import regex as re
from pyspark.sql import Row
from pyspark.sql.functions import udf, when
from pyspark.sql.types import BooleanType

formats = [r'^(?:\(\d{3}\)-)\d{3}-\d{4}$',
           r'^(?:\d{3}-)\d{3}-\d{4}$', r'^(?:\d{3}-)\d{7}$', r'^\d{10}$']


def validate_format(number):
    length = len(number)
    if length == 14:
        if (re.match(formats[0], number)):
            return True
        raise ValueError
    if length == 12:
        if (re.match(formats[1], number)):
            return True
        raise ValueError
    if length == 11:
        if (re.match(formats[2], number)):
            return True
        raise ValueError
    if length == 10:
        if (re.match(formats[3], number)):
            return True
        raise ValueError
    raise ValueError


def process_df(spark):
    my_cols = Row("Column1", "Column2", "Column3", "Column4")
    row_1 = my_cols('(617)-283-3811', 'Salah', 'Messi', None)
    row_2 = my_cols('617-2833811', 'Messi', 'Virgil', 'Messi')
    row_3 = my_cols('617-283-3811', 'Ronaldo', 'Messi', 'Ronaldo')
    row_seq = [row_1, row_2, row_3]
    df = spark.createDataFrame(row_seq)
    validate_format_udf = udf(validate_format, BooleanType())
    try:
        df2 = df.select("Column2", validate_format_udf(
            "Column2").alias("something"))
        df2.show()
    except ValueError:
        df = df.drop_column("Column2")
    print(df.show())