14
Is there a source code level debugger with breakpoints, single-stepping, etc.?
15
------------------------------------------------------------------------------
19
The pdb module is a simple but adequate console-mode debugger for Python. It is
20
part of the standard Python library, and is :mod:`documented in the Library
21
Reference Manual <pdb>`. You can also write your own debugger by using the code
22
for pdb as an example.
24
The IDLE interactive development environment, which is part of the standard
25
Python distribution (normally available as Tools/scripts/idle), includes a
26
graphical debugger. There is documentation for the IDLE debugger at
27
http://www.python.org/idle/doc/idle2.html#Debugger.
29
PythonWin is a Python IDE that includes a GUI debugger based on pdb. The
30
Pythonwin debugger colors breakpoints and has quite a few cool features such as
31
debugging non-Pythonwin programs. Pythonwin is available as part of the `Python
32
for Windows Extensions <http://sourceforge.net/projects/pywin32/>`__ project and
33
as a part of the ActivePython distribution (see
34
http://www.activestate.com/Products/ActivePython/index.html).
36
`Boa Constructor <http://boa-constructor.sourceforge.net/>`_ is an IDE and GUI
37
builder that uses wxWidgets. It offers visual frame creation and manipulation,
38
an object inspector, many views on the source like object browsers, inheritance
39
hierarchies, doc string generated html documentation, an advanced debugger,
40
integrated help, and Zope support.
42
`Eric <http://www.die-offenbachs.de/eric/index.html>`_ is an IDE built on PyQt
43
and the Scintilla editing component.
45
Pydb is a version of the standard Python debugger pdb, modified for use with DDD
46
(Data Display Debugger), a popular graphical debugger front end. Pydb can be
47
found at http://bashdb.sourceforge.net/pydb/ and DDD can be found at
48
http://www.gnu.org/software/ddd.
50
There are a number of commercial Python IDEs that include graphical debuggers.
53
* Wing IDE (http://wingware.com/)
54
* Komodo IDE (http://www.activestate.com/Products/Komodo)
57
Is there a tool to help find bugs or perform static analysis?
58
-------------------------------------------------------------
62
PyChecker is a static analysis tool that finds bugs in Python source code and
63
warns about code complexity and style. You can get PyChecker from
64
http://pychecker.sf.net.
66
`Pylint <http://www.logilab.org/projects/pylint>`_ is another tool that checks
67
if a module satisfies a coding standard, and also makes it possible to write
68
plug-ins to add a custom feature. In addition to the bug checking that
69
PyChecker performs, Pylint offers some additional features such as checking line
70
length, whether variable names are well-formed according to your coding
71
standard, whether declared interfaces are fully implemented, and more.
72
http://www.logilab.org/card/pylint_manual provides a full list of Pylint's
76
How can I create a stand-alone binary from a Python script?
77
-----------------------------------------------------------
79
You don't need the ability to compile Python to C code if all you want is a
80
stand-alone program that users can download and run without having to install
81
the Python distribution first. There are a number of tools that determine the
82
set of modules required by a program and bind these modules together with a
83
Python binary to produce a single executable.
85
One is to use the freeze tool, which is included in the Python source tree as
86
``Tools/freeze``. It converts Python byte code to C arrays; a C compiler you can
87
embed all your modules into a new program, which is then linked with the
88
standard Python modules.
90
It works by scanning your source recursively for import statements (in both
91
forms) and looking for the modules in the standard Python path as well as in the
92
source directory (for built-in modules). It then turns the bytecode for modules
93
written in Python into C code (array initializers that can be turned into code
94
objects using the marshal module) and creates a custom-made config file that
95
only contains those built-in modules which are actually used in the program. It
96
then compiles the generated C code and links it with the rest of the Python
97
interpreter to form a self-contained binary which acts exactly like your script.
99
Obviously, freeze requires a C compiler. There are several other utilities
100
which don't. One is Thomas Heller's py2exe (Windows only) at
102
http://www.py2exe.org/
104
Another is Christian Tismer's `SQFREEZE <http://starship.python.net/crew/pirx>`_
105
which appends the byte code to a specially-prepared Python interpreter that can
106
find the byte code in the executable.
108
Other tools include Fredrik Lundh's `Squeeze
109
<http://www.pythonware.com/products/python/squeeze>`_ and Anthony Tuininga's
110
`cx_Freeze <http://starship.python.net/crew/atuining/cx_Freeze/index.html>`_.
113
Are there coding standards or a style guide for Python programs?
114
----------------------------------------------------------------
116
Yes. The coding style required for standard library modules is documented as
123
Why am I getting an UnboundLocalError when the variable has a value?
124
--------------------------------------------------------------------
126
It can be a surprise to get the UnboundLocalError in previously working
127
code when it is modified by adding an assignment statement somewhere in
128
the body of a function.
138
works, but this code:
145
results in an UnboundLocalError:
148
Traceback (most recent call last):
150
UnboundLocalError: local variable 'x' referenced before assignment
152
This is because when you make an assignment to a variable in a scope, that
153
variable becomes local to that scope and shadows any similarly named variable
154
in the outer scope. Since the last statement in foo assigns a new value to
155
``x``, the compiler recognizes it as a local variable. Consequently when the
156
earlier ``print(x)`` attempts to print the uninitialized local variable and
159
In the example above you can access the outer scope variable by declaring it
170
This explicit declaration is required in order to remind you that (unlike the
171
superficially analogous situation with class and instance variables) you are
172
actually modifying the value of the variable in the outer scope:
177
You can do a similar thing in a nested scope using the :keyword:`nonlocal`
193
What are the rules for local and global variables in Python?
194
------------------------------------------------------------
196
In Python, variables that are only referenced inside a function are implicitly
197
global. If a variable is assigned a new value anywhere within the function's
198
body, it's assumed to be a local. If a variable is ever assigned a new value
199
inside the function, the variable is implicitly local, and you need to
200
explicitly declare it as 'global'.
202
Though a bit surprising at first, a moment's consideration explains this. On
203
one hand, requiring :keyword:`global` for assigned variables provides a bar
204
against unintended side-effects. On the other hand, if ``global`` was required
205
for all global references, you'd be using ``global`` all the time. You'd have
206
to declare as global every reference to a built-in function or to a component of
207
an imported module. This clutter would defeat the usefulness of the ``global``
208
declaration for identifying side-effects.
211
Why do lambdas defined in a loop with different values all return the same result?
