Installing unicode csv for Python?

Question

I'm new to github and I'm trying to install unicodecsv (https://github.com/jdunck/python-unicodecsv).

I'm trying

sudo pip install -e git://github.com/jdunck/python-unicodecsv.git#egg=unicodecsv

But I'm getting an error message. I'm probably doing someone basic wrong, can someone help?

Obtaining unicodecsv from git+git://github.com/jdunck/python-unicodecsv.git#egg=unicodecsv
  Cloning git://github.com/jdunck/python-unicodecsv.git to ./src/unicodecsv
  Running setup.py egg_info for package unicodecsv
    Traceback (most recent call last):
      File "<string>", line 14, in <module>
      File "/Users/dave/Dropbox/Promoter/working/src/unicodecsv/setup.py", line 5, in <module>
        version = __import__('unicodecsv').__version__
      File "unicodecsv/__init__.py", line 49
        except TypeError, e:
                        ^
    SyntaxError: invalid syntax
    Complete output from command python setup.py egg_info:
    Traceback (most recent call last):

  File "<string>", line 14, in <module>

  File "/Users/dave/Dropbox/Promoter/working/src/unicodecsv/setup.py", line 5, in <module>

    version = __import__('unicodecsv').__version__

  File "unicodecsv/__init__.py", line 49

    except TypeError, e:

                    ^

SyntaxError: invalid syntax

----------------------------------------
Command python setup.py egg_info failed with error code 1
Storing complete log in /Users/dave/.pip/pip.log
Davids-MacBook-Air:working dave$ sudo pip install -e git://github.com/jdunck/python-unicodecsv.git#egg=unicodecsv
Obtaining unicodecsv from git+git://github.com/jdunck/python-unicodecsv.git#egg=unicodecsv
  Updating ./src/unicodecsv clone
^[  Running setup.py egg_info for package unicodecsv
    Traceback (most recent call last):
      File "<string>", line 14, in <module>
      File "/Users/dave/Dropbox/Promoter/working/src/unicodecsv/setup.py", line 5, in <module>
        version = __import__('unicodecsv').__version__
      File "unicodecsv/__init__.py", line 49
        except TypeError, e:
                        ^
    SyntaxError: invalid syntax
    Complete output from command python setup.py egg_info:
    Traceback (most recent call last):

  File "<string>", line 14, in <module>

  File "/Users/dave/Dropbox/Promoter/working/src/unicodecsv/setup.py", line 5, in <module>

    version = __import__('unicodecsv').__version__

  File "unicodecsv/__init__.py", line 49

    except TypeError, e:

                    ^

SyntaxError: invalid syntax

----------------------------------------
Command python setup.py egg_info failed with error code 1
Storing complete log in /Users/dave/.pip/pip.log

Solution

The problem is that you're using Python 3.0 or later, and trying to use a library which is only compatible with 2.7 and earlier.

The specific problem is the line the traceback points at:

except TypeError, e:

This syntax was deprecated in 2.6, in favor of (more flexible and more consistent) new syntax:

except TypeError as e:

In 3.0 and later, the deprecated syntax is no longer allowed at all.

So, if you want to use this library, someone will have to port it—you, the author, or someone else. It may just be a matter of running 2to3, or fixing each except statement manually—but it may be a lot more to do than that, especially considering this library is all about Unicode.

---

However, it's worth noting that Python 3.x doesn't have the same problem as 2.x did. You can pass the csv module text/Unicode file objects, and it will just handle them. Adapting the example from the unicodecsv docs:

>>> import csv
>>> from io import StringIO
>>> f = StringIO()
>>> w = csv.writer(f)
>>> w.writerow(('é', 'ñ'))
>>> f.seek(0)
>>> r = csv.reader(f)
>>> row = r.next()
>>> print row[0], row[1]
é ñ

Note that I didn't even have to specify utf-8, because StringIO is a Unicode str buffer, not a bytes buffer. You don't have to worry about coding at all.

---

If you didn't even know you were running Python 3.x (as in, you knew you installed it, but you were careful to keep Apple's pre-installed Python 2.7 higher on the PATH), there are three common reasons this can happen.

• Apple's Python doesn't come with pip; all of the popular Python 3.x installers and packages do. So, if you haven't installed pip for 2.7, the only one you have is 3.x.

• Apple's Python installs scripts like pip to /usr/local/bin. So do some of the popular Python 3.x installers and packages. So, whichever you installed more recently wins. (The 3.x one should also be available as /usr/local/bin/pip3, so overwriting its /usr/local/bin/pip with 2.7's is usually fine… unless pip3 is a symlink to pip.)

• sudo deliberately discards most of your user environment, so 2.7 may be higher on your PATH when running as yourself, but not when running with sudo.

Comments

[Ignacio Vazquez-Abrams]

That's not what the new syntax looks like.

[abarnert]

@IgnacioVazquez-Abrams: Oops, I originally wrote about the new raise and except syntax, then decided that was more detail than necessary and removed the raise parts, then edited it incorrectly. Thanks for the catch!

[David Bailey]

Thanks very much for this. The issue I have is that I am reading data from the CSV with ? replacing special characters. For example, "São Paulo" reads as "S?o Paulo".

[abarnert]

@DavidBailey: Yes, in Python 2.7 the module is useful; that's why I said "Python 3.x doesn't have the same problem as 2.x did". Most likely, your python is actually /usr/bin/python, which is 2.7.2, while your pip is /usr/local/bin/pip or /Library/Frameworks/Python.framework/Versions/3.3/bin/pip is something, which is 3.3, which is going to confuse you. That's why I wrote a whole section on this in the answer, which you should read.

[David Bailey]

Sorry, reread and installed pip for 2.7, now have safely installed. Thanks and my bad.