modal 1.1.2.dev27__tar.gz → 1.1.2.dev28__tar.gz

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: modal
-Version: 1.1.2.dev27
+Version: 1.1.2.dev28
 Summary: Python client library for Modal
 Author-email: Modal Labs <support@modal.com>
 License: Apache-2.0

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/modal/_utils/blob_utils.py

@@ -444,14 +444,24 @@ def get_file_upload_spec_from_fileobj(fp: BinaryIO, mount_filename: PurePosixPat
 _FileUploadSource2 = Callable[[], ContextManager[BinaryIO]]
 
 
+@dataclasses.dataclass
+class FileUploadBlock:
+    # The start (byte offset, inclusive) of the block within the file
+    start: int
+    # The end (byte offset, exclusive) of the block, after having removed any trailing zeroes
+    end: int
+    # Raw (unencoded 32 byte) SHA256 sum of the block, not including trailing zeroes
+    contents_sha256: bytes
+
+
 @dataclasses.dataclass
 class FileUploadSpec2:
     source: _FileUploadSource2
     source_description: Union[str, Path]
 
     path: str
-    # Raw (unencoded 32 byte) SHA256 sum per 8MiB file block
-    blocks_sha256: list[bytes]
+    # 8MiB file blocks
+    blocks: list[FileUploadBlock]
     mode: int  # file permission bits (last 12 bits of st_mode)
     size: int
 
@@ -522,53 +532,102 @@ class FileUploadSpec2:
             source_fp.seek(0, os.SEEK_END)
             size = source_fp.tell()
 
-        blocks_sha256 = await hash_blocks_sha256(source, size, hash_semaphore)
+        blocks = await _gather_blocks(source, size, hash_semaphore)
 
         return FileUploadSpec2(
             source=source,
             source_description=source_description,
             path=mount_filename.as_posix(),
-            blocks_sha256=blocks_sha256,
+            blocks=blocks,
             mode=mode & 0o7777,
             size=size,
         )
 
 
-async def hash_blocks_sha256(
+async def _gather_blocks(
     source: _FileUploadSource2,
     size: int,
     hash_semaphore: asyncio.Semaphore,
-) -> list[bytes]:
+) -> list[FileUploadBlock]:
     def ceildiv(a: int, b: int) -> int:
         return -(a // -b)
 
     num_blocks = ceildiv(size, BLOCK_SIZE)
 
-    def blocking_hash_block_sha256(block_idx: int) -> bytes:
-        sha256_hash = hashlib.sha256()
-        block_start = block_idx * BLOCK_SIZE
+    async def gather_block(block_idx: int) -> FileUploadBlock:
+        async with hash_semaphore:
+            return await asyncio.to_thread(_gather_block, source, block_idx)
 
-        with source() as block_fp:
-            block_fp.seek(block_start)
+    tasks = (gather_block(idx) for idx in range(num_blocks))
+    return await asyncio.gather(*tasks)
 
-            num_bytes_read = 0
-            while num_bytes_read < BLOCK_SIZE:
-                chunk = block_fp.read(BLOCK_SIZE - num_bytes_read)
 
-                if not chunk:
-                    break
+def _gather_block(source: _FileUploadSource2, block_idx: int) -> FileUploadBlock:
+    start = block_idx * BLOCK_SIZE
+    end = _find_end_of_block(source, start, start + BLOCK_SIZE)
+    contents_sha256 = _hash_range_sha256(source, start, end)
+    return FileUploadBlock(start=start, end=end, contents_sha256=contents_sha256)
 
-                num_bytes_read += len(chunk)
-                sha256_hash.update(chunk)
 
-        return sha256_hash.digest()
+def _hash_range_sha256(source: _FileUploadSource2, start, end):
+    sha256_hash = hashlib.sha256()
+    range_size = end - start
 
