splitfile.py

import os, math
import fileinput

class FileSplit(object):

    MAX_BUFFER_SIZE = 102400

    def __init__(self, fileName, splitAmount):
        self.fileName = fileName
        self.splitAmount = splitAmount
        self.fileSize = None
        self.splitSize = None
        self.bufferSize = None
        self.splitReadAmount = None

    def __fileSize(self):
        if self.fileSize is None:
            file = open(self.fileName, "rb", 0)
            file.seek(0, os.SEEK_END)
            self.fileSize = file.tell()
            file.close()
        return self.fileSize

    def __splitSize(self):
        if self.splitSize is None:
            self.splitSize = int(math.ceil(self.__fileSize() / self.splitAmount))
        return self.splitSize

    def __bufferSize(self):
        if self.bufferSize is None:
            self.bufferSize = self.__splitSize() if FileSplit.MAX_BUFFER_SIZE > self.__splitSize() else FileSplit.MAX_BUFFER_SIZE
        return self.bufferSize

    def __splitReadAmount(self):
        if self.splitReadAmount is None:
            self.splitReadAmount = int(math.floor(self.__splitSize() / self.__bufferSize()))
        return self.splitReadAmount

    def process(self):
        splitAmount = 1
        input = open(self.fileName, "rb", 0)
        while splitAmount <= self.splitAmount:
            output = open(self.fileName + '.' + str(splitAmount) + '.CSV', "wb")
            output.write(b'NSUDF;STATUS\r\n')
            splitReadAmount = 1
            while splitReadAmount < self.__splitReadAmount():
                rb = input.read(self.__bufferSize())
                output.write(rb)
                splitReadAmount += 1
            rb = None
            if splitAmount != self.splitAmount:
                rb = input.read(self.__bufferSize())
                lastLinePos = rb.rfind(b'\n') + 1
                input.seek(-self.__bufferSize(), 1)
                rb = input.read(lastLinePos)
            elif splitReadAmount == self.__splitReadAmount():
                rb = input.read()
            output.write(rb)
            splitAmount += 1
            output.close()
        input.close()

    def desc(self):
        print('Tamanho total', self.__fileSize(), 'bytes')
        print('Quantidade divisoes', self.splitAmount)
        print('Tamanho divisao', self.__splitSize(), 'bytes')
        print('Tamanho buffer', self.__bufferSize(), 'bytes')
        print('Quantidade leitura por bloco (buffer)', self.__splitReadAmount())

with fileinput.FileInput("NSUDF.TXT", inplace=True, backup='.bak') as file:
    for line in file:
        print(line.replace('\n', ';1\n'), end='')

split = FileSplit("NSUDF.TXT", 20)
split.desc()
split.process()