__ = "0.1.52"