Execuable to find all Qt modules referenced by a set of C++ source files

FindQtModules.py

"""
This executable allows to find all Qt modules referenced by a set of C++ source files.

A Qt module (e.g QtNetwork) aggregates together a collection of Qt classes and is
associated with a set of public and private headers.

For example, to select multiple files and identify associated Qt modules:

    QT_INCLUDE_DIR=/path/to/Qt5.15.2/5.15.2/gcc_64/include
    python FindQtModules.py $QT_INCLUDE_DIR $(find /path/to/Awesome/Lib/ -type f)

Limitations:
* Preprocessor directives are not evaluated.
* Tested with Qt 5.x

This code is licensed under the terms of the Apache License, Version 2.0.
"""

import argparse
import os
import re


def configure_parser(parser):
    parser.add_argument(
        "qt_include_dir", type=str, metavar="QT_INCLUDE_DIR",
        help="Path to the Qt include directory (e.g /path/to/Qt5.15.2/5.15.2/gcc_64/include)"
    )
    parser.add_argument(
        "source_files", type=str, nargs="+", metavar="SOURCE_FILE",
        help="Path of the source file to inspect"
    )


def collect_qt_headers(qt_include_dir):
    """Return a map of Qt header to Qt module name.

    The Qt modules are identified as the direct sub-directory of ``qt_include_dir``.
    """
    qt_modules = os.listdir(qt_include_dir)
    if "QtCore" not in qt_modules:
        raise RuntimeException("Invalid QT_INCLUDE_DIR: Missing QtCore sub-directory")

    qt_header_without_extension_pattern = re.compile(r'^[qQ]\w+(?!\.h)$')

    qt_headers_to_modules = dict()

    for qt_module in qt_modules:
        for dirname, dirnames, filenames in os.walk(os.path.join(qt_include_dir, qt_module)):
            for filename in filenames:
                if filename.endswith(".h") or qt_header_without_extension_pattern.match(filename):
                    qt_headers_to_modules[filename] = qt_module

    return qt_headers_to_modules


def is_binary_file(filepath):
    """Return True if ``filepath`` exists and is a binary file.
 
    Adapted from https://stackoverflow.com/questions/898669/how-can-i-detect-if-a-file-is-binary-non-text-in-python/7392391#7392391
    """
    textchars = bytearray({7,8,9,10,12,13,27} | set(range(0x20, 0x100)) - {0x7f})
    is_binary_string = lambda bytes: bool(bytes.translate(None, textchars))
    with open(filepath, 'rb') as fp:
        return is_binary_string(fp.read(1024))


def inspect_source_file(source_file, qt_headers_to_modules):
    """Return a map of Qt module to header referenced by ``source_file``.
    """
    referenced_qt_modules = dict()

    include_pattern = re.compile(r'\#\s*include\s*[\<\"]([\w\.]+)[\>\"]')
    try:
        with open(source_file) as contents:
            for line in contents:
                match_object = include_pattern.search(line)
                if match_object is not None:
                    header_filename = match_object.group(1)
                    if header_filename in qt_headers_to_modules:
                        qt_module = qt_headers_to_modules[header_filename]
                        referenced_qt_modules[qt_module] = header_filename
    except UnicodeDecodeError as exc:
        print(f"Ignoring {source_file}: {exc}")

    return referenced_qt_modules

def main(argv=None):
    parser = argparse.ArgumentParser(description=__doc__)
    configure_parser(parser)
    args = parser.parse_args(argv[1:] if argv else None)
 
    referenced_qt_modules = dict()

    qt_headers_to_modules = collect_qt_headers(args.qt_include_dir)

    for source_file in args.source_files:
        if is_binary_file(source_file):
            continue
        referenced_qt_modules = {**referenced_qt_modules, **inspect_source_file(source_file, qt_headers_to_modules)}

    for qt_module in referenced_qt_modules:
        print(qt_module)


if __name__ == '__main__':
    main()