212
----------------------------------------------------------------------------------
214
Assume you use a for loop to define a few different lambdas (or even plain
218
>>> for x in range(5):
219
... squares.append(lambda: x**2)
221
This gives you a list that contains 5 lambdas that calculate ``x**2``. You
222
might expect that, when called, they would return, respectively, ``0``, ``1``,
223
``4``, ``9``, and ``16``. However, when you actually try you will see that
224
they all return ``16``::
231
This happens because ``x`` is not local to the lambdas, but is defined in
232
the outer scope, and it is accessed when the lambda is called --- not when it
233
is defined. At the end of the loop, the value of ``x`` is ``4``, so all the
234
functions now return ``4**2``, i.e. ``16``. You can also verify this by
235
changing the value of ``x`` and see how the results of the lambdas change::
241
In order to avoid this, you need to save the values in variables local to the
242
lambdas, so that they don't rely on the value of the global ``x``::
245
>>> for x in range(5):
246
... squares.append(lambda n=x: n**2)
248
Here, ``n=x`` creates a new variable ``n`` local to the lambda and computed
249
when the lambda is defined so that it has the same value that ``x`` had at
250
that point in the loop. This means that the value of ``n`` will be ``0``
251
in the first lambda, ``1`` in the second, ``2`` in the third, and so on.
252
Therefore each lambda will now return the correct result::
259
Note that this behaviour is not peculiar to lambdas, but applies to regular
263
How do I share global variables across modules?
264
------------------------------------------------
266
The canonical way to share information across modules within a single program is
267
to create a special module (often called config or cfg). Just import the config
268
module in all modules of your application; the module then becomes available as
269
a global name. Because there is only one instance of each module, any changes
270
made to the module object get reflected everywhere. For example:
274
x = 0 # Default value of the 'x' configuration setting
287
Note that using a module is also the basis for implementing the Singleton design
288
pattern, for the same reason.
291
What are the "best practices" for using import in a module?
292
-----------------------------------------------------------
294
In general, don't use ``from modulename import *``. Doing so clutters the
295
importer's namespace. Some people avoid this idiom even with the few modules
296
that were designed to be imported in this manner. Modules designed in this
297
manner include :mod:`tkinter`, and :mod:`threading`.
299
Import modules at the top of a file. Doing so makes it clear what other modules
300
your code requires and avoids questions of whether the module name is in scope.
301
Using one import per line makes it easy to add and delete module imports, but
302
using multiple imports per line uses less screen space.
304
It's good practice if you import modules in the following order:
306
1. standard library modules -- e.g. ``sys``, ``os``, ``getopt``, ``re``
307
2. third-party library modules (anything installed in Python's site-packages
308
directory) -- e.g. mx.DateTime, ZODB, PIL.Image, etc.
309
3. locally-developed modules
311
Never use relative package imports. If you're writing code that's in the
312
``package.sub.m1`` module and want to import ``package.sub.m2``, do not just
313
write ``from . import m2``, even though it's legal. Write ``from package.sub
314
import m2`` instead. See :pep:`328` for details.
316
It is sometimes necessary to move imports to a function or class to avoid
317
problems with circular imports. Gordon McMillan says:
319
Circular imports are fine where both modules use the "import <module>" form
320
of import. They fail when the 2nd module wants to grab a name out of the
321
first ("from module import name") and the import is at the top level. That's
322
because names in the 1st are not yet available, because the first module is
323
busy importing the 2nd.
325
In this case, if the second module is only used in one function, then the import
326
can easily be moved into that function. By the time the import is called, the
327
first module will have finished initializing, and the second module can do its
330
It may also be necessary to move imports out of the top level of code if some of
331
the modules are platform-specific. In that case, it may not even be possible to
332
import all of the modules at the top of the file. In this case, importing the
333
correct modules in the corresponding platform-specific code is a good option.
335
Only move imports into a local scope, such as inside a function definition, if
336
it's necessary to solve a problem such as avoiding a circular import or are
337
trying to reduce the initialization time of a module. This technique is
338
especially helpful if many of the imports are unnecessary depending on how the
339
program executes. You may also want to move imports into a function if the
340
modules are only ever used in that function. Note that loading a module the
341
first time may be expensive because of the one time initialization of the
342
module, but loading a module multiple times is virtually free, costing only a
343
couple of dictionary lookups. Even if the module name has gone out of scope,
344
the module is probably available in :data:`sys.modules`.
346
If only instances of a specific class use a module, then it is reasonable to
347
import the module in the class's ``__init__`` method and then assign the module
348
to an instance variable so that the module is always available (via that
349
instance variable) during the life of the object. Note that to delay an import
350
until the class is instantiated, the import must be inside a method. Putting
351
the import inside the class but outside of any method still causes the import to
352
occur when the module is initialized.
355
How can I pass optional or keyword parameters from one function to another?
356
---------------------------------------------------------------------------
358
Collect the arguments using the ``*`` and ``**`` specifiers in the function's
359
parameter list; this gives you the positional arguments as a tuple and the
360
keyword arguments as a dictionary. You can then pass these arguments when
361
calling another function by using ``*`` and ``**``::
363
def f(x, *args, **kwargs):
365
kwargs['width'] = '14.3c'
367
g(x, *args, **kwargs)
371
single: argument; difference from parameter
372
single: parameter; difference from argument
374
.. _faq-argument-vs-parameter:
376
What is the difference between arguments and parameters?
377
--------------------------------------------------------
379
:term:`Parameters <parameter>` are defined by the names that appear in a
380
function definition, whereas :term:`arguments <argument>` are the values
381
actually passed to a function when calling it. Parameters define what types of
382
arguments a function can accept. For example, given the function definition::
384
def func(foo, bar=None, **kwargs):
387
*foo*, *bar* and *kwargs* are parameters of ``func``. However, when calling
388
``func``, for example::
390
func(42, bar=314, extra=somevar)
392
the values ``42``, ``314``, and ``somevar`` are arguments.
395
How do I write a function with output parameters (call by reference)?
396
---------------------------------------------------------------------
398
Remember that arguments are passed by assignment in Python. Since assignment
399
just creates references to objects, there's no alias between an argument name in
400
the caller and callee, and so no call-by-reference per se. You can achieve the
401
desired effect in a number of ways.
403
1) By returning a tuple of the results::
406
a = 'new-value' # a and b are local names
407
b = b + 1 # assigned to new objects
408
return a, b # return new values
410
x, y = 'old-value', 99
412
print(x, y) # output: new-value 100
414
This is almost always the clearest solution.
416
2) By using global variables. This isn't thread-safe, and is not recommended.
418
3) By passing a mutable (changeable in-place) object::
421
a[0] = 'new-value' # 'a' references a mutable list
422
a[1] = a[1] + 1 # changes a shared object
424
args = ['old-value', 99]
426
print(args[0], args[1]) # output: new-value 100
428
4) By passing in a dictionary that gets mutated::
431
args['a'] = 'new-value' # args is a mutable dictionary
432
args['b'] = args['b'] + 1 # change it in-place
434
args = {'a':' old-value', 'b': 99}
436
print(args['a'], args['b'])
438
5) Or bundle up values in a class instance::
441
def __init__(self, **args):
442
for (key, value) in args.items():
443
setattr(self, key, value)
446
args.a = 'new-value' # args is a mutable callByRef
447
args.b = args.b + 1 # change object in-place
449
args = callByRef(a='old-value', b=99)
451
print(args.a, args.b)
454
There's almost never a good reason to get this complicated.
