csv_file.py 文件源码

def __parse_file(self):
        filename = self.options['filename']
        if not filename:
            raise IOError
        sep = self.options['column_separator']
        quote = self.options['quote_character']
        values = []

        with open(filename, 'rU') as infile:
            # if sep is not a one character string, csv.reader will raise a TypeError
            if not quote:
                csvreader = csv.reader(infile, delimiter=str(sep), quoting=csv.QUOTE_NONE)
            else:
                csvreader = csv.reader(infile, delimiter=str(sep), quotechar=str(quote))

            # get each line from the file and separate it into columns based on sep
            for row in csvreader:
                # append all lines as-is case-wise
                # unicode(str, errors='ignore') causes all invalid characters to be stripped out
                values.append([unicode(value.strip(), errors='ignore') for value in row])
                # ensure the number of columns in each row is the same as the previous row
                if len(values) > 1:
                    assert len(values[-1]) == len(values[-2])

        return values