thds.adls 4.1.20250703020842__py3-none-any.whl → 4.1.20250709185906__py3-none-any.whl

thds/adls/download.py

@@ -320,6 +320,18 @@ def _excs_to_retry() -> ty.Callable[[Exception], bool]:
     )
 
 
+def _log_nonfatal_hash_error_exc(exc: Exception, url: str) -> None:
+    """Azure exceptions are very noisy."""
+    msg = "Unable to set hash for %s: %s"
+    exception_txt = str(exc)
+    log, extra_txt = (
+        (logger.debug, type(exc).__name__)
+        if ("AuthorizationPermissionMismatch" in exception_txt or "ConditionNotMet" in exception_txt)
+        else (logger.warning, exception_txt)
+    )
+    log(msg, url, extra_txt)
+
+
 @_dl_scope.bound
 def download_or_use_verified(
     fs_client: FileSystemClient,
@@ -362,7 +374,7 @@ def download_or_use_verified(
                 assert file_properties
                 dl_file_client.set_metadata(meta, **etag.match_etag(file_properties))
             except (HttpResponseError, ResourceModifiedError) as ex:
-                logger.info(f"Unable to set Hash for {remote_key}: {ex}")
+                _log_nonfatal_hash_error_exc(ex, dl_file_client.url)
         return si.value.hit
     except AzureError as err:
         errors.translate_azure_error(fs_client, remote_key, err)
@@ -420,7 +432,7 @@ async def async_download_or_use_verified(
                 await dl_file_client.set_metadata(meta, **etag.match_etag(file_properties))  # type: ignore[misc]
                 # TODO - check above type ignore
             except (HttpResponseError, ResourceModifiedError) as ex:
-                logger.info(f"Unable to set Hash for {remote_key}: {ex}")
+                _log_nonfatal_hash_error_exc(ex, dl_file_client.url)
         return si.value.hit
     except AzureError as err:
         errors.translate_azure_error(fs_client, remote_key, err)

thds/adls/errors.py

@@ -27,6 +27,10 @@ class ContentLengthMismatchError(BlobPropertiesValidationError):
     """Raised when the content length of a file does not match the expected value as retrieved from the server."""
 
 
+class NotADirectoryError(ValueError):
+    """Raised when a path was expected to be a directory but is not."""
+
+
 def is_blob_not_found(exc: Exception) -> bool:
     return (isinstance(exc, HttpResponseError) and exc.status_code == 404) or isinstance(
         exc, BlobNotFoundError

thds/adls/impl.py

@@ -34,7 +34,7 @@ from thds.core import lazy, log
 from ._upload import async_upload_decision_and_metadata
 from .conf import CONNECTION_TIMEOUT, UPLOAD_CHUNK_SIZE
 from .download import async_download_or_use_verified
-from .errors import translate_azure_error
+from .errors import NotADirectoryError, translate_azure_error
 from .file_properties import is_directory
 from .ro_cache import from_cache_path_to_local, global_cache
 from .shared_credential import get_credential_kwargs
@@ -247,9 +247,10 @@ class ADLSFileSystem:
         """
         # normalize remote path to a standard relative dir path -
         # this ensures correctness of strip_prefix() below
-        remote_path = remote_path.strip("/") + "/"
+        stripped_remote_path = remote_path.strip("/")
+        remote_path = stripped_remote_path + "/"
         dir_path = self._local_path_for(remote_path, local_path)
-        dir_path.mkdir(exist_ok=True, parents=True)
+        made_dir = False
         path_filter_ = _true if path_filter is None else path_filter
 
         # remove the remote directory prefix to determine a relative path for creation under dir_path
@@ -262,8 +263,22 @@ class ADLSFileSystem:
             if not path.is_directory and path_filter_(path)
         )
 
+        # shim generator to check for file vs directory, to prevent confusing errors that happen lower down
+        async def validated_paths() -> AsyncIterator[PathPair]:
+            async for path_pair in paths:
+                if path_pair.remote_path == stripped_remote_path:
+                    raise NotADirectoryError(
+                        f"Path '{stripped_remote_path}' points to a file, not a directory. "
+                        f"Use fetch_file() instead."
+                    )
+                nonlocal made_dir
+                if not made_dir:
+                    dir_path.mkdir(exist_ok=True, parents=True)
+                    made_dir = True
+                yield path_pair
+
         local_paths = []
-        async for batch in self._async_batch(paths, batch_size):
+        async for batch in self._async_batch(validated_paths(), batch_size):
             local_paths.extend(
                 await asyncio.gather(
                     *[

{thds_adls-4.1.20250703020842.dist-info → thds_adls-4.1.20250709185906.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: thds.adls
-Version: 4.1.20250703020842
+Version: 4.1.20250709185906
 Summary: ADLS tools
 Author-email: Trilliant Health <info@trillianthealth.com>
 License: MIT

{thds_adls-4.1.20250703020842.dist-info → thds_adls-4.1.20250709185906.dist-info}/RECORD

