rutile-game/tools/shuriken.py

#!/usr/bin/env python3

############################################
# shuriken - A ninja-based C++ build tool
# ------------------------------------------
# Version: 0.1.0 (modified)
# Author:  binaryDiv
# License: MIT License
# https://git.0xbd.space/binaryDiv/shuriken
############################################

import argparse
import json
import shlex
import subprocess
import sys
from collections import namedtuple
from dataclasses import asdict, dataclass, field
from pathlib import Path
from typing import Any, Self

import yaml


class ConfigValidationException(Exception):
    pass


@dataclass(kw_only=True)
class TargetConfig:
    cpp_compiler: str | None = None
    cpp_standard: str | None = None
    cpp_flags: str | None = None
    cpp_flags_extra: str | None = None

    linker_flags: str | None = None
    linker_flags_extra: str | None = None
    linker_args: str | None = None
    linker_args_extra: str | None = None

    output_file: str | None = None
    run_command: str | None = None

    def update(self, data: dict) -> None:
        for key, value in data.items():
            if hasattr(self, key) is True and value is not None:
                setattr(self, key, str(value))

    @property
    def merged_cpp_flags(self) -> str:
        return f"{self.cpp_flags or ''} {self.cpp_flags_extra or ''}".strip()

    @property
    def merged_linker_flags(self) -> str:
        return f"{self.linker_flags or ''} {self.linker_flags_extra or ''}".strip()

    @property
    def merged_linker_args(self) -> str:
        return f"{self.linker_args or ''} {self.linker_args_extra or ''}".strip()


@dataclass(kw_only=True)
class MergedTargetConfig(TargetConfig):
    cpp_compiler: str = ''
    cpp_standard: str = ''
    cpp_flags: str = ''
    cpp_flags_extra: str = ''

    linker_flags: str = ''
    linker_flags_extra: str = ''
    linker_args: str = ''
    linker_args_extra: str = ''

    output_file: str = ''
    run_command: str = ''

    @classmethod
    def create_from(cls, target_defaults: TargetConfig, target_config: TargetConfig) -> Self:
        merged_config = cls()
        merged_config.update(asdict(target_defaults))
        merged_config.update(asdict(target_config))
        return merged_config

    def validate(self) -> None:
        # Validate that all required fields are set
        if not self.cpp_compiler:
            raise ConfigValidationException('cpp_compiler must be set')
        if not self.cpp_standard:
            raise ConfigValidationException('cpp_standard must be set')
        if not self.output_file:
            raise ConfigValidationException('output_file must be set')


@dataclass(kw_only=True)
class ShurikenConfig:
    __path_fields = ['source_dir', 'build_dir']
    __list_fields = []
    __bool_fields = ['enable_import_std']

    source_dir: Path = Path('src')
    build_dir: Path = Path('build')

    defaults: TargetConfig = field(default_factory=lambda: TargetConfig(
        cpp_compiler='clang++',
        cpp_standard='c++20',
    ))

    enable_import_std: bool = False

    default_target: str = ''
    targets: dict[str, TargetConfig] = field(default_factory=dict)

    # Merged target configs, generated from defaults and targets
    _merged_target_configs = {}

    def update(self, data: dict) -> None:
        for key, value in data.items():
            # Ignore unknown config keys
            if not hasattr(self, key):
                continue

            if key == 'defaults':
                assert type(value) is dict
                self.defaults.update(value)
            elif key == 'targets':
                assert type(value) is dict
                for target_name, target_config in value.items():
                    assert type(target_name) is str and len(target_name) > 0
                    assert type(target_config) is dict or target_config is None

                    # Ignore "hidden" targets starting with a dot (can be used for YAML anchors)
                    if target_name[0] == '.':
                        continue

                    if target_name not in self.targets:
                        self.targets[target_name] = TargetConfig()
                    if target_config is not None:
                        self.targets[target_name].update(target_config)
            elif key in self.__path_fields:
                setattr(self, key, Path(value))
            elif key in self.__list_fields:
                assert type(value) is list
                setattr(self, key, list(str(item) for item in value))
            elif key in self.__bool_fields:
                assert type(value) is bool
                setattr(self, key, value)
            else:
                setattr(self, key, str(value))

