Introduction

    This document gives coding conventions for the Python code comprising the

    standard library in the main Python distribution.  Please see the

    companion informational PEP describing style guidelines for the C code in

    the C implementation of Python[1].

    This document was adapted from Guido's original Python Style Guide

    essay[2], with some additions from Barry's style guide[5].  Where there's

    conflict, Guido's style rules for the purposes of this PEP.  This PEP may

    still be incomplete (in fact, it may never be finished <wink>).

A Foolish Consistency is the Hobgoblin of Little Minds

    One of Guido's key insights is that code is read much more often than it

    is written.  The guidelines provided here are intended to improve the

    readability of code and make it consistent across the wide spectrum of

    Python code.  As PEP 20 [6] says, "Readability counts".

    A style guide is about consistency.  Consistency with this style guide is

    important.  Consistency within a project is more important. Consistency

    within one module or function is most important.

    But most importantly: know when to be inconsistent -- sometimes the style

    guide just doesn't apply.  When in doubt, use your best judgment.  Look

    at other examples and decide what looks best.  And don't hesitate to ask!

    Two good reasons to break a particular rule:

    (1) When applying the rule would make the code less readable, even for

        someone who is used to reading code that follows the rules.

    (2) To be consistent with surrounding code that also breaks it (maybe for

        historic reasons) -- although this is also an opportunity to clean up

        someone else's mess (in true XP style).

Code lay-out

  Indentation

    Use 4 spaces per indentation level.

    For really old code that you don't want to mess up, you can continue to

    use 8-space tabs.

    Continuation lines should align wrapped elements either vertically using

    Python's implicit line joining inside parentheses, brackets and braces, or

    using a hanging indent.  When using a hanging indent the following

    considerations should be applied; there should be no arguments on the

    first line and further indentation should be used to clearly distinguish

    itself as a continuation line.

    Yes:  # Aligned with opening delimiter

          foo = long_function_name(var_one, var_two,

                                   var_three, var_four)

          # More indentation included to distinguish this from the rest.

          def long_function_name(

                  var_one, var_two, var_three,

                  var_four):

              print(var_one)

    No:   # Arguments on first line forbidden when not using vertical alignment

          foo = long_function_name(var_one, var_two,

              var_three, var_four)

          # Further indentation required as indentation is not distinguishable

          def long_function_name(

              var_one, var_two, var_three,

              var_four):

              print(var_one)

    Optional:

          # Extra indentation is not necessary.

          foo = long_function_name(

            var_one, var_two,

            var_three, var_four)

  Tabs or Spaces?

    Never mix tabs and spaces.

    The most popular way of indenting Python is with spaces only.  The

    second-most popular way is with tabs only.  Code indented with a mixture

    of tabs and spaces should be converted to using spaces exclusively.  When

    invoking the Python command line interpreter with the -t option, it issues

    warnings about code that illegally mixes tabs and spaces.  When using -tt

    these warnings become errors.  These options are highly recommended!

    For new projects, spaces-only are strongly recommended over tabs.  Most

    editors have features that make this easy to do.

  Maximum Line Length

    Limit all lines to a maximum of 79 characters.

    There are still many devices around that are limited to 80 character

    lines; plus, limiting windows to 80 characters makes it possible to have

    several windows side-by-side.  The default wrapping on such devices

    disrupts the visual structure of the code, making it more difficult to

    understand.  Therefore, please limit all lines to a maximum of 79

    characters.  For flowing long blocks of text (docstrings or comments),

    limiting the length to 72 characters is recommended.

    The preferred way of wrapping long lines is by using Python's implied line

    continuation inside parentheses, brackets and braces.  Long lines can be

    broken over multiple lines by wrapping expressions in parentheses. These

    should be used in preference to using a backslash for line continuation.

    Make sure to indent the continued line appropriately.  The preferred place

    to break around a binary operator is *after* the operator, not before it.

    Some examples:

    class Rectangle(Blob):

        def __init__(self, width, height,

                     color='black', emphasis=None, highlight=0):

            if (width == 0 and height == 0 and

                color == 'red' and emphasis == 'strong' or

                highlight > 100):

                raise ValueError("sorry, you lose")

            if width == 0 and height == 0 and (color == 'red' or

                                               emphasis is None):

                raise ValueError("I don't think so -- values are %s, %s" %

                                 (width, height))

            Blob.__init__(self, width, height,

                          color, emphasis, highlight)

  Blank Lines

    Separate top-level function and class definitions with two blank lines.

