Copy C or C++ include header dependencies recursively

copy_headers.py

# Example command:
# python copy_headers.py --headers tensorflow/lite/model.h tensorflow/lite/interpreter.h \
# --destination MyAwesomeProject/include
# --include-dirs ../flatbuffers/include  # Note that these are relative to --source-dir
# --source-dir ../tensorflow/

import argparse
import os
import subprocess


def get_structure(file, directories):
    for directory in directories:
        if file.startswith(directory):
            structure = os.path.dirname(file[len(directory) + (not directory.endswith('/')):])
            return structure
    return os.path.dirname(file)


def copy(src, dst):
    with open(src, 'rb') as sf, open(dst, 'wb') as df:
        df.write(sf.read())


if __name__ == '__main__':
    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter,
                                     description="Copy header dependencies recuresively")
    parser.add_argument("--include-dirs", nargs="*",
                        help="Header include directories, relative to source-dir")
    parser.add_argument("--headers", nargs="*", required=True,
                        help="Header paths whose dependencies will be copied, relative to source-dir")
    parser.add_argument("--destination", help="destination directory to copy files", required=True)
    parser.add_argument("--dry-run", help="Just log, don't copy", action='store_true')
    parser.add_argument("--gcc-args", help="gcc args to be passed", default="--std=c++11")
    parser.add_argument("--source-dir", help="gcc args to be passed", default=".")
    args = parser.parse_args()

    include_dirs = args.include_dirs if args.include_dirs else []
    source_dir = args.source_dir
    include_dirs.append('.')
    include_args = []
    for include_dir in include_dirs:
        include_args += ["-I", include_dir]

    headers = args.headers
    destination = args.destination
    dry_run = args.dry_run

    copy_files = dict()  # filepath => directory structure under include dir
    for header in headers:
        cmd = ["g++", args.gcc_args] + include_args + ["-MM", header]
        p = subprocess.Popen(cmd, cwd=source_dir, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
        out, err = p.communicate()
        out = out.decode('utf8')
        files = out[out.index(':') + 1:].replace('\\', '').split()
        for file in files:
            copy_files[os.path.abspath(os.path.join(source_dir, file))] = get_structure(file, include_dirs)

    for file, file_dir in copy_files.items():
        file_destination = os.path.join(destination, file_dir, os.path.basename(file))
        if dry_run:
            print(f'Would copy: {file} => {file_destination}')
        else:
            destination_dir = os.path.dirname(file_destination)
            if not os.path.exists(destination_dir):
                os.makedirs(destination_dir)
            copy(file, file_destination)