Add black pre-commit hook

.pre-commit-config.yaml

@@ -14,11 +14,11 @@ repos:
     hooks:
       - id: clang-format
 
-  - repo: https://github.com/pre-commit/mirrors-autopep8
-    rev: v2.0.4
+  - repo: https://github.com/psf/black
+    rev: 25.1.0
     hooks:
-      - id: autopep8
-        files: ^misc/codegen/.*\.py
+      - id: black
+        files: ^(misc/codegen/.*|misc/scripts/models-as-data/bulk_generate_mad)\.py$
 
   - repo: local
     hooks:

misc/codegen/codegen.py

@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-""" Driver script to run all code generation """
+"""Driver script to run all code generation"""
 
 import argparse
 import logging
@@ -9,7 +9,7 @@
 import typing
 import shlex
 
-if 'BUILD_WORKSPACE_DIRECTORY' not in os.environ:
+if "BUILD_WORKSPACE_DIRECTORY" not in os.environ:
     # we are not running with `bazel run`, set up module search path
     _repo_root = pathlib.Path(__file__).resolve().parents[2]
     sys.path.append(str(_repo_root))
@@ -29,57 +29,105 @@ def _parse_args() -> argparse.Namespace:
         conf = None
 
     p = argparse.ArgumentParser(description="Code generation suite")
-    p.add_argument("--generate", type=lambda x: x.split(","),
-                   help="specify what targets to generate as a comma separated list, choosing among dbscheme, ql, "
-                        "trap, cpp and rust")
-    p.add_argument("--verbose", "-v", action="store_true", help="print more information")
+    p.add_argument(
+        "--generate",
+        type=lambda x: x.split(","),
+        help="specify what targets to generate as a comma separated list, choosing among dbscheme, ql, "
+        "trap, cpp and rust",
+    )
+    p.add_argument(
+        "--verbose", "-v", action="store_true", help="print more information"
+    )
     p.add_argument("--quiet", "-q", action="store_true", help="only print errors")
-    p.add_argument("--configuration-file", "-c", type=_abspath, default=conf,
-                   help="A configuration file to load options from. By default, the first codegen.conf file found by "
-                        "going up directories from the current location. If present all paths provided in options are "
-                        "considered relative to its directory")
-    p.add_argument("--root-dir", type=_abspath,
-                   help="the directory that should be regarded as the root of the language pack codebase. Used to "
-                        "compute QL imports and in some comments and as root for relative paths provided as options. "
-                        "If not provided it defaults to the directory of the configuration file, if any")
+    p.add_argument(
+        "--configuration-file",
+        "-c",
+        type=_abspath,
+        default=conf,
+        help="A configuration file to load options from. By default, the first codegen.conf file found by "
+        "going up directories from the current location. If present all paths provided in options are "
+        "considered relative to its directory",
+    )
+    p.add_argument(
+        "--root-dir",
+        type=_abspath,
+        help="the directory that should be regarded as the root of the language pack codebase. Used to "
+        "compute QL imports and in some comments and as root for relative paths provided as options. "
+        "If not provided it defaults to the directory of the configuration file, if any",
+    )
     path_arguments = [
-        p.add_argument("--schema",
-                       help="input schema file (default schema.py)"),
-        p.add_argument("--dbscheme",
-                       help="output file for dbscheme generation, input file for trap generation"),
-        p.add_argument("--ql-output",
-                       help="output directory for generated QL files"),
-        p.add_argument("--ql-stub-output",
-                       help="output directory for QL stub/customization files. Defines also the "
-                            "generated qll file importing every class file"),
-        p.add_argument("--ql-test-output",
-                       help="output directory for QL generated extractor test files"),
-        p.add_argument("--ql-cfg-output",
-                       help="output directory for QL CFG layer (optional)."),
-        p.add_argument("--cpp-output",
-                       help="output directory for generated C++ files, required if trap or cpp is provided to "
-                            "--generate"),
-        p.add_argument("--rust-output",
-                       help="output directory for generated Rust files, required if rust is provided to "
-                            "--generate"),
-        p.add_argument("--generated-registry",
-                       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("--schema", help="input schema file (default schema.py)"),
+        p.add_argument(
+            "--dbscheme",
+            help="output file for dbscheme generation, input file for trap generation",
+        ),
+        p.add_argument("--ql-output", help="output directory for generated QL files"),
+        p.add_argument(
+            "--ql-stub-output",
+            help="output directory for QL stub/customization files. Defines also the "
+            "generated qll file importing every class file",
+        ),
+        p.add_argument(
+            "--ql-test-output",
+            help="output directory for QL generated extractor test files",
+        ),
+        p.add_argument(
+            "--ql-cfg-output", help="output directory for QL CFG layer (optional)."
+        ),
+        p.add_argument(
+            "--cpp-output",
+            help="output directory for generated C++ files, required if trap or cpp is provided to "
+            "--generate",
+        ),
+        p.add_argument(
+            "--rust-output",
+            help="output directory for generated Rust files, required if rust is provided to "
+            "--generate",
+        ),
+        p.add_argument(
+            "--generated-registry",
+            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 "
-                        "script relative to the root directory")
-    p.add_argument("--trap-library",
-                   help="path to the trap library from an include directory, required if generating C++ trap bindings"),
-    p.add_argument("--ql-format", action="store_true", default=True,
-                   help="use codeql to autoformat QL files (which is the default)")
-    p.add_argument("--no-ql-format", action="store_false", dest="ql_format", help="do not format QL files")
-    p.add_argument("--codeql-binary", default="codeql", help="command to use for QL formatting (default %(default)s)")
-    p.add_argument("--force", "-f", action="store_true",
-                   help="generate all files without skipping unchanged files and overwriting modified ones")
-    p.add_argument("--use-current-directory", action="store_true",
-                   help="do not consider paths as relative to --root-dir or the configuration directory")
+    p.add_argument(
+        "--script-name",
+        help="script name to put in header comments of generated files. By default, the path of this "
+        "script relative to the root directory",
+    )
+    p.add_argument(
+        "--trap-library",
+        help="path to the trap library from an include directory, required if generating C++ trap bindings",
+    ),
+    p.add_argument(
+        "--ql-format",
+        action="store_true",
+        default=True,
+        help="use codeql to autoformat QL files (which is the default)",
+    )
+    p.add_argument(
+        "--no-ql-format",
+        action="store_false",
+        dest="ql_format",
+        help="do not format QL files",
+    )
+    p.add_argument(
+        "--codeql-binary",
+        default="codeql",
+        help="command to use for QL formatting (default %(default)s)",
+    )
+    p.add_argument(
+        "--force",
+        "-f",
+        action="store_true",
+        help="generate all files without skipping unchanged files and overwriting modified ones",
+    )
+    p.add_argument(
+        "--use-current-directory",
+        action="store_true",
+        help="do not consider paths as relative to --root-dir or the configuration directory",
+    )
     opts = p.parse_args()
     if opts.configuration_file is not None:
         with open(opts.configuration_file) as config:
@@ -97,7 +145,15 @@ def _parse_args() -> argparse.Namespace:
     for arg in path_arguments:
         path = getattr(opts, arg.dest)
         if path is not None:
-            setattr(opts, arg.dest, _abspath(path) if opts.use_current_directory else (opts.root_dir / path))
+            setattr(
+                opts,
+                arg.dest,
+                (
+                    _abspath(path)
+                    if opts.use_current_directory
+                    else (opts.root_dir / path)
+                ),
+            )
     if not opts.script_name:
         opts.script_name = paths.exe_file.relative_to(opts.root_dir)
     return opts
@@ -115,7 +171,7 @@ def run():
         log_level = logging.ERROR
     else:
         log_level = logging.INFO
-    logging.basicConfig(format="{levelname} {message}", style='{', level=log_level)
+    logging.basicConfig(format="{levelname} {message}", style="{", level=log_level)
     for target in opts.generate:
         generate(target, opts, render.Renderer(opts.script_name))
 