-    async def hash_block_sha256(block_idx: int) -> bytes:
-        async with hash_semaphore:
-            return await asyncio.to_thread(blocking_hash_block_sha256, block_idx)
+    with source() as fp:
+        fp.seek(start)
+
+        num_bytes_read = 0
+        while num_bytes_read < range_size:
+            chunk = fp.read(range_size - num_bytes_read)
+
+            if not chunk:
+                break
+
+            num_bytes_read += len(chunk)
+            sha256_hash.update(chunk)
+
+    return sha256_hash.digest()
+
+
+def _find_end_of_block(source: _FileUploadSource2, start: int, end: int) -> Optional[int]:
+    """Finds the appropriate end of a block, which is the index of the byte just past the last non-zero byte in the
+    block.
+
+    >>> _find_end_of_block(lambda: BytesIO(b"abc123\0\0\0"), 0, 1024)
+    6
+    >>> _find_end_of_block(lambda: BytesIO(b"abc123\0\0\0"), 3, 1024)
+    6
+    >>> _find_end_of_block(lambda: BytesIO(b"abc123\0\0\0"), 0, 3)
+    4
+    >>> _find_end_of_block(lambda: BytesIO(b"abc123\0\0\0a"), 0, 9)
+    6
+    >>> _find_end_of_block(lambda: BytesIO(b"\0\0\0"), 0, 3)
+    0
+    >>> _find_end_of_block(lambda: BytesIO(b"\0\0\0\0\0\0"), 3, 6)
+    3
+    >>> _find_end_of_block(lambda: BytesIO(b""), 0, 1024)
+    0
+    """
+    size = end - start
+    new_end = start
 
-    tasks = (hash_block_sha256(idx) for idx in range(num_blocks))
-    return await asyncio.gather(*tasks)
+    with source() as block_fp:
+        block_fp.seek(start)
+
+        num_bytes_read = 0
+        while num_bytes_read < size:
+            chunk = block_fp.read(size - num_bytes_read)
+
+            if not chunk:
+                break
+
+            stripped_chunk = chunk.rstrip(b"\0")
+            if stripped_chunk:
+                new_end = start + num_bytes_read + len(stripped_chunk)
+
+            num_bytes_read += len(chunk)
+
+    return new_end
 
 
 def use_md5(url: str) -> bool:

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/modal/client.pyi

@@ -33,7 +33,7 @@ class _Client:
         server_url: str,
         client_type: int,
         credentials: typing.Optional[tuple[str, str]],
-        version: str = "1.1.2.dev27",
+        version: str = "1.1.2.dev28",
     ):
         """mdmd:hidden
         The Modal client object is not intended to be instantiated directly by users.
@@ -164,7 +164,7 @@ class Client:
         server_url: str,
         client_type: int,
         credentials: typing.Optional[tuple[str, str]],
-        version: str = "1.1.2.dev27",
+        version: str = "1.1.2.dev28",
     ):
         """mdmd:hidden
         The Modal client object is not intended to be instantiated directly by users.

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/modal/functions.pyi

@@ -433,7 +433,7 @@ class Function(
 
     _call_generator: ___call_generator_spec[typing_extensions.Self]
 
-    class __remote_spec(typing_extensions.Protocol[ReturnType_INNER, P_INNER, SUPERSELF]):
+    class __remote_spec(typing_extensions.Protocol[P_INNER, ReturnType_INNER, SUPERSELF]):
         def __call__(self, /, *args: P_INNER.args, **kwargs: P_INNER.kwargs) -> ReturnType_INNER:
             """Calls the function remotely, executing it with the given arguments and returning the execution's result."""
             ...
@@ -442,7 +442,7 @@ class Function(
             """Calls the function remotely, executing it with the given arguments and returning the execution's result."""
             ...
 
-    remote: __remote_spec[modal._functions.ReturnType, modal._functions.P, typing_extensions.Self]
+    remote: __remote_spec[modal._functions.P, modal._functions.ReturnType, typing_extensions.Self]
 
     class __remote_gen_spec(typing_extensions.Protocol[SUPERSELF]):
         def __call__(self, /, *args, **kwargs) -> typing.Generator[typing.Any, None, None]:
@@ -469,7 +469,7 @@ class Function(
         """
         ...
 
-    class ___experimental_spawn_spec(typing_extensions.Protocol[ReturnType_INNER, P_INNER, SUPERSELF]):
+    class ___experimental_spawn_spec(typing_extensions.Protocol[P_INNER, ReturnType_INNER, SUPERSELF]):
         def __call__(self, /, *args: P_INNER.args, **kwargs: P_INNER.kwargs) -> FunctionCall[ReturnType_INNER]:
             """[Experimental] Calls the function with the given arguments, without waiting for the results.
 
@@ -493,7 +493,7 @@ class Function(
             ...
 
