import os
from pathlib import Path

THIRD_PARTY_DIR_NAMES = {
    'node_modules', 'vendor', 'third_party', 'third-party', 'extern', 'external', 'deps',
    'Cesium', 'Cesium-1.132', 'dist', 'build'
}

CODE_EXTENSIONS = {
    '.py', '.js', '.ts', '.tsx', '.jsx', '.css', '.scss', '.html', '.c', '.h', '.cpp', '.hpp',
    '.cc', '.hh', '.m', '.mm', '.java', '.go', '.rs', '.cs', '.vue', '.svelte'
}

SKIP_DIR_NAMES = {
    '.git', '.hg', '.svn', '__pycache__', '.mypy_cache', '.pytest_cache', '.idea', '.vscode',
    '.venv', 'venv', 'env', '.tox', '.cache', 'Resources', 'icons', 'tex', 'terminal'
}

MAX_FILE_SIZE_BYTES = 2 * 1024 * 1024  # 2 MiB


def is_third_party(path: Path) -> bool:
    for part in path.parts:
        # normalize case on Windows
        name = part.lower()
        if name in {n.lower() for n in THIRD_PARTY_DIR_NAMES}:
            return True
    return False


def should_skip_dir(dirname: str) -> bool:
    return dirname.lower() in {n.lower() for n in SKIP_DIR_NAMES}


def is_code_file(path: Path) -> bool:
    return path.suffix.lower() in CODE_EXTENSIONS


def count_lines(path: Path) -> int:
    try:
        if path.stat().st_size > MAX_FILE_SIZE_BYTES:
            return 0
        # Try text read with universal newlines, fallback to binary
        try:
            with path.open('r', encoding='utf-8', errors='ignore') as f:
                return sum(1 for _ in f)
        except Exception:
            with path.open('rb') as f:
                return f.read().count(b'\n')
    except Exception:
        return 0


def main(root: Path) -> None:
    third_party_loc = 0
    first_party_loc = 0
    third_party_files = 0
    first_party_files = 0

    for dirpath, dirnames, filenames in os.walk(root):
        # prune directories
        dirnames[:] = [d for d in dirnames if not should_skip_dir(d)]

        current = Path(dirpath)
        current_is_third = is_third_party(current)

        for fn in filenames:
            p = current / fn
            if not is_code_file(p):
                continue
            loc = count_lines(p)
            if current_is_third:
                third_party_loc += loc
                third_party_files += 1
            else:
                first_party_loc += loc
                first_party_files += 1

    total_loc = first_party_loc + third_party_loc
    open_source_rate = (third_party_loc / total_loc) * 100 if total_loc else 0.0

    print('Open-Source Rate (by LOC): {:.2f}%'.format(open_source_rate))
    print('\nBreakdown:')
    print('  First-party LOC : {} ({} files)'.format(first_party_loc, first_party_files))
    print('  Third-party LOC : {} ({} files)'.format(third_party_loc, third_party_files))
    print('  Total LOC       : {}'.format(total_loc))
    print('\nNotes:')
    print('  - Third-party directories are heuristically detected by common names: {}'.format(', '.join(sorted(THIRD_PARTY_DIR_NAMES))))
    print('  - Static asset directories are skipped: {}'.format(', '.join(sorted(SKIP_DIR_NAMES))))
    print('  - Counted code extensions: {}'.format(', '.join(sorted(CODE_EXTENSIONS))))


if __name__ == '__main__':
    main(Path('.').resolve())