csv342 1.0.1

Python 3 like CSV module for Python 2

csv342 is a Python module similar to the the csv module in the standard library. Under Python 3, it just calls the standard csv module. Under Python 2, it provides a Python 3 like interface to reading and writing CSV files, in particular concerning non ASCII characters.

It is distributed under the BSD license. The source code is available from https://github.com/roskakori/csv342.

Installation

To install, simply run:

$ pip install --upgrade csv342

Alternatively you can download the distributin archive from http://pypi.python.org/pypi/csv342/, unpack it and copy csv342.py into your application path.

Examples

First, consider changing all string literals in you source code to unicode instead of str under Python 2 using to avoid the messy u string prefix.

>>> from __future__ import unicode_literals

The following examples assume you did that, making your Python 2 code look like Python 3 even more.

Once you import csv342 using

>>> import csv342 as csv

your code can call CSV functions the same way independent of whether it runs under Python 2 or 3. First, let’s write a few test data to a io.StringIO:

>>> import io
>>> csv_stream = io.StringIO()
>>> csv_writer = csv.writer(csv_stream)
>>> csv_writer.writerow(['a', 'b'])
>>> csv_writer.writerow(['1', '"x"'])
>>> csv_content = csv_stream.getvalue()
>>> str(csv_content)
'a,b\r\n1,"""x"""\r\n'

To read data from a io.StringIO use:

>>> csv_stream = io.StringIO('a,b\r\n1,"""x"""\r\n')
>>> csv_reader = csv.reader(csv_stream)
>>> for row in csv_reader:
>>>     print(row)

To read a UTF-8 encoded CSV file with non ASCII characters use:

>>> csv_path = os.path.join('test', 'utf-8.csv')
>>> with io.open(csv_path, encoding='utf-8', newline='') as csv_file:
>>>     csv_reader = csv.reader(csv_file, delimiter=',')
>>>     for row in csv_reader:
>>>         print('row {0:d}: data={1}'.format(csv_reader.line_num, row))

Features

• Supports Python 2’s unicode strings.

• Provides reader, writer, DictReader and DictWriter.

• Supports reading and writing with files, io.StringIO etc.

• Rejects attempts to read or write with cStringIO or StringIO.StringIO (which do not really work with unicode); use io.StringIO instead.

Limitations

• All limitations of the standard csv module apply.

• Uses the standard csv.Sniffer under Python 2.

• Requires Python 2.6 or later.

Performance

Processing a CSV with Python 2 using csv342 is about 30% slower than processing it with Python 3. This is probably due the fact that under Python 2 there is an intermediate translation to UTF-8 using pure Python code while in Python 3 uses mostly native code.

Provided you have both Python 2 and 3 installed on the same machine, you can test this yourself running:

python3 test/performance.py
python2 test/performance.py

On an ancient laptop with a core 2 duo Intel CPU and Ubuntu 14.04, this takes 10.2 seconds respectively 13.6 seconds.

License

Copyright (c) 2016-2020, Thomas Aglassinger All rights reserved.

Distributed under the BSD License. For more information, see LICENSE.txt.

Version history

Version 1.0.1, 2020-05-05

• Fixed inconsistent license information in setup.py (contributed by Stephen DiCato).

Version 1.0.0, 2018-04-02

• Fixed DictReader so that it can now read from a list of lines (issue #1, contributed by Lucas Wiman).

• Fixed AttributeError in StringIO test (contributed by Lucas Wiman).

Version 0.2, 2016-04-17

• Fixed version number when running under Python 3.

• Fixed helper constant IS_PYHTON2 which always was False.

• Added remaining symbols from Python 2’s csv module.

Version 0.1, 2016-04-16

• Initial release.