[feat] Expose containing module for servergen (#105)

Why
===

We have some `sed` commands that get run against generated code.
Instead, surface config options so we don't need to do that.

We also don't need to lint generated sources.

What changed
============

Added `replit_river.codegen server --module ...`
`# ruff: noqa` at the top of generated files

Test plan
=========

Manual testing resulted in the same file after replacing `sed` with
`--module ...`

replit_river/codegen/client.py

@@ -466,6 +466,7 @@ def generate_river_client_module(
     chunks: List[str] = [
         dedent(
             """\
+        # ruff: noqa
         # Code generated by river.codegen. DO NOT EDIT.
         from collections.abc import AsyncIterable, AsyncIterator
         import datetime

replit_river/codegen/run.py

@@ -13,6 +13,9 @@ def main() -> None:
     server = subparsers.add_parser(
         "server", help="Codegen a River server from gRPC protos"
     )
+    server.add_argument(
+        "--module", dest="module_name", help="output module", default="."
+    )
     server.add_argument("--output", help="output directory", required=True)
     server.add_argument("proto", help="proto file")
 
@@ -39,7 +42,7 @@ def main() -> None:
     if args.command == "server":
         proto_path = os.path.abspath(args.proto)
         target_directory = os.path.abspath(args.output)
-        proto_to_river_server_codegen(proto_path, target_directory)
+        proto_to_river_server_codegen(args.module_name, proto_path, target_directory)
     elif args.command == "server-schema":
         proto_path = os.path.abspath(args.proto)
         target_directory = os.path.abspath(args.output)

replit_river/codegen/server.py

@@ -1,6 +1,7 @@
 import collections
 import os.path
 import tempfile
+from textwrap import dedent
 from typing import DefaultDict, List, Sequence
 
 import black
@@ -220,20 +221,27 @@ def message_encoder(
 
 def generate_river_module(
     module_name: str,
+    pb_module_name: str,
     fds: descriptor_pb2.FileDescriptorSet,
 ) -> Sequence[str]:
     """Generates the lines of a River module."""
     chunks: List[str] = [
-        "# Code generated by river.codegen. DO NOT EDIT.",
-        "import datetime",
-        "from typing import Any, Dict, Mapping, Tuple",
-        "",
-        "from google.protobuf import timestamp_pb2",
-        "from google.protobuf.wrappers_pb2 import BoolValue",
+        dedent(
+            f"""\
+        # Code generated by river.codegen. DO NOT EDIT.
+        import datetime
+        from typing import Any, Dict, Mapping, Tuple
+
+        from google.protobuf import timestamp_pb2
+        from google.protobuf.wrappers_pb2 import BoolValue
+
+        import replit_river as river
+
+        from {module_name} import {pb_module_name}_pb2, {pb_module_name}_pb2_grpc
+        """
+        ),
         "",
-        "import replit_river as river",
         "",
-        f"from . import {module_name}_pb2, {module_name}_pb2_grpc\n\n",
     ]
     for pd in fds.file:
 
@@ -242,15 +250,15 @@ def _remove_namespace(name: str) -> str:
 
         # Generate the message encoders/decoders.
         for message in pd.message_type:
-            chunks.extend(message_encoder(module_name, message))
-            chunks.extend(message_decoder(module_name, message))
+            chunks.extend(message_encoder(pb_module_name, message))
+            chunks.extend(message_decoder(pb_module_name, message))
 
         # Generate the service stubs.
         for service in pd.service:
             chunks.extend(
                 [
                     f"""def add_{service.name}Servicer_to_server(
-                    servicer: {module_name}_pb2_grpc.{service.name}Servicer,
+                    servicer: {pb_module_name}_pb2_grpc.{service.name}Servicer,
                     server: river.Server,
                 ) -> None:""",
                     (
@@ -301,7 +309,11 @@ def _remove_namespace(name: str) -> str:
     return chunks
 
 
-def proto_to_river_server_codegen(proto_path: str, target_directory: str) -> None:
+def proto_to_river_server_codegen(
+    module_name: str,
+    proto_path: str,
+    target_directory: str,
+) -> None:
     fds = descriptor_pb2.FileDescriptorSet()
     with tempfile.TemporaryDirectory() as tempdir:
         descriptor_path = os.path.join(tempdir, "descriptor.pb")
@@ -317,12 +329,12 @@ def proto_to_river_server_codegen(proto_path: str, target_directory: str) -> Non
         )
         with open(descriptor_path, "rb") as f:
             fds.ParseFromString(f.read())
-    module_name = os.path.splitext(os.path.basename(proto_path))[0]
+    pb_module_name = os.path.splitext(os.path.basename(proto_path))[0]
     contents = black.format_str(
-        "\n".join(generate_river_module(module_name, fds)),
+        "\n".join(generate_river_module(module_name, pb_module_name, fds)),
         mode=black.FileMode(string_normalization=False),
     )
     os.makedirs(target_directory, exist_ok=True)
-    output_path = f"{target_directory}/{module_name}_river.py"
+    output_path = f"{target_directory}/{pb_module_name}_river.py"
     with open(output_path, "w") as f:
         f.write(contents)