456
Your best choice is to return a tuple containing the multiple results.
459
How do you make a higher order function in Python?
460
--------------------------------------------------
462
You have two choices: you can use nested scopes or you can use callable objects.
463
For example, suppose you wanted to define ``linear(a,b)`` which returns a
464
function ``f(x)`` that computes the value ``a*x+b``. Using nested scopes::
471
Or using a callable object::
475
def __init__(self, a, b):
476
self.a, self.b = a, b
478
def __call__(self, x):
479
return self.a * x + self.b
483
taxes = linear(0.3, 2)
485
gives a callable object where ``taxes(10e6) == 0.3 * 10e6 + 2``.
487
The callable object approach has the disadvantage that it is a bit slower and
488
results in slightly longer code. However, note that a collection of callables
489
can share their signature via inheritance::
491
class exponential(linear):
493
def __call__(self, x):
494
return self.a * (x ** self.b)
496
Object can encapsulate state for several methods::
506
self.value = self.value + 1
509
self.value = self.value - 1
512
inc, dec, reset = count.up, count.down, count.set
514
Here ``inc()``, ``dec()`` and ``reset()`` act like functions which share the
515
same counting variable.
518
How do I copy an object in Python?
519
----------------------------------
521
In general, try :func:`copy.copy` or :func:`copy.deepcopy` for the general case.
522
Not all objects can be copied, but most can.
524
Some objects can be copied more easily. Dictionaries have a :meth:`~dict.copy`
527
newdict = olddict.copy()
529
Sequences can be copied by slicing::
534
How can I find the methods or attributes of an object?
535
------------------------------------------------------
537
For an instance x of a user-defined class, ``dir(x)`` returns an alphabetized
538
list of the names containing the instance attributes and methods and attributes
539
defined by its class.
542
How can my code discover the name of an object?
543
-----------------------------------------------
545
Generally speaking, it can't, because objects don't really have names.
546
Essentially, assignment always binds a name to a value; The same is true of
547
``def`` and ``class`` statements, but in that case the value is a
548
callable. Consider the following code::
558
<__main__.A object at 0x16D07CC>
560
<__main__.A object at 0x16D07CC>
562
Arguably the class has a name: even though it is bound to two names and invoked
563
through the name B the created instance is still reported as an instance of
564
class A. However, it is impossible to say whether the instance's name is a or
565
b, since both names are bound to the same value.
567
Generally speaking it should not be necessary for your code to "know the names"
568
of particular values. Unless you are deliberately writing introspective
569
programs, this is usually an indication that a change of approach might be
572
In comp.lang.python, Fredrik Lundh once gave an excellent analogy in answer to
575
The same way as you get the name of that cat you found on your porch: the cat
576
(object) itself cannot tell you its name, and it doesn't really care -- so
577
the only way to find out what it's called is to ask all your neighbours
578
(namespaces) if it's their cat (object)...
580
....and don't be surprised if you'll find that it's known by many names, or
584
What's up with the comma operator's precedence?
585
-----------------------------------------------
587
Comma is not an operator in Python. Consider this session::
592
Since the comma is not an operator, but a separator between expressions the
593
above is evaluated as if you had entered::
601
The same is true of the various assignment operators (``=``, ``+=`` etc). They
602
are not truly operators but syntactic delimiters in assignment statements.
605
Is there an equivalent of C's "?:" ternary operator?
606
----------------------------------------------------
608
Yes, there is. The syntax is as follows::
610
[on_true] if [expression] else [on_false]
613
small = x if x < y else y
615
Before this syntax was introduced in Python 2.5, a common idiom was to use
618
[expression] and [on_true] or [on_false]
620
However, this idiom is unsafe, as it can give wrong results when *on_true*
621
has a false boolean value. Therefore, it is always better to use
622
the ``... if ... else ...`` form.
625
Is it possible to write obfuscated one-liners in Python?
626
--------------------------------------------------------
628
Yes. Usually this is done by nesting :keyword:`lambda` within
629
:keyword:`lambda`. See the following three examples, due to Ulf Bartelt::
631
from functools import reduce
634
print(list(filter(None,map(lambda y:y*reduce(lambda x,y:x*y!=0,
635
map(lambda x,y=y:y%x,range(2,int(pow(y,0.5)+1))),1),range(2,1000)))))
637
# First 10 Fibonacci numbers
638
print(list(map(lambda x,f=lambda x,f:(f(x-1,f)+f(x-2,f)) if x>1 else 1:
642
print((lambda Ru,Ro,Iu,Io,IM,Sx,Sy:reduce(lambda x,y:x+y,map(lambda y,
643
Iu=Iu,Io=Io,Ru=Ru,Ro=Ro,Sy=Sy,L=lambda yc,Iu=Iu,Io=Io,Ru=Ru,Ro=Ro,i=IM,
644
Sx=Sx,Sy=Sy:reduce(lambda x,y:x+y,map(lambda x,xc=Ru,yc=yc,Ru=Ru,Ro=Ro,
645
i=i,Sx=Sx,F=lambda xc,yc,x,y,k,f=lambda xc,yc,x,y,k,f:(k<=0)or (x*x+y*y
646
>=4.0) or 1+f(xc,yc,x*x-y*y+xc,2.0*x*y+yc,k-1,f):f(xc,yc,x,y,k,f):chr(
647
64+F(Ru+x*(Ro-Ru)/Sx,yc,0,0,i)),range(Sx))):L(Iu+y*(Io-Iu)/Sy),range(Sy
648
))))(-2.1, 0.7, -1.2, 1.2, 30, 80, 24))
649
# \___ ___/ \___ ___/ | | |__ lines on screen
650
# V V | |______ columns on screen
651
# | | |__________ maximum of "iterations"
652
# | |_________________ range on y axis
653
# |____________________________ range on x axis
655
Don't try this at home, kids!
661
How do I specify hexadecimal and octal integers?
662
------------------------------------------------
664
To specify an octal digit, precede the octal value with a zero, and then a lower
665
or uppercase "o". For example, to set the variable "a" to the octal value "10"
666
(8 in decimal), type::
672
Hexadecimal is just as easy. Simply precede the hexadecimal number with a zero,
673
and then a lower or uppercase "x". Hexadecimal digits can be specified in lower
674
or uppercase. For example, in the Python interpreter::
684
Why does -22 // 10 return -3?
685
-----------------------------
687
It's primarily driven by the desire that ``i % j`` have the same sign as ``j``.