    Method definitions inside a class are separated by a single blank line.

    Extra blank lines may be used (sparingly) to separate groups of related

    functions.  Blank lines may be omitted between a bunch of related

    one-liners (e.g. a set of dummy implementations).

    Use blank lines in functions, sparingly, to indicate logical sections.

    Python accepts the control-L (i.e. ^L) form feed character as whitespace;

    Many tools treat these characters as page separators, so you may use them

    to separate pages of related sections of your file.  Note, some editors

    and web-based code viewers may not recognize control-L as a form feed

    and will show another glyph in its place.

  Encodings (PEP 263)

    Code in the core Python distribution should always use the ASCII or

    Latin-1 encoding (a.k.a. ISO-8859-1).  For Python 3.0 and beyond,

    UTF-8 is preferred over Latin-1, see PEP 3120.

    Files using ASCII should not have a coding cookie.  Latin-1 (or

    UTF-8) should only be used when a comment or docstring needs to

    mention an author name that requires Latin-1; otherwise, using

    \x, \u or \U escapes is the preferred way to include non-ASCII

    data in string literals.

    For Python 3.0 and beyond, the following policy is prescribed for

    the standard library (see PEP 3131): All identifiers in the Python

    standard library MUST use ASCII-only identifiers, and SHOULD use

    English words wherever feasible (in many cases, abbreviations and

    technical terms are used which aren't English). In addition,

    string literals and comments must also be in ASCII. The only

    exceptions are (a) test cases testing the non-ASCII features, and

    (b) names of authors. Authors whose names are not based on the

    latin alphabet MUST provide a latin transliteration of their

    names.

    Open source projects with a global audience are encouraged to

    adopt a similar policy.

Imports

    - Imports should usually be on separate lines, e.g.:

        Yes: import os

             import sys

        No:  import sys, os

      it's okay to say this though:

        from subprocess import Popen, PIPE

    - Imports are always put at the top of the file, just after any module

      comments and docstrings, and before module globals and constants.

      Imports should be grouped in the following order:

      1. standard library imports

      2. related third party imports

      3. local application/library specific imports

      You should put a blank line between each group of imports.

      Put any relevant __all__ specification after the imports.

    - Relative imports for intra-package imports are highly discouraged.

      Always use the absolute package path for all imports.

      Even now that PEP 328 [7] is fully implemented in Python 2.5,

      its style of explicit relative imports is actively discouraged;

      absolute imports are more portable and usually more readable.

    - When importing a class from a class-containing module, it's usually okay

      to spell this

        from myclass import MyClass

        from foo.bar.yourclass import YourClass

      If this spelling causes local name clashes, then spell them

        import myclass

        import foo.bar.yourclass

      and use "myclass.MyClass" and "foo.bar.yourclass.YourClass"

Whitespace in Expressions and Statements

  Pet Peeves

    Avoid extraneous whitespace in the following situations:

    - Immediately inside parentheses, brackets or braces.

      Yes: spam(ham[1], {eggs: 2})

      No:  spam( ham[ 1 ], { eggs: 2 } )

    - Immediately before a comma, semicolon, or colon:

      Yes: if x == 4: print x, y; x, y = y, x

      No:  if x == 4 : print x , y ; x , y = y , x

    - Immediately before the open parenthesis that starts the argument

      list of a function call:

      Yes: spam(1)

      No:  spam (1)

    - Immediately before the open parenthesis that starts an indexing or

      slicing:

      Yes: dict['key'] = list[index]

      No:  dict ['key'] = list [index]

    - More than one space around an assignment (or other) operator to

      align it with another.

      Yes:

          x = 1

          y = 2

          long_variable = 3

      No:

          x             = 1

          y             = 2

          long_variable = 3

  Other Recommendations

    - Always surround these binary operators with a single space on

      either side: assignment (=), augmented assignment (+=, -= etc.),

      comparisons (==, <, >, !=, <>, <=, >=, in, not in, is, is not),

      Booleans (and, or, not).