@@ -7,15 +7,15 @@ thds/adls/conf.py,sha256=nTw3X1ilC3A_905jZH-rWXFsESeHAKQn5IghvfX2VIo,1991
 thds/adls/copy.py,sha256=jUWbGvTpb4B3yRGS0nhGSbDzqRPzUqYgH0z1lFRJB3k,6365
 thds/adls/dbfs.py,sha256=pPAjbIZRKJsaXKQljDMUgqS_zy1yKeEZHGMueXbuv3g,2219
 thds/adls/defaults.py,sha256=GGq5Pn4r-8cX4bZItp4nnwWAAz7S07pzPoOegw0y5Fw,676
-thds/adls/download.py,sha256=u-ckaExnTMAVUlhMbnxM_urBggbZkgXMQSSfp1czPeQ,18350
+thds/adls/download.py,sha256=WOpMXGUbWImBdkM4tSW7qnCbu7G_cRXKF5pFQVLPPxs,18772
 thds/adls/download_lock.py,sha256=tgT48l4C5_qmArGeq05gl7VlxT22dZBH2Xwxx0itE9o,3176
-thds/adls/errors.py,sha256=6cLg2E4SB8ic46PBzA3ynRH4b1oR8qRb07RBgKGJRxY,1783
+thds/adls/errors.py,sha256=6NMLHtVNsWBRDXaes9yzHj9cwKOD9t1dwL4BltdtjhU,1895
 thds/adls/etag.py,sha256=ct7jpHhNFcKzbekn5rZ3m6DhjK48A7qOZGwDiHkc-pc,242
 thds/adls/file_properties.py,sha256=C9Kl3a5wuBNWYgZYnZbkH04u8uxadEcjVJIm3UevUM0,1912
 thds/adls/fqn.py,sha256=0zHmHhBWN7GEfKRB3fBC1NVhaiIHHifBdCRanyT01X8,5822
 thds/adls/global_client.py,sha256=f4VJw5y_Yh__8gQUcdSYTh1aU6iEPlauMchVirSAwDQ,3716
 thds/adls/hashes.py,sha256=RDQS-C38wskUhxXGFGLJ4ox8vm7ofurxSsUk13Ywijo,5309
-thds/adls/impl.py,sha256=4rZAGlhU_UojPy1FC7j3KEFIj6BWSbCDAVV1FCyki3s,42496
+thds/adls/impl.py,sha256=cNf1vmeS46X_wvyVdDJ8qFfowHn2QwtU5C80BmDtu5Y,43247
 thds/adls/md5.py,sha256=hGT8AIX32VUsnRCbm8cel9OlxAiRrgjwNWQTqRDHM_k,374
 thds/adls/named_roots.py,sha256=7SLbAoQQpV_mrFZaUPjYoS-F9dxQxN5Hg4M3YPirF_w,751
 thds/adls/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -35,8 +35,8 @@ thds/adls/azcopy/upload.py,sha256=0l5FzV9IgZ2iQhm4eKZjTdw4SO17bHd8VnwcTev1lUs,27
 thds/adls/tools/download.py,sha256=CW2cWbCRdUqisVVVoqqvqk5Ved7pPGTkwnZj3uV0jy4,1587
 thds/adls/tools/ls.py,sha256=OgEaIfTK359twlZIj-A0AW_nv81Z6zi0b9Tw6OJJfWA,1083
 thds/adls/tools/upload.py,sha256=5WyWkpuVp2PETZ3O3ODlq8LXszSHU73ZMnIDZXPJdC8,442
-thds_adls-4.1.20250703020842.dist-info/METADATA,sha256=zWALY04Mrs1zZNpje2AyaiU-B6tN7La5ldtAbVEXxac,587
-thds_adls-4.1.20250703020842.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-thds_adls-4.1.20250703020842.dist-info/entry_points.txt,sha256=uTqreT1AIwqJboMfLv5w6sviM8mNbAkln765gIjzoA4,152
-thds_adls-4.1.20250703020842.dist-info/top_level.txt,sha256=LTZaE5SkWJwv9bwOlMbIhiS-JWQEEIcjVYnJrt-CriY,5
-thds_adls-4.1.20250703020842.dist-info/RECORD,,
+thds_adls-4.1.20250709185906.dist-info/METADATA,sha256=xmQtReaLdgxixhJlbPzjQGhgClkAvRB15p2vSGi_BtY,587
+thds_adls-4.1.20250709185906.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+thds_adls-4.1.20250709185906.dist-info/entry_points.txt,sha256=uTqreT1AIwqJboMfLv5w6sviM8mNbAkln765gIjzoA4,152
+thds_adls-4.1.20250709185906.dist-info/top_level.txt,sha256=LTZaE5SkWJwv9bwOlMbIhiS-JWQEEIcjVYnJrt-CriY,5
+thds_adls-4.1.20250709185906.dist-info/RECORD,,