xref: /netbsd-src/external/gpl3/gcc/dist/libstdc++-v3/doc/html/manual/strings.html (revision a3e9eb183fc82afd2c22c9f48bcba8e00d943a4f)
14fee23f9Smrg<?xml version="1.0" encoding="UTF-8" standalone="no"?>
2d79abf08Smrg<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Chapter 7.  Strings</title><meta name="generator" content="DocBook XSL Stylesheets Vsnapshot" /><meta name="keywords" content="ISO C++, library" /><meta name="keywords" content="ISO C++, runtime, library" /><link rel="home" href="../index.html" title="The GNU C++ Library" /><link rel="up" href="std_contents.html" title="Part II.  Standard Contents" /><link rel="prev" href="traits.html" title="Traits" /><link rel="next" href="localization.html" title="Chapter 8.  Localization" /></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Chapter 7. 
34fee23f9Smrg  Strings
44fee23f9Smrg
548fb7bfaSmrg</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="traits.html">Prev</a> </td><th width="60%" align="center">Part II. 
648fb7bfaSmrg    Standard Contents
748fb7bfaSmrg  </th><td width="20%" align="right"> <a accesskey="n" href="localization.html">Next</a></td></tr></table><hr /></div><div class="chapter"><div class="titlepage"><div><div><h2 class="title"><a id="std.strings"></a>Chapter 7. 
84fee23f9Smrg  Strings
94d5abbe8Smrg  <a id="id-1.3.4.5.1.1.1" class="indexterm"></a>
1048fb7bfaSmrg</h2></div></div></div><div class="toc"><p><strong>Table of Contents</strong></p><dl class="toc"><dt><span class="section"><a href="strings.html#std.strings.string">String Classes</a></span></dt><dd><dl><dt><span class="section"><a href="strings.html#strings.string.simple">Simple Transformations</a></span></dt><dt><span class="section"><a href="strings.html#strings.string.case">Case Sensitivity</a></span></dt><dt><span class="section"><a href="strings.html#strings.string.character_types">Arbitrary Character Types</a></span></dt><dt><span class="section"><a href="strings.html#strings.string.token">Tokenizing</a></span></dt><dt><span class="section"><a href="strings.html#strings.string.shrink">Shrink to Fit</a></span></dt><dt><span class="section"><a href="strings.html#strings.string.Cstring">CString (MFC)</a></span></dt></dl></dd></dl></div><div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="std.strings.string"></a>String Classes</h2></div></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="strings.string.simple"></a>Simple Transformations</h3></div></div></div><p>
1148fb7bfaSmrg      Here are Standard, simple, and portable ways to perform common
1248fb7bfaSmrg      transformations on a <code class="code">string</code> instance, such as
1348fb7bfaSmrg      "convert to all upper case." The word transformations
1448fb7bfaSmrg      is especially apt, because the standard template function
1548fb7bfaSmrg      <code class="code">transform&lt;&gt;</code> is used.