        # Generate merged target configs (we generate all of them so that we can validate the full config)
        for target_name, target_config in self.targets.items():
            self._merged_target_configs[target_name] = MergedTargetConfig.create_from(self.defaults, target_config)

    def update_from_yaml(self, file_path: Path) -> None:
        with file_path.open('r') as file:
            self.update(yaml.safe_load(file))

    def validate(self) -> None:
        # Validate root elements
        if not self.default_target:
            raise ConfigValidationException('default_target must be set')
        if not self.source_dir or not self.source_dir.is_dir():
            raise ConfigValidationException('source_dir must be set to an existing directory')
        if not self.build_dir:
            # build_dir will be automatically created if it doesn't exist
            raise ConfigValidationException('build_dir must be set')

        # Validate that targets and default target are defined
        if not self.targets:
            raise ConfigValidationException('At least one target must be defined')
        if self.default_target not in self.targets:
            raise ConfigValidationException('default_target is not defined in targets')

        # Validate individual build targets
        for target_name, merged_target_config in self._merged_target_configs.items():
            try:
                merged_target_config.validate()
            except ConfigValidationException as exc:
                raise ConfigValidationException(f'Target "{target_name}": {str(exc)}')

    def get_ninja_file_path(self, target: str) -> Path:
        return self.build_dir / f'build.{target}.ninja'

    def get_merged_target_config(self, target: str) -> MergedTargetConfig:
        assert target in self.targets, f'Target "{target}" not defined!'
        return self._merged_target_configs[target]


class ShurikenArgumentParser:
    argument_parser: argparse.ArgumentParser

    default_config_path: Path = Path('shuriken.yaml')
    default_config_override_path: Path = Path('shuriken.override.yaml')

    def __init__(self):
        self.argument_parser = argparse.ArgumentParser()
        self.argument_parser.add_argument(
            '-c', '--config',
            type=Path,
            help=f'Config file (default: {self.default_config_path})',
            metavar='FILE',
        )
        self.argument_parser.add_argument(
            '-o', '--config-override',
            type=Path,
            help=f'Config file to override config values (default: {self.default_config_override_path})',
            metavar='FILE',
        )
        self.argument_parser.add_argument(
            '-t', '--target',
            help=f'Build target to use (default: see "default_target" in config)',
        )

        # Define subcommands
        subparsers = self.argument_parser.add_subparsers(dest='command', title='Subcommands')

        subparser_generate = subparsers.add_parser('generate', help='Generate Ninja build files')
        subparser_generate.add_argument(
            '--stdout',
            action='store_true',
            dest='generate_stdout',
            help='Print generated Ninja file to stdout instead of writing to a file',
        )

        _subparser_build = subparsers.add_parser('build', help='Build project (default command)')
        _subparser_compdb = subparsers.add_parser('compdb', help='Generate compilation database')

        subparser_run = subparsers.add_parser('run', help='Build and run project')
        subparser_run.add_argument(
            'run_args',
            nargs='*',
            help='Arguments that are passed to the application',
        )

        subparser_lint = subparsers.add_parser('lint', help='Lint the project')
        subparser_lint.add_argument(
            'lint_files',
            nargs='*',
            help='Files to lint (defaults to all source files)',
        )

        subparser_clean = subparsers.add_parser('clean', help='Remove all build files of the current target')
        subparser_clean.add_argument(
            '--all',
            action='store_true',
            dest='clean_all',
            help='Remove all generated files from all targets',
        )

        _subparser_dump_config = subparsers.add_parser(
            'dump-config',
            help='Dumps the parsed config as well as the effective build target config (for debugging)',
        )
        _subparser_dyndep = subparsers.add_parser(
            'p1689-to-dyndeps',
            help='(Internal command) Parse p1689 file and generate Ninja dyndeps',
        )

    def parse_args(self, args: list[str] | None = None) -> argparse.Namespace:
        parsed_args = self.argument_parser.parse_args(args)
        parsed_args.config = self.ensure_valid_path_or_default(
            parsed_args.config,
            self.default_config_path,
            required=True,
        )
        parsed_args.config_override = self.ensure_valid_path_or_default(
            parsed_args.config_override,
            self.default_config_override_path,
        )
        return parsed_args

