feat: handle interrupted downloads with decompressive transcoding

google/resumable_media/_helpers.py

@@ -32,6 +32,7 @@
 
 RANGE_HEADER = "range"
 CONTENT_RANGE_HEADER = "content-range"
+CONTENT_ENCODING_HEADER = "content-encoding"
 
 _SLOW_CRC32C_WARNING = (
     "Currently using crcmod in pure python form. This is a slow "
@@ -40,6 +41,8 @@
 )
 _GENERATION_HEADER = "x-goog-generation"
 _HASH_HEADER = "x-goog-hash"
+_STORED_CONTENT_ENCODING_HEADER = "x-goog-stored-content-encoding"
+
 _MISSING_CHECKSUM = """\
 No {checksum_type} checksum was returned from the service while downloading {}
 (which happens for composite objects), so client-side content integrity
@@ -369,6 +372,23 @@ def add_query_parameters(media_url, query_params):
     return urlunsplit((scheme, netloc, path, query, frag))
 
 
+def _is_decompressive_transcoding(response, get_headers):
+    """Returns True if the object was served decompressed. This happens when the
+    "x-goog-stored-content-encoding" header is "gzip" and "content-encoding" header
+    is not "gzip". See more at: https://cloud.google.com/storage/docs/transcoding#transcoding_and_gzip
+    Args:
+        response (~requests.Response): The HTTP response object.
+        get_headers (callable: response->dict): returns response headers.
+    Returns:
+        bool: Returns True if decompressive transcoding has occurred; otherwise, False.
+    """
+    headers = get_headers(response)
+    return (
+        headers.get(_STORED_CONTENT_ENCODING_HEADER) == "gzip"
+        and headers.get(CONTENT_ENCODING_HEADER) != "gzip"
+    )
+
+
 class _DoNothingHash(object):
     """Do-nothing hash object.
 

google/resumable_media/requests/download.py

@@ -36,6 +36,13 @@
   {}
 """
 
+_STREAM_SEEK_ERROR = """\
+Incomplete download for:
+{}
+Error writing to stream while handling a gzip-compressed file download.
+Please restart the download.
+"""
+
 
 class Download(_request_helpers.RequestsMixin, _download.Download):
     """Helper to manage downloading a resource from a Google API.
@@ -206,7 +213,18 @@ def retriable_request():
 
             self._process_response(result)
 
+            # With decompressive transcoding, GCS serves back the whole file regardless of the range request,
+            # thus we reset the stream position to the start of the stream.
+            # See: https://cloud.google.com/storage/docs/transcoding#range
             if self._stream is not None:
+                if _helpers._is_decompressive_transcoding(result, self._get_headers):
+                    try:
+                        self._stream.seek(0)
+                    except Exception as exc:
+                        msg = _STREAM_SEEK_ERROR.format(url)
+                        raise Exception(msg) from exc
+                    self._bytes_downloaded = 0
+
                 self._write_to_stream(result)
 
             return result
@@ -379,7 +397,18 @@ def retriable_request():
 
             self._process_response(result)
 
+            # With decompressive transcoding, GCS serves back the whole file regardless of the range request,
+            # thus we reset the stream position to the start of the stream.
+            # See: https://cloud.google.com/storage/docs/transcoding#range
             if self._stream is not None:
+                if _helpers._is_decompressive_transcoding(result, self._get_headers):
+                    try:
+                        self._stream.seek(0)
+                    except Exception as exc:
+                        msg = _STREAM_SEEK_ERROR.format(url)
+                        raise Exception(msg) from exc
+                    self._bytes_downloaded = 0
+
                 self._write_to_stream(result)
 
             return result

tests/system/requests/test_download.py

@@ -294,6 +294,26 @@ def test_download_to_stream(self, add_files, authorized_transport):
             assert stream.getvalue() == actual_contents
             check_tombstoned(download, authorized_transport)
 
+    def test_download_gzip_w_stored_content_headers(
+        self, add_files, authorized_transport
+    ):
+        # Retrieve the gzip compressed file
+        info = ALL_FILES[-1]
+        actual_contents = self._get_contents(info)
+        blob_name = get_blob_name(info)
+
+        # Create the actual download object.
+        media_url = utils.DOWNLOAD_URL_TEMPLATE.format(blob_name=blob_name)
+        stream = io.BytesIO()
+        download = self._make_one(media_url, stream=stream)
+        # Consume the resource.
+        response = download.consume(authorized_transport)
+        assert response.status_code == http.client.OK
+        assert response.headers.get(_helpers._STORED_CONTENT_ENCODING_HEADER) == "gzip"
+        assert response.headers.get("X-Goog-Stored-Content-Length") is not None
+        assert stream.getvalue() == actual_contents
+        check_tombstoned(download, authorized_transport)
+
     def test_extra_headers(self, authorized_transport, secret_file):
         blob_name, data, headers = secret_file
         # Create the actual download object.

tests/unit/requests/test_download.py

@@ -395,6 +395,45 @@ def test_consume_w_bytes_downloaded(self):
         range_bytes = "bytes={:d}-{:d}".format(offset, end)
         assert download._headers["range"] == range_bytes
 
