/lincs/io/config_stats.py

# -*- coding: utf-8 -*-
""" Code for parsing the configuration file used for
    running statistics on Lincs experimental treatments.
"""
import yaml
import os
import pandas


REQUIRED_FIELDS = (
    ('experiment_name', str),
    ('treatments', dict),
    ('signals', list),
    ('on_threshold', int),
)


class ConfigParseError(Exception):
    """docstring for ConfigParseError"""


def check_config(config):
    """ Check to make sure all config values are present and in the correct
        format. Check to make sure Excel files are in order.

    """
    for field, obj_type in REQUIRED_FIELDS:

        if field not in config:
            raise ConfigParseError("Missing field in config: {0}".format(
                field
            ))
        if not isinstance(config[field], obj_type):
            msg = "Field '{0}' should be of type {1}".format(field, obj_type)
            raise ConfigParseError(msg)

    if "control" not in config["treatments"]:
        raise ConfigParseError("Configuration missing a control treatment!")

    for k, v in config["treatments"].iteritems():

        # Make sure each treatment has an Excel file listed
        #
        if "path" not in v or not v["path"]:
            msg = "Treatment {0} is missing a path to data file".format(k)
            raise ConfigParseError(msg)

        # Make sure that file exists
        #
        if not os.path.isfile(v["path"]):
            msg = "Treatment {0} data file {1} not found!".format(k, v["path"])
            raise ConfigParseError(msg)

        # Open the file.  Make sure the required sheet exists.
        # Set the required sheet if it's the default.
        #
        xls_file = pandas.ExcelFile(v['path'])
        if "sheet" not in v:
            config["treatments"][k]["sheet"] = xls_file.sheet_names[0]
        elif v["sheet"] not in xls_file.sheet_names:
            msg = "Sheet {0} not found in {1}!".format(v["sheet"], v["path"])
            raise ConfigParseError(msg)

        # Ensure there is a cell_count_column
        #
        if "cell_count_column" not in v:
            config["treatments"][k]["cell_count_column"] = "cell count"

        # Make sure there is a column for each signal and the cell_count
        #
        data = xls_file.parse(v["sheet"])
        needed_columns = config["signals"] \
            + [config["treatments"][k]["cell_count_column"]]
        for col in needed_columns:
            if col not in data.columns:
                msg = "Column {0} is missing in sheet {1} of {2}".format(
                    col,
                    v["sheet"],
                    v["path"]
                )
                raise ConfigParseError(msg)


def read_config(file_path):
    """ Read the configuration file

        :param file_path: Path to the config file
        :type file_path: string.
        :returns:  dictionary version of config file.
    """
    with open(file_path) as fh:
        config = yaml.safe_load(fh)

    check_config(config)

    return config