How to split values in a CSV file

There are two tasks here. Firstly, filter the data, secondly parse the data to get the required values.

Filtering can be done with the built-in filter function (or a list comprehension or generator expression if you prefer). For the parsing a small function can be used as we need to process two dates per row.

operator.itemgetter is used to extract data from rows efficiently.

import csv
import operator

# Get the value of the 'take' column
type_getter = operator.itemgetter(4)
# Extract the columns we want to work with
columns_getter = operator.itemgetter(0, 4, 5)


def process_datestring(datestring):
    # Assumes offset is always positive; consider
    # using re.split if it could be positive or negative.
    date, _, offset = datestring.partition('+')
    dateparts = date.split('-')
    td, *_ = offset.partition(':')
    dateparts.append(td)
    return dateparts


headers = ['YEAR', 'MONTH', 'DAY', 'TD', 'TYPE', 'YEAR', 'MONTH', 'DAY', 'TD']
with open(file_path, 'r') as my_file, open('outputfile.txt', 'w') as outer:
    reader = csv.reader(my_file, delimiter = ',')
    next(reader) #Skip the first row because it's just the header
    writer = csv.writer(outer, delimiter = '\t')

    writer.writerow(headers)

    filtered = filter(lambda r: type_getter(r) == 'TAKE', reader)
    for row in filtered:
        date1, type_, date2 = columns_getter(row)
        out_row = process_datestring(date1)
        out_row.append(type_)
        out_row.extend(process_datestring(date2))
        writer.writerow(out_row)

You could write a parse function. Something like this:

def parser(row):
    if "TAKE" in row[4]: #Checks for your 'TAKE' statement
        year = row[0][0:4]
        month = row[0][5:7]
        ...
        year_2 = row[5][0:4]
        ...
        return [year, month, day, td, row[4], year_2, month_2,...]

And just give this whole thing to your writer:

for row in reader:
    converted = parser(row)
    writer.writerow(converted)

This should work. If you want to be more carefull, you will have to convert the string to a datetime.datetime object using strptime and then extract the dates, but if chopping the string is good enough, you could do this.