cran-package-metadata-parser.py

def from_packages_list(data: str) -> Generator:
    """Parses CRAN package metadata from
    https://cran.r-project.org/src/contrib/PACKAGES
    and returns the list of dictionaries.

    Args:
        data (str): raw text from the package list

    Returns:
        (Generator): each entry from packages as dictionary
    """
    fields = set()
    tmp = {}

    # We want to iterate over each line and accumulate
    # keys in dictionary, once we meet the same key
    # in our dictionary we have a single package
    # metadata parsed so we yield and repeat again.
    for line in data.splitlines():
        line = str(line)
        if not line.strip():
            continue

        if ":" in line:
            parts = line.split(":")
            field = str(parts[0].strip())
            value = str("".join(parts[1:]).strip())

            if field in fields:
                fields = {field}
                result = {**tmp}
                tmp = {field: value}
                if result:
                    yield result
            else:
                # Here we want to parse dangling lines
                # like the ones with long dependency
                # list, `R (>= 2.15.0), xtable, pbapply ... \n    and more`
                tmp[field] = str(value)
                fields.add(field)
        else:
            pairs = list(tmp.items())
            if pairs:
                last_field = str(pairs[-1][0])
                tmp[last_field] += f" {line.strip()}"

    # We also need to return the metadata for
    # the last parsed package.
    if tmp:
        yield tmp


def to_cran_format(metadata: Dict) -> Optional[str]:
    """
    Dump dictionary into the following form

        Package: A3
        Version: 1.0.0
        Depends: R (>= 2.15.0), xtable, pbapply
        Suggests: randomForest, e1071
        License: GPL (>= 2)
        MD5sum: 027ebdd8affce8f0effaecfcd5f5ade2
        NeedsCompilation: no

    Args:
        metadata (Dict): Converts metadata dictionary to deb format

    Returns:
        (Optional[str]): package record as deb format
    """
    return "\n".join([
        f"{key}: {value}"
        for key, value in metadata.items()
    ])


def from_cran_format(metadata: str) -> Dict:
    """Parse package metadata
    Note: it is a shorthand to `from_packages_list`
          then extracts the first value from it.

    Input should be in the following format
    which is R package metadata description
    see: https://cran.r-project.org/src/contrib/PACKAGES

        Package: A3
        Version: 1.0.0
        Depends: R (>= 2.15.0), xtable, pbapply
        Suggests: randomForest, e1071
        License: GPL (>= 2)
        MD5sum: 027ebdd8affce8f0effaecfcd5f5ade2
        NeedsCompilation: no

    Args:
        metadata (str): metadata text information

    Returns:
        (Dict): Parse deb format and return dictionary
    """
    [package] = list(from_packages_list(metadata))
    return package