688
If you want that, and also want::
690
i == (i // j) * j + (i % j)
692
then integer division has to return the floor. C also requires that identity to
693
hold, and then compilers that truncate ``i // j`` need to make ``i % j`` have
694
the same sign as ``i``.
696
There are few real use cases for ``i % j`` when ``j`` is negative. When ``j``
697
is positive, there are many, and in virtually all of them it's more useful for
698
``i % j`` to be ``>= 0``. If the clock says 10 now, what did it say 200 hours
699
ago? ``-190 % 12 == 2`` is useful; ``-190 % 12 == -10`` is a bug waiting to
703
How do I convert a string to a number?
704
--------------------------------------
706
For integers, use the built-in :func:`int` type constructor, e.g. ``int('144')
707
== 144``. Similarly, :func:`float` converts to floating-point,
708
e.g. ``float('144') == 144.0``.
710
By default, these interpret the number as decimal, so that ``int('0144') ==
711
144`` and ``int('0x144')`` raises :exc:`ValueError`. ``int(string, base)`` takes
712
the base to convert from as a second optional argument, so ``int('0x144', 16) ==
713
324``. If the base is specified as 0, the number is interpreted using Python's
714
rules: a leading '0' indicates octal, and '0x' indicates a hex number.
716
Do not use the built-in function :func:`eval` if all you need is to convert
717
strings to numbers. :func:`eval` will be significantly slower and it presents a
718
security risk: someone could pass you a Python expression that might have
719
unwanted side effects. For example, someone could pass
720
``__import__('os').system("rm -rf $HOME")`` which would erase your home
723
:func:`eval` also has the effect of interpreting numbers as Python expressions,
724
so that e.g. ``eval('09')`` gives a syntax error because Python does not allow
725
leading '0' in a decimal number (except '0').
728
How do I convert a number to a string?
729
--------------------------------------
731
To convert, e.g., the number 144 to the string '144', use the built-in type
732
constructor :func:`str`. If you want a hexadecimal or octal representation, use
733
the built-in functions :func:`hex` or :func:`oct`. For fancy formatting, see
734
the :ref:`string-formatting` section, e.g. ``"{:04d}".format(144)`` yields
735
``'0144'`` and ``"{:.3f}".format(1/3)`` yields ``'0.333'``.
738
How do I modify a string in place?
739
----------------------------------
741
You can't, because strings are immutable. In most situations, you should
742
simply construct a new string from the various parts you want to assemble
743
it from. However, if you need an object with the ability to modify in-place
744
unicode data, try using a :class:`io.StringIO` object or the :mod:`array`
748
>>> s = "Hello, world"
749
>>> sio = io.StringIO(s)
754
>>> sio.write("there!")
760
>>> a = array.array('u', s)
762
array('u', 'Hello, world')
765
array('u', 'yello, world')
770
How do I use strings to call functions/methods?
771
-----------------------------------------------
773
There are various techniques.
775
* The best is to use a dictionary that maps strings to functions. The primary
776
advantage of this technique is that the strings do not need to match the names
777
of the functions. This is also the primary technique used to emulate a case
786
dispatch = {'go': a, 'stop': b} # Note lack of parens for funcs
788
dispatch[get_input()]() # Note trailing parens to call function
790
* Use the built-in function :func:`getattr`::
793
getattr(foo, 'bar')()
795
Note that :func:`getattr` works on any object, including classes, class
796
instances, modules, and so on.
798
This is used in several places in the standard library, like this::
807
f = getattr(foo_instance, 'do_' + opname)
811
* Use :func:`locals` or :func:`eval` to resolve the function name::
824
Note: Using :func:`eval` is slow and dangerous. If you don't have absolute
825
control over the contents of the string, someone could pass a string that
826
resulted in an arbitrary function being executed.
828
Is there an equivalent to Perl's chomp() for removing trailing newlines from strings?
829
-------------------------------------------------------------------------------------
831
You can use ``S.rstrip("\r\n")`` to remove all occurrences of any line
832
terminator from the end of the string ``S`` without removing other trailing
833
whitespace. If the string ``S`` represents more than one line, with several
834
empty lines at the end, the line terminators for all the blank lines will
837
>>> lines = ("line 1 \r\n"
840
>>> lines.rstrip("\n\r")
843
Since this is typically only desired when reading text one line at a time, using
844
``S.rstrip()`` this way works well.
847
Is there a scanf() or sscanf() equivalent?
848
------------------------------------------
852
For simple input parsing, the easiest approach is usually to split the line into
853
whitespace-delimited words using the :meth:`~str.split` method of string objects
854
and then convert decimal strings to numeric values using :func:`int` or
855
:func:`float`. ``split()`` supports an optional "sep" parameter which is useful
856
if the line uses something other than whitespace as a separator.
858
For more complicated input parsing, regular expressions are more powerful
859
than C's :c:func:`sscanf` and better suited for the task.
862
What does 'UnicodeDecodeError' or 'UnicodeEncodeError' error mean?
863
-------------------------------------------------------------------
865
See the :ref:`unicode-howto`.
871
My program is too slow. How do I speed it up?
872
---------------------------------------------
874
That's a tough one, in general. First, here are a list of things to
875
remember before diving further:
877
* Performance characteristics vary across Python implementations. This FAQ
878
focusses on :term:`CPython`.
879
* Behaviour can vary across operating systems, especially when talking about
880
I/O or multi-threading.
881
* You should always find the hot spots in your program *before* attempting to
882
optimize any code (see the :mod:`profile` module).
883
* Writing benchmark scripts will allow you to iterate quickly when searching
884
for improvements (see the :mod:`timeit` module).
885
* It is highly recommended to have good code coverage (through unit testing
886
or any other technique) before potentially introducing regressions hidden
887
in sophisticated optimizations.
889
That being said, there are many tricks to speed up Python code. Here are
890
some general principles which go a long way towards reaching acceptable
893
* Making your algorithms faster (or changing to faster ones) can yield
894
much larger benefits than trying to sprinkle micro-optimization tricks
897
* Use the right data structures. Study documentation for the :ref:`bltin-types`
898
and the :mod:`collections` module.
900
* When the standard library provides a primitive for doing something, it is
901
likely (although not guaranteed) to be faster than any alternative you
902
may come up with. This is doubly true for primitives written in C, such
903
as builtins and some extension types. For example, be sure to use
904
either the :meth:`list.sort` built-in method or the related :func:`sorted`
905
function to do sorting (and see the
906
`sorting mini-HOWTO <http://wiki.python.org/moin/HowTo/Sorting>`_ for examples
907
of moderately advanced usage).
909
* Abstractions tend to create indirections and force the interpreter to work
910
more. If the levels of indirection outweigh the amount of useful work
911
done, your program will be slower. You should avoid excessive abstraction,
912
especially under the form of tiny functions or methods (which are also often
913
detrimental to readability).