     _experimental_spawn: ___experimental_spawn_spec[
-        modal._functions.ReturnType, modal._functions.P, typing_extensions.Self
+        modal._functions.P, modal._functions.ReturnType, typing_extensions.Self
     ]
 
     class ___spawn_map_inner_spec(typing_extensions.Protocol[P_INNER, SUPERSELF]):
@@ -502,7 +502,7 @@ class Function(
 
     _spawn_map_inner: ___spawn_map_inner_spec[modal._functions.P, typing_extensions.Self]
 
-    class __spawn_spec(typing_extensions.Protocol[ReturnType_INNER, P_INNER, SUPERSELF]):
+    class __spawn_spec(typing_extensions.Protocol[P_INNER, ReturnType_INNER, SUPERSELF]):
         def __call__(self, /, *args: P_INNER.args, **kwargs: P_INNER.kwargs) -> FunctionCall[ReturnType_INNER]:
             """Calls the function with the given arguments, without waiting for the results.
 
@@ -523,7 +523,7 @@ class Function(
             """
             ...
 
-    spawn: __spawn_spec[modal._functions.ReturnType, modal._functions.P, typing_extensions.Self]
+    spawn: __spawn_spec[modal._functions.P, modal._functions.ReturnType, typing_extensions.Self]
 
     def get_raw_f(self) -> collections.abc.Callable[..., typing.Any]:
         """Return the inner Python object wrapped by this Modal Function."""

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/modal/volume.py

@@ -1122,9 +1122,9 @@ class _VolumeUploadContextManager2(_AbstractVolumeUploadContextManager):
             for file_spec in file_specs:
                 blocks = [
                     api_pb2.VolumePutFiles2Request.Block(
-                        contents_sha256=block_sha256, put_response=put_responses.get(block_sha256)
+                        contents_sha256=block.contents_sha256, put_response=put_responses.get(block.contents_sha256)
                     )
-                    for block_sha256 in file_spec.blocks_sha256
+                    for block in file_spec.blocks
                 ]
                 files.append(
                     api_pb2.VolumePutFiles2Request.File(
@@ -1181,7 +1181,7 @@ async def _put_missing_blocks(
         # TODO(dflemstr): Type is `api_pb2.VolumePutFiles2Response.MissingBlock` but synchronicity gets confused
         # by the nested class (?)
         missing_block,
-    ) -> (bytes, bytes):
+    ) -> tuple[bytes, bytes]:
         # Lazily import to keep the eager loading time of this module down
         from ._utils.bytes_io_segment_payload import BytesIOSegmentPayload
 
@@ -1190,9 +1190,7 @@ async def _put_missing_blocks(
         file_spec = file_specs[missing_block.file_index]
         # TODO(dflemstr): What if the underlying file has changed here in the meantime; should we check the
         #  hash again just to be sure?
-        block_sha256 = file_spec.blocks_sha256[missing_block.block_index]
-        block_start = missing_block.block_index * BLOCK_SIZE
-        block_length = min(BLOCK_SIZE, file_spec.size - block_start)
+        block = file_spec.blocks[missing_block.block_index]
 
         if file_spec.path not in file_progresses:
             file_task_id = progress_cb(name=file_spec.path, size=file_spec.size)
@@ -1216,8 +1214,8 @@ async def _put_missing_blocks(
             with file_spec.source() as source_fp:
                 payload = BytesIOSegmentPayload(
                     source_fp,
-                    block_start,
-                    block_length,
+                    block.start,
+                    block.end - block.start,
                     # limit chunk size somewhat to not keep event loop busy for too long
                     chunk_size=256 * 1024,
                     progress_report_cb=task_progress_cb,
@@ -1229,7 +1227,7 @@ async def _put_missing_blocks(
         if len(file_progress.pending_blocks) == 0:
             task_progress_cb(complete=True)
 
-        return block_sha256, resp_data
+        return block.contents_sha256, resp_data
 
     tasks = [asyncio.create_task(put_missing_block(missing_block)) for missing_block in missing_blocks]
     for task_result in asyncio.as_completed(tasks):

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/modal.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: modal
-Version: 1.1.2.dev27
+Version: 1.1.2.dev28
 Summary: Python client library for Modal
 Author-email: Modal Labs <support@modal.com>
 License: Apache-2.0

{modal-1.1.2.dev27 → modal-1.1.2.dev28}/modal_version/__init__.py

@@ -1,4 +1,4 @@
 # Copyright Modal Labs 2025
 """Supplies the current version of the modal client library."""
 
-__version__ = "1.1.2.dev27"
+__version__ = "1.1.2.dev28"