2
0
mirror of https://github.com/boostorg/python.git synced 2026-01-26 06:42:27 +00:00

Initial commit of indexing suite documentation

[SVN r19314]
This commit is contained in:
Joel de Guzman
2003-07-26 05:48:59 +00:00
parent a6440a3fa6
commit e41abb6e92

278
doc/v2/indexing.html Normal file
View File

@@ -0,0 +1,278 @@
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0//EN"
"http://www.w3.org/TR/REC-html40/strict.dtd">
<link rel="stylesheet" type="text/css" href="../boost.css">
<title>Boost.Python Indexing Support</title>
<div> <img src="../../../../c++boost.gif"
alt="c++boost.gif (8819 bytes)"
align="center"
width="277" height="86">
<hr>
<h1>Boost.Python Indexing Support</h1>
<p>Indexing is a Python module for easy exportation of indexable C++ containers
to Python. Indexable containers are containers that allow random access through
the operator[] (e.g. std::vector).
<p>While Boost Python has all the facilities to expose indexable C++ containers
such as the ubiquitous std::vector to Python, the procedure is not as straightforward
as we'd like it to be. When in Python, do as the Pythonians do. Yet, Python
containers do not map easily to C++ containers. Emulating Python containers
in C++ (see Python Reference Manual, <a href="http://www.python.org/doc/current/ref/sequence-types.html">Emulating
container types</a>) using Boost Python is non trivial. There are a lot of
issues to consider before we can map a C++ container to Python which involves
implementing wrapper functions for the methods <strong>__len__</strong>, <strong>__getitem__</strong>,
<strong>__setitem__</strong>, <strong>__delitem__, </strong><strong>__iter__
</strong>and<strong> __contains</strong>.
<p>The goals:</div>
<ul>
<li>
<div>Make indexable C++ containers behave exactly as one would expect a Python
container to behave.</div>
</li>
<li>Provide default reference semantics for container element indexing (__getitem__)
such that c[i] can be mutable. Require: <tt>c[i].m() == c[i] </tt> where m
is a non-const (mutating) member function (method).</li>
<li>Return safe references from __getitem__ such that subsequent adds and deletes
to and from the container will not result in dangling references (will not
crash Python).</li>
<li>Support slice indexes.</li>
<li>Allow interoperability with Python containers (e.g lists, tuples).</li>
<li>Allow for extensibility through redefinable policy classes.</li>
<li>Provide predefined support for the most common STL and STL like indexable
containers.</li>
</ul>
<div>
<hr>
<h2>The Boost.Python Indexing Interface</h2>
<h3>indexing_suite</h3>
<p>The <tt>indexing_suite</tt> class is the base protocol class for the management
of C++ containers intended to be integrated to Python. The objective is make
a C++ container look and feel and behave exactly as we'd expect a Python container.
The class automatically wraps these special Python methods:</p>
</div>
<dl>
<dt>
<div><strong>__len__(self)</strong></div>
</dt>
<dd>
<div>Called to implement the built-in function len() Should return the length
of the object, an integer &gt;= 0. Also, an object that doesn't define a
__nonzero__() method and whose __len__() method returns zero is considered
to be false in a Boolean context.</div>
</dd>
<dt>&nbsp;</dt>
<dd>
<div></div>
</dd>
<dt><strong>__getitem__(self, key)</strong></dt>
<dd>Called to implement evaluation of self[key]. For sequence types, the accepted
keys should be integers and slice objects. Note that the special interpretation
of negative indexes (if the class wishes to emulate a sequence type) is up
to the __getitem__() method. If key is of an inappropriate type, TypeError
may be raised; if of a value outside the set of indexes for the sequence (after
any special interpretation of negative values), IndexError should be raised.
Note: for loops expect that an IndexError will be raised for illegal indexes
to allow proper detection of the end of the sequence.</dd>
</dl>
<dl>
<dt>
<div><strong>__setitem__(self, key, value)</strong></div>
</dt>
<dd>
<div> Called to implement assignment to self[key]. Same note as for __getitem__().
This should only be implemented for mappings if the objects support changes
to the values for keys, or if new keys can be added, or for sequences if
elements can be replaced. The same exceptions should be raised for improper
key values as for the __getitem__() method.<br>
</div>
</dd>
<dt><strong>__delitem__(self, key)</strong></dt>
<dd>Called to implement deletion of self[key]. Same note as for __getitem__().