    def error(self, message: str) -> None:
        self.argument_parser.error(message)

    def ensure_valid_path_or_default(self, path: Path | None, default: Path, *, required: bool = False) -> Path | None:
        if path is None:
            if required or default.exists():
                path = default
        if path is not None and not path.exists():
            self.error(f'File not found: {path}')
        return path


class ShurikenCli:
    args: argparse.Namespace
    config: ShurikenConfig

    def __init__(self, input_args: list[str] | None = None):
        # Parse command line arguments
        argument_parser = ShurikenArgumentParser()
        self.args = argument_parser.parse_args(input_args)

        # Parse config files
        self.config = ShurikenConfig()
        if self.args.config:
            self.config.update_from_yaml(self.args.config)
        if self.args.config_override:
            self.config.update_from_yaml(self.args.config_override)

        # Validate parsed config
        try:
            self.config.validate()
        except ConfigValidationException as exc:
            self.log_error(f'Config validation error: {exc}')
            exit(1)

        # Check that --target is a valid target
        if self.args.target is not None and self.args.target not in self.config.targets:
            argument_parser.error(f'Target "{self.args.target}" not defined in config')

    def get_selected_target(self) -> str:
        return self.args.target if self.args.target is not None else self.config.default_target

    @staticmethod
    def log_info(message: str) -> None:
        print(f'shuriken: {message}')

    @staticmethod
    def log_error(message: str) -> None:
        print(f'shuriken: Error: {message}')

    def run(self) -> None:
        # Get build target (specified with --target or default_target from config)
        selected_target = self.get_selected_target()

        # Default command: build
        command = self.args.command or 'build'

        match command:
            case 'generate':
                self.generate_ninja_file(selected_target, write_to_stdout=self.args.generate_stdout)

            case 'compdb':
                # Always generate compilation database for default target
                # (avoids issues with clang-scan-deps in different environments like emscripten)
                self.generate_compdb(self.config.default_target)

            case 'build':
                self.build_project(selected_target)

            case 'run':
                self.run_project(selected_target)

            case 'lint':
                self.lint_project(selected_target)

            case 'clean':
                if self.args.clean_all:
                    self.log_info('Cleaning up *all* build targets')
                    for target in self.config.targets:
                        self.clean_project(target)
                else:
                    self.clean_project(selected_target)

            case 'dump-config':
                self.dump_config()

            case 'p1689-to-dyndeps':
                self.generate_dyndeps_from_p1689(selected_target)

            case _:
                raise Exception(f'Unknown subcommand "{self.args.command}"')

    def call_ninja(self, target: str, *args: str) -> None:
        run_result = subprocess.run([
            'ninja',
            '-f',
            self.config.get_ninja_file_path(target),
            '-v',
            *args,
        ])

        if run_result.returncode > 0:
            self.log_error(f'Ninja exited with return code {run_result.returncode}')
            exit(1)

    def generate_ninja_file(self, target: str, *, write_to_stdout: bool = False) -> None:
        generator = NinjaFileGenerator(config=self.config, target=target)
        ninja_file_content = generator.generate_file()

        if write_to_stdout:
            print(ninja_file_content)
            return

        # Create build directory for target if it doesn't exist yet
        self.config.build_dir.joinpath(target).mkdir(parents=True, exist_ok=True)

        ninja_file_path = self.config.get_ninja_file_path(target)
        ninja_file_content += '\n'

        # Check if file content has changed before writing it (to preserve the modified timestamp if unmodified)
        if ninja_file_path.exists() and ninja_file_path.read_text() == ninja_file_content:
            self.log_info(f'Ninja file for target {target} is up to date')
        else:
            # Write Ninja file
            self.log_info(f'Generating Ninja file for target {target}')
            with ninja_file_path.open('w') as ninja_file:
                ninja_file.write(ninja_file_content)