915
If you have reached the limit of what pure Python can allow, there are tools
916
to take you further away. For example, `Cython <http://cython.org>`_ can
917
compile a slightly modified version of Python code into a C extension, and
918
can be used on many different platforms. Cython can take advantage of
919
compilation (and optional type annotations) to make your code significantly
920
faster than when interpreted. If you are confident in your C programming
921
skills, you can also :ref:`write a C extension module <extending-index>`
925
The wiki page devoted to `performance tips
926
<http://wiki.python.org/moin/PythonSpeed/PerformanceTips>`_.
928
.. _efficient_string_concatenation:
930
What is the most efficient way to concatenate many strings together?
931
--------------------------------------------------------------------
933
:class:`str` and :class:`bytes` objects are immutable, therefore concatenating
934
many strings together is inefficient as each concatenation creates a new
935
object. In the general case, the total runtime cost is quadratic in the
938
To accumulate many :class:`str` objects, the recommended idiom is to place
939
them into a list and call :meth:`str.join` at the end::
944
result = ''.join(chunks)
946
(another reasonably efficient idiom is to use :class:`io.StringIO`)
948
To accumulate many :class:`bytes` objects, the recommended idiom is to extend
949
a :class:`bytearray` object using in-place concatenation (the ``+=`` operator)::
952
for b in my_bytes_objects:
956
Sequences (Tuples/Lists)
957
========================
959
How do I convert between tuples and lists?
960
------------------------------------------
962
The type constructor ``tuple(seq)`` converts any sequence (actually, any
963
iterable) into a tuple with the same items in the same order.
965
For example, ``tuple([1, 2, 3])`` yields ``(1, 2, 3)`` and ``tuple('abc')``
966
yields ``('a', 'b', 'c')``. If the argument is a tuple, it does not make a copy
967
but returns the same object, so it is cheap to call :func:`tuple` when you
968
aren't sure that an object is already a tuple.
970
The type constructor ``list(seq)`` converts any sequence or iterable into a list
971
with the same items in the same order. For example, ``list((1, 2, 3))`` yields
972
``[1, 2, 3]`` and ``list('abc')`` yields ``['a', 'b', 'c']``. If the argument
973
is a list, it makes a copy just like ``seq[:]`` would.
976
What's a negative index?
977
------------------------
979
Python sequences are indexed with positive numbers and negative numbers. For
980
positive numbers 0 is the first index 1 is the second index and so forth. For
981
negative indices -1 is the last index and -2 is the penultimate (next to last)
982
index and so forth. Think of ``seq[-n]`` as the same as ``seq[len(seq)-n]``.
984
Using negative indices can be very convenient. For example ``S[:-1]`` is all of
985
the string except for its last character, which is useful for removing the
986
trailing newline from a string.
989
How do I iterate over a sequence in reverse order?
990
--------------------------------------------------
992
Use the :func:`reversed` built-in function, which is new in Python 2.4::
994
for x in reversed(sequence):
995
... # do something with x...
997
This won't touch your original sequence, but build a new copy with reversed
998
order to iterate over.
1000
With Python 2.3, you can use an extended slice syntax::
1002
for x in sequence[::-1]:
1003
... # do something with x...
1006
How do you remove duplicates from a list?
1007
-----------------------------------------
1009
See the Python Cookbook for a long discussion of many ways to do this:
1011
http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/52560
1013
If you don't mind reordering the list, sort it and then scan from the end of the
1014
list, deleting duplicates as you go::
1019
for i in range(len(mylist)-2, -1, -1):
1020
if last == mylist[i]:
1025
If all elements of the list may be used as set keys (i.e. they are all
1026
:term:`hashable`) this is often faster ::
1028
mylist = list(set(mylist))
1030
This converts the list into a set, thereby removing duplicates, and then back
1034
How do you make an array in Python?
1035
-----------------------------------
1039
["this", 1, "is", "an", "array"]
1041
Lists are equivalent to C or Pascal arrays in their time complexity; the primary
1042
difference is that a Python list can contain objects of many different types.
1044
The ``array`` module also provides methods for creating arrays of fixed types
1045
with compact representations, but they are slower to index than lists. Also
1046
note that the Numeric extensions and others define array-like structures with
1047
various characteristics as well.
1049
To get Lisp-style linked lists, you can emulate cons cells using tuples::
1051
lisp_list = ("like", ("this", ("example", None) ) )
1053
If mutability is desired, you could use lists instead of tuples. Here the
1054
analogue of lisp car is ``lisp_list[0]`` and the analogue of cdr is
1055
``lisp_list[1]``. Only do this if you're sure you really need to, because it's
1056
usually a lot slower than using Python lists.
1059
How do I create a multidimensional list?
1060
----------------------------------------
1062
You probably tried to make a multidimensional array like this::
1064
>>> A = [[None] * 2] * 3
1066
This looks correct if you print it::
1069
[[None, None], [None, None], [None, None]]
1071
But when you assign a value, it shows up in multiple places:
1075
[[5, None], [5, None], [5, None]]
1077
The reason is that replicating a list with ``*`` doesn't create copies, it only
1078
creates references to the existing objects. The ``*3`` creates a list
1079
containing 3 references to the same list of length two. Changes to one row will
1080
show in all rows, which is almost certainly not what you want.
1082
The suggested approach is to create a list of the desired length first and then
1083
fill in each element with a newly created list::
1089
This generates a list containing 3 different lists of length two. You can also
1090
use a list comprehension::
1093
A = [[None] * w for i in range(h)]
1095
Or, you can use an extension that provides a matrix datatype; `Numeric Python
1096
<http://www.numpy.org/>`_ is the best known.
1099
How do I apply a method to a sequence of objects?
1100
-------------------------------------------------
1102
Use a list comprehension::
1104
result = [obj.method() for obj in mylist]
1107
Why does a_tuple[i] += ['item'] raise an exception when the addition works?
1108
---------------------------------------------------------------------------
1110
This is because of a combination of the fact that augmented assignment
1111
operators are *assignment* operators, and the difference between mutable and
1112
immutable objects in Python.
1114
This discussion applies in general when augmented assignment operators are
1115
applied to elements of a tuple that point to mutable objects, but we'll use
1116
a ``list`` and ``+=`` as our exemplar.
