Add raw binary fixture support for gossip harness

src/test_suite/fuzz_context.py

@@ -98,6 +98,7 @@
     context_human_encode_fn=gossip_codec.encode_input,
     context_human_decode_fn=gossip_codec.decode_input,
     effects_human_encode_fn=gossip_codec.encode_output,
+    raw_binary_io=True,
 )
 
 

src/test_suite/fuzz_interface.py

@@ -84,6 +84,9 @@ class HarnessCtx:
     effects_human_decode_fn: Callable[[EffectsType], None] = generic_human_decode
     regenerate_transformation_fn: Callable[[FixtureType], None] = generic_transform
     supports_flatbuffers: bool = False
+    raw_binary_io: bool = (
+        False  # context.data passed as raw bytes; output is a raw single-byte bool
+    )
     fixture_type: Type[FixtureType] = field(init=False)
     context_type: Type[ContextType] = field(init=False)
     effects_type: Type[EffectsType] = field(init=False)

src/test_suite/multiprocessing_utils.py

@@ -176,12 +176,13 @@ def process_target(
         - invoke_pb.InstrEffects | None: Result of instruction execution.
     """
 
-    serialized_instruction_context = context.SerializeToString(deterministic=True)
-    if serialized_instruction_context is None:
-        return None
-
-    # Prepare input data and output buffers
-    in_data = serialized_instruction_context
+    if harness_ctx.raw_binary_io:
+        in_data = context.data
+    else:
+        serialized_instruction_context = context.SerializeToString(deterministic=True)
+        if serialized_instruction_context is None:
+            return None
+        in_data = serialized_instruction_context
     in_ptr = (ctypes.c_uint8 * len(in_data))(*in_data)
     in_sz = len(in_data)
     out_sz = ctypes.c_uint64(OUTPUT_BUFFER_SIZE)
@@ -209,7 +210,14 @@ def process_target(
     # Process the output
     output_data = bytearray(globals.output_buffer_pointer[: out_sz.value])
     output_object = harness_ctx.effects_type()
-    output_object.ParseFromString(output_data)
+
+    if harness_ctx.raw_binary_io and len(output_data) == 1:
+        for field_desc in output_object.DESCRIPTOR.fields:
+            if field_desc.type == field_desc.TYPE_BOOL:
+                setattr(output_object, field_desc.name, output_data[0] != 0)
+                break
+    else:
+        output_object.ParseFromString(output_data)
 
     return output_object
 

src/test_suite/test_suite.py

@@ -1539,6 +1539,18 @@ def fetch_repros(client):
     if num_duplicates > 0:
         print(f"Removed {num_duplicates} duplicate(s)")
 
+    for section_name in section_names_list:
+        harness = _infer_raw_binary_harness(section_name)
+        if harness is not None:
+            converted = _convert_raw_crashes_to_contexts(globals.inputs_dir, harness)
+            if converted > 0:
+                default_harness_ctx = next(
+                    name for name, obj in HARNESS_MAP.items() if obj is harness
+                )
+                print(
+                    f"Converted {converted} raw crash file(s) to {harness.context_extension} context(s)"
+                )
+
     create_fixtures_dir = globals.output_dir / "create_fixtures"
     if create_fixtures_dir.exists():
         shutil.rmtree(create_fixtures_dir)
@@ -1587,6 +1599,54 @@ def fetch_repros(client):
     )
 