        # Create symlink build.ninja to default target ninja file for convenience
        if target == self.config.default_target:
            self.symlink_build_ninja_file(ninja_file_path)

    def symlink_build_ninja_file(self, symlink_target: Path) -> None:
        symlink_path = Path('build.ninja')

        if symlink_path.exists() and not symlink_path.is_symlink():
            self.log_error('Cannot symlink build.ninja: File exists and is not a symlink')
            return

        if symlink_path.is_symlink() and symlink_path.resolve() == symlink_target.absolute():
            # Symlink already correct
            return

        # Remove existing symlink if it exists and create a new symlink
        self.log_info(f'Updating build.ninja symlink to {symlink_target}')
        symlink_path.unlink(missing_ok=True)
        symlink_path.symlink_to(symlink_target)

    def generate_compdb(self, target: str) -> None:
        # Regenerate Ninja file if necessary
        self.generate_ninja_file(target)

        self.log_info('Generating compilation database')
        self.call_ninja(target, 'compdb')

    def build_project(self, target: str) -> None:
        # Regenerate Ninja file if necessary
        self.generate_ninja_file(target)

        self.log_info(f'Building project for target {target}')
        self.call_ninja(target)

    def run_project(self, target: str) -> None:
        # Build project using Ninja if necessary
        self.build_project(target)

        merged_target_config = self.config.get_merged_target_config(target)
        output_file_path = self.config.build_dir / target / merged_target_config.output_file

        # Get run command from config or default to running the output file
        run_command = merged_target_config.run_command or './{out} {args}'

        # Replace placeholders in run command
        run_command = run_command.format(
            out=output_file_path,
            args=shlex.join(self.args.run_args),
        )

        try:
            # Run command
            self.log_info(f'Running command for target {target}:')
            self.log_info(f'  {run_command}')
            run_result = subprocess.run(run_command, shell=True)

            if run_result.returncode > 0:
                self.log_error(f'Run command exited with return code {run_result.returncode}')
                exit(1)
        except KeyboardInterrupt:
            self.log_info('Keyboard interrupt')

    def lint_project(self, target: str) -> None:
        # Build project using Ninja (necessary before linting)
        self.build_project(target)

        # Collect files to lint
        if self.args.lint_files:
            files_args = shlex.join(self.args.lint_files)
        else:
            files_args = f"$(find {self.config.source_dir} -name '*.cpp' -or -name '*.cppm')"

        self.log_info('Running clangd-tidy')
        run_result = subprocess.run(
            f'clangd-tidy -p {self.config.build_dir} --allow-extensions cpp,cppm {files_args}',
            shell=True,
        )

        if run_result.returncode > 0:
            self.log_error(f'clangd-tidy exited with return code {run_result.returncode}')
            exit(1)

    def clean_project(self, target: str) -> None:
        ninja_file_path = self.config.get_ninja_file_path(target)

        if not ninja_file_path.exists():
            self.log_info(f'Ninja file {ninja_file_path} not found, nothing to clean up')
            return

        self.log_info(f'Cleaning up build files for target {target}')
        self.call_ninja(target, '-t', 'clean')

        self.log_info(f'Removing Ninja file {ninja_file_path}')
        ninja_file_path.unlink(missing_ok=True)

        # Remove build.ninja symlink if it points to the same Ninja file
        ninja_symlink = Path('build.ninja')
        if ninja_symlink.is_symlink() and ninja_symlink.resolve() == ninja_file_path.absolute():
            self.log_info(f'Removing build.ninja symlink')
            ninja_symlink.unlink()

    def dump_config(self) -> None:
        target = self.get_selected_target()

        self.log_info('Dumping fully parsed config:')
        print(json.dumps(asdict(self.config), indent=4, default=str))

        self.log_info(f'Dumping effective build target config for target "{target}":')
        merged_target_config = self.config.get_merged_target_config(target)
        print(json.dumps(asdict(merged_target_config), indent=4, default=str))

    def generate_dyndeps_from_p1689(self, target: str) -> None:
        # Parse P1689 build dependency JSON file from stdin (generated by clang-scan-deps)
        parsed_p1689 = json.load(sys.stdin)

        # Parse rules and get module dependencies
        build_dependency_parser = BuildDependencyParser(config=self.config, target=target)
        build_dependencies = build_dependency_parser.parse_build_dependencies(parsed_p1689)

