<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <!-- Copyright David Abrahams 2006. Distributed under the Boost --> <!-- Software License, Version 1.0. (See accompanying --> <!-- file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) --> <html> <head> <meta name="generator" content= "HTML Tidy for Windows (vers 1st February 2003), see www.w3.org"> <meta http-equiv="Content-Type" content="text/html; charset=us-ascii"> <link rel="stylesheet" type="text/css" href="../boost.css"> <title> Indexing Support </title> </head> <body> <table border="0" cellpadding="7" cellspacing="0" width="100%" summary="header"> <tr> <td valign="top" width="300"> <h3> <a href="../../../../index.htm"><img height="86" width="277" alt="C++ Boost" src="../../../../boost.png" border= "0"></a> </h3> </td> <td valign="top"> <h1 align="center"> <a href="../index.html">Boost.Python</a> </h1> <h2> Headers <boost/python/indexing/indexing_suite.hpp><br> <boost/python/indexing/vector_indexing_suite.hpp></h2> </td> </tr> </table> <hr> <h2> Contents </h2> <dl class="page-index"> <dt> <a href="#introduction">Introduction</a> </dt> <dt> <a href="#interface">Interface</a> </dt> <dd> <dl class="page-index"> <dt> <a href="#indexing_suite">indexing_suite</a> </dt> <dt> <a href="#indexing_suite_subclasses">indexing_suite sub-classes</a> </dt> <dd> <dl class="page-index"> <dt> <a href="#vector_indexing_suite">vector_indexing_suite</a> </dt> </dl> </dd> </dl> </dd> </dl> <dl> <dt> <a href="#indexing_suite_class">indexing_suite class</a> </dt> <dt> <a href="#vector_indexing_suite_class">vector_indexing_suite class<br> </a><a href="#map_indexing_suite_class">map_indexing_suite class</a> </dt> </dl> <hr> <h2> <a name="introduction" id="introduction"></a>Introduction </h2> <p> Indexing is a Boost Python facility 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> <p> While Boost Python has all the facilities needed 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. 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. These involve 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> <p> The goals: </p> <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 (<tt>__getitem__</tt>) such that <tt>c[i]</tt> can be mutable. Require: <div> <pre> val = c[i] c[i].m() val == c[i] </pre> </div>where <tt>m</tt> is a non-const (mutating) member function (method). </li> <li> Return safe references from <tt>__getitem__</tt> 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> Accept Python container arguments (e.g. lists, tuples) wherever appropriate. </li> <li> Allow for extensibility through re-definable policy classes. </li> <li> Provide predefined support for the most common STL and STL like indexable containers. </li> </ul> <hr> <h2> <a name="interface"></a>The Boost.Python Indexing Interface</h2> <h3> <a name="indexing_suite"></a>indexing_suite [ Header <boost/python/indexing/indexing_suite.hpp> ]</h3> <p> The <tt>indexing_suite</tt> class is the base 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 (taken from the Python reference: <a href= "http://www.python.org/doc/current/ref/sequence-types.html">Emulating container types</a>): </p> <dl> <dd> <dl> <dt> <b><a name="l2h-126"><tt class= "method">__len__</tt></a></b>(<var>self</var>) </dt> <dd> Called to implement the built-in function <tt class= "function">len()</tt><a name="l2h-134"> </a> Should return the length of the object, an integer <code>>=</code> 0. Also, an object that doesn't define a <tt class= "method">__nonzero__()</tt> method and whose <tt class= "method">__len__()</tt> method returns zero is considered to be false in a Boolean context. <a name="l2h-128"> </a> </dd> </dl> <dl> <dt> <b><a name="l2h-129"><tt class= "method">__getitem__</tt></a></b>(<var>self, key</var>) </dt> <dd> Called to implement evaluation of <code><var>self</var>[<var>key</var>]</code>. For sequence types, the accepted keys should be integers and slice objects.