This should only be implemented for mappings if the objects support removal
of keys, or for sequences if elements can be removed from the sequence. The
same exceptions should be raised for improper key values as for the __getitem__()
method.</dd>
<dt>
<div><br>
<strong>__iter__(self)</strong></div>
</dt>
<dd>
<div>This method is called when an iterator is required for a container. This
method should return a new iterator object that can iterate over all the
objects in the container. For mappings, it should iterate over the keys
of the container, and should also be made available as the method iterkeys().</div>
</dd>
<dt>&nbsp;</dt>
<dd>
<div></div>
</dd>
<dt>
<div><strong>__contains__(self, item)</strong></div>
</dt>
<dd>
<div>Called to implement membership test operators. Should return true if
item is in self, false otherwise. For mapping objects, this should consider
the keys of the mapping rather than the values or the key-item pairs. </div>
</dd>
</dl>
<h3>indexing_suite sub-classes</h3>
<p>The <tt>indexing_suite</tt> is not meant to be used as is. A couple of policy
functions must be supplied by subclasses of <tt>indexing_suite</tt>. However,
a set of <tt>indexing_suite</tt> subclasses for the standard indexable STL containers
will be provided, In most cases, we can simply use the available predefined
suites. In some cases, if needed, we can refine them to suit our needs.</p>
<h3>vector_indexing_suite</h3>
<p>The <tt>vector_indexing_suite</tt> class is a predefined <tt>indexing_suite</tt>
derived class for wrapping <tt>std::vector</tt> (and <tt>std::vector</tt> like)
classes (currently, this is the only predefined suite available). It provides
all the policies required by the <tt>indexing_suite</tt>.</p>
<p> Example usage:</p>
<pre> class X {...};
...
class_&lt;std::vector&lt;X&gt; &gt;(&quot;XVec&quot;)
.def(vector_indexing_suite&lt;std::vector&lt;X&gt; &gt;())
;
</pre>
<p>That's it! <tt>XVec</tt> is now a full-fledged Python container (see the <a href="../../test/vector_indexing_suite.cpp">example
in full</a>, along with its <a href="../../test/vector_indexing_suite.py">python
test</a>).</p>
<hr>
<h2>indexing_suite class</h2>
<pre> template &lt;<br> class Container<br> , class DerivedPolicies<br> , bool NoProxy = false<br> , class Element = typename Container::value_type<br> , class Key = typename Container::value_type<br> , class Index = typename Container::size_type<br> &gt;<br> class indexing_suite <br> : public def_arg&lt;<br> indexing_suite&lt;<br> Container<br> , DerivedPolicies<br> , NoProxy<br> , Element<br> , Key<br> , Index<br> &gt; &gt;<br> {<br> public:</pre>
<pre> template &lt;class Class&gt;<br> void visit(Class&amp; cl) const;<br> };</pre>
<dl>
<dt> <strong><tt>Container</tt></strong></dt>
<dd>The contianer type to be wrapped to Python.</dd>
<dt>&nbsp;</dt>
<dt><strong><tt>DerivedPolicies</tt></strong></dt>
<dd>Derived classes provide the hooks needed by the <tt>indexing_suite:</tt></dd>
</dl>
<pre> static element_type&amp; <br> get_item(Container&amp; container, index_type i);
static object <br> get_slice(Container&amp; container, index_type from, index_type to);
static void <br> set_item(Container&amp; container, index_type i, element_type const&amp; v);
static void <br> set_slice(<br> Container&amp; container, index_type from, <br> index_type to, element_type const&amp; v<br> );
template &lt;class Iter&gt;<br> static void <br> set_slice(Container&amp; container, index_type from, <br> index_type to, Iter first, Iter last<br> );
static void <br> delete_item(Container&amp; container, index_type i);
static void <br> delete_slice(Container&amp; container, index_type from, index_type to);
static size_t<br> size(Container&amp; container);
template &lt;class T&gt;<br> static bool<br> contains(Container&amp; container, T const&amp; val);
static index_type<br> convert_index(Container&amp; container, PyObject* i);
static index_type<br> adjust_index(index_type current, index_type from, <br> index_type to, size_type len<br> );
</pre>
<blockquote>
<p>Most of these policies are self explanatory. <tt><strong>convert_index</strong></tt>
and <tt><strong>adjust_index</strong></tt>, however, deserves some explanation.</p>
<p><strong><tt>convert_index</tt></strong> converts an Python index into a C++
index that the container can handle. For instance, negative indexes in Python,
by convention, indexes from the right (e.g. <tt>C[-1]</tt> indexes the rightmost
element in <tt>C</tt>). <strong><tt>convert_index</tt></strong> should handle
the necessary conversion for the C++ container (e.g. convert <tt>-1</tt> to