        # Generate Ninja dyndep file and write to stdout
        print('ninja_dyndep_version = 1\n')
        for out, deps in build_dependencies.items():
            if deps:
                print(f'build {out}: dyndep | {" ".join(str(dep) for dep in deps)}')
            else:
                print(f'build {out}: dyndep')


CompileTargetCpp = namedtuple('CompileTargetCpp', ['src', 'obj'])
CompileTargetCppm = namedtuple('CompileTargetCppm', ['src', 'obj', 'pcm'])
CompileTargetStdModule = namedtuple('CompileTargetStdModule', ['original_src', 'generated_src', 'pcm'])


class NinjaFileGenerator:
    config: ShurikenConfig
    target: str
    target_config: TargetConfig

    compile_targets_cpp: list[CompileTargetCpp]
    compile_targets_cppm: list[CompileTargetCppm]
    compile_targets_std_module: list[CompileTargetStdModule]
    link_target_obj_files: list[str]

    def __init__(self, *, config: ShurikenConfig, target: str):
        self.config = config
        self.target = target
        self.target_config = config.get_merged_target_config(target)

    def generate_file(self) -> str:
        self.collect_compile_targets()
        return self.render_file()

    def find_source_files(self, glob: str) -> list[Path]:
        return sorted(self.config.source_dir.glob(glob))

    def collect_compile_targets(self) -> None:
        self.compile_targets_cpp = []
        self.compile_targets_cppm = []
        self.compile_targets_std_module = []

        # Compile targets for .cpp files
        for src_file in self.find_source_files('**/*.cpp'):
            relative_name = src_file.relative_to(self.config.source_dir)
            self.compile_targets_cpp.append(CompileTargetCpp(
                src=str(src_file),
                obj=f"$builddir/obj/{relative_name.with_suffix('.o')}",
            ))

        # Compile targets for .cppm files
        for src_file in self.find_source_files('**/*.cppm'):
            relative_name = src_file.relative_to(self.config.source_dir)
            module_name = '.'.join(relative_name.with_suffix('.pcm').parts)
            self.compile_targets_cppm.append(CompileTargetCppm(
                src=str(src_file),
                obj=f"$builddir/obj/{relative_name.with_suffix('.o')}",
                pcm=f"$builddir/pcm/{module_name}",
            ))

        # Compile targets for standard library modules ("import std")
        if self.config.enable_import_std:
            # NOTE: Add std.compat to the list (here and a few other places) if we ever need it. For now, std is enough.
            std_module_source_finder = StdModuleSourceFinder(config=self.config, target=self.target)
            std_module_sources = std_module_source_finder.find_module_sources(['std'])
            for module_name, module_src_path in std_module_sources.items():
                self.compile_targets_std_module.append(CompileTargetStdModule(
                    original_src=str(module_src_path),
                    generated_src=f"$builddir/generated/stdlib/{module_name}.cppm",
                    pcm=f"$builddir/pcm/{module_name}.pcm",
                ))

        # Link target for output file (executable): gather all object files
        self.link_target_obj_files = sorted(
            target.obj for target in self.compile_targets_cpp + self.compile_targets_cppm
        )

    def render_file(self) -> str:
        # Get script path to run shuriken
        shuriken_script_path = sys.argv[0]

        # Shortcut variables
        config = self.config
        target = self.target
        target_config = self.target_config
        build_dir = config.build_dir
        ninja_file_path = config.get_ninja_file_path(config.default_target)
        cpp_compiler = target_config.cpp_compiler

        # Generate list of all source input files to be used as implicit dependencies for the dyndep file
        dyndep_file_implicit_inputs = ' '.join(sorted(
            t.src for t in self.compile_targets_cpp + self.compile_targets_cppm
        ))

        # Generate list of all .dep files that are generated implicitly during dyndep file generation.
        # This is mostly necessary so that Ninja can create the directories for these files beforehand.
        dyndep_file_implicit_outputs = ' '.join(sorted(
            f'{t.obj}.dep' for t in self.compile_targets_cpp + self.compile_targets_cppm
        ))

