utah_junk_row_step

from datetime import datetime
import parsekit

class FilterRows(parsekit.Step):

    start_row = parsekit.Argument(
        "The row to start emitting from.",
        required=True,
        type=int)

    skip_empty_row = parsekit.Argument(
        "Skip empty row.",
        type=bool)

    skip_day_start = parsekit.Argument(
        "Skip row starting with day.",
        type=bool)

    def configure(self, options):
        self.row_counter = 0

    def run(self, record, metadata):
        self.row_counter += 1
        if self.row_counter >= self.options.start_row:

            if self.options.skip_empty_row:
                if any(record):

                    if self.options.skip_day_start:
                        first_val = record[0].split(',')[0]
                        try:
                            datetime.strptime(first_val, '%A')
                        except:
                            return record, metadata
                    else:
                        return record, metadata
            else:
                if self.options.skip_day_start:
                    first_val = record[0].split(',')[0]
                    try:
                        datetime.strptime(first_val, '%A')
                    except:
                        return record, metadata
                else:
                    return record, metadata