<tt>C.size()-1</tt>). <tt><strong>convert_index</strong></tt> should also
be able to convert the type of the index (A dynamic Python type) to the actual
type that the C++ container expects.</p>
<p>When a container expands or contracts, held indexes to its elements must
be adjusted to follow the movement of data. For instance, if we erase 3 elements,
starting from index 0 from a 5 element vector, what used to be at index 4
will now be at index 1:</p>
<pre>
[a][b][c][d][e] ---&gt; [d][e]
^ ^
4 1</pre>
<p> <strong><tt>adjust_index</tt></strong> takes care of the adjustment. Given
a current index, the function should return the adjusted index when data in
the container at index <tt>from</tt>..<tt>to</tt> is replaced by <tt>len</tt>
elements. </p>
</blockquote>
<dl>
<dt><strong><tt>NoProxy</tt></strong></dt>
<dd>By default indexed elements have Python reference semantics and are returned
by proxy. This can be disabled by supplying <strong>true</strong> in the <tt>NoProxy</tt>
template parameter.</dd>
<dt><tt><strong><br>
Element</strong></tt></dt>
<dd><tt><strong></strong></tt>The container's element type. Defaults to <tt>Container::value_type</tt></dd>
<dt><tt><strong></strong></tt><tt><strong><br>
Key</strong></tt></dt>
<dd><tt><strong></strong></tt>The container's key type. Defaults to <tt>Container::value_type</tt></dd>
<dt><tt><strong></strong></tt><tt><strong><br>
Index</strong></tt></dt>
<dd><tt><strong></strong></tt>The container's index type. Defaults to <tt>Container::size_type</tt></dd>
</dl>
<table width="75%" border="1" align="center">
<tr>
<td><h2><strong><tt>def_arg</tt></strong></h2>
<p>The Boost Python <tt><strong>class_</strong></tt> interface provides
a generic visitation interface to avoid cluttering the class interface
through the <tt><strong>def_arg</strong></tt> mechanism. indexing_suite
derives from the <tt>def_arg</tt> base class and provides the requisite
<strong> <tt>visit</tt></strong> member function taking in a const reference
to the class. </p></td>
</tr>
</table>
<br>
<div>
<hr>
<h2>vector_indexing_suite class</h2>
<pre>
template &lt;<br> class Container, <br> bool NoProxy = false,<br> class DerivedPolicies = unspecified_default<br> class vector_indexing_suite <br> : public indexing_suite&lt;Container, DerivedPolicies, NoProxy&gt;<br> {<br> public:<br> <br> typedef typename Container::value_type element_type;<br> typedef typename Container::value_type key_type;<br> typedef typename Container::size_type index_type;<br> typedef typename Container::size_type size_type;<br> typedef typename Container::difference_type difference_type;<br> <br> static element_type&amp; <br> get_item(Container&amp; container, index_type i);
static object
get_slice(Container&amp; container, index_type from, index_type to);
static void <br> set_item(Container&amp; container, index_type i, element_type const&amp; v);
static void
set_slice(Container&amp; container, index_type from,
index_type to, element_type const&amp; v);
template &lt;class Iter&gt;<br> static void <br> set_slice(Container&amp; container, index_type from, <br> index_type to, Iter first, Iter last);
static void
delete_item(Container&amp; container, index_type i);
static void
delete_slice(Container&amp; container, index_type from, index_type to);<br>
static size_t
size(Container&amp; container);
static bool
contains(Container&amp; container, key_type const&amp; key);
static index_type
convert_index(Container&amp; container, PyObject* i);
static index_type
adjust_index(index_type current, index_type from,
index_type to, size_type len);
};</pre>
<dl>
<dt><strong><tt>Container</tt></strong></dt>
<dd>The contianer type to be wrapped to Python.</dd>
<dt>&nbsp;</dt>
<dt><strong><tt>NoProxy</tt></strong></dt>
<dd>By default indexed elements have Python reference semantics and are returned
by proxy. This can be disabled by supplying <strong>true</strong> in the
<tt>NoProxy</tt> template parameter.</dd>
<dt><br>
<strong><tt>DerivedPolicies</tt></strong></dt>
<dd>The <tt>vector_indexing_suite</tt> may still be derived to further tweak
any of the predefined policies. Static polymorphism through CRTP (James
Coplien. &quot;Curiously Recurring Template Pattern&quot;. C++ Report, Feb.
1995) enables the base <tt>indexing_suite</tt> class to call policy function
of the most derived class.</dd>
</dl>
<hr>
&copy; Copyright Joel de Guzman 2003. Permission to copy, use, modify, sell
and distribute this document is granted provided this copyright notice appears
in all copies. This document is provided "as is" without express or implied
warranty, and with no claim as to its suitability for any purpose.</div>