        # Generate list of generated source files (currently just the copied std module file)
        generated_source_files = ' '.join(sorted(
            t.generated_src for t in self.compile_targets_std_module
        ))

        # Generate Ninja file
        return f"""
ninja_required_version = 1.10

# Build directory and paths
builddir = {build_dir / target}
compdb_file = {build_dir}/compile_commands.json
dyndep_file = $builddir/.ninja_dyndep

# Compiler flags
cpp_base_flags = -std={target_config.cpp_standard} -fprebuilt-module-path=$builddir/pcm/
cpp_flags = $cpp_base_flags {target_config.merged_cpp_flags}
cpp_std_module_flags = $cpp_base_flags -Wno-reserved-module-identifier -x c++-module
linker_flags = {target_config.merged_linker_flags}
linker_args = {target_config.merged_linker_args}

# -- RULES

# Rule to copy an arbitrary file (or multiple) from in to out
rule copy
    command = cp $in $out

# Rule to compile a .cpp file to a .o file
rule cpp
    depfile = $out.dep
    command = {cpp_compiler} $cpp_flags -MMD -MF $out.dep -c $in -o $out

# Rule to compile a .cppm file (C++20 module) to a .o file ($out) and a .pcm file ($pcm_out)
rule cppm
    depfile = $out.dep
    command = {cpp_compiler} $cpp_flags -MMD -MF $out.dep -c $in -o $out -fmodule-output=$pcm_out

# Rule to compile a standard library module to a .pcm file (e.g. "import std")
rule std_module
    command = {cpp_compiler} $cpp_std_module_flags --precompile $in -o $out

# Rule to link several .o files to an executable
rule link
    command = {cpp_compiler} $linker_flags -o $out $in $linker_args

# Rule to generate a compilation database (JSON file) from the build targets defined in a Ninja file
rule generate_compdb
    command = ninja -f $in -t compdb cpp cppm std_module > $out

# Rule to generate a Ninja dyndep file for C++ module dependencies based on a compilation database
rule generate_dyndeps
    command = clang-scan-deps -format=p1689 -compilation-database=$in | {shuriken_script_path} -t {target} p1689-to-dyndeps > $out


# -- STATIC BUILD TARGETS

# Generate compilation database from default target Ninja file
build $compdb_file: generate_compdb {ninja_file_path} | generate_source_files

# Shortcut alias to generate compilation database
build compdb: phony $compdb_file

# Generate Ninja dyndep file from compilation database
build $dyndep_file | {dyndep_file_implicit_outputs}: generate_dyndeps $compdb_file | {dyndep_file_implicit_inputs}


# -- GENERATED BUILD TARGETS

# Link output file (default target)
build $builddir/{target_config.output_file}: link {' '.join(self.link_target_obj_files)}
default $builddir/{target_config.output_file}

# Phony target that depends on all generated source files
build generate_source_files: phony {generated_source_files}

{self.render_compile_targets()}

# End of build.{target}.ninja
""".strip()

    def render_compile_targets(self) -> str:
        rendered_targets = list(map(self.render_compile_target_std_module, self.compile_targets_std_module))
        rendered_targets += list(map(self.render_compile_target_cpp, self.compile_targets_cpp))
        rendered_targets += list(map(self.render_compile_target_cppm, self.compile_targets_cppm))
        return '\n\n'.join(rendered_targets)

    @staticmethod
    def render_compile_target_cpp(target: CompileTargetCpp) -> str:
        return f"""
build {target.obj}: cpp {target.src} || $dyndep_file
    dyndep = $dyndep_file
""".strip()

    @staticmethod
    def render_compile_target_cppm(target: CompileTargetCppm) -> str:
        return f"""
build {target.obj} | {target.pcm}: cppm {target.src} || $dyndep_file
    dyndep = $dyndep_file
    pcm_out = {target.pcm}
""".strip()