    - Use spaces around arithmetic operators:

      Yes:

          i = i + 1

          submitted += 1

          x = x * 2 - 1

          hypot2 = x * x + y * y

          c = (a + b) * (a - b)

      No:

          i=i+1

          submitted +=1

          x = x*2 - 1

          hypot2 = x*x + y*y

          c = (a+b) * (a-b)

    - Don't use spaces around the '=' sign when used to indicate a

      keyword argument or a default parameter value.

      Yes:

          def complex(real, imag=0.0):

              return magic(r=real, i=imag)

      No:

          def complex(real, imag = 0.0):

              return magic(r = real, i = imag)

    - Compound statements (multiple statements on the same line) are

      generally discouraged.

      Yes:

          if foo == 'blah':

              do_blah_thing()

          do_one()

          do_two()

          do_three()

      Rather not:

          if foo == 'blah': do_blah_thing()

          do_one(); do_two(); do_three()

    - While sometimes it's okay to put an if/for/while with a small

      body on the same line, never do this for multi-clause

      statements.  Also avoid folding such long lines!

      Rather not:

          if foo == 'blah': do_blah_thing()

          for x in lst: total += x

          while t < 10: t = delay()

      Definitely not:

          if foo == 'blah': do_blah_thing()

          else: do_non_blah_thing()

          try: something()

          finally: cleanup()

          do_one(); do_two(); do_three(long, argument,

                                       list, like, this)

          if foo == 'blah': one(); two(); three()

Comments

    Comments that contradict the code are worse than no comments.  Always make

    a priority of keeping the comments up-to-date when the code changes!

    Comments should be complete sentences.  If a comment is a phrase or

    sentence, its first word should be capitalized, unless it is an identifier

    that begins with a lower case letter (never alter the case of

    identifiers!).

    If a comment is short, the period at the end can be omitted.  Block

    comments generally consist of one or more paragraphs built out of complete

    sentences, and each sentence should end in a period.

    You should use two spaces after a sentence-ending period.

    When writing English, Strunk and White apply.

    Python coders from non-English speaking countries: please write

    your comments in English, unless you are 120% sure that the code

    will never be read by people who don't speak your language.

  Block Comments

    Block comments generally apply to some (or all) code that follows them,

    and are indented to the same level as that code.  Each line of a block

    comment starts with a # and a single space (unless it is indented text

    inside the comment).

    Paragraphs inside a block comment are separated by a line containing a

    single #.

  Inline Comments

    Use inline comments sparingly.

    An inline comment is a comment on the same line as a statement.  Inline

    comments should be separated by at least two spaces from the statement.

    They should start with a # and a single space.

    Inline comments are unnecessary and in fact distracting if they state

    the obvious.  Don't do this:

        x = x + 1                 # Increment x

    But sometimes, this is useful:

        x = x + 1                 # Compensate for border

Documentation Strings

    Conventions for writing good documentation strings (a.k.a. "docstrings")

    are immortalized in PEP 257 [3].

    - Write docstrings for all public modules, functions, classes, and

      methods.  Docstrings are not necessary for non-public methods, but you

      should have a comment that describes what the method does.  This comment

      should appear after the "def" line.

    - PEP 257 describes good docstring conventions.  Note that most

      importantly, the """ that ends a multiline docstring should be on a line

      by itself, and preferably preceded by a blank line, e.g.:

      """Return a foobang

      Optional plotz says to frobnicate the bizbaz first.

      """

    - For one liner docstrings, it's okay to keep the closing """ on the same

      line.

Version Bookkeeping

    If you have to have Subversion, CVS, or RCS crud in your source file, do

    it as follows.

        __version__ = "$Revision: 00f8e3bb1197 $"

        # $Source$

    These lines should be included after the module's docstring, before any

    other code, separated by a blank line above and below.

Naming Conventions

    The naming conventions of Python's library are a bit of a mess, so we'll

    never get this completely consistent -- nevertheless, here are the

    currently recommended naming standards.  New modules and packages

    (including third party frameworks) should be written to these standards,

    but where an existing library has a different style, internal consistency

    is preferred.

  Descriptive: Naming Styles

    There are a lot of different naming styles.  It helps to be able to

    recognize what naming style is being used, independently from what they

    are used for.

