mmapped csv

import csv
import codecs
import cStringIO

class CsvIter(Object):

    def __init__(self, csvFileName):

        self.csvFileName = csvFileName
        self.fileEncoding_ = self._get_encoding(self.csvFileName)
        self.csvfile = codecs.open(self.csvFileName, "r+", encoding="utf-8")
        self.data = self.mapfile(self.csvfile)
        self.lookup = self._get_row_lookup(self.data)

    def __getitem__(self, key):
        """return an item from a memory-mapped csv file"""
        try:
            if key == 0:
                start, end = 0, self.lookup[key]
            else:
                start, end = self.lookup[key - 1], self.lookup[key]
        except KeyError:
            raise IndexError("index out of range")
        return next(csv.reader(cStringIO.StringIO(self.data[start: end])))

    def mapfile(self, fileObj):
        size = os.path.getsize(fileObj.name)
        return mmap.mmap(fileObj.fileno(), size)

    def _get_row_lookup(self, data):
        lino, record_start, lookup = 0, 0, {}
        while True:
            line = data.readline()
            record_start += len(line)
            lookup[lino] = record_start
            lino += 1
            if not line:
                break
        return lookup

    # source: https://docs.python.org/2/library/csv.html (bottom of page)
    def unicode_csv_reader(self, unicode_csv_data, dialect, encoding, **kwargs):
        csv_reader = csv.reader(self.utf_8_encoder(unicode_csv_data, encoding),
                                dialect=dialect, **kwargs)
        for row in csv_reader:
            yield [unicode(cell, encoding) for cell in row]

    def utf_8_encoder(self, unicode_csv_data, encoding):
        for line in unicode_csv_data:
            yield line.encode(encoding)