python-peps/peps/pep-0357.rst

PEP: 357
Title: Allowing Any Object to be Used for Slicing
Author: Travis Oliphant <oliphant@ee.byu.edu>
Status: Final
Type: Standards Track
Content-Type: text/x-rst
Created: 09-Feb-2006
Python-Version: 2.5
Post-History:


Abstract
========

This PEP proposes adding an ``nb_index`` slot in ``PyNumberMethods`` and an
``__index__`` special method so that arbitrary objects can be used
whenever integers are explicitly needed in Python, such as in slice
syntax (from which the slot gets its name).


Rationale
=========

Currently integers and long integers play a special role in
slicing in that they are the only objects allowed in slice
syntax. In other words, if X is an object implementing the
sequence protocol, then ``X[obj1:obj2]`` is only valid if ``obj1`` and
``obj2`` are both integers or long integers.  There is no way for ``obj1``
and ``obj2`` to tell Python that they could be reasonably used as
indexes into a sequence.  This is an unnecessary limitation.

In NumPy, for example, there are 8 different integer scalars
corresponding to unsigned and signed integers of 8, 16, 32, and 64
bits.  These type-objects could reasonably be used as integers in
many places where Python expects true integers but cannot inherit from
the Python integer type because of incompatible memory layouts.
There should be some way to be able to tell Python that an object can
behave like an integer.

It is not possible to use the ``nb_int`` (and ``__int__`` special method)
for this purpose because that method is used to *coerce* objects
to integers.  It would be inappropriate to allow every object that
can be coerced to an integer to be used as an integer everywhere
Python expects a true integer.  For example, if ``__int__`` were used
to convert an object to an integer in slicing, then float objects
would be allowed in slicing and ``x[3.2:5.8]`` would not raise an error
as it should.


Proposal
========

Add an ``nb_index`` slot to ``PyNumberMethods``, and a corresponding
``__index__`` special method.  Objects could define a function to
place in the ``nb_index`` slot that returns a Python integer
(either an int or a long). This integer can
then be appropriately converted to a ``Py_ssize_t`` value whenever
Python needs one such as in ``PySequence_GetSlice``,
``PySequence_SetSlice``, and ``PySequence_DelSlice``.


Specification
=============

1) The ``nb_index`` slot will have the following signature::

       PyObject *index_func (PyObject *self)

   The returned object must be a Python ``IntType`` or
   Python ``LongType``. NULL should be returned on
   error with an appropriate error set.

2) The ``__index__`` special method will have the signature::

       def __index__(self):
           return obj

   where obj must be either an int or a long.

3) 3 new abstract C-API functions will be added

   a) The first checks to see if the object supports the index
      slot and if it is filled in.

      ::

          int PyIndex_Check(obj)

      This will return true if the object defines the ``nb_index``
      slot.

   b) The second is a simple wrapper around the ``nb_index`` call that
      raises ``PyExc_TypeError`` if the call is not available or if it
      doesn't return an int or long.  Because the
      ``PyIndex_Check`` is performed inside the ``PyNumber_Index`` call
      you can call it directly and manage any error rather than
      check for compatibility first.

      ::

          PyObject *PyNumber_Index (PyObject *obj)

   c) The third call helps deal with the common situation of
      actually needing a ``Py_ssize_t`` value from the object to use for
      indexing or other needs.

      ::

          Py_ssize_t PyNumber_AsSsize_t(PyObject *obj, PyObject *exc)

      The function calls the ``nb_index`` slot of obj if it is
      available and then converts the returned Python integer into
      a ``Py_ssize_t`` value.  If this goes well, then the value is
      returned.  The second argument allows control over what
      happens if the integer returned from ``nb_index`` cannot fit
      into a ``Py_ssize_t`` value.

      If exc is NULL, then the returned value will be clipped to
      ``PY_SSIZE_T_MAX`` or ``PY_SSIZE_T_MIN`` depending on whether the
      ``nb_index`` slot of obj returned a positive or negative
      integer.  If exc is non-NULL, then it is the error object
      that will be set to replace the ``PyExc_OverflowError`` that was
      raised when the Python integer or long was converted to ``Py_ssize_t``.