    The following naming styles are commonly distinguished:

    - b (single lowercase letter)

    - B (single uppercase letter)

    - lowercase

    - lower_case_with_underscores

    - UPPERCASE

    - UPPER_CASE_WITH_UNDERSCORES

    - CapitalizedWords (or CapWords, or CamelCase -- so named because

      of the bumpy look of its letters[4]).  This is also sometimes known as

      StudlyCaps.

      Note: When using abbreviations in CapWords, capitalize all the letters

      of the abbreviation.  Thus HTTPServerError is better than

      HttpServerError.

    - mixedCase (differs from CapitalizedWords by initial lowercase

      character!)

    - Capitalized_Words_With_Underscores (ugly!)

    There's also the style of using a short unique prefix to group related

    names together.  This is not used much in Python, but it is mentioned for

    completeness.  For example, the os.stat() function returns a tuple whose

    items traditionally have names like st_mode, st_size, st_mtime and so on.

    (This is done to emphasize the correspondence with the fields of the

    POSIX system call struct, which helps programmers familiar with that.)

    The X11 library uses a leading X for all its public functions.  In Python,

    this style is generally deemed unnecessary because attribute and method

    names are prefixed with an object, and function names are prefixed with a

    module name.

    In addition, the following special forms using leading or trailing

    underscores are recognized (these can generally be combined with any case

    convention):

    - _single_leading_underscore: weak "internal use" indicator.  E.g. "from M

      import *" does not import objects whose name starts with an underscore.

    - single_trailing_underscore_: used by convention to avoid conflicts with

      Python keyword, e.g.

      Tkinter.Toplevel(master, class_='ClassName')

    - __double_leading_underscore: when naming a class attribute, invokes name

      mangling (inside class FooBar, __boo becomes _FooBar__boo; see below).

    - __double_leading_and_trailing_underscore__: "magic" objects or

      attributes that live in user-controlled namespaces.  E.g. __init__,

      __import__ or __file__.  Never invent such names; only use them

      as documented.

  Prescriptive: Naming Conventions

    Names to Avoid

      Never use the characters 'l' (lowercase letter el), 'O' (uppercase

      letter oh), or 'I' (uppercase letter eye) as single character variable

      names.

      In some fonts, these characters are indistinguishable from the numerals

      one and zero.  When tempted to use 'l', use 'L' instead.

    Package and Module Names

      Modules should have short, all-lowercase names.  Underscores can be used

      in the module name if it improves readability.  Python packages should

      also have short, all-lowercase names, although the use of underscores is

      discouraged.

      Since module names are mapped to file names, and some file systems are

      case insensitive and truncate long names, it is important that module

      names be chosen to be fairly short -- this won't be a problem on Unix,

      but it may be a problem when the code is transported to older Mac or

      Windows versions, or DOS.

      When an extension module written in C or C++ has an accompanying Python

      module that provides a higher level (e.g. more object oriented)

      interface, the C/C++ module has a leading underscore (e.g. _socket).

    Class Names

      Almost without exception, class names use the CapWords convention.

      Classes for internal use have a leading underscore in addition.

    Exception Names

      Because exceptions should be classes, the class naming convention

      applies here.  However, you should use the suffix "Error" on your

      exception names (if the exception actually is an error).

    Global Variable Names

      (Let's hope that these variables are meant for use inside one module

      only.)  The conventions are about the same as those for functions.

      Modules that are designed for use via "from M import *" should use the

      __all__ mechanism to prevent exporting globals, or use the older

      convention of prefixing such globals with an underscore (which you might

      want to do to indicate these globals are "module non-public").

    Function Names

      Function names should be lowercase, with words separated by underscores

      as necessary to improve readability.

      mixedCase is allowed only in contexts where that's already the

      prevailing style (e.g. threading.py), to retain backwards compatibility.

    Function and method arguments

      Always use 'self' for the first argument to instance methods.

      Always use 'cls' for the first argument to class methods.

      If a function argument's name clashes with a reserved keyword, it is

      generally better to append a single trailing underscore rather than use

      an abbreviation or spelling corruption.  Thus "print_" is better than

      "prnt".  (Perhaps better is to avoid such clashes by using a synonym.)

    Method Names and Instance Variables

      Use the function naming rules: lowercase with words separated by

      underscores as necessary to improve readability.