    @staticmethod
    def render_compile_target_std_module(target: CompileTargetStdModule) -> str:
        return f"""
build {target.generated_src}: copy {target.original_src}

build {target.pcm}: std_module {target.generated_src} || $dyndep_file
    dyndep = $dyndep_file
""".strip()


class BuildDependencyParser:
    config: ShurikenConfig
    target: str

    def __init__(self, *, config: ShurikenConfig, target: str):
        self.config = config
        self.target = target

    def parse_build_dependencies(self, parsed_p1689: dict[str, Any]) -> dict[Path, list[Path]]:
        # Shortcut variables
        source_dir = self.config.source_dir
        target_build_dir = self.config.build_dir / self.target
        default_target_build_dir = self.config.build_dir / self.config.default_target

        # Map of module names to PCM file paths
        module_map: dict[str, Path] = {}

        # Validate module name and path convention, construct map of module names to PCM file paths
        for rule in parsed_p1689['rules']:
            provides = rule.get('provides', [])
            if provides:
                assert len(provides) == 1, f"Rule provides more than one module: {rule}"
                assert provides[0]['is-interface'] is True, f"Rule provides non-interface module: {rule}"

                module_name: str = provides[0]['logical-name']
                assert module_name not in module_map, f'Module "{module_name}" is provided more than once: {rule}'

                if module_name != 'std':
                    expected_source_path = str(Path(source_dir, module_name.replace('.', '/') + '.cppm'))
                    assert provides[0]['source-path'] == expected_source_path, \
                        f"Module name does not match source path: {rule}"

                module_map[module_name] = Path(target_build_dir, 'pcm', module_name + '.pcm')

        # Collect all module dependencies to generate dyndeps
        build_dependencies: dict[Path, list[Path]] = {}

        # Parse rules and get module dependencies
        for rule in parsed_p1689['rules']:
            # Compilation database contains "build/{default_target}" paths, we need to map them to "build/{target}"
            out = target_build_dir / Path(rule['primary-output']).relative_to(default_target_build_dir)

            build_dependencies[out] = []
            for require in rule.get('requires', []):
                module_name = require['logical-name']
                assert module_name in module_map, f'Module "{module_name}" not found in module map'
                build_dependencies[out].append(module_map[module_name])

        return build_dependencies


class StdModuleSourceFinder:
    cpp_compiler: str

    def __init__(self, *, config: ShurikenConfig, target: str):
        self.cpp_compiler = config.get_merged_target_config(target).cpp_compiler

    def find_module_sources(self, module_names: list[str]) -> dict[str, Path]:
        # Get normalized path of standard library "module manifest" (JSON file containing source paths for std modules)
        module_manifest_path = self.get_module_manifest_path()
        module_manifest_base_dir = module_manifest_path.parent

        # Parse module manifest JSON file
        with module_manifest_path.open('r') as f:
            module_manifest_json = json.load(f)

        # Collect the module source paths
        std_module_sources: dict[str, Path] = {}
        for modules_item in module_manifest_json['modules']:
            # Ignore modules we don't need (probably just std.compat)
            module_name = modules_item['logical-name']
            if module_name not in module_names:
                continue

            # The source path is usually relative to the module manifest file, convert to absolute path
            module_src_path = Path(modules_item['source-path'])
            if not module_src_path.is_absolute():
                module_src_path = module_manifest_base_dir / module_src_path
            if not module_src_path.is_file():
                raise Exception(f'Source path for module "{module_name}" is not a file: {module_src_path}')

            # Resolve symlinks and normalize path
            std_module_sources[module_name] = module_src_path.resolve()

        return std_module_sources

    def get_module_manifest_path(self) -> Path:
        # Use compiler to get the path to the module manifest JSON file
        clang_cmd = [self.cpp_compiler, '-print-library-module-manifest-path']
        clang_cmd_result = subprocess.run(clang_cmd, capture_output=True)

        if clang_cmd_result.returncode != 0:
            cmd_stderr = clang_cmd_result.stderr.decode().strip()
            raise Exception(f'Subprocess "{" ".join(clang_cmd)}" failed. stderr:\n{cmd_stderr}')

        # Read path from subprocess stdout
        module_manifest_path = Path(clang_cmd_result.stdout.decode().strip())

        if not module_manifest_path.is_file():
            raise Exception(f'Could not locate stdlib module manifest: {module_manifest_path}')

        # Resolve symlinks and normalize path
        return module_manifest_path.resolve()


if __name__ == '__main__':
    ShurikenCli().run()