4) A new ``operator.index(obj)`` function will be added that calls
   equivalent of ``obj.__index__()`` and raises an error if obj does not implement
   the special method.


Implementation Plan
===================

1) Add the ``nb_index`` slot in ``object.h`` and modify ``typeobject.c`` to
   create the ``__index__`` method

2) Change the ``ISINT`` macro in ``ceval.c`` to ``ISINDEX`` and alter it to
   accommodate objects with the index slot defined.

3) Change the ``_PyEval_SliceIndex`` function to accommodate objects
   with the index slot defined.

4) Change all builtin objects (e.g. lists) that use the ``as_mapping``
   slots for subscript access and use a special-check for integers to
   check for the slot as well.

5) Add the ``nb_index`` slot to integers and long_integers
   (which just return themselves)

6) Add ``PyNumber_Index`` C-API to return an integer from any
   Python Object that has the ``nb_index`` slot.

7) Add the ``operator.index(x)`` function.

8) Alter ``arrayobject.c`` and ``mmapmodule.c`` to use the new C-API for their
   sub-scripting and other needs.

9) Add unit-tests


Discussion Questions
====================

Speed
-----

Implementation should not slow down Python because integers and long
integers used as indexes will complete in the same number of
instructions.  The only change will be that what used to generate
an error will now be acceptable.

Why not use ``nb_int`` which is already there?
----------------------------------------------

The ``nb_int`` method is used for coercion and so means something
fundamentally different than what is requested here.  This PEP
proposes a method for something that *can* already be thought of as
an integer communicate that information to Python when it needs an
integer.  The biggest example of why using ``nb_int`` would be a bad
thing is that float objects already define the ``nb_int`` method, but
float objects *should not* be used as indexes in a sequence.

Why the name ``__index__``?
---------------------------

Some questions were raised regarding the name ``__index__`` when other
interpretations of the slot are possible.  For example, the slot
can be used any time Python requires an integer internally (such
as in ``"mystring" * 3``).  The name was suggested by Guido because
slicing syntax is the biggest reason for having such a slot and
in the end no better name emerged. See the discussion thread [1]_
for examples of names that were suggested such as "``__discrete__``" and
"``__ordinal__``".

Why return ``PyObject *`` from ``nb_index``?
--------------------------------------------

Initially ``Py_ssize_t`` was selected as the return type for the
``nb_index`` slot.  However, this led to an inability to track and
distinguish overflow and underflow errors without ugly and brittle
hacks. As the ``nb_index`` slot is used in at least 3 different ways
in the Python core (to get an integer, to get a slice end-point,
and to get a sequence index), there is quite a bit of flexibility
needed to handle all these cases.  The importance of having the
necessary flexibility to handle all the use cases is critical.
For example, the initial implementation that returned ``Py_ssize_t`` for
``nb_index`` led to the discovery that on a 32-bit machine with >=2GB of RAM
``s = 'x' * (2**100)`` works but ``len(s)`` was clipped at 2147483647.
Several fixes were suggested but eventually it was decided that
``nb_index`` needed to return a Python Object similar to the ``nb_int``
and ``nb_long`` slots in order to handle overflow correctly.

Why can't ``__index__`` return any object with the ``nb_index`` method?
-----------------------------------------------------------------------

This would allow infinite recursion in many different ways that are not
easy to check for.  This restriction is similar to the requirement that
``__nonzero__`` return an int or a bool.


Reference Implementation
========================

Submitted as patch 1436368 to SourceForge.


References
==========

.. [1] Travis Oliphant, PEP for adding an sq_index slot so that any object, a
       or b, can be used in X[a:b] notation,

       https://mail.python.org/pipermail/python-dev/2006-February/thread.html#60594

Copyright
=========

This document is placed in the public domain.