1648fb7bfaSmrg   </p><p>
1748fb7bfaSmrg     This code will go through some iterations.  Here's a simple
1848fb7bfaSmrg     version:
1948fb7bfaSmrg   </p><pre class="programlisting">
2048fb7bfaSmrg   #include &lt;string&gt;
2148fb7bfaSmrg   #include &lt;algorithm&gt;
2248fb7bfaSmrg   #include &lt;cctype&gt;      // old &lt;ctype.h&gt;
2348fb7bfaSmrg
2448fb7bfaSmrg   struct ToLower
2548fb7bfaSmrg   {
2648fb7bfaSmrg     char operator() (char c) const  { return std::tolower(c); }
2748fb7bfaSmrg   };
2848fb7bfaSmrg
2948fb7bfaSmrg   struct ToUpper
3048fb7bfaSmrg   {
3148fb7bfaSmrg     char operator() (char c) const  { return std::toupper(c); }
3248fb7bfaSmrg   };
3348fb7bfaSmrg
3448fb7bfaSmrg   int main()
3548fb7bfaSmrg   {
3648fb7bfaSmrg     std::string  s ("Some Kind Of Initial Input Goes Here");
3748fb7bfaSmrg
3848fb7bfaSmrg     // Change everything into upper case
3948fb7bfaSmrg     std::transform (s.begin(), s.end(), s.begin(), ToUpper());
4048fb7bfaSmrg
4148fb7bfaSmrg     // Change everything into lower case
4248fb7bfaSmrg     std::transform (s.begin(), s.end(), s.begin(), ToLower());
4348fb7bfaSmrg
4448fb7bfaSmrg     // Change everything back into upper case, but store the
4548fb7bfaSmrg     // result in a different string
4648fb7bfaSmrg     std::string  capital_s;
4748fb7bfaSmrg     capital_s.resize(s.size());
4848fb7bfaSmrg     std::transform (s.begin(), s.end(), capital_s.begin(), ToUpper());
4948fb7bfaSmrg   }
5048fb7bfaSmrg   </pre><p>
5148fb7bfaSmrg     <span class="emphasis"><em>Note</em></span> that these calls all
5248fb7bfaSmrg      involve the global C locale through the use of the C functions
5348fb7bfaSmrg      <code class="code">toupper/tolower</code>.  This is absolutely guaranteed to work --
5448fb7bfaSmrg      but <span class="emphasis"><em>only</em></span> if the string contains <span class="emphasis"><em>only</em></span> characters
5548fb7bfaSmrg      from the basic source character set, and there are <span class="emphasis"><em>only</em></span>
5648fb7bfaSmrg      96 of those.  Which means that not even all English text can be
5748fb7bfaSmrg      represented (certain British spellings, proper names, and so forth).
5848fb7bfaSmrg      So, if all your input forevermore consists of only those 96
5948fb7bfaSmrg      characters (hahahahahaha), then you're done.
6048fb7bfaSmrg   </p><p><span class="emphasis"><em>Note</em></span> that the
6148fb7bfaSmrg      <code class="code">ToUpper</code> and <code class="code">ToLower</code> function objects
6248fb7bfaSmrg      are needed because <code class="code">toupper</code> and <code class="code">tolower</code>
6348fb7bfaSmrg      are overloaded names (declared in <code class="code">&lt;cctype&gt;</code> and
6448fb7bfaSmrg      <code class="code">&lt;locale&gt;</code>) so the template-arguments for
6548fb7bfaSmrg      <code class="code">transform&lt;&gt;</code> cannot be deduced, as explained in
6648fb7bfaSmrg      <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2002-11/msg00180.html" target="_top">this
6748fb7bfaSmrg      message</a>.
6848fb7bfaSmrg
6948fb7bfaSmrg      At minimum, you can write short wrappers like
7048fb7bfaSmrg   </p><pre class="programlisting">
7148fb7bfaSmrg   char toLower (char c)
7248fb7bfaSmrg   {
73*a3e9eb18Smrg      // std::tolower(c) is undefined if c &lt; 0 so cast to unsigned char.
74*a3e9eb18Smrg      return std::tolower((unsigned char)c);
7548fb7bfaSmrg   } </pre><p>(Thanks to James Kanze for assistance and suggestions on all of this.)