<a name="l2h-135"> </a> Note that the special interpretation of negative indexes (if the class wishes to emulate a sequence type) is up to the <tt class= "method">__getitem__()</tt> method. If <var>key</var> is of an inappropriate type, <tt class="exception">TypeError</tt> may be raised; if of a value outside the set of indexes for the sequence (after any special interpretation of negative values), <tt class="exception">IndexError</tt> should be raised. <span class="note"><b class="label">Note:</b> <tt class="keyword">for</tt> loops expect that an <tt class= "exception">IndexError</tt> will be raised for illegal indexes to allow proper detection of the end of the sequence.</span> </dd> </dl> <dl> <dt> <b><a name="l2h-130"><tt class= "method">__setitem__</tt></a></b>(<var>self, key, value</var>) </dt> <dd> Called to implement assignment to <code><var>self</var>[<var>key</var>]</code>. Same note as for <tt class="method">__getitem__()</tt>. 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 <var>key</var> values as for the <tt class= "method">__getitem__()</tt> method. </dd> </dl> <dl> <dt> <b><a name="l2h-131"><tt class= "method">__delitem__</tt></a></b>(<var>self, key</var>) </dt> <dd> Called to implement deletion of <code><var>self</var>[<var>key</var>]</code>. Same note as for <tt class="method">__getitem__()</tt>. 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 <var>key</var> values as for the <tt class="method">__getitem__()</tt> method. </dd> </dl> <dl> <dt> <b><a name="l2h-132"><tt class= "method">__iter__</tt></a></b>(<var>self</var>) </dt> <dd> 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 <tt class= "method">iterkeys()</tt>. <p> Iterator objects also need to implement this method; they are required to return themselves. For more information on iterator objects, see ``<a class="ulink" href= "http://www.python.org/doc/current/lib/typeiter.html">Iterator Types</a>'' in the <em class="citetitle"><a href= "http://www.python.org/doc/current/lib/lib.html" title= "Python Library Reference">Python Library Reference</a></em>. </p> </dd> </dl> <dl> <dt> <b><a name="l2h-133"><tt class= "method">__contains__</tt></a></b>(<var>self, item</var>) </dt> <dd> Called to implement membership test operators. Should return true if <var>item</var> is in <var>self</var>, false otherwise. For mapping objects, this should consider the keys of the mapping rather than the values or the key-item pairs. </dd> </dl> </dd> </dl> <h3> <a name="indexing_suite_subclasses"></a>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, we can refine the predefined suites to suit our needs. </p> <h3> <a name="vector_indexing_suite"></a>vector_indexing_suite [ Header <boost/python/indexing/vector_indexing_suite.hpp> ] </h3> <p> The <tt>vector_indexing_suite</tt> class is a predefined <tt>indexing_suite</tt> derived class designed to wrap <tt>std::vector</tt> (and <tt>std::vector</tt> like [i.e. a class with std::vector interface]) classes. It provides all the policies required by the <tt>indexing_suite</tt>. </p> <p> Example usage: </p> <pre> class X {...}; ... class_<std::vector<X> >("XVec") .def(vector_indexing_suite<std::vector<X> >()) ; </pre> <p> <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> <h3><a name="map_indexing_suite" id="map_indexing_suite"></a>map_indexing_suite [ Header <boost/python/indexing/map_indexing_suite.hpp> ] </h3> <p> The <tt>map_indexing_suite</tt> class is a predefined <tt>indexing_suite</tt> derived class designed to wrap <tt>std::map</tt> (and <tt>std::map</tt> like [i.e. a class with std::map interface]) classes. It provides all the policies required by the <tt>indexing_suite</tt>. </p> <p> Example usage: </p> <pre> class X {...}; ... class_<std::map<X> >("XMap") .def(map_indexing_suite<std::map<X> >()) ; </pre> <p> By default indexed elements are returned by proxy. This can be disabled by supplying <tt>true</tt> in the NoProxy template parameter. <tt>XMap</tt> is now a full-fledged Python container (see the <a href="../../test/map_indexing_suite.cpp">example in full</a>, along with its <a href="../../test/map_indexing_suite.py">python test</a>).