misc/codegen/generators/cppgen.py

@@ -49,7 +49,11 @@ def _get_trap_name(cls: schema.Class, p: schema.Property) -> str | None:
     return inflection.pluralize(trap_name)
 
 
-def _get_field(cls: schema.Class, p: schema.Property, add_or_none_except: typing.Optional[str] = None) -> cpp.Field:
+def _get_field(
+    cls: schema.Class,
+    p: schema.Property,
+    add_or_none_except: typing.Optional[str] = None,
+) -> cpp.Field:
     args = dict(
         field_name=p.name + ("_" if p.name in cpp.cpp_keywords else ""),
         base_type=_get_type(p.type, add_or_none_except),
@@ -83,14 +87,15 @@ def _get_class(self, name: str) -> cpp.Class:
             bases=[self._get_class(b) for b in cls.bases],
             fields=[
                 _get_field(cls, p, self._add_or_none_except)
-                for p in cls.properties if "cpp_skip" not in p.pragmas and not p.synth
+                for p in cls.properties
+                if "cpp_skip" not in p.pragmas and not p.synth
             ],
             final=not cls.derived,
             trap_name=trap_name,
         )
 
     def get_classes(self):
-        ret = {'': []}
+        ret = {"": []}
         for k, cls in self._classmap.items():
             if not cls.synth:
                 ret.setdefault(cls.group, []).append(self._get_class(cls.name))
@@ -102,6 +107,12 @@ def generate(opts, renderer):
     processor = Processor(schemaloader.load_file(opts.schema))
     out = opts.cpp_output
     for dir, classes in processor.get_classes().items():
-        renderer.render(cpp.ClassList(classes, opts.schema,
-                                      include_parent=bool(dir),
-                                      trap_library=opts.trap_library), out / dir / "TrapClasses")
+        renderer.render(
+            cpp.ClassList(
+                classes,
+                opts.schema,
+                include_parent=bool(dir),
+                trap_library=opts.trap_library,
+            ),
+            out / dir / "TrapClasses",
+        )