ProcessCollectionsData.py 文件源码

 复制代码def standardize_variable_names(table, RULES):
    """
    Script to standardize the variable names in the tables
    PARAM DataFrame table: A table returned from pd.read_sql
    PARAM list[tuples]: A list of tuples with string replacements, i.e., (string, replacement)
    RET table
    """
    variableNames = list(table.columns.values)
    standardizedNames = {} # Pandas renames columns with a dictionary object
    for v in variableNames:
        f = v
        for r in RULES:
            f = re.sub(r[0],r[1],f)
        print '%s to %s' %(v,f)
        standardizedNames[v] = f
    table = table.rename(columns=standardizedNames)
    return table