diff options
Diffstat (limited to 'r_qtl/r_qtl2.py')
| -rw-r--r-- | r_qtl/r_qtl2.py | 33 |
1 files changed, 28 insertions, 5 deletions
diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py index 9da4081..c6307f5 100644 --- a/r_qtl/r_qtl2.py +++ b/r_qtl/r_qtl2.py @@ -18,6 +18,10 @@ FILE_TYPES = ( "geno", "founder_geno", "pheno", "covar", "phenocovar", "gmap", "pmap", "phenose") +__CONTROL_FILE_ERROR_MESSAGE__ = ( + "The zipped bundle that was provided does not contain a valid control file " + "in either JSON or YAML format.") + def __special_file__(filename): """ @@ -72,6 +76,8 @@ def transpose_csv( def __read_by_line__(_path): with open(_path, "r", encoding="utf8") as infile: for line in infile: + if line.startswith("#"): + continue yield line transposed_data= (f"{linejoinerfn(items)}\n" for items in zip(*( @@ -112,7 +118,7 @@ def __control_data_from_zipfile__(zfile: ZipFile) -> dict: or filename.endswith(".json")))) num_files = len(files) if num_files == 0: - raise InvalidFormat("Expected a json or yaml control file.") + raise InvalidFormat(__CONTROL_FILE_ERROR_MESSAGE__) if num_files > 1: raise InvalidFormat("Found more than one possible control file.") @@ -129,7 +135,6 @@ def __control_data_from_zipfile__(zfile: ZipFile) -> dict: else yaml.safe_load(zfile.read(files[0]))) } - def __control_data_from_dirpath__(dirpath: Path): """Load control data from a given directory path.""" files = tuple(path for path in dirpath.iterdir() @@ -137,7 +142,7 @@ def __control_data_from_dirpath__(dirpath: Path): and (path.suffix in (".yaml", ".json")))) num_files = len(files) if num_files == 0: - raise InvalidFormat("Expected a json or yaml control file.") + raise InvalidFormat(__CONTROL_FILE_ERROR_MESSAGE__) if num_files > 1: raise InvalidFormat("Found more than one possible control file.") @@ -200,8 +205,8 @@ def control_data(control_src: Union[Path, ZipFile]) -> dict: if control_src.is_dir(): return __cleanup__(__control_data_from_dirpath__(control_src)) raise InvalidFormat( - "Expects either a zipfile.ZipFile object or a pathlib.Path object " - "pointing to a directory containing the R/qtl2 bundle.") + "Expects either a zipped bundle of files or a path-like object " + "pointing to the zipped R/qtl2 bundle.") def replace_na_strings(cdata, val): @@ -549,3 +554,21 @@ def load_samples(zipfilepath: Union[str, Path], pass return tuple(samples) + + + +def read_text_file(filepath: Union[str, Path]) -> Iterator[str]: + """Read the raw text from a text file.""" + with open(filepath, "r", encoding="utf8") as _file: + for line in _file: + yield line + + +def read_csv_file(filepath: Union[str, Path], + separator: str = ",", + comment_char: str = "#") -> Iterator[tuple[str, ...]]: + """Read a file as a csv file.""" + for line in read_text_file(filepath): + if line.startswith(comment_char): + continue + yield tuple(field.strip() for field in line.split(separator)) |