Add BufferedAsyncIteratorToSyncStream util

homeassistant/components/backblaze/backup.py

@@ -18,6 +18,7 @@ from homeassistant.core import HomeAssistant, callback
 
 from . import BackblazeConfigEntry
 from .const import DATA_BACKUP_AGENT_LISTENERS, DOMAIN, SEPARATOR
+from .util import BufferedAsyncIteratorToSyncStream
 
 
 async def async_get_backup_agents(
@@ -98,6 +99,8 @@ class BackblazeBackupAgent(BackupAgent):
         """Upload a backup."""
 
         # Prepare file info metadata to store with the backup in Backblaze
+        # Backblaze can only store a mapping of strings to strings, so we need
+        # to serialize the metadata into a string format.
         file_info = {
             "backup_id": backup.backup_id,
             "database_included": str(backup.database_included).lower(),
@@ -119,12 +122,15 @@ class BackblazeBackupAgent(BackupAgent):
         if backup.folders:
             file_info["folders"] = ",".join(folder.value for folder in backup.folders)
 
-        stream: AsyncIterator[bytes] = await open_stream()
-
+        iterator = await open_stream()
+        stream = BufferedAsyncIteratorToSyncStream(
+            iterator,
+            buffer_size=8 * 1024 * 1024,  # Buffer up to 8MB
+        )
         try:
             await self._hass.async_add_executor_job(
-                self._bucket.upload_bytes,
-                b"".join([chunk async for chunk in stream]),
+                self._bucket.upload_unbound_stream,
+                stream,
                 f"{backup.backup_id}.tar",
                 "application/octet-stream",
                 file_info,

homeassistant/components/backblaze/util.py

@@ -0,0 +1,87 @@
+"""Utilities for the Backblaze B2 integration."""
+
+import asyncio
+from collections.abc import AsyncIterator
+from concurrent.futures import Future
+import io
+
+
+class BufferedAsyncIteratorToSyncStream(io.RawIOBase):
+    """An wrapper to make an AsyncIterator[bytes] a buffered synchronous readable stream."""
+
+    _done: bool = False
+    _read_future: Future[bytes] | None = None
+
+    def __init__(self, iterator: AsyncIterator[bytes], buffer_size: int = 1024) -> None:
+        """Initialize the stream."""
+        self._buffer = bytearray()
+        self._buffer_size = buffer_size
+        self._iterator = iterator
+        self._loop = asyncio.get_running_loop()
+
+    def readable(self) -> bool:
+        """Mark the stream as readable."""
+        return True
+
+    def _load_next_chunk(self) -> None:
+        """Load the next chunk into the buffer."""
+        if self._done:
+            return
+
+        if not self._read_future:
+            # Fetch a larger chunk asynchronously
+            self._read_future = asyncio.run_coroutine_threadsafe(
+                self._fetch_next_chunk(), self._loop
+            )
+
+        if self._read_future.done():
+            try:
+                data = self._read_future.result()
+                if data:
+                    self._buffer.extend(data)
+                else:
+                    self._done = True
+            except StopAsyncIteration:
+                self._done = True
+            except Exception as err:  # noqa: BLE001
+                raise io.BlockingIOError(f"Failed to load chunk: {err}") from err
+            finally:
+                self._read_future = None
+
+    async def _fetch_next_chunk(self) -> bytes:
+        """Fetch multiple chunks until buffer size is filled."""
+        chunks = []
+        total_size = 0
+
+        try:
+            # Fill the buffer up to the specified size
+            while total_size < self._buffer_size:
+                chunk = await anext(self._iterator)
+                chunks.append(chunk)
+                total_size += len(chunk)
+        except StopAsyncIteration:
+            pass  # The end, return what we have
+
+        return b"".join(chunks)
+
+    def read(self, size: int = -1) -> bytes:
+        """Read bytes."""
+        if size == -1:
+            # Read all remaining data
+            while not self._done:
+                self._load_next_chunk()
+            size = len(self._buffer)
+
+        # Ensure enough data in the buffer
+        while len(self._buffer) < size and not self._done:
+            self._load_next_chunk()
+
+        # Return requested data
+        data = self._buffer[:size]
+        self._buffer = self._buffer[size:]
+        return bytes(data)
+
+    def close(self) -> None:
+        """Close the stream."""
+        self._done = True
+        super().close()