7648fb7bfaSmrg   </p><p>Another common operation is trimming off excess whitespace.  Much
7748fb7bfaSmrg      like transformations, this task is trivial with the use of string's
7848fb7bfaSmrg      <code class="code">find</code> family.  These examples are broken into multiple
7948fb7bfaSmrg      statements for readability:
8048fb7bfaSmrg   </p><pre class="programlisting">
8148fb7bfaSmrg   std::string  str (" \t blah blah blah    \n ");
8248fb7bfaSmrg
8348fb7bfaSmrg   // trim leading whitespace
8448fb7bfaSmrg   string::size_type  notwhite = str.find_first_not_of(" \t\n");
8548fb7bfaSmrg   str.erase(0,notwhite);
8648fb7bfaSmrg
8748fb7bfaSmrg   // trim trailing whitespace
8848fb7bfaSmrg   notwhite = str.find_last_not_of(" \t\n");
8948fb7bfaSmrg   str.erase(notwhite+1); </pre><p>Obviously, the calls to <code class="code">find</code> could be inserted directly
9048fb7bfaSmrg      into the calls to <code class="code">erase</code>, in case your compiler does not
9148fb7bfaSmrg      optimize named temporaries out of existence.
9248fb7bfaSmrg   </p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="strings.string.case"></a>Case Sensitivity</h3></div></div></div><p>
9348fb7bfaSmrg    </p><p>The well-known-and-if-it-isn't-well-known-it-ought-to-be
9448fb7bfaSmrg      <a class="link" href="http://www.gotw.ca/gotw/" target="_top">Guru of the Week</a>
9548fb7bfaSmrg      discussions held on Usenet covered this topic in January of 1998.
9648fb7bfaSmrg      Briefly, the challenge was, <span class="quote">“<span class="quote">write a 'ci_string' class which
9748fb7bfaSmrg      is identical to the standard 'string' class, but is
9848fb7bfaSmrg      case-insensitive in the same way as the (common but nonstandard)
9948fb7bfaSmrg      C function stricmp()</span>”</span>.
10048fb7bfaSmrg   </p><pre class="programlisting">
10148fb7bfaSmrg   ci_string s( "AbCdE" );
10248fb7bfaSmrg
10348fb7bfaSmrg   // case insensitive
10448fb7bfaSmrg   assert( s == "abcde" );
10548fb7bfaSmrg   assert( s == "ABCDE" );
10648fb7bfaSmrg
10748fb7bfaSmrg   // still case-preserving, of course
10848fb7bfaSmrg   assert( strcmp( s.c_str(), "AbCdE" ) == 0 );
10948fb7bfaSmrg   assert( strcmp( s.c_str(), "abcde" ) != 0 ); </pre><p>The solution is surprisingly easy.  The original answer was
11048fb7bfaSmrg   posted on Usenet, and a revised version appears in Herb Sutter's
11148fb7bfaSmrg   book <span class="emphasis"><em>Exceptional C++</em></span> and on his website as <a class="link" href="http://www.gotw.ca/gotw/029.htm" target="_top">GotW 29</a>.
11248fb7bfaSmrg   </p><p>See?  Told you it was easy!</p><p>
11348fb7bfaSmrg     <span class="emphasis"><em>Added June 2000:</em></span> The May 2000 issue of C++
11448fb7bfaSmrg     Report contains a fascinating <a class="link" href="http://lafstern.org/matt/col2_new.pdf" target="_top"> article</a> by
11548fb7bfaSmrg     Matt Austern (yes, <span class="emphasis"><em>the</em></span> Matt Austern) on why
11648fb7bfaSmrg     case-insensitive comparisons are not as easy as they seem, and
11748fb7bfaSmrg     why creating a class is the <span class="emphasis"><em>wrong</em></span> way to go
11848fb7bfaSmrg     about it in production code.  (The GotW answer mentions one of
11948fb7bfaSmrg     the principle difficulties; his article mentions more.)
12048fb7bfaSmrg   </p><p>Basically, this is "easy" only if you ignore some things,
12148fb7bfaSmrg      things which may be too important to your program to ignore.  (I chose
12248fb7bfaSmrg      to ignore them when originally writing this entry, and am surprised
12348fb7bfaSmrg      that nobody ever called me on it...)  The GotW question and answer
12448fb7bfaSmrg      remain useful instructional tools, however.
12548fb7bfaSmrg   </p><p><span class="emphasis"><em>Added September 2000:</em></span>  James Kanze provided a link to a
12648fb7bfaSmrg      <a class="link" href="http://www.unicode.org/reports/tr21/tr21-5.html" target="_top">Unicode
12748fb7bfaSmrg      Technical Report discussing case handling</a>, which provides some
12848fb7bfaSmrg      very good information.
12948fb7bfaSmrg   </p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="strings.string.character_types"></a>Arbitrary Character Types</h3></div></div></div><p>
13048fb7bfaSmrg    </p><p>The <code class="code">std::basic_string</code> is tantalizingly general, in that
13148fb7bfaSmrg      it is parameterized on the type of the characters which it holds.
13248fb7bfaSmrg      In theory, you could whip up a Unicode character class and instantiate
13348fb7bfaSmrg      <code class="code">std::basic_string&lt;my_unicode_char&gt;</code>, or assuming
13448fb7bfaSmrg      that integers are wider than characters on your platform, maybe just
13548fb7bfaSmrg      declare variables of type <code class="code">std::basic_string&lt;int&gt;</code>.
13648fb7bfaSmrg   </p><p>That's the theory.  Remember however that basic_string has additional
13748fb7bfaSmrg      type parameters, which take default arguments based on the character
13848fb7bfaSmrg      type (called <code class="code">CharT</code> here):
13948fb7bfaSmrg   </p><pre class="programlisting">
14048fb7bfaSmrg      template &lt;typename CharT,
14148fb7bfaSmrg		typename Traits = char_traits&lt;CharT&gt;,
14248fb7bfaSmrg		typename Alloc = allocator&lt;CharT&gt; &gt;
14348fb7bfaSmrg      class basic_string { .... };</pre><p>Now, <code class="code">allocator&lt;CharT&gt;</code> will probably Do The Right
14448fb7bfaSmrg      Thing by default, unless you need to implement your own allocator
14548fb7bfaSmrg      for your characters.
14648fb7bfaSmrg   </p><p>But <code class="code">char_traits</code> takes more work.  The char_traits
14748fb7bfaSmrg      template is <span class="emphasis"><em>declared</em></span> but not <span class="emphasis"><em>defined</em></span>.
14848fb7bfaSmrg      That means there is only
14948fb7bfaSmrg   </p><pre class="programlisting">
15048fb7bfaSmrg      template &lt;typename CharT&gt;
15148fb7bfaSmrg	struct char_traits
15248fb7bfaSmrg	{
15348fb7bfaSmrg	    static void foo (type1 x, type2 y);
15448fb7bfaSmrg	    ...
15548fb7bfaSmrg	};</pre><p>and functions such as char_traits&lt;CharT&gt;::foo() are not
15648fb7bfaSmrg      actually defined anywhere for the general case.  The C++ standard
15748fb7bfaSmrg      permits this, because writing such a definition to fit all possible
15848fb7bfaSmrg      CharT's cannot be done.
15948fb7bfaSmrg   </p><p>The C++ standard also requires that char_traits be specialized for
16048fb7bfaSmrg      instantiations of <code class="code">char</code> and <code class="code">wchar_t</code>, and it
16148fb7bfaSmrg      is these template specializations that permit entities like
16248fb7bfaSmrg      <code class="code">basic_string&lt;char,char_traits&lt;char&gt;&gt;</code> to work.
16348fb7bfaSmrg   </p><p>If you want to use character types other than char and wchar_t,
16448fb7bfaSmrg      such as <code class="code">unsigned char</code> and <code class="code">int</code>, you will
16548fb7bfaSmrg      need suitable specializations for them.  For a time, in earlier
16648fb7bfaSmrg      versions of GCC, there was a mostly-correct implementation that
16748fb7bfaSmrg      let programmers be lazy but it broke under many situations, so it
16848fb7bfaSmrg      was removed.  GCC 3.4 introduced a new implementation that mostly
16948fb7bfaSmrg      works and can be specialized even for <code class="code">int</code> and other
17048fb7bfaSmrg      built-in types.
