Function to create the delta for a file

kittens/transfer/librsync.py

@@ -3,34 +3,39 @@
 # License: GPLv3 Copyright: 2021, Kovid Goyal <kovid at kovidgoyal.net>
 
 import os
-from typing import Generator, Iterator, Optional
+from typing import IO, Generator, Iterator, Optional
 
 from .rsync import (
-    IO_BUFFER_SIZE, RsyncError, SignatureCapsule, begin_create_signature,
-    begin_load_signature, iter_job, make_hash_table
+    IO_BUFFER_SIZE, JobCapsule, RsyncError, SignatureCapsule,
+    begin_create_delta, begin_create_signature, begin_load_signature, iter_job,
+    make_hash_table
 )
 
 
+def drive_job_on_file(f: IO[bytes], job: JobCapsule) -> Iterator[bytes]:
+    finished = False
+    prev_unused_input = b''
+    while not finished:
+        input_data = f.read(IO_BUFFER_SIZE)
+        no_more_data = not input_data
+        if prev_unused_input:
+            input_data = prev_unused_input + input_data
+            prev_unused_input = b''
+        output, finished, sz_of_unused_input = iter_job(job, input_data, no_more_data)
+        if sz_of_unused_input > 0 and not finished:
+            if no_more_data:
+                raise RsyncError(f"{sz_of_unused_input} bytes of input data were not used")
+            prev_unused_input = input_data[-sz_of_unused_input:]
+        yield output
+
+
 def signature_of_file(path: str) -> Iterator[bytes]:
     with open(path, 'rb') as f:
         f.seek(0, os.SEEK_END)
         fsz = f.tell()
         job = begin_create_signature(fsz)
         f.seek(0)
-        finished = False
-        prev_unused_input = b''
-        while not finished:
-            input_data = f.read(IO_BUFFER_SIZE)
-            no_more_data = not input_data
-            if prev_unused_input:
-                input_data = prev_unused_input + input_data
-                prev_unused_input = b''
-            output, finished, sz_of_unused_input = iter_job(job, input_data, no_more_data)
-            if sz_of_unused_input > 0 and not finished:
-                if no_more_data:
-                    raise RsyncError(f"{sz_of_unused_input} bytes of input data were not used")
-                prev_unused_input = input_data[-sz_of_unused_input:]
-            yield output
+        yield from drive_job_on_file(f, job)
 
 
 def load_signature() -> Generator[Optional[SignatureCapsule], bytes, None]:
@@ -50,3 +55,9 @@ def load_signature() -> Generator[Optional[SignatureCapsule], bytes, None]:
             prev_unused_input = input_data[-sz_of_unused_input:]
     make_hash_table(signature)
     yield signature
+
+
+def delta_for_file(path: str, sig: SignatureCapsule) -> Iterator[bytes]:
+    job = begin_create_delta(sig)
+    with open(path, 'rb') as f:
+        yield from drive_job_on_file(f, job)

kittens/transfer/rsync.c

@@ -123,7 +123,7 @@ build_hash_table(PyObject *self UNUSED, PyObject *args) {
 static PyObject*
 begin_create_delta(PyObject *self UNUSED, PyObject *args) {
     PyObject *sig_capsule;
-    if (!PyArg_ParseTuple(args, "O!y#|p", &PyCapsule_Type, &sig_capsule)) return NULL;
+    if (!PyArg_ParseTuple(args, "O!", &PyCapsule_Type, &sig_capsule)) return NULL;
     GET_SIG_FROM_CAPSULE;
     rs_job_t *job = rs_delta_begin(sig);
     if (!job) return PyErr_NoMemory();

kittens/transfer/rsync.pyi

@@ -27,5 +27,9 @@ def make_hash_table(sig: SignatureCapsule) -> None:
     pass
 
 
+def begin_create_delta(sig: SignatureCapsule) -> JobCapsule:
+    pass
+
+
 def iter_job(job_capsule: JobCapsule, input_data: bytes, eof: Optional[bool] = None, expecting_output: bool = True) -> Tuple[bytes, bool, int]:
     pass