fix: filter download_kwargs in BlobReader (#1411)

cojenco · web-flow · commit 0c2121045031 · 2025-01-22T14:53:41.000-08:00
diff --git a/google/cloud/storage/fileio.py b/google/cloud/storage/fileio.py
@@ -89,6 +89,7 @@ class BlobReader(io.BufferedIOBase):
         configuration changes for Retry objects such as delays and deadlines
         are respected.
 
+    :type download_kwargs: dict
     :param download_kwargs:
         Keyword arguments to pass to the underlying API calls.
         The following arguments are supported:
@@ -98,9 +99,10 @@ class BlobReader(io.BufferedIOBase):
         - ``if_metageneration_match``
         - ``if_metageneration_not_match``
         - ``timeout``
+        - ``raw_download``
 
-        Note that download_kwargs are also applied to blob.reload(), if a reload
-        is needed during seek().
+        Note that download_kwargs (excluding ``raw_download``) are also applied to blob.reload(),
+        if a reload is needed during seek().
     """
 
     def __init__(self, blob, chunk_size=None, retry=DEFAULT_RETRY, **download_kwargs):
@@ -175,7 +177,10 @@ def seek(self, pos, whence=0):
         self._checkClosed()  # Raises ValueError if closed.
 
         if self._blob.size is None:
-            self._blob.reload(**self._download_kwargs)
+            reload_kwargs = {
+                k: v for k, v in self._download_kwargs.items() if k != "raw_download"
+            }
+            self._blob.reload(**reload_kwargs)
 
         initial_offset = self._pos + self._buffer.tell()
 
@@ -272,6 +277,7 @@ class BlobWriter(io.BufferedIOBase):
         configuration changes for Retry objects such as delays and deadlines
         are respected.
 
+    :type upload_kwargs: dict
     :param upload_kwargs:
         Keyword arguments to pass to the underlying API
         calls. The following arguments are supported:
diff --git a/tests/system/test_fileio.py b/tests/system/test_fileio.py
@@ -116,3 +116,26 @@ def test_blobwriter_exit(
     blobs_to_delete.append(blob)
     # blob should have been uploaded
     assert blob.exists()
+
+
+def test_blobreader_w_raw_download(
+    shared_bucket,
+    blobs_to_delete,
+    file_data,
+):
+    blob = shared_bucket.blob("LargeFile")
+    info = file_data["big"]
+    with open(info["path"], "rb") as file_obj:
+        with blob.open("wb", chunk_size=256 * 1024, if_generation_match=0) as writer:
+            writer.write(file_obj.read())
+        blobs_to_delete.append(blob)
+
+    # Test BlobReader read and seek handles raw downloads.
+    with open(info["path"], "rb") as file_obj:
+        with blob.open("rb", chunk_size=256 * 1024, raw_download=True) as reader:
+            reader.seek(0)
+            file_obj.seek(0)
+            assert file_obj.read() == reader.read()
+            # End of file reached; further reads should be blank but not
+            # raise an error.
+            assert reader.read() == b""