 
+def _infer_raw_binary_harness(lineage: str) -> "HarnessCtx | None":
+    """If *lineage* maps to a raw_binary_io harness, return it.
+    Returns None otherwise (including for normal protobuf harnesses)."""
+    for entrypoint, harness in ENTRYPOINT_HARNESS_MAP.items():
+        if not harness.raw_binary_io:
+            continue
+        core = (
+            entrypoint.removeprefix("sol_compat_")
+            .removesuffix("_v1")
+            .removesuffix("_v2")
+        )
+        if core and core in lineage:
+            return harness
+    return None
+
+
+def _convert_raw_crashes_to_contexts(inputs_dir: Path, harness: "HarnessCtx") -> int:
+    """Convert raw (non-protobuf) .fix files in *inputs_dir* into context
+    protobuf files that a raw_binary_io harness can consume.  Returns
+    the number of files converted."""
+    from test_suite.multiprocessing_utils import _MetadataOnlyFixture
+
+    converted = 0
+    for fix_file in list(inputs_dir.rglob(f"*{FIXTURE_EXTENSION}")):
+        with open(fix_file, "rb") as f:
+            raw = f.read()
+
+        try:
+            meta = _MetadataOnlyFixture()
+            meta.ParseFromString(raw)
+            if meta.HasField("metadata") and meta.metadata.fn_entrypoint:
+                continue
+        except Exception:
+            pass
+
+        try:
+            ctx = harness.context_type()
+            ctx.data = raw
+            ctx_path = fix_file.with_suffix(harness.context_extension)
+            with open(ctx_path, "wb") as f:
+                f.write(ctx.SerializeToString(deterministic=True))
+            fix_file.unlink()
+            converted += 1
+        except Exception as e:
+            print(f"  Warning: failed to convert {fix_file.name}: {e}")
+    return converted
+
+
 @app.command(help="Debug a single repro by hash.")
 def debug_mismatch(
     repro_hash: str = typer.Argument(
@@ -1679,6 +1739,24 @@ def debug_mismatch(
                 raise typer.Exit(code=1)
             print(f"{result}\n")
 
+        # Convert raw crash files to context protobufs if needed.
+        # Targets like gossip (raw_binary_io) use raw binary inputs, not
+        # protobuf fixtures.
+        harness_ctx_for_lineage = _infer_raw_binary_harness(lineage)
+        if harness_ctx_for_lineage is not None:
+            converted = _convert_raw_crashes_to_contexts(
+                globals.inputs_dir, harness_ctx_for_lineage
+            )
+            if converted > 0:
+                default_harness_ctx = next(
+                    name
+                    for name, obj in HARNESS_MAP.items()
+                    if obj is harness_ctx_for_lineage
+                )
+                print(
+                    f"Converted {converted} raw crash file(s) to {harness_ctx_for_lineage.context_extension} context(s)"
+                )
+
         # Deduplicate
         print("Deduplicating fixtures...")
         num_duplicates = deduplicate_fixtures_by_hash(globals.inputs_dir)

tests/test_scripts.py

@@ -123,6 +123,47 @@ def test_buf_works(self, buf_path):
         assert result.returncode == 0
 
 
+class TestRawBinaryIO:
+    """Tests for the raw_binary_io harness flag."""
+
+    def test_convert_raw_crash_to_context(self, tmp_path):
+        """Raw .fix crash files are converted to .gossipctx context files."""
+        from test_suite.test_suite import _convert_raw_crashes_to_contexts
+        from test_suite.fuzz_context import GossipHarness, FIXTURE_EXTENSION
+        import test_suite.protos.gossip_pb2 as gossip_pb
+
+        crash_data = b"\x01\x00\x00\x00" + b"\x00" * 156
+        fix_path = tmp_path / f"test_crash{FIXTURE_EXTENSION}"
+        fix_path.write_bytes(crash_data)
+
+        converted = _convert_raw_crashes_to_contexts(tmp_path, GossipHarness)
+        assert converted == 1
+        assert not fix_path.exists(), ".fix file should have been removed"
+
+        ctx_path = tmp_path / "test_crash.gossipctx"
+        assert ctx_path.exists(), ".gossipctx file should have been created"
+
+        ctx = gossip_pb.GossipMessageBinary()
+        ctx.ParseFromString(ctx_path.read_bytes())
+        assert ctx.data == crash_data
+
+    def test_convert_skips_valid_fixtures(self, tmp_path):
+        """Valid protobuf fixtures are not converted."""
+        from test_suite.test_suite import _convert_raw_crashes_to_contexts
+        from test_suite.fuzz_context import GossipHarness, FIXTURE_EXTENSION
+        import test_suite.protos.gossip_pb2 as gossip_pb
+
+        fixture = gossip_pb.GossipMessageFixture()
+        fixture.metadata.fn_entrypoint = "sol_compat_gossip_message_deserialize_v1"
+        fixture.input.data = b"\x04\x00\x00\x00"
+        fix_path = tmp_path / f"valid{FIXTURE_EXTENSION}"
+        fix_path.write_bytes(fixture.SerializeToString(deterministic=True))
+
+        converted = _convert_raw_crashes_to_contexts(tmp_path, GossipHarness)
+        assert converted == 0
+        assert fix_path.exists(), "valid .fix file should NOT have been removed"
+
+
 class TestGeneratedCode:
     """Tests for generated code (when available)."""