1120
>>> a_tuple = (1, 2)
1122
Traceback (most recent call last):
1124
TypeError: 'tuple' object does not support item assignment
1126
The reason for the exception should be immediately clear: ``1`` is added to the
1127
object ``a_tuple[0]`` points to (``1``), producing the result object, ``2``,
1128
but when we attempt to assign the result of the computation, ``2``, to element
1129
``0`` of the tuple, we get an error because we can't change what an element of
1132
Under the covers, what this augmented assignment statement is doing is
1133
approximately this::
1135
>>> result = a_tuple[0] + 1
1136
>>> a_tuple[0] = result
1137
Traceback (most recent call last):
1139
TypeError: 'tuple' object does not support item assignment
1141
It is the assignment part of the operation that produces the error, since a
1144
When you write something like::
1146
>>> a_tuple = (['foo'], 'bar')
1147
>>> a_tuple[0] += ['item']
1148
Traceback (most recent call last):
1150
TypeError: 'tuple' object does not support item assignment
1152
The exception is a bit more surprising, and even more surprising is the fact
1153
that even though there was an error, the append worked::
1158
To see why this happens, you need to know that (a) if an object implements an
1159
``__iadd__`` magic method, it gets called when the ``+=`` augmented assignment
1160
is executed, and its return value is what gets used in the assignment statement;
1161
and (b) for lists, ``__iadd__`` is equivalent to calling ``extend`` on the list
1162
and returning the list. That's why we say that for lists, ``+=`` is a
1163
"shorthand" for ``list.extend``::
1170
This is equivalent to::
1172
>>> result = a_list.__iadd__([1])
1175
The object pointed to by a_list has been mutated, and the pointer to the
1176
mutated object is assigned back to ``a_list``. The end result of the
1177
assignment is a no-op, since it is a pointer to the same object that ``a_list``
1178
was previously pointing to, but the assignment still happens.
1180
Thus, in our tuple example what is happening is equivalent to::
1182
>>> result = a_tuple[0].__iadd__(['item'])
1183
>>> a_tuple[0] = result
1184
Traceback (most recent call last):
1186
TypeError: 'tuple' object does not support item assignment
1188
The ``__iadd__`` succeeds, and thus the list is extended, but even though
1189
``result`` points to the same object that ``a_tuple[0]`` already points to,
1190
that final assignment still results in an error, because tuples are immutable.
1196
How can I get a dictionary to display its keys in a consistent order?
1197
---------------------------------------------------------------------
1199
You can't. Dictionaries store their keys in an unpredictable order, so the
1200
display order of a dictionary's elements will be similarly unpredictable.
1202
This can be frustrating if you want to save a printable version to a file, make
1203
some changes and then compare it with some other printed dictionary. In this
1204
case, use the ``pprint`` module to pretty-print the dictionary; the items will
1205
be presented in order sorted by the key.
1207
A more complicated solution is to subclass ``dict`` to create a
1208
``SortedDict`` class that prints itself in a predictable order. Here's one
1209
simpleminded implementation of such a class::
1211
class SortedDict(dict):
1213
keys = sorted(self.keys())
1214
result = ("{!r}: {!r}".format(k, self[k]) for k in keys)
1215
return "{{{}}}".format(", ".join(result))
1219
This will work for many common situations you might encounter, though it's far
1220
from a perfect solution. The largest flaw is that if some values in the
1221
dictionary are also dictionaries, their values won't be presented in any
1225
I want to do a complicated sort: can you do a Schwartzian Transform in Python?
1226
------------------------------------------------------------------------------
1228
The technique, attributed to Randal Schwartz of the Perl community, sorts the
1229
elements of a list by a metric which maps each element to its "sort value". In
1230
Python, just use the ``key`` argument for the ``sort()`` method::
1233
Isorted.sort(key=lambda s: int(s[10:15]))
1235
The ``key`` argument is new in Python 2.4, for older versions this kind of
1236
sorting is quite simple to do with list comprehensions. To sort a list of
1237
strings by their uppercase values::
1239
tmp1 = [(x.upper(), x) for x in L] # Schwartzian transform
1241
Usorted = [x[1] for x in tmp1]
1243
To sort by the integer value of a subfield extending from positions 10-15 in
1246
tmp2 = [(int(s[10:15]), s) for s in L] # Schwartzian transform
1248
Isorted = [x[1] for x in tmp2]
1250
For versions prior to 3.0, Isorted may also be computed by ::
1253
return int(s[10:15])
1256
return cmp(intfield(s1), intfield(s2))
1261
but since this method calls ``intfield()`` many times for each element of L, it
1262
is slower than the Schwartzian Transform.
1265
How can I sort one list by values from another list?
1266
----------------------------------------------------
1268
Merge them into an iterator of tuples, sort the resulting list, and then pick
1269
out the element you want. ::
1271
>>> list1 = ["what", "I'm", "sorting", "by"]
1272
>>> list2 = ["something", "else", "to", "sort"]
1273
>>> pairs = zip(list1, list2)
1274
>>> pairs = sorted(pairs)
1276
[("I'm", 'else'), ('by', 'sort'), ('sorting', 'to'), ('what', 'something')]
1277
>>> result = [x[1] for x in pairs]
1279
['else', 'sort', 'to', 'something']
1282
An alternative for the last step is::
1285
>>> for p in pairs: result.append(p[1])
1287
If you find this more legible, you might prefer to use this instead of the final
1288
list comprehension. However, it is almost twice as slow for long lists. Why?
1289
First, the ``append()`` operation has to reallocate memory, and while it uses
1290
some tricks to avoid doing that each time, it still has to do it occasionally,
1291
and that costs quite a bit. Second, the expression "result.append" requires an
1292
extra attribute lookup, and third, there's a speed reduction from having to make
1293
all those function calls.
1302
A class is the particular object type created by executing a class statement.
1303
Class objects are used as templates to create instance objects, which embody
1304
both the data (attributes) and code (methods) specific to a datatype.
1306
A class can be based on one or more other classes, called its base class(es). It
1307
then inherits the attributes and methods of its base classes. This allows an
1308
object model to be successively refined by inheritance. You might have a
1309
generic ``Mailbox`` class that provides basic accessor methods for a mailbox,
1310
and subclasses such as ``MboxMailbox``, ``MaildirMailbox``, ``OutlookMailbox``
1311
that handle various specific mailbox formats.
1317
A method is a function on some object ``x`` that you normally call as
1318
``x.name(arguments...)``. Methods are defined as functions inside the class
1322
def meth (self, arg):
1323
return arg * 2 + self.attribute
1329
Self is merely a conventional name for the first argument of a method. A method
1330
defined as ``meth(self, a, b, c)`` should be called as ``x.meth(a, b, c)`` for
1331
some instance ``x`` of the class in which the definition occurs; the called
1332
method will think it is called as ``meth(x, a, b, c)``.
1334
See also :ref:`why-self`.
1337
How do I check if an object is an instance of a given class or of a subclass of it?
1338
-----------------------------------------------------------------------------------
1340
Use the built-in function ``isinstance(obj, cls)``. You can check if an object
1341
is an instance of any of a number of classes by providing a tuple instead of a
1342
single class, e.g. ``isinstance(obj, (class1, class2, ...))``, and can also
1343
check whether an object is one of Python's built-in types, e.g.
1344
``isinstance(obj, str)`` or ``isinstance(obj, (int, float, complex))``.
