2to3 is a Python program that reads Python 2.x source code and applies a series
of fixers to transform it into valid Python 3.x code. The standard library
contains a rich set of fixers that will handle almost all code. 2to3 supporting
library lib2to3 is, however, a flexible and generic library, so it is
possible to write your own fixers for 2to3. lib2to3 could also be
adapted to custom applications in which Python code needs to be edited
automatically.
2to3 will usually be installed with the Python interpreter as a script. It is
also located in the Tools/scripts directory of the Python root.
2to3’s basic arguments are a list of files or directories to transform. The directories are recursively traversed for Python sources.
Here is a sample Python 2.x source file, example.py:
def greet(name):
print "Hello, {0}!".format(name)
print "What's your name?"
name = raw_input()
greet(name)
It can be converted to Python 3.x code via 2to3 on the command line:
$ 2to3 example.py
A diff against the original source file is printed. 2to3 can also write the
needed modifications right back to the source file. (A backup of the original
file is made unless -n is also given.) Writing the changes back is
enabled with the -w flag:
$ 2to3 -w example.py
After transformation, example.py looks like this:
def greet(name):
print("Hello, {0}!".format(name))
print("What's your name?")
name = input()
greet(name)
Comments and exact indentation are preserved throughout the translation process.
By default, 2to3 runs a set of predefined fixers. The
-l flag lists all available fixers. An explicit set of fixers to run
can be given with -f. Likewise the -x explicitly disables a
fixer. The following example runs only the imports and has_key fixers:
$ 2to3 -f imports -f has_key example.py
This command runs every fixer except the apply fixer:
$ 2to3 -x apply example.py
Some fixers are explicit, meaning they aren’t run by default and must be
listed on the command line to be run. Here, in addition to the default fixers,
the idioms fixer is run:
$ 2to3 -f all -f idioms example.py
Notice how passing all enables all default fixers.
Sometimes 2to3 will find a place in your source code that needs to be changed, but 2to3 cannot fix automatically. In this case, 2to3 will print a warning beneath the diff for a file. You should address the warning in order to have compliant 3.x code.
2to3 can also refactor doctests. To enable this mode, use the -d
flag. Note that only doctests will be refactored. This also doesn’t require
the module to be valid Python. For example, doctest like examples in a reST
document could also be refactored with this option.
The -v option enables output of more information on the translation
process.
Since some print statements can be parsed as function calls or statements, 2to3
cannot always read files containing the print function. When 2to3 detects the
presence of the from __future__ import print_function compiler directive, it
modifies its internal grammar to interpret print() as a function. This
change can also be enabled manually with the -p flag. Use
-p to run fixers on code that already has had its print statements
converted.
The -o or --output-dir option allows specification of an
alternate directory for processed output files to be written to. The
-n flag is required when using this as backup files do not make sense
when not overwriting the input files.
New in version 3.2.3: The -o option was added.
The -W or --write-unchanged-files flag tells 2to3 to always
write output files even if no changes were required to the file. This is most
useful with -o so that an entire Python source tree is copied with
translation from one directory to another.
This option implies the -w flag as it would not make sense otherwise.
New in version 3.2.3: The -W flag was added.
The --add-suffix option specifies a string to append to all output
filenames. The -n flag is required when specifying this as backups
are not necessary when writing to different filenames. Example:
$ 2to3 -n -W --add-suffix=3 example.py
Will cause a converted file named example.py3 to be written.
New in version 3.2.3: The --add-suffix option was added.
To translate an entire project from one directory tree to another use:
$ 2to3 --output-dir=python3-version/mycode -W -n python2-version/mycode
Each step of transforming code is encapsulated in a fixer. The command 2to3 -l lists them. As documented above, each can be turned on
and off individually. They are described here in more detail.
applyapply(). For example apply(function, *args, **kwargs) is converted to function(*args, **kwargs).assertsReplaces deprecated unittest method names with the correct ones.
From |
To |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
basestringbasestring to str.bufferbuffer to memoryview. This fixer is optional because the memoryview API is similar but not exactly the same as that of buffer.dictdict.iteritems() is converted to dict.items(), dict.iterkeys() to dict.keys(), and dict.itervalues() to dict.values(). Similarly, dict.viewitems(), dict.viewkeys() and dict.viewvalues() are converted respectively to dict.items(), dict.keys() and dict.values(). It also wraps existing usages of dict.items(), dict.keys(), and dict.values() in a call to list.exceptexcept X, T to except X as T.execexec statement to the exec() function.execfileexecfile(). The argument to execfile() is wrapped in calls to open(), compile(), and exec().exitfuncsys.exitfunc to use of the atexit module.funcattrsmy_function.func_closure is converted to my_function.__closure__.futurefrom __future__ import new_feature statements.getcwduos.getcwdu() to os.getcwd().has_keydict.has_key(key) to key in dict.idiomsThis optional fixer performs several transformations that make Python code
more idiomatic. Type comparisons like type(x) is SomeClass and
type(x) == SomeClass are converted to isinstance(x, SomeClass).
while 1 becomes while True. This fixer also tries to make use of
sorted() in appropriate places. For example, this block
L = list(some_iterable)
L.sort()
is changed to
L = sorted(some_iterable)
importimportsimports2imports fixer only because of technical limitations.inputinput(prompt) to eval(input(prompt)).internintern() to sys.intern().isinstanceisinstance(). For example, isinstance(x, (int, int)) is converted to isinstance(x, int) and isinstance(x, (int, float, int)) is converted to isinstance(x, (int, float)).itertools_importsitertools.ifilter(), itertools.izip(), and itertools.imap(). Imports of itertools.ifilterfalse() are also changed to itertools.filterfalse().itertoolsitertools.ifilter(), itertools.izip(), and itertools.imap() to their built-in equivalents. itertools.ifilterfalse() is changed to itertools.filterfalse().longlong to int.mapmap() in a list call. It also changes map(None, x) to list(x). Using from future_builtins import map disables this fixer.metaclass__metaclass__ = Meta in the class body) to the new (class X(metaclass=Meta)).methodattrsmeth.im_func is converted to meth.__func__.ne<>, to !=.nextnext() methods to the next() function. It also renames next() methods to __next__().nonzero__nonzero__() to __bool__().numliteralsoperatorConverts calls to various functions in the operator module to other,
but equivalent, function calls. When needed, the appropriate import
statements are added, e.g. import collections.abc. The following mapping
are made:
From |
To |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
paren[x for x in 1, 2] becomes [x for x in (1, 2)].printprint statement to the print() function.raiseraise E, V to raise E(V), and raise E, V, T to raise E(V).with_traceback(T). If E is a tuple, the translation will be incorrect because substituting tuples for exceptions has been removed in 3.0.raw_inputraw_input() to input().reducereduce() to functools.reduce().reloadreload() to importlib.reload().renamessys.maxint to sys.maxsize.reprrepr() function.set_literalset constructor with set literals. This fixer is optional.standarderrorStandardError to Exception.sys_excsys.exc_value, sys.exc_type, sys.exc_traceback to use sys.exc_info().throwthrow() method.tuple_paramstypestypes module.unicodeunicode to str.ws_commaxreadlinesfor x in file.xreadlines() to for x in file.zipzip() usage in a list call. This is disabled when from future_builtins import zip appears.