Merge pull request #13056 from github/redsun82/swift-mark-generated-files

Codegen: mark generated checked in files as such

Author: redsun82

misc/codegen/codegen.py

@@ -58,7 +58,9 @@ def _parse_args() -> argparse.Namespace:
                        help="output directory for generated C++ files, required if trap or cpp is provided to "
                             "--generate"),
         p.add_argument("--generated-registry",
-                       help="registry file containing information about checked-in generated code"),
+                       help="registry file containing information about checked-in generated code. A .gitattributes"
+                            "file is generated besides it to mark those files with linguist-generated=true. Must"
+                            "be in a directory containing all generated code."),
     ]
     p.add_argument("--script-name",
                    help="script name to put in header comments of generated files. By default, the path of this "
@@ -108,7 +110,7 @@ def run():
         log_level = logging.INFO
     logging.basicConfig(format="{levelname} {message}", style='{', level=log_level)
     for target in opts.generate:
-        generate(target, opts, render.Renderer(opts.script_name, opts.root_dir))
+        generate(target, opts, render.Renderer(opts.script_name))
 
 
 if __name__ == "__main__":

misc/codegen/lib/render.py

@@ -25,14 +25,10 @@ class Error(Exception):
 class Renderer:
     """ Template renderer using mustache templates in the `templates` directory """
 
-    def __init__(self, generator: pathlib.Path, root_dir: pathlib.Path):
+    def __init__(self, generator: pathlib.Path):
         self._r = pystache.Renderer(search_dirs=str(paths.templates_dir), escape=lambda u: u)
-        self._root_dir = root_dir
         self._generator = generator
 
-    def _get_path(self, file: pathlib.Path):
-        return file.relative_to(self._root_dir)
-
     def render(self, data: object, output: pathlib.Path):
         """ Render `data` to `output`.
 
@@ -60,7 +56,7 @@ def _do_write(self, mnemonic: str, contents: str, output: pathlib.Path):
 
     def manage(self, generated: typing.Iterable[pathlib.Path], stubs: typing.Iterable[pathlib.Path],
                registry: pathlib.Path, force: bool = False) -> "RenderManager":
-        return RenderManager(self._generator, self._root_dir, generated, stubs, registry, force)
+        return RenderManager(self._generator, generated, stubs, registry, force)
 
 
 class RenderManager(Renderer):
@@ -85,10 +81,10 @@ class Hashes:
         pre: str
         post: typing.Optional[str] = None
 
-    def __init__(self, generator: pathlib.Path, root_dir: pathlib.Path, generated: typing.Iterable[pathlib.Path],
+    def __init__(self, generator: pathlib.Path, generated: typing.Iterable[pathlib.Path],
                  stubs: typing.Iterable[pathlib.Path],
                  registry: pathlib.Path, force: bool = False):
-        super().__init__(generator, root_dir)
+        super().__init__(generator)
         self._registry_path = registry
         self._force = force
         self._hashes = {}
@@ -117,10 +113,13 @@ def __exit__(self, exc_type, exc_val, exc_tb):
                 self._hashes.pop(self._get_path(f), None)
         # clean up the registry from files that do not exist any more
         for f in list(self._hashes):
-            if not (self._root_dir / f).exists():
+            if not (self._registry_path.parent / f).exists():
                 self._hashes.pop(f)
         self._dump_registry()
 
+    def _get_path(self, file: pathlib.Path):
+        return file.relative_to(self._registry_path.parent)
+
     def _do_write(self, mnemonic: str, contents: str, output: pathlib.Path):
         hash = self._hash_string(contents)
         rel_output = self._get_path(output)
@@ -186,13 +185,17 @@ def _load_registry(self):
         try:
             with open(self._registry_path) as reg:
                 for line in reg:
-                    filename, prehash, posthash = line.split()
-                    self._hashes[pathlib.Path(filename)] = self.Hashes(prehash, posthash)
+                    if line.strip():
+                        filename, prehash, posthash = line.split()
+                        self._hashes[pathlib.Path(filename)] = self.Hashes(prehash, posthash)
         except FileNotFoundError:
             pass
 
     def _dump_registry(self):
         self._registry_path.parent.mkdir(parents=True, exist_ok=True)
-        with open(self._registry_path, 'w') as out:
+        with open(self._registry_path, 'w') as out, open(self._registry_path.parent / ".gitattributes", "w") as attrs:
+            print(f"/{self._registry_path.name}", "linguist-generated", file=attrs)
+            print("/.gitattributes", "linguist-generated", file=attrs)
             for f, hashes in sorted(self._hashes.items()):
                 print(f, hashes.pre, hashes.post, file=out)
+                print(f"/{f}", "linguist-generated", file=attrs)

misc/codegen/test/test_qlgen.py

@@ -26,7 +26,7 @@ def ql_output_path(): return paths.root_dir / "ql/lib/other/path"
 def ql_test_output_path(): return paths.root_dir / "ql/test/path"
 
 
-def generated_registry_path(): return paths.root_dir / "registry.list"
+def generated_registry_path(): return paths.root_dir / "ql/registry.list"
 
 
 def import_file(): return stub_path().with_suffix(".qll")