1346
Note that most programs do not use :func:`isinstance` on user-defined classes
1347
very often. If you are developing the classes yourself, a more proper
1348
object-oriented style is to define methods on the classes that encapsulate a
1349
particular behaviour, instead of checking the object's class and doing a
1350
different thing based on what class it is. For example, if you have a function
1351
that does something::
1354
if isinstance(obj, Mailbox):
1355
# ... code to search a mailbox
1356
elif isinstance(obj, Document):
1357
# ... code to search a document
1360
A better approach is to define a ``search()`` method on all the classes and just
1365
# ... code to search a mailbox
1369
# ... code to search a document
1377
Delegation is an object oriented technique (also called a design pattern).
1378
Let's say you have an object ``x`` and want to change the behaviour of just one
1379
of its methods. You can create a new class that provides a new implementation
1380
of the method you're interested in changing and delegates all other methods to
1381
the corresponding method of ``x``.
1383
Python programmers can easily implement delegation. For example, the following
1384
class implements a class that behaves like a file but converts all written data
1389
def __init__(self, outfile):
1390
self._outfile = outfile
1393
self._outfile.write(s.upper())
1395
def __getattr__(self, name):
1396
return getattr(self._outfile, name)
1398
Here the ``UpperOut`` class redefines the ``write()`` method to convert the
1399
argument string to uppercase before calling the underlying
1400
``self.__outfile.write()`` method. All other methods are delegated to the
1401
underlying ``self.__outfile`` object. The delegation is accomplished via the
1402
``__getattr__`` method; consult :ref:`the language reference <attribute-access>`
1403
for more information about controlling attribute access.
1405
Note that for more general cases delegation can get trickier. When attributes
1406
must be set as well as retrieved, the class must define a :meth:`__setattr__`
1407
method too, and it must do so carefully. The basic implementation of
1408
:meth:`__setattr__` is roughly equivalent to the following::
1412
def __setattr__(self, name, value):
1413
self.__dict__[name] = value
1416
Most :meth:`__setattr__` implementations must modify ``self.__dict__`` to store
1417
local state for self without causing an infinite recursion.
1420
How do I call a method defined in a base class from a derived class that overrides it?
1421
--------------------------------------------------------------------------------------
1423
Use the built-in :func:`super` function::
1425
class Derived(Base):
1427
super(Derived, self).meth()
1429
For version prior to 3.0, you may be using classic classes: For a class
1430
definition such as ``class Derived(Base): ...`` you can call method ``meth()``
1431
defined in ``Base`` (or one of ``Base``'s base classes) as ``Base.meth(self,
1432
arguments...)``. Here, ``Base.meth`` is an unbound method, so you need to
1433
provide the ``self`` argument.
1436
How can I organize my code to make it easier to change the base class?
1437
----------------------------------------------------------------------
1439
You could define an alias for the base class, assign the real base class to it
1440
before your class definition, and use the alias throughout your class. Then all
1441
you have to change is the value assigned to the alias. Incidentally, this trick
1442
is also handy if you want to decide dynamically (e.g. depending on availability
1443
of resources) which base class to use. Example::
1445
BaseAlias = <real base class>
1447
class Derived(BaseAlias):
1449
BaseAlias.meth(self)
1453
How do I create static class data and static class methods?
1454
-----------------------------------------------------------
1456
Both static data and static methods (in the sense of C++ or Java) are supported
1459
For static data, simply define a class attribute. To assign a new value to the
1460
attribute, you have to explicitly use the class name in the assignment::
1463
count = 0 # number of times C.__init__ called
1466
C.count = C.count + 1
1469
return C.count # or return self.count
1471
``c.count`` also refers to ``C.count`` for any ``c`` such that ``isinstance(c,
1472
C)`` holds, unless overridden by ``c`` itself or by some class on the base-class
1473
search path from ``c.__class__`` back to ``C``.
1475
Caution: within a method of C, an assignment like ``self.count = 42`` creates a
1476
new and unrelated instance named "count" in ``self``'s own dict. Rebinding of a
1477
class-static data name must always specify the class whether inside a method or
1482
Static methods are possible::
1486
def static(arg1, arg2, arg3):
1487
# No 'self' parameter!
1490
However, a far more straightforward way to get the effect of a static method is
1491
via a simple module-level function::
1496
If your code is structured so as to define one class (or tightly related class
1497
hierarchy) per module, this supplies the desired encapsulation.
1500
How can I overload constructors (or methods) in Python?
1501
-------------------------------------------------------
1503
This answer actually applies to all methods, but the question usually comes up
1504
first in the context of constructors.
1511
C() { cout << "No arguments\n"; }
1512
C(int i) { cout << "Argument is " << i << "\n"; }
1515
In Python you have to write a single constructor that catches all cases using
1516
default arguments. For example::
1519
def __init__(self, i=None):
1521
print("No arguments")
1523
print("Argument is", i)
1525
This is not entirely equivalent, but close enough in practice.
1527
You could also try a variable-length argument list, e.g. ::
1529
def __init__(self, *args):
1532
The same approach works for all method definitions.
1535
I try to use __spam and I get an error about _SomeClassName__spam.
1536
------------------------------------------------------------------
1538
Variable names with double leading underscores are "mangled" to provide a simple
1539
but effective way to define class private variables. Any identifier of the form
1540
``__spam`` (at least two leading underscores, at most one trailing underscore)
1541
is textually replaced with ``_classname__spam``, where ``classname`` is the
1542
current class name with any leading underscores stripped.
1544
This doesn't guarantee privacy: an outside user can still deliberately access
1545
the "_classname__spam" attribute, and private values are visible in the object's
1546
``__dict__``. Many Python programmers never bother to use private variable
1550
My class defines __del__ but it is not called when I delete the object.
1551
-----------------------------------------------------------------------
1553
There are several possible reasons for this.
1555
The del statement does not necessarily call :meth:`__del__` -- it simply
1556
decrements the object's reference count, and if this reaches zero
1557
:meth:`__del__` is called.
1559
If your data structures contain circular links (e.g. a tree where each child has
1560
a parent reference and each parent has a list of children) the reference counts
1561
will never go back to zero. Once in a while Python runs an algorithm to detect
1562
such cycles, but the garbage collector might run some time after the last
1563
reference to your data structure vanishes, so your :meth:`__del__` method may be
1564
called at an inconvenient and random time. This is inconvenient if you're trying
1565
to reproduce a problem. Worse, the order in which object's :meth:`__del__`
1566
methods are executed is arbitrary. You can run :func:`gc.collect` to force a
1567
collection, but there *are* pathological cases where objects will never be
1570
Despite the cycle collector, it's still a good idea to define an explicit
1571
``close()`` method on objects to be called whenever you're done with them. The
1572
``close()`` method can then remove attributes that refer to subobjecs. Don't
1573
call :meth:`__del__` directly -- :meth:`__del__` should call ``close()`` and
1574
``close()`` should make sure that it can be called more than once for the same
1577
Another way to avoid cyclical references is to use the :mod:`weakref` module,
1578
which allows you to point to objects without incrementing their reference count.
