Merge pull request 'Keep Python wrapper key buffers alive in WriteRange/ReadRange' (#43) from weaselbot/conflict-set:weaselbot/issue-42 into main

Reviewed-on: #43
Reviewed-by: andrew <andrew@weaselab.dev>

conflict_set.py

@@ -27,23 +27,37 @@ class Result(enum.Enum):
     TOO_OLD = 2
 
 
-def write(begin: bytes, end: Optional[bytes] = None) -> WriteRange:
-    b = (ctypes.c_ubyte * len(begin)).from_buffer(bytearray(begin))
+def _make_key(buf: bytes) -> tuple[_Key, bytearray]:
+    """Create a _Key and a backing bytearray that must be kept alive."""
+    backing = bytearray(buf)
+    array = (ctypes.c_ubyte * len(backing)).from_buffer(backing)
+    return _Key(array, len(array)), backing
 
+
+def write(begin: bytes, end: Optional[bytes] = None) -> WriteRange:
+    begin_key, begin_buf = _make_key(begin)
     if end is None:
-        e = (ctypes.c_ubyte * 0)()
+        end_key = _Key((ctypes.c_ubyte * 0)(), 0)
+        end_buf = None
     else:
-        e = (ctypes.c_ubyte * len(end)).from_buffer(bytearray(end))
-    return WriteRange(_Key(b, len(b)), _Key(e, len(e)))
+        end_key, end_buf = _make_key(end)
+    result = WriteRange(begin_key, end_key)
+    result._begin_buf = begin_buf
+    result._end_buf = end_buf
+    return result
 
 
 def read(version: int, begin: bytes, end: Optional[bytes] = None) -> ReadRange:
-    b = (ctypes.c_ubyte * len(begin)).from_buffer(bytearray(begin))
+    begin_key, begin_buf = _make_key(begin)
     if end is None:
-        e = (ctypes.c_ubyte * 0)()
+        end_key = _Key((ctypes.c_ubyte * 0)(), 0)
+        end_buf = None
     else:
-        e = (ctypes.c_ubyte * len(end)).from_buffer(bytearray(end))
-    return ReadRange(_Key(b, len(b)), _Key(e, len(e)), version)
+        end_key, end_buf = _make_key(end)
+    result = ReadRange(begin_key, end_key, version)
+    result._begin_buf = begin_buf
+    result._end_buf = end_buf
+    return result
 
 
 class ConflictSet:

test_conflict_set.py

@@ -57,6 +57,42 @@ def test_conflict_set():
         assert cs.check(read(0, key), read(1, key)) == [Result.TOO_OLD, Result.COMMIT]
 
 
+def test_write_read_without_outer_reference():
+    # Regression test for issue #42: WriteRange/ReadRange must keep their
+    # backing key buffers alive, because the C library reads the pointer
+    # stored in _Key while addWrites/check run.
+    with DebugConflictSet() as cs:
+        # The bytes literal is not referenced after this expression.
+        cs.addWrites(1, write(b"key"))
+        assert cs.check(read(0, b"key")) == [Result.CONFLICT]
+
+        cs.addWrites(2, write(b"a", b"z"))
+        assert cs.check(read(1, b"a", b"z")) == [Result.CONFLICT]
+        assert cs.check(read(1, b"b")) == [Result.CONFLICT]
+        assert cs.check(read(1, b"0")) == [Result.COMMIT]
+
+
+def test_range_keeps_key_buffers_alive():
+    # Verify the fix for issue #42: returned range objects must retain a
+    # reference to the backing bytearray so the C pointer stays valid after
+    # the helper returns.
+    w = write(b"key")
+    assert w._begin_buf == bytearray(b"key")
+    assert w._end_buf is None
+
+    w2 = write(b"a", b"z")
+    assert w2._begin_buf == bytearray(b"a")
+    assert w2._end_buf == bytearray(b"z")
+
+    r = read(0, b"key")
+    assert r._begin_buf == bytearray(b"key")
+    assert r._end_buf is None
+
+    r2 = read(1, b"a", b"z")
+    assert r2._begin_buf == bytearray(b"a")
+    assert r2._end_buf == bytearray(b"z")
+
+
 def test_update_zero_should_commit():
     with DebugConflictSet() as cs1:
         with DebugConflictSet() as cs2: