diff options
Diffstat (limited to 'r_qtl/r_qtl2.py')
| -rw-r--r-- | r_qtl/r_qtl2.py | 45 |
1 files changed, 31 insertions, 14 deletions
diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py index dfa84ba..0ef487f 100644 --- a/r_qtl/r_qtl2.py +++ b/r_qtl/r_qtl2.py @@ -16,7 +16,7 @@ from r_qtl.exceptions import InvalidFormat, MissingFileException FILE_TYPES = ( "geno", "founder_geno", "pheno", "covar", "phenocovar", "gmap", "pmap", - "phenose") + "phenose", "phenonum") __CONTROL_FILE_ERROR_MESSAGE__ = ( "The zipped bundle that was provided does not contain a valid control file " @@ -411,22 +411,19 @@ def file_data(zfile: ZipFile, try: if isinstance(cdata[member_key], list): - for row in (line for lines in + yield from (line for lines in (file_data( zfile, member_key, {**cdata, member_key: innerfile}, process_value, process_transposed_value) for innerfile in cdata[member_key]) - for line in lines): - yield row + for line in lines) return if not cdata.get(f"{member_key}_transposed", False): - for row in with_non_transposed(zfile, member_key, cdata, process_value): - yield row + yield from with_non_transposed(zfile, member_key, cdata, process_value) return - for row in with_transposed( - zfile, member_key, cdata, process_transposed_value): - yield row + yield from with_transposed( + zfile, member_key, cdata, process_transposed_value) except KeyError as exc: raise MissingFileException(*exc.args) from exc @@ -477,8 +474,7 @@ def raw_file_data(zipfilepath: Union[str, Path], with (ZipFile(str(zipfilepath), "r") as zfile, zfile.open(memberfilename) as innerfile): wrappedfile = io.TextIOWrapper(innerfile) - for line in wrappedfile: - yield line + yield from wrappedfile def strip_comments(rawdata: Iterator[str], commentchar) -> Iterator[str]: """Remove comments from raw text.""" @@ -568,15 +564,36 @@ def load_samples(zipfilepath: Union[str, Path], 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 + yield from _file def read_csv_file(filepath: Union[str, Path], separator: str = ",", comment_char: str = "#") -> Iterator[tuple[str, ...]]: - """Read a file as a csv file.""" + """Read a file as a csv file. This does not process the N/A values.""" for line in read_text_file(filepath): if line.startswith(comment_char): continue yield tuple(field.strip() for field in line.split(separator)) + + +def read_csv_file_headers( + filepath: Union[str, Path], + transposed: bool, + separator: str = ",", + comment_char: str = "#" +) -> tuple[str, ...]: + """Read the 'true' headers of a CSV file.""" + headers = tuple() + for line in read_text_file(filepath): + if line.startswith(comment_char): + continue + + line = tuple(field.strip() for field in line.split(separator)) + if not transposed: + return line + + headers = headers + (line[0],) + continue + + return headers |