</p> <hr> <h2> <a name="indexing_suite_class"></a>indexing_suite class </h2> <h2> <tt>indexing_suite<<br> </tt><tt>class Container<br> , class DerivedPolicies<font color="#007F00"><br> </font></tt> <tt>, bool NoProxy<br> , bool NoSlice<br> </tt><tt>, class Data<br> , class Index<br> </tt><tt>, class Key</tt></h2> <table width="100%" border="1"> <tr> <td> <strong>Template Parameter</strong><br> </td> <td> <strong>Requirements</strong> </td> <td> <strong>Semantics</strong> </td> <td> <strong>Default</strong> </td> </tr> <tr> <td> <font color="#007F00"><tt>Container</tt></font> </td> <td> A class type </td> <td> The container type to be wrapped to Python. </td> <td> </td> </tr> <tr> <td> <font color="#007F00"><tt>DerivedPolicies</tt></font> </td> <td> A subclass of indexing_suite </td> <td> Derived classes provide the policy hooks. See <a href= "#DerivedPolicies">DerivedPolicies</a> below. </td> <td> </td> </tr> <tr> <td> <font color="#007F00"><tt>NoProxy</tt></font> </td> <td> A boolean </td> <td> 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. </td> <td> false </td> </tr> <tr> <td> <font color="#007F00"><tt>NoSlice</tt></font> </td> <td> A boolean </td> <td> Do not allow slicing. </td> <td> false </td> </tr> <tr> <td> <font color="#007F00"><tt>Data</tt></font> </td> <td> </td> <td> The container's data type. </td> <td> <tt>Container::value_type</tt> </td> </tr> <tr> <td> <font color="#007F00"><tt>Index</tt></font> </td> <td> </td> <td> The container's index type. </td> <td> <tt>Container::size_type</tt> </td> </tr> <tr> <td> <font color="#007F00"><tt>Key</tt></font> </td> <td> </td> <td> The container's key type. </td> <td> <tt>Container::value_type</tt> </td> </tr> </table> <pre> template <<br> class Container , class DerivedPolicies , bool NoProxy = false<br> , bool NoSlice = false , class Data = typename Container::value_type , class Index = typename Container::size_type , class Key = typename Container::value_type ><br> class indexing_suite : unspecified { public: indexing_suite(); // default constructor } </pre> <h2> <tt><a name="DerivedPolicies"></a>DerivedPolicies</tt> </h2> <dl> <dd> Derived classes provide the hooks needed by the <tt>indexing_suite:</tt> </dd> </dl> <pre> data_type& get_item(Container& container, index_type i); static object get_slice(Container& container, index_type from, index_type to); static void set_item(Container& container, index_type i, data_type const& v); static void set_slice( Container& container, index_type from, index_type to, data_type const& v ); template <class Iter> static void<br> set_slice(Container& container, index_type from, index_type to, Iter first, Iter last ); static void delete_item(Container& container, index_type i); static void delete_slice(Container& container, index_type from, index_type to); static size_t size(Container& container); template <class T> static bool contains(Container& container, T const& val); static index_type convert_index(Container& container, PyObject* i); static index_type adjust_index(index_type current, index_type from, index_type to, size_type len ); </pre> <blockquote> <p> Most of these policies are self explanatory. <tt>However, <strong>convert_index</strong></tt> and <tt><strong>adjust_index</strong></tt> deserve some explanation. </p> <p> <strong><tt>convert_index</tt></strong> converts a Python index into a C++ index that the container can handle. For instance, negative indexes in Python, by convention, start counting 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] ---> [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> <div> <hr> <h2> <a name="vector_indexing_suite_class"></a>vector_indexing_suite class </h2> <h3> Class template <tt><br> vector_indexing_suite<<br> class <font color="#007F00">Container</font><br> , bool <font color="#007F00">NoProxy</font><br> , class <font color="#007F00">DerivedPolicies</font>></tt> </h3> <table width="100%" border="1"> <tr> <td> <strong>Template Parameter</strong><br> </td> <td> <strong>Requirements</strong> </td> <td> <strong>Semantics</strong> </td> <td> <strong>Default</strong> </td> </tr> <tr> <td> <font color="#007F00"><tt>Container</tt></font> </td> <td> A class type </td> <td> The container type to be wrapped to Python. </td> <td> </td> </tr> <tr> <td> <font color="#007F00"><tt>NoProxy</tt></font> </td> <td> A