1579
Tree data structures, for instance, should use weak references for their parent
1580
and sibling references (if they need them!).
1582
.. XXX relevant for Python 3?
1584
If the object has ever been a local variable in a function that caught an
1585
expression in an except clause, chances are that a reference to the object
1586
still exists in that function's stack frame as contained in the stack trace.
1587
Normally, calling :func:`sys.exc_clear` will take care of this by clearing
1588
the last recorded exception.
1590
Finally, if your :meth:`__del__` method raises an exception, a warning message
1591
is printed to :data:`sys.stderr`.
1594
How do I get a list of all instances of a given class?
1595
------------------------------------------------------
1597
Python does not keep track of all instances of a class (or of a built-in type).
1598
You can program the class's constructor to keep track of all instances by
1599
keeping a list of weak references to each instance.
1602
Why does the result of ``id()`` appear to be not unique?
1603
--------------------------------------------------------
1605
The :func:`id` builtin returns an integer that is guaranteed to be unique during
1606
the lifetime of the object. Since in CPython, this is the object's memory
1607
address, it happens frequently that after an object is deleted from memory, the
1608
next freshly created object is allocated at the same position in memory. This
1609
is illustrated by this example:
1616
The two ids belong to different integer objects that are created before, and
1617
deleted immediately after execution of the ``id()`` call. To be sure that
1618
objects whose id you want to examine are still alive, create another reference
1621
>>> a = 1000; b = 2000
1631
How do I create a .pyc file?
1632
----------------------------
1634
When a module is imported for the first time (or when the source is more recent
1635
than the current compiled file) a ``.pyc`` file containing the compiled code
1636
should be created in the same directory as the ``.py`` file.
1638
One reason that a ``.pyc`` file may not be created is permissions problems with
1639
the directory. This can happen, for example, if you develop as one user but run
1640
as another, such as if you are testing with a web server. Creation of a .pyc
1641
file is automatic if you're importing a module and Python has the ability
1642
(permissions, free space, etc...) to write the compiled module back to the
1645
Running Python on a top level script is not considered an import and no
1646
``.pyc`` will be created. For example, if you have a top-level module
1647
``foo.py`` that imports another module ``xyz.py``, when you run ``foo``,
1648
``xyz.pyc`` will be created since ``xyz`` is imported, but no ``foo.pyc`` file
1649
will be created since ``foo.py`` isn't being imported.
1651
If you need to create ``foo.pyc`` -- that is, to create a ``.pyc`` file for a module
1652
that is not imported -- you can, using the :mod:`py_compile` and
1653
:mod:`compileall` modules.
1655
The :mod:`py_compile` module can manually compile any module. One way is to use
1656
the ``compile()`` function in that module interactively::
1658
>>> import py_compile
1659
>>> py_compile.compile('foo.py') # doctest: +SKIP
1661
This will write the ``.pyc`` to the same location as ``foo.py`` (or you can
1662
override that with the optional parameter ``cfile``).
1664
You can also automatically compile all files in a directory or directories using
1665
the :mod:`compileall` module. You can do it from the shell prompt by running
1666
``compileall.py`` and providing the path of a directory containing Python files
1669
python -m compileall .
1672
How do I find the current module name?
1673
--------------------------------------
1675
A module can find out its own module name by looking at the predefined global
1676
variable ``__name__``. If this has the value ``'__main__'``, the program is
1677
running as a script. Many modules that are usually used by importing them also
1678
provide a command-line interface or a self-test, and only execute this code
1679
after checking ``__name__``::
1682
print('Running test...')
1685
if __name__ == '__main__':
1689
How can I have modules that mutually import each other?
1690
-------------------------------------------------------
1692
Suppose you have the following modules:
1696
from bar import bar_var
1701
from foo import foo_var
1704
The problem is that the interpreter will perform the following steps:
1707
* Empty globals for foo are created
1708
* foo is compiled and starts executing
1710
* Empty globals for bar are created
1711
* bar is compiled and starts executing
1712
* bar imports foo (which is a no-op since there already is a module named foo)
1713
* bar.foo_var = foo.foo_var
1715
The last step fails, because Python isn't done with interpreting ``foo`` yet and
1716
the global symbol dictionary for ``foo`` is still empty.
1718
The same thing happens when you use ``import foo``, and then try to access
1719
``foo.foo_var`` in global code.
1721
There are (at least) three possible workarounds for this problem.
1723
Guido van Rossum recommends avoiding all uses of ``from <module> import ...``,
1724
and placing all code inside functions. Initializations of global variables and
1725
class variables should use constants or built-in functions only. This means
1726
everything from an imported module is referenced as ``<module>.<name>``.
1728
Jim Roskind suggests performing steps in the following order in each module:
1730
* exports (globals, functions, and classes that don't need imported base
1732
* ``import`` statements
1733
* active code (including globals that are initialized from imported values).
1735
van Rossum doesn't like this approach much because the imports appear in a
1736
strange place, but it does work.
1738
Matthias Urlichs recommends restructuring your code so that the recursive import
1739
is not necessary in the first place.
1741
These solutions are not mutually exclusive.
1744
__import__('x.y.z') returns <module 'x'>; how do I get z?
1745
---------------------------------------------------------
1749
__import__('x.y.z').y.z
1751
For more realistic situations, you may have to do something like ::
1754
for i in s.split(".")[1:]:
1757
See :mod:`importlib` for a convenience function called
1758
:func:`~importlib.import_module`.
1762
When I edit an imported module and reimport it, the changes don't show up. Why does this happen?
1763
-------------------------------------------------------------------------------------------------
1765
For reasons of efficiency as well as consistency, Python only reads the module
1766
file on the first time a module is imported. If it didn't, in a program
1767
consisting of many modules where each one imports the same basic module, the
1768
basic module would be parsed and re-parsed many times. To force re-reading of a
1769
changed module, do this::
1773
importlib.reload(modname)
1775
Warning: this technique is not 100% fool-proof. In particular, modules
1776
containing statements like ::
1778
from modname import some_objects
1780
will continue to work with the old version of the imported objects. If the
1781
module contains class definitions, existing class instances will *not* be
1782
updated to use the new class definition. This can result in the following
1783
paradoxical behaviour:
1785
>>> import importlib
1787
>>> c = cls.C() # Create an instance of C
1788
>>> importlib.reload(cls)
1789
<module 'cls' from 'cls.py'>
1790
>>> isinstance(c, cls.C) # isinstance is false?!?
1793
The nature of the problem is made clear if you print out the "identity" of the
1796
>>> hex(id(c.__class__))