17148fb7bfaSmrg   </p><p>If you want to use your own special character class, then you have
17248fb7bfaSmrg      <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2002-08/msg00163.html" target="_top">a lot
17348fb7bfaSmrg      of work to do</a>, especially if you with to use i18n features
17448fb7bfaSmrg      (facets require traits information but don't have a traits argument).
17548fb7bfaSmrg   </p><p>Another example of how to specialize char_traits was given <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2002-08/msg00260.html" target="_top">on the
17648fb7bfaSmrg      mailing list</a> and at a later date was put into the file <code class="code">
17748fb7bfaSmrg      include/ext/pod_char_traits.h</code>.  We agree
17848fb7bfaSmrg      that the way it's used with basic_string (scroll down to main())
17948fb7bfaSmrg      doesn't look nice, but that's because <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2002-08/msg00236.html" target="_top">the
18048fb7bfaSmrg      nice-looking first attempt</a> turned out to <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2002-08/msg00242.html" target="_top">not
18148fb7bfaSmrg      be conforming C++</a>, due to the rule that CharT must be a POD.
18248fb7bfaSmrg      (See how tricky this is?)
18348fb7bfaSmrg   </p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="strings.string.token"></a>Tokenizing</h3></div></div></div><p>
18448fb7bfaSmrg    </p><p>The Standard C (and C++) function <code class="code">strtok()</code> leaves a lot to
18548fb7bfaSmrg      be desired in terms of user-friendliness.  It's unintuitive, it
18648fb7bfaSmrg      destroys the character string on which it operates, and it requires
18748fb7bfaSmrg      you to handle all the memory problems.  But it does let the client
18848fb7bfaSmrg      code decide what to use to break the string into pieces; it allows
18948fb7bfaSmrg      you to choose the "whitespace," so to speak.
19048fb7bfaSmrg   </p><p>A C++ implementation lets us keep the good things and fix those
19148fb7bfaSmrg      annoyances.  The implementation here is more intuitive (you only
19248fb7bfaSmrg      call it once, not in a loop with varying argument), it does not
19348fb7bfaSmrg      affect the original string at all, and all the memory allocation
19448fb7bfaSmrg      is handled for you.
19548fb7bfaSmrg   </p><p>It's called stringtok, and it's a template function. Sources are
19648fb7bfaSmrg   as below, in a less-portable form than it could be, to keep this
19748fb7bfaSmrg   example simple (for example, see the comments on what kind of
19848fb7bfaSmrg   string it will accept).
19948fb7bfaSmrg   </p><pre class="programlisting">
20048fb7bfaSmrg#include &lt;string&gt;
20148fb7bfaSmrgtemplate &lt;typename Container&gt;
20248fb7bfaSmrgvoid
20348fb7bfaSmrgstringtok(Container &amp;container, string const &amp;in,
20448fb7bfaSmrg	  const char * const delimiters = " \t\n")
20548fb7bfaSmrg{
20648fb7bfaSmrg    const string::size_type len = in.length();
20748fb7bfaSmrg	  string::size_type i = 0;
20848fb7bfaSmrg
20948fb7bfaSmrg    while (i &lt; len)
21048fb7bfaSmrg    {
21148fb7bfaSmrg	// Eat leading whitespace
21248fb7bfaSmrg	i = in.find_first_not_of(delimiters, i);
21348fb7bfaSmrg	if (i == string::npos)
21448fb7bfaSmrg	  return;   // Nothing left but white space
21548fb7bfaSmrg
21648fb7bfaSmrg	// Find the end of the token
21748fb7bfaSmrg	string::size_type j = in.find_first_of(delimiters, i);
