Chris@16: // ----------------------------------------------------------------------------
Chris@16: // Copyright (C) 2002-2006 Marcin Kalicinski
Chris@16: //
Chris@16: // Distributed under the Boost Software License, Version 1.0. 
Chris@16: // (See accompanying file LICENSE_1_0.txt or copy at 
Chris@16: // http://www.boost.org/LICENSE_1_0.txt)
Chris@16: //
Chris@16: // For more information, see www.boost.org
Chris@16: // ----------------------------------------------------------------------------
Chris@16: #ifndef BOOST_PROPERTY_TREE_DETAIL_XML_PARSER_UTILS_HPP_INCLUDED
Chris@16: #define BOOST_PROPERTY_TREE_DETAIL_XML_PARSER_UTILS_HPP_INCLUDED
Chris@16: 
Chris@16: #include <boost/property_tree/detail/ptree_utils.hpp>
Chris@16: #include <boost/property_tree/detail/xml_parser_error.hpp>
Chris@16: #include <boost/property_tree/detail/xml_parser_writer_settings.hpp>
Chris@16: #include <string>
Chris@16: #include <algorithm>
Chris@16: #include <locale>
Chris@16: 
Chris@16: namespace boost { namespace property_tree { namespace xml_parser
Chris@16: {
Chris@16: 
Chris@101:     template<class Str>
Chris@101:     Str condense(const Str &s)
Chris@16:     {
Chris@101: 	typedef typename Str::value_type Ch;
Chris@101: 	Str r;
Chris@16:         std::locale loc;
Chris@16:         bool space = false;
Chris@101:         typename Str::const_iterator end = s.end();
Chris@101:         for (typename Str::const_iterator it = s.begin();
Chris@16:              it != end; ++it)
Chris@16:         {
Chris@16:             if (isspace(*it, loc) || *it == Ch('\n'))
Chris@16:             {
Chris@16:                 if (!space)
Chris@16:                     r += Ch(' '), space = true;
Chris@16:             }
Chris@16:             else
Chris@16:                 r += *it, space = false;
Chris@16:         }
Chris@16:         return r;
Chris@16:     }
Chris@16: 
Chris@101: 
Chris@101:     template<class Str>
Chris@101:     Str encode_char_entities(const Str &s)
Chris@16:     {
Chris@16:         // Don't do anything for empty strings.
Chris@16:         if(s.empty()) return s;
Chris@16: 
Chris@101:         typedef typename Str::value_type Ch;
Chris@101: 
Chris@16:         Str r;
Chris@16:         // To properly round-trip spaces and not uglify the XML beyond
Chris@16:         // recognition, we have to encode them IF the text contains only spaces.
Chris@16:         Str sp(1, Ch(' '));
Chris@16:         if(s.find_first_not_of(sp) == Str::npos) {
Chris@16:             // The first will suffice.
Chris@101:             r = detail::widen<Str>("&#32;");
Chris@16:             r += Str(s.size() - 1, Ch(' '));
Chris@16:         } else {
Chris@16:             typename Str::const_iterator end = s.end();
Chris@16:             for (typename Str::const_iterator it = s.begin(); it != end; ++it)
Chris@16:             {
Chris@16:                 switch (*it)
Chris@16:                 {
Chris@101:                     case Ch('<'): r += detail::widen<Str>("&lt;"); break;
Chris@101:                     case Ch('>'): r += detail::widen<Str>("&gt;"); break;
Chris@101:                     case Ch('&'): r += detail::widen<Str>("&amp;"); break;
Chris@101:                     case Ch('"'): r += detail::widen<Str>("&quot;"); break;
Chris@101:                     case Ch('\''): r += detail::widen<Str>("&apos;"); break;
Chris@101:                     case Ch('\t'): r += detail::widen<Str>("&#9;"); break;
Chris@101:                     case Ch('\n'): r += detail::widen<Str>("&#10;"); break;
Chris@16:                     default: r += *it; break;
Chris@16:                 }
Chris@16:             }
Chris@16:         }
Chris@16:         return r;
Chris@16:     }
Chris@16:     
Chris@101:     template<class Str>
Chris@101:     Str decode_char_entities(const Str &s)
Chris@16:     {
Chris@101: 	typedef typename Str::value_type Ch;
Chris@16:         Str r;
Chris@16:         typename Str::const_iterator end = s.end();
Chris@16:         for (typename Str::const_iterator it = s.begin(); it != end; ++it)
Chris@16:         {
Chris@16:             if (*it == Ch('&'))
Chris@16:             {
Chris@16:                 typename Str::const_iterator semicolon = std::find(it + 1, end, Ch(';'));
Chris@16:                 if (semicolon == end)
Chris@16:                     BOOST_PROPERTY_TREE_THROW(xml_parser_error("invalid character entity", "", 0));
Chris@16:                 Str ent(it + 1, semicolon);
Chris@101:                 if (ent == detail::widen<Str>("lt")) r += Ch('<');
Chris@101:                 else if (ent == detail::widen<Str>("gt")) r += Ch('>');
Chris@101:                 else if (ent == detail::widen<Str>("amp")) r += Ch('&');
Chris@101:                 else if (ent == detail::widen<Str>("quot")) r += Ch('"');
Chris@101:                 else if (ent == detail::widen<Str>("apos")) r += Ch('\'');
Chris@16:                 else
Chris@16:                     BOOST_PROPERTY_TREE_THROW(xml_parser_error("invalid character entity", "", 0));
Chris@16:                 it = semicolon;
Chris@16:             }
Chris@16:             else
Chris@16:                 r += *it;
Chris@16:         }
Chris@16:         return r;
Chris@16:     }
Chris@16:     
Chris@101:     template<class Str>
Chris@101:     const Str &xmldecl()
Chris@16:     {
Chris@101:         static Str s = detail::widen<Str>("<?xml>");
Chris@16:         return s;
Chris@16:     }
Chris@16: 
Chris@101:     template<class Str>
Chris@101:     const Str &xmlattr()
Chris@16:     {
Chris@101:         static Str s = detail::widen<Str>("<xmlattr>");
Chris@16:         return s;
Chris@16:     }
Chris@16: 
Chris@101:     template<class Str>
Chris@101:     const Str &xmlcomment()
Chris@16:     {
Chris@101:         static Str s = detail::widen<Str>("<xmlcomment>");
Chris@16:         return s;
Chris@16:     }
Chris@16: 
Chris@101:     template<class Str>
Chris@101:     const Str &xmltext()
Chris@16:     {
Chris@101:         static Str s = detail::widen<Str>("<xmltext>");
Chris@16:         return s;
Chris@16:     }
Chris@16: 
Chris@16: } } }
Chris@16: 
Chris@16: #endif