Tool to simlink CMake external data files on Windows

externaldata.py

"""Tools for dealing with data registered with
CMake external data:
  - Symlink from external data cache to build binary location.
    Currently assumes the files have been downloaded and does not deal with
    references that change.
"""
from itertools import chain
from pathlib import Path
from typing import Iterator, Sequence

# Constants.
# todo: Make them script arguments
CMAKE_SOURCE_DIR = Path.home() / Path('Code/mantid')
CMAKE_BINARY_DIR = CMAKE_SOURCE_DIR / 'build'
CMAKE_EXTERNALDATA_BINARY_SUBDIR = 'ExternalData'
CHECKSUM_ALG = 'md5'
CMAKE_EXTERNALDATA_CACHE_DIR = Path('C:/MantidExternalData')


def find_checksum_files(base_dirs: Sequence[Path], checksum_alg: str) -> Iterator[Path]:
    """
    Find files named *.checksum_alg, starting from base_dirs
    """
    return chain.from_iterable([base_dir.glob(f'**/*.{checksum_alg}') for base_dir in base_dirs])

def link_cache_file(checksum_filepath: Path, cmake_source_dir:Path ,
                    cmake_binary_dir: Path, externaldata_cache_dir: Path):
    """
    Create a symlink from an external data cache, where files are named after their
    checksum value, to a file with the real name given by the basename of the checksum
    file. The contents of the checksum file given the checksum value.
    """
    with open(checksum_filepath, 'r') as checksum_file_handle:
        checksum_value = checksum_file_handle.readline().strip()

    # The final suffix denotes the checksum algorithm, e.g Testing\\Data\\DocTest\\164198.nxs.md5
    # contains an MD5 checksum value and the final real filename is 164198.nxs in the 'Testing\\Data\\DocTest'
    # directory.
    checksum_suffixes = checksum_filepath.suffixes
    checksum_alg = checksum_suffixes[-1][1:]  # final suffix and remove the leading '.' character
    cache_filepath = externaldata_cache_dir / checksum_alg.upper() / checksum_value

    real_filename = checksum_filepath.stem
    real_subbdir = checksum_filepath.parent.relative_to(cmake_source_dir)
    real_filepath = cmake_binary_dir / CMAKE_EXTERNALDATA_BINARY_SUBDIR / real_subbdir / real_filename
    if real_filepath.exists():
        print(f"'{real_filepath}' exists. Skipping.")
        return

    print(f"Linking '{real_filepath}' to '{cache_filepath}'")
    real_filepath.parent.mkdir(parents=True, exist_ok=True)
    real_filepath.symlink_to(cache_filepath)


def main() -> int:
    """Main entry point for the program"""

    for path in find_checksum_files([CMAKE_SOURCE_DIR / 'Testing'], CHECKSUM_ALG):
        try:
            link_cache_file(path, CMAKE_SOURCE_DIR, CMAKE_BINARY_DIR, CMAKE_EXTERNALDATA_CACHE_DIR)
        except OSError as exc:
            print(f"Error linking cached data for '{path}': {str(exc)}")

    return 0

if __name__ == '__main__':
    main()