21848fb7bfaSmrg
21948fb7bfaSmrg	// Push token
22048fb7bfaSmrg	if (j == string::npos)
22148fb7bfaSmrg	{
22248fb7bfaSmrg	  container.push_back(in.substr(i));
22348fb7bfaSmrg	  return;
22448fb7bfaSmrg	}
22548fb7bfaSmrg	else
22648fb7bfaSmrg	  container.push_back(in.substr(i, j-i));
22748fb7bfaSmrg
22848fb7bfaSmrg	// Set up for next loop
22948fb7bfaSmrg	i = j + 1;
23048fb7bfaSmrg    }
23148fb7bfaSmrg}
23248fb7bfaSmrg</pre><p>
23348fb7bfaSmrg     The author uses a more general (but less readable) form of it for
23448fb7bfaSmrg     parsing command strings and the like.  If you compiled and ran this
23548fb7bfaSmrg     code using it:
23648fb7bfaSmrg   </p><pre class="programlisting">
23748fb7bfaSmrg   std::list&lt;string&gt;  ls;
23848fb7bfaSmrg   stringtok (ls, " this  \t is\t\n  a test  ");
23948fb7bfaSmrg   for (std::list&lt;string&gt;const_iterator i = ls.begin();
24048fb7bfaSmrg	i != ls.end(); ++i)
24148fb7bfaSmrg   {
24248fb7bfaSmrg       std::cerr &lt;&lt; ':' &lt;&lt; (*i) &lt;&lt; ":\n";
24348fb7bfaSmrg   } </pre><p>You would see this as output:
24448fb7bfaSmrg   </p><pre class="programlisting">
24548fb7bfaSmrg   :this:
24648fb7bfaSmrg   :is:
24748fb7bfaSmrg   :a:
24848fb7bfaSmrg   :test: </pre><p>with all the whitespace removed.  The original <code class="code">s</code> is still
24948fb7bfaSmrg      available for use, <code class="code">ls</code> will clean up after itself, and
25048fb7bfaSmrg      <code class="code">ls.size()</code> will return how many tokens there were.
25148fb7bfaSmrg   </p><p>As always, there is a price paid here, in that stringtok is not
25248fb7bfaSmrg      as fast as strtok.  The other benefits usually outweigh that, however.
25348fb7bfaSmrg   </p><p><span class="emphasis"><em>Added February 2001:</em></span>  Mark Wilden pointed out that the
25448fb7bfaSmrg      standard <code class="code">std::getline()</code> function can be used with standard
25548fb7bfaSmrg      <code class="code">istringstreams</code> to perform
25648fb7bfaSmrg      tokenizing as well.  Build an istringstream from the input text,
25748fb7bfaSmrg      and then use std::getline with varying delimiters (the three-argument
25848fb7bfaSmrg      signature) to extract tokens into a string.
25948fb7bfaSmrg   </p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="strings.string.shrink"></a>Shrink to Fit</h3></div></div></div><p>
26048fb7bfaSmrg    </p><p>From GCC 3.4 calling <code class="code">s.reserve(res)</code> on a
26148fb7bfaSmrg      <code class="code">string s</code> with <code class="code">res &lt; s.capacity()</code> will
26248fb7bfaSmrg      reduce the string's capacity to <code class="code">std::max(s.size(), res)</code>.
26348fb7bfaSmrg   </p><p>This behaviour is suggested, but not required by the standard. Prior
26448fb7bfaSmrg      to GCC 3.4 the following alternative can be used instead
26548fb7bfaSmrg   </p><pre class="programlisting">
26648fb7bfaSmrg      std::string(str.data(), str.size()).swap(str);
26748fb7bfaSmrg   </pre><p>This is similar to the idiom for reducing
26848fb7bfaSmrg      a <code class="code">vector</code>'s memory usage
26948fb7bfaSmrg      (see <a class="link" href="../faq.html#faq.size_equals_capacity" title="7.8.">this FAQ
27048fb7bfaSmrg      entry</a>) but the regular copy constructor cannot be used
2714d5abbe8Smrg      because libstdc++'s <code class="code">string</code> is Copy-On-Write in GCC 3.