      Use one leading underscore only for non-public methods and instance

      variables.

      To avoid name clashes with subclasses, use two leading underscores to

      invoke Python's name mangling rules.

      Python mangles these names with the class name: if class Foo has an

      attribute named __a, it cannot be accessed by Foo.__a.  (An insistent

      user could still gain access by calling Foo._Foo__a.)  Generally, double

      leading underscores should be used only to avoid name conflicts with

      attributes in classes designed to be subclassed.

      Note: there is some controversy about the use of __names (see below).

    Constants

       Constants are usually defined on a module level and written in all

       capital letters with underscores separating words.  Examples include

       MAX_OVERFLOW and TOTAL.

    Designing for inheritance

      Always decide whether a class's methods and instance variables

      (collectively: "attributes") should be public or non-public.  If in

      doubt, choose non-public; it's easier to make it public later than to

      make a public attribute non-public.

      Public attributes are those that you expect unrelated clients of your

      class to use, with your commitment to avoid backward incompatible

      changes.  Non-public attributes are those that are not intended to be

      used by third parties; you make no guarantees that non-public attributes

      won't change or even be removed.

      We don't use the term "private" here, since no attribute is really

      private in Python (without a generally unnecessary amount of work).

      Another category of attributes are those that are part of the "subclass

      API" (often called "protected" in other languages).  Some classes are

      designed to be inherited from, either to extend or modify aspects of the

      class's behavior.  When designing such a class, take care to make

      explicit decisions about which attributes are public, which are part of

      the subclass API, and which are truly only to be used by your base

      class.

      With this in mind, here are the Pythonic guidelines:

      - Public attributes should have no leading underscores.

      - If your public attribute name collides with a reserved keyword, append

        a single trailing underscore to your attribute name.  This is

        preferable to an abbreviation or corrupted spelling.  (However,

        notwithstanding this rule, 'cls' is the preferred spelling for any

        variable or argument which is known to be a class, especially the

        first argument to a class method.)

        Note 1: See the argument name recommendation above for class methods.

      - For simple public data attributes, it is best to expose just the

        attribute name, without complicated accessor/mutator methods.  Keep in

        mind that Python provides an easy path to future enhancement, should

        you find that a simple data attribute needs to grow functional

        behavior.  In that case, use properties to hide functional

        implementation behind simple data attribute access syntax.

        Note 1: Properties only work on new-style classes.

        Note 2: Try to keep the functional behavior side-effect free, although

        side-effects such as caching are generally fine.

        Note 3: Avoid using properties for computationally expensive

        operations; the attribute notation makes the caller believe

        that access is (relatively) cheap.

      - If your class is intended to be subclassed, and you have attributes

        that you do not want subclasses to use, consider naming them with

        double leading underscores and no trailing underscores.  This invokes

        Python's name mangling algorithm, where the name of the class is

        mangled into the attribute name.  This helps avoid attribute name

        collisions should subclasses inadvertently contain attributes with the

        same name.

        Note 1: Note that only the simple class name is used in the mangled

        name, so if a subclass chooses both the same class name and attribute

        name, you can still get name collisions.

        Note 2: Name mangling can make certain uses, such as debugging and

        __getattr__(), less convenient.  However the name mangling algorithm

        is well documented and easy to perform manually.

        Note 3: Not everyone likes name mangling.  Try to balance the

        need to avoid accidental name clashes with potential use by

        advanced callers.

Programming Recommendations

    - Code should be written in a way that does not disadvantage other

      implementations of Python (PyPy, Jython, IronPython, Pyrex, Psyco,

      and such).

      For example, do not rely on CPython's efficient implementation of

      in-place string concatenation for statements in the form a+=b or a=a+b.

      Those statements run more slowly in Jython.  In performance sensitive

      parts of the library, the ''.join() form should be used instead.  This

      will ensure that concatenation occurs in linear time across various

      implementations.

    - Comparisons to singletons like None should always be done with

      'is' or 'is not', never the equality operators.

      Also, beware of writing "if x" when you really mean "if x is not None"

      -- e.g. when testing whether a variable or argument that defaults to

      None was set to some other value.  The other value might have a type

      (such as a container) that could be false in a boolean context!