+    def test_consume_gzip_reset_stream_w_bytes_downloaded(self):
+        stream = io.BytesIO()
+        chunks = (b"up down ", b"charlie ", b"brown")
+        end = 65536
+
+        download = download_mod.Download(
+            EXAMPLE_URL, stream=stream, end=end, headers=None, checksum="md5"
+        )
+        transport = mock.Mock(spec=["request"])
+
+        # Mock a decompressive transcoding retry operation with bytes already downloaded in the stream
+        headers = {_helpers._STORED_CONTENT_ENCODING_HEADER: "gzip"}
+        transport.request.return_value = _mock_response(chunks=chunks, headers=headers)
+        offset = 16
+        download._bytes_downloaded = offset
+        download.consume(transport)
+
+        assert stream.getvalue() == b"".join(chunks)
+        assert download._bytes_downloaded == len(b"".join(chunks))
+
+    def test_consume_gzip_reset_stream_error(self):
+        stream = io.BytesIO()
+        chunks = (b"up down ", b"charlie ", b"brown")
+        end = 65536
+
+        download = download_mod.Download(
+            EXAMPLE_URL, stream=stream, end=end, headers=None, checksum="md5"
+        )
+        transport = mock.Mock(spec=["request"])
+
+        # Mock a stream seek error while resuming a decompressive transcoding download
+        stream.seek = mock.Mock(side_effect=OSError("mock stream seek error"))
+        headers = {_helpers._STORED_CONTENT_ENCODING_HEADER: "gzip"}
+        transport.request.return_value = _mock_response(chunks=chunks, headers=headers)
+        offset = 16
+        download._bytes_downloaded = offset
+        with pytest.raises(Exception):
+            download.consume(transport)
+
 
 class TestRawDownload(object):
     def test__write_to_stream_no_hash_check(self):
@@ -772,6 +811,49 @@ def test_consume_w_bytes_downloaded(self):
         range_bytes = "bytes={:d}-{:d}".format(offset, end)
         assert download._headers["range"] == range_bytes
 
+    def test_consume_gzip_reset_stream_w_bytes_downloaded(self):
+        stream = io.BytesIO()
+        chunks = (b"up down ", b"charlie ", b"brown")
+        end = 65536
+
+        download = download_mod.RawDownload(
+            EXAMPLE_URL, stream=stream, end=end, headers=None, checksum="md5"
+        )
+        transport = mock.Mock(spec=["request"])
+
+        # Mock a decompressive transcoding retry operation with bytes already downloaded in the stream
+        headers = {_helpers._STORED_CONTENT_ENCODING_HEADER: "gzip"}
+        transport.request.return_value = _mock_raw_response(
+            chunks=chunks, headers=headers
+        )
+        offset = 16
+        download._bytes_downloaded = offset
+        download.consume(transport)
+
+        assert stream.getvalue() == b"".join(chunks)
+        assert download._bytes_downloaded == len(b"".join(chunks))
+
+    def test_consume_gzip_reset_stream_error(self):
+        stream = io.BytesIO()
+        chunks = (b"up down ", b"charlie ", b"brown")
+        end = 65536
+
+        download = download_mod.RawDownload(
+            EXAMPLE_URL, stream=stream, end=end, headers=None, checksum="md5"
+        )
+        transport = mock.Mock(spec=["request"])
+
+        # Mock a stream seek error while resuming a decompressive transcoding download
+        stream.seek = mock.Mock(side_effect=OSError("mock stream seek error"))
+        headers = {_helpers._STORED_CONTENT_ENCODING_HEADER: "gzip"}
+        transport.request.return_value = _mock_raw_response(
+            chunks=chunks, headers=headers
+        )
+        offset = 16
+        download._bytes_downloaded = offset
+        with pytest.raises(Exception):
+            download.consume(transport)
+
 
 class TestChunkedDownload(object):
     @staticmethod

tests/unit/test__helpers.py

@@ -425,6 +425,31 @@ def test_header_value(self):
         assert generation_header == self.GENERATION_VALUE
 
 
+class Test__is_decompressive_transcoding(object):
+    def test_empty_value(self):
+        headers = {}
+        response = _mock_response(headers=headers)
+        assert _helpers._is_decompressive_transcoding(response, _get_headers) is False
+
+    def test_gzip_in_headers(self):
+        headers = {_helpers._STORED_CONTENT_ENCODING_HEADER: "gzip"}
+        response = _mock_response(headers=headers)
+        assert _helpers._is_decompressive_transcoding(response, _get_headers) is True
+
+    def test_gzip_not_in_headers(self):
+        headers = {_helpers._STORED_CONTENT_ENCODING_HEADER: "identity"}
+        response = _mock_response(headers=headers)
+        assert _helpers._is_decompressive_transcoding(response, _get_headers) is False
+
+    def test_gzip_w_content_encoding_in_headers(self):
+        headers = {
+            _helpers._STORED_CONTENT_ENCODING_HEADER: "gzip",
+            _helpers.CONTENT_ENCODING_HEADER: "gzip",
+        }
+        response = _mock_response(headers=headers)
+        assert _helpers._is_decompressive_transcoding(response, _get_headers) is False
+
+
 class Test__get_generation_from_url(object):
 
     GENERATION_VALUE = 1641590104888641