27248fb7bfaSmrg   </p><p>In <a class="link" href="status.html#status.iso.2011" title="C++ 2011">C++11</a> mode you can call
27348fb7bfaSmrg      <code class="code">s.shrink_to_fit()</code> to achieve the same effect as
27448fb7bfaSmrg      <code class="code">s.reserve(s.size())</code>.
27548fb7bfaSmrg   </p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="strings.string.Cstring"></a>CString (MFC)</h3></div></div></div><p>
27648fb7bfaSmrg    </p><p>A common lament seen in various newsgroups deals with the Standard
27748fb7bfaSmrg      string class as opposed to the Microsoft Foundation Class called
27848fb7bfaSmrg      CString.  Often programmers realize that a standard portable
27948fb7bfaSmrg      answer is better than a proprietary nonportable one, but in porting
28048fb7bfaSmrg      their application from a Win32 platform, they discover that they
28148fb7bfaSmrg      are relying on special functions offered by the CString class.
28248fb7bfaSmrg   </p><p>Things are not as bad as they seem.  In
28348fb7bfaSmrg      <a class="link" href="http://gcc.gnu.org/ml/gcc/1999-04n/msg00236.html" target="_top">this
28448fb7bfaSmrg      message</a>, Joe Buck points out a few very important things:
28548fb7bfaSmrg   </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>The Standard <code class="code">string</code> supports all the operations
28648fb7bfaSmrg	     that CString does, with three exceptions.
28748fb7bfaSmrg	 </p></li><li class="listitem"><p>Two of those exceptions (whitespace trimming and case
28848fb7bfaSmrg	     conversion) are trivial to implement.  In fact, we do so
28948fb7bfaSmrg	     on this page.
29048fb7bfaSmrg	 </p></li><li class="listitem"><p>The third is <code class="code">CString::Format</code>, which allows formatting
29148fb7bfaSmrg	     in the style of <code class="code">sprintf</code>.  This deserves some mention:
29248fb7bfaSmrg	 </p></li></ul></div><p>
29348fb7bfaSmrg      The old libg++ library had a function called form(), which did much
29448fb7bfaSmrg      the same thing.  But for a Standard solution, you should use the
29548fb7bfaSmrg      stringstream classes.  These are the bridge between the iostream
29648fb7bfaSmrg      hierarchy and the string class, and they operate with regular
29748fb7bfaSmrg      streams seamlessly because they inherit from the iostream
29848fb7bfaSmrg      hierarchy.  An quick example:
29948fb7bfaSmrg   </p><pre class="programlisting">
30048fb7bfaSmrg   #include &lt;iostream&gt;
30148fb7bfaSmrg   #include &lt;string&gt;
30248fb7bfaSmrg   #include &lt;sstream&gt;
30348fb7bfaSmrg
30448fb7bfaSmrg   string f (string&amp; incoming)     // incoming is "foo  N"
30548fb7bfaSmrg   {
30648fb7bfaSmrg       istringstream   incoming_stream(incoming);
30748fb7bfaSmrg       string          the_word;
30848fb7bfaSmrg       int             the_number;
30948fb7bfaSmrg
31048fb7bfaSmrg       incoming_stream &gt;&gt; the_word        // extract "foo"
31148fb7bfaSmrg		       &gt;&gt; the_number;     // extract N
31248fb7bfaSmrg
31348fb7bfaSmrg       ostringstream   output_stream;
31448fb7bfaSmrg       output_stream &lt;&lt; "The word was " &lt;&lt; the_word
31548fb7bfaSmrg		     &lt;&lt; " and 3*N was " &lt;&lt; (3*the_number);
31648fb7bfaSmrg
31748fb7bfaSmrg       return output_stream.str();
31848fb7bfaSmrg   } </pre><p>A serious problem with CString is a design bug in its memory
31948fb7bfaSmrg      allocation.  Specifically, quoting from that same message:
32048fb7bfaSmrg   </p><pre class="programlisting">