    - When implementing ordering operations with rich comparisons, it is best to

      implement all six operations (__eq__, __ne__, __lt__, __le__, __gt__,

      __ge__) rather than relying on other code to only exercise a particular

      comparison.

      To minimize the effort involved, the functools.total_ordering() decorator

      provides a tool to generate missing comparison methods.

      PEP 207 indicates that reflexivity rules *are* assumed by Python.  Thus,

      the interpreter may swap y>x with x<y, y>=x with x<=y, and may swap the

      arguments of x==y and x!=y.  The sort() and min() operations are

      guaranteed to use the < operator and the max() function uses the >

      operator.  However, it is best to implement all six operations so that

      confusion doesn't arise in other contexts.

    - Use class-based exceptions.

      String exceptions in new code are forbidden, because this language

      feature is being removed in Python 2.6.

      Modules or packages should define their own domain-specific base

      exception class, which should be subclassed from the built-in Exception

      class.  Always include a class docstring.  E.g.:

        class MessageError(Exception):

            """Base class for errors in the email package."""

      Class naming conventions apply here, although you should add the suffix

      "Error" to your exception classes, if the exception is an error.

      Non-error exceptions need no special suffix.

    - When raising an exception, use "raise ValueError('message')" instead of

      the older form "raise ValueError, 'message'".

      The paren-using form is preferred because when the exception arguments

      are long or include string formatting, you don't need to use line

      continuation characters thanks to the containing parentheses.  The older

      form will be removed in Python 3000.

    - When catching exceptions, mention specific exceptions

      whenever possible instead of using a bare 'except:' clause.

      For example, use:

          try:

              import platform_specific_module

          except ImportError:

              platform_specific_module = None

      A bare 'except:' clause will catch SystemExit and KeyboardInterrupt

      exceptions, making it harder to interrupt a program with Control-C,

      and can disguise other problems.  If you want to catch all

      exceptions that signal program errors, use 'except Exception:'.

      A good rule of thumb is to limit use of bare 'except' clauses to two

      cases:

         1) If the exception handler will be printing out or logging

            the traceback; at least the user will be aware that an

            error has occurred.

         2) If the code needs to do some cleanup work, but then lets

            the exception propagate upwards with 'raise'.

            'try...finally' is a better way to handle this case.

    - Additionally, for all try/except clauses, limit the 'try' clause

      to the absolute minimum amount of code necessary.  Again, this

      avoids masking bugs.

      Yes:

          try:

              value = collection[key]

          except KeyError:

              return key_not_found(key)

          else:

              return handle_value(value)

      No:

          try:

              # Too broad!

              return handle_value(collection[key])

          except KeyError:

              # Will also catch KeyError raised by handle_value()

              return key_not_found(key)

    - Use string methods instead of the string module.

      String methods are always much faster and share the same API with

      unicode strings.  Override this rule if backward compatibility with

      Pythons older than 2.0 is required.

    - Use ''.startswith() and ''.endswith() instead of string slicing to check

      for prefixes or suffixes.

      startswith() and endswith() are cleaner and less error prone.  For

      example:

        Yes: if foo.startswith('bar'):

        No:  if foo[:3] == 'bar':

      The exception is if your code must work with Python 1.5.2 (but let's

      hope not!).

    - Object type comparisons should always use isinstance() instead

      of comparing types directly.

        Yes: if isinstance(obj, int):

        No:  if type(obj) is type(1):

      When checking if an object is a string, keep in mind that it might be a

      unicode string too!  In Python 2.3, str and unicode have a common base

      class, basestring, so you can do:

        if isinstance(obj, basestring):

    - For sequences, (strings, lists, tuples), use the fact that empty

      sequences are false.

      Yes: if not seq:

           if seq:

      No: if len(seq)

          if not len(seq)

    - Don't write string literals that rely on significant trailing

      whitespace.  Such trailing whitespace is visually indistinguishable and

      some editors (or more recently, reindent.py) will trim them.

    - Don't compare boolean values to True or False using ==

        Yes:   if greeting:

        No:    if greeting == True:

        Worse: if greeting is True:

Rules that apply only to the standard library

    - Do not use function type annotations in the standard library.

      These are reserved for users and third-party modules.  See

      PEP 3107 and the bug 10899 for details.