Common porting errors wiki

I think it could be helpful to someone (at the very least it would be helpful to future me) if i mention here the problems and errors i faced in porting sugar to python3 and their solution according to me.
So here are the errors: 

SyntaxError: invalid syntax
Line of error: except Exception, e:
Explanation: The correct syntax for exception in Python3 use as instead of a comma
Correct code: except Exception as e:

ImportError: No module named 'urllib2'
Line of error: import urllib2
Explanation: In Python3 urllib2 has been merged with urllib , so there is no module named urllib2. You can directly import the required module like urllib.request or urllib.error .
http://legacy.python.org/dev/peps/pep-3108/#urllib-package
Correct Code: import urllib.request, urllib.error, urllib.parse

ImportError: No module named 'StringIO'
Line of error: import StringIO
Explanation : The StringIO and cStringIO modules are gone. Instead, import the io module and use io.StringIO or io.BytesIO for text and data respectively.
Correct code: import io

ValueError: can't have unbuffered text I/O
Line of Error: sys.stdout = os.fdopen(sys.stdout.fileno(), 'w', 0)
Explanation: As the error says, we cannot have unbuffered text I/O in Python3. So i have removed this line for now, although there are some other ways to get unbuffered text I/O like using the -u argument.
Correct Code:

/usr/bin/env: python3: No such file or directory
Explanation : On some debugging, i found out that in my case the error was due to the virtualenv . As the virtualenv was running using python2 , it couldn't find python3 inside the virtual environment. So this error may vary from case to case and we would need to find out why the interpreter cannot find the python3 directory
Line of error: subprocess.check_call(["python2.7",
                           os.path.join(source_dir, "virtualenv.py"),
                           "-q", "--system-site-packages",
                            get_virtualenv_dir()])
Correct Code: subprocess.check_call(["python3.3",
                           os.path.join(source_dir, "virtualenv.py"),
                           "-q", "--system-site-packages",
                            get_virtualenv_dir()])

TypeError: Unicode-objects must be encoded before hashing
 Line of Error: path_hash.update(self._config_path)
Explanation: Unlike Python2, in Python3 all the hashlib fucntions take in bytes rather than strings so we need to encode the string before passing.
Correct Code: path_hash.update((self._config_path).encode('utf-8'))

TypeError: expected bytes, bytearray or buffer compatible object
Line of error: base64_hash = base64_hash.replace("+", "0")
Explanation: The .replace method requires a string input , hence we need to decode what we encodeed in the previous error.
Correct Code: base64_hash = base64_hash.decode('utf-8')
                        base64_hash = base64_hash.replace("+", "0")

TypeError: Type str doesn't support the buffer API
Line of Error: for mounted in mount_output.split("\n"):
Explanation: This is also a type error like the previous one which means we are again confusing bytes and strings. On debugging, i found out that mount_output is actually a byte object and thus string method split is not compatible with it, so we first need to decode it.
Correct Code: mount_output = mount_output.decode('utf-8')
                        for mounted in mount_output.split("\n"):

SyntaxError: assignment to keyword

Line of Error: False, True = 0, 1

Explanation: In Python3 True and False are keywords instead of literals , so they cannot be reassigned.

Refer to this article for the history of True/False keywords: http://python-history.blogspot.in/2013/11/story-of-none-true-false.html

Correct code: Just remove this line because True and False are already defined as keywords

UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 79: ordinal not in range(128)

Explanation: seems to be another one of those strings vs unicode vs bytes problem.
Correct code : b'' instead of ''

AttributeError: 'module' object has no attribute 'find'
Line Of Error: if string.find(thing, '=') >= 0:
Explanation: The string.find method has been deprecated in python3 , use the .find() method of strings instead
Correct code: if thing.find('=') >= 0:

FileNotFoundError: [Errno 2] No such file or directory: '/home/curiousguy13/py3sugar/sugar-build/build/commands/common.pyc'
Line Of Error: os.unlink(os.path.join(common.commands_dir, "common.pyc"))
Explanation: In python3 the the compiled header files are stored in __pycache__ instead of the same directory. see more here : http://stackoverflow.com/questions/154443/how-to-avoid-pyc-files
So, just changing the directory address to __pycache__ in common.py solved the issue. 

PS: The list is not exhaustive and i will keep on adding new errors and problems that i face with their solutions

WARNING: Use the above post at your own risk. The explanation and the solution of errors are according to my own knowledge of Python and some online research and they might be wrong in which case you can point out the mistake in the comments and i can correct it.

Porting to Python3 Progress

Last week I tried to solve the telepathy issue. The telepathy issue is basically that the telepathy library used in sugar is deprecated and not compatible with Python3 so we had to do something about it.

On researching, I found out that telepathy-glib is the new library which has taken over for telepathy-python and we need to write the telepathy-glib code in gobject convention in order for it to be compatible with introspection and also telepathy-glib is compatible with Python3.

But i had no idea what all these were, so i started reading up on telepathy,telepathy-glib,gobject,introspection,dbus etc. So all this new stuff was a little overwhelming and was taking too much of my time so in friday's sugar meeting, Walter Bender suggested to leave telepathy for now and port other modules first and that is exactly what i am going to do this week and i'll try to port as much code as possible this week and try to increase the test coverage too and after that we'll take care of telepathy.

Also side by side I continued working on porting sugar-build. 

Porting Sugar to Python3 Progress

I started with porting sugar-build to Python3 as it is used to build the rest of the sugar environment.

Until now, I have used 2to3[1] script with all fixers enabled on the sugar-build module which contained an osbuild script, python scripts for various commands and the test. And now I am executing the osbuild script and removing the errors one by one by doing the necessary changed in the code. Right now, it is taking time to deal with the errors which mainly include the proper usage and handling of strings in Python3[2].
It is worth noting though that as I learn how to deal with these errors, the porting further is going to get easier and easier.

I have cloned the sugar-build repo on GitHub and am working on it.
So although, it doesn't work perfectly right now you can follow my progress here:

https://github.com/curiousguy13/sugar-build/tree/py3

[1]: https://docs.python.org/2/library/2to3.html
[2]: http://www.diveintopython3.net/strings.html

Virtual OS trouble

While I was just getting ready to get started with the porting of sugar, the network configuration in my virtual OS (fedora) changed and I could not connect to the internet anymore. I looked for solutions online and even  asked on the vmware irc but still no solution . So, after a couple of days of trying everything from changing network setting in the vmware and fedora to doing a system restore, I still couldn't figure out the problem , so I finally decided to reinstall fedora. Yesterday , I reinstalled fedora and it's updates and also had to reinstall sugar. So, most of the time yesterday went in setting up the development environment again. :(