32148fb7bfaSmrg   CString suffers from a common programming error that results in
32248fb7bfaSmrg   poor performance.  Consider the following code:
32348fb7bfaSmrg
32448fb7bfaSmrg   CString n_copies_of (const CString&amp; foo, unsigned n)
32548fb7bfaSmrg   {
32648fb7bfaSmrg	   CString tmp;
32748fb7bfaSmrg	   for (unsigned i = 0; i &lt; n; i++)
32848fb7bfaSmrg		   tmp += foo;
32948fb7bfaSmrg	   return tmp;
33048fb7bfaSmrg   }
33148fb7bfaSmrg
33248fb7bfaSmrg   This function is O(n^2), not O(n).  The reason is that each +=
33348fb7bfaSmrg   causes a reallocation and copy of the existing string.  Microsoft
33448fb7bfaSmrg   applications are full of this kind of thing (quadratic performance
33548fb7bfaSmrg   on tasks that can be done in linear time) -- on the other hand,
33648fb7bfaSmrg   we should be thankful, as it's created such a big market for high-end
33748fb7bfaSmrg   ix86 hardware. :-)
33848fb7bfaSmrg
33948fb7bfaSmrg   If you replace CString with string in the above function, the
34048fb7bfaSmrg   performance is O(n).
34148fb7bfaSmrg   </pre><p>Joe Buck also pointed out some other things to keep in mind when
34248fb7bfaSmrg      comparing CString and the Standard string class:
34348fb7bfaSmrg   </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>CString permits access to its internal representation; coders
34448fb7bfaSmrg	     who exploited that may have problems moving to <code class="code">string</code>.
34548fb7bfaSmrg	 </p></li><li class="listitem"><p>Microsoft ships the source to CString (in the files
34648fb7bfaSmrg	     MFC\SRC\Str{core,ex}.cpp), so you could fix the allocation
34748fb7bfaSmrg	     bug and rebuild your MFC libraries.
34848fb7bfaSmrg	     <span class="emphasis"><em><span class="emphasis"><em>Note:</em></span> It looks like the CString shipped
34948fb7bfaSmrg	     with VC++6.0 has fixed this, although it may in fact have been
35048fb7bfaSmrg	     one of the VC++ SPs that did it.</em></span>
35148fb7bfaSmrg	 </p></li><li class="listitem"><p><code class="code">string</code> operations like this have O(n) complexity
35248fb7bfaSmrg	     <span class="emphasis"><em>if the implementors do it correctly</em></span>.  The libstdc++
35348fb7bfaSmrg	     implementors did it correctly.  Other vendors might not.
3544d5abbe8Smrg	 </p></li><li class="listitem"><p>While parts of the SGI STL are used in libstdc++, their
35548fb7bfaSmrg	     string class is not.  The SGI <code class="code">string</code> is essentially
35648fb7bfaSmrg	     <code class="code">vector&lt;char&gt;</code> and does not do any reference
35748fb7bfaSmrg	     counting like libstdc++'s does.  (It is O(n), though.)
35848fb7bfaSmrg	     So if you're thinking about SGI's string or rope classes,
35948fb7bfaSmrg	     you're now looking at four possibilities:  CString, the
36048fb7bfaSmrg	     libstdc++ string, the SGI string, and the SGI rope, and this
36148fb7bfaSmrg	     is all before any allocator or traits customizations!  (More
36248fb7bfaSmrg	     choices than you can shake a stick at -- want fries with that?)
36348fb7bfaSmrg	 </p></li></ul></div></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="traits.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="std_contents.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="localization.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Traits </td><td width="20%" align="center"><a accesskey="h" href="../index.html">Home</a></td><td width="40%" align="right" valign="top"> Chapter 8. 
36448fb7bfaSmrg  Localization
36548fb7bfaSmrg
36648fb7bfaSmrg</td></tr></table></div></body></html>