boolean </td> <td> 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. </td> <td> false </td> </tr> <tr> <td> <font color="#007F00"><tt>DerivedPolicies</tt></font> </td> <td> A subclass of indexing_suite </td> <td> The <tt>vector_indexing_suite</tt> may still be derived to further tweak any of the predefined policies. Static polymorphism through CRTP (James Coplien. "Curiously Recurring Template Pattern". C++ Report, Feb. 1995) enables the base <tt>indexing_suite</tt> class to call policy function of the most derived class </td> <td> </td> </tr> </table> <pre> template <<br> class Container,<br> bool NoProxy = false,<br> class DerivedPolicies = unspecified_default<br> class vector_indexing_suite : unspecified_base<br> {<br> public:<br><br> typedef typename Container::value_type data_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> data_type&<br> get_item(Container& container, index_type i); static object get_slice(Container& container, index_type from, index_type to); static void<br> set_item(Container& container, index_type i, data_type const& v); static void set_slice(Container& container, index_type from, index_type to, data_type const& v); template <class Iter><br> static void<br> set_slice(Container& container, index_type from,<br> index_type to, Iter first, Iter last); static void delete_item(Container& container, index_type i); static void delete_slice(Container& container, index_type from, index_type to);<br> static size_t size(Container& container); static bool contains(Container& container, key_type const& key); static index_type convert_index(Container& container, PyObject* i); static index_type adjust_index(index_type current, index_type from, index_type to, size_type len); }; </pre> <h2><a name="vector_indexing_suite_class"></a>map_indexing_suite class </h2> <h3> Class template <tt><br> map_indexing_suite<<br> class <font color="#007F00">Container</font><br> , bool <font color="#007F00">NoProxy</font><br> , class <font color="#007F00">DerivedPolicies</font>></tt> </h3> <table width="100%" border="1"> <tr> <td> <strong>Template Parameter</strong><br> </td> <td> <strong>Requirements</strong> </td> <td> <strong>Semantics</strong> </td> <td> <strong>Default</strong> </td> </tr> <tr> <td> <font color="#007F00"><tt>Container</tt></font> </td> <td> A class type </td> <td> The container type to be wrapped to Python. </td> <td> </td> </tr> <tr> <td> <font color="#007F00"><tt>NoProxy</tt></font> </td> <td> A boolean </td> <td> 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. </td> <td> false </td> </tr> <tr> <td> <font color="#007F00"><tt>DerivedPolicies</tt></font> </td> <td> A subclass of indexing_suite </td> <td> The <tt>vector_indexing_suite</tt> may still be derived to further tweak any of the predefined policies. Static polymorphism through CRTP (James Coplien. "Curiously Recurring Template Pattern". C++ Report, Feb. 1995) enables the base <tt>indexing_suite</tt> class to call policy function of the most derived class </td> <td> </td> </tr> </table> <pre> template <<br> class Container,<br> bool NoProxy = false,<br> class DerivedPolicies = unspecified_default<br> class map_indexing_suite : unspecified_base<br> {<br> public:<br><br> typedef typename Container::value_type value_type;<br> typedef typename Container::value_type::second_type data_type;<br> typedef typename Container::key_type key_type;<br> typedef typename Container::key_type index_type;<br> typedef typename Container::size_type size_type;<br> typedef typename Container::difference_type difference_type;<br><br> static data_type&<br> get_item(Container& container, index_type i); static void<br> set_item(Container& container, index_type i, data_type const& v); static void delete_item(Container& container, index_type i);<br> static size_t size(Container& container); static bool contains(Container& container, key_type const& key); static bool<br> compare_index(Container& container, index_type a, index_type b); <br> static index_type convert_index(Container& container, PyObject* i); }; </pre> <hr> © 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> </body> </html>