property_tree/detail/xml_parser_utils.hpp

0001 // ----------------------------------------------------------------------------
0002 // Copyright (C) 2002-2006 Marcin Kalicinski
0003 //
0004 // Distributed under the Boost Software License, Version 1.0.
0005 // (See accompanying file LICENSE_1_0.txt or copy at
0006 // http://www.boost.org/LICENSE_1_0.txt)
0007 //
0008 // For more information, see www.boost.org
0009 // ----------------------------------------------------------------------------
0010 #ifndef BOOST_PROPERTY_TREE_DETAIL_XML_PARSER_UTILS_HPP_INCLUDED
0011 #define BOOST_PROPERTY_TREE_DETAIL_XML_PARSER_UTILS_HPP_INCLUDED
0012
0013 #include <boost/property_tree/detail/ptree_utils.hpp>
0014 #include <boost/property_tree/detail/xml_parser_error.hpp>
0015 #include <boost/property_tree/detail/xml_parser_writer_settings.hpp>
0016 #include <string>
0017 #include <algorithm>
0018 #include <locale>
0019
0020 namespace boost { namespace property_tree { namespace xml_parser
0021 {
0022
0023     template<class Str>
0024     Str condense(const Str &s)
0025     {
0026         typedef typename Str::value_type Ch;
0027         Str r;
0028         std::locale loc;
0029         bool space = false;
0030         typename Str::const_iterator end = s.end();
0031         for (typename Str::const_iterator it = s.begin();
0032              it != end; ++it)
0033         {
0034             if (isspace(*it, loc) || *it == Ch('\n'))
0035             {
0036                 if (!space)
0037                     r += Ch(' '), space = true;
0038             }
0039             else
0040                 r += *it, space = false;
0041         }
0042         return r;
0043     }
0044
0045
0046     template<class Str>
0047     Str encode_char_entities(const Str &s)
0048     {
0049         // Don't do anything for empty strings.
0050         if(s.empty()) return s;
0051
0052         typedef typename Str::value_type Ch;
0053
0054         Str r;
0055         // To properly round-trip spaces and not uglify the XML beyond
0056         // recognition, we have to encode them IF the text contains only spaces.
0057         Str sp(1, Ch(' '));
0058         if(s.find_first_not_of(sp) == Str::npos) {
0059             // The first will suffice.
0060             r = detail::widen<Str>("&#32;");
0061             r += Str(s.size() - 1, Ch(' '));
0062         } else {
0063             typename Str::const_iterator end = s.end();
0064             for (typename Str::const_iterator it = s.begin(); it != end; ++it)
0065             {
0066                 switch (*it)
0067                 {
0068                     case Ch('<'): r += detail::widen<Str>("&lt;"); break;
0069                     case Ch('>'): r += detail::widen<Str>("&gt;"); break;
0070                     case Ch('&'): r += detail::widen<Str>("&amp;"); break;
0071                     case Ch('"'): r += detail::widen<Str>("&quot;"); break;
0072                     case Ch('\''): r += detail::widen<Str>("&apos;"); break;
0073                     default: r += *it; break;
0074                 }
0075             }
0076         }
0077         return r;
0078     }
0079
0080     template<class Str>
0081     Str decode_char_entities(const Str &s)
0082     {
0083         typedef typename Str::value_type Ch;
0084         Str r;
0085         typename Str::const_iterator end = s.end();
0086         for (typename Str::const_iterator it = s.begin(); it != end; ++it)
0087         {
0088             if (*it == Ch('&'))
0089             {
0090                 typename Str::const_iterator semicolon = std::find(it + 1, end, Ch(';'));
0091                 if (semicolon == end)
0092                     BOOST_PROPERTY_TREE_THROW(xml_parser_error("invalid character entity", "", 0));
0093                 Str ent(it + 1, semicolon);
0094                 if (ent == detail::widen<Str>("lt")) r += Ch('<');
0095                 else if (ent == detail::widen<Str>("gt")) r += Ch('>');
0096                 else if (ent == detail::widen<Str>("amp")) r += Ch('&');
0097                 else if (ent == detail::widen<Str>("quot")) r += Ch('"');
0098                 else if (ent == detail::widen<Str>("apos")) r += Ch('\'');
0099                 else
0100                     BOOST_PROPERTY_TREE_THROW(xml_parser_error("invalid character entity", "", 0));
0101                 it = semicolon;
0102             }
0103             else
0104                 r += *it;
0105         }
0106         return r;
0107     }
0108
0109     template<class Str>
0110     const Str &xmldecl()
0111     {
0112         static Str s = detail::widen<Str>("<?xml>");
0113         return s;
0114     }
0115
0116     template<class Str>
0117     const Str &xmlattr()
0118     {
0119         static Str s = detail::widen<Str>("<xmlattr>");
0120         return s;
0121     }
0122
0123     template<class Str>
0124     const Str &xmlcomment()
0125     {
0126         static Str s = detail::widen<Str>("<xmlcomment>");
0127         return s;
0128     }
0129
0130     template<class Str>
0131     const Str &xmltext()
0132     {
0133         static Str s = detail::widen<Str>("<xmltext>");
0134         return s;
0135     }
0136
0137 } } }
0138
0139 #endif