mini_xml1.cpp 6.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242
  1. /*=============================================================================
  2. Copyright (c) 2001-2010 Joel de Guzman
  3. Distributed under the Boost Software License, Version 1.0. (See accompanying
  4. file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
  5. =============================================================================*/
  6. ///////////////////////////////////////////////////////////////////////////////
  7. //
  8. // A mini XML-like parser
  9. //
  10. // [ JDG March 25, 2007 ] spirit2
  11. //
  12. ///////////////////////////////////////////////////////////////////////////////
  13. #include <boost/config/warning_disable.hpp>
  14. #include <boost/spirit/include/qi.hpp>
  15. #include <boost/spirit/include/phoenix_core.hpp>
  16. #include <boost/spirit/include/phoenix_operator.hpp>
  17. #include <boost/spirit/include/phoenix_fusion.hpp>
  18. #include <boost/spirit/include/phoenix_stl.hpp>
  19. #include <boost/fusion/include/adapt_struct.hpp>
  20. #include <boost/variant/recursive_variant.hpp>
  21. #include <boost/foreach.hpp>
  22. #include <iostream>
  23. #include <fstream>
  24. #include <string>
  25. #include <vector>
  26. namespace client
  27. {
  28. namespace fusion = boost::fusion;
  29. namespace phoenix = boost::phoenix;
  30. namespace qi = boost::spirit::qi;
  31. namespace ascii = boost::spirit::ascii;
  32. ///////////////////////////////////////////////////////////////////////////
  33. // Our mini XML tree representation
  34. ///////////////////////////////////////////////////////////////////////////
  35. //[tutorial_xml1_structures
  36. struct mini_xml;
  37. typedef
  38. boost::variant<
  39. boost::recursive_wrapper<mini_xml>
  40. , std::string
  41. >
  42. mini_xml_node;
  43. struct mini_xml
  44. {
  45. std::string name; // tag name
  46. std::vector<mini_xml_node> children; // children
  47. };
  48. //]
  49. }
  50. // We need to tell fusion about our mini_xml struct
  51. // to make it a first-class fusion citizen
  52. //[tutorial_xml1_adapt_structures
  53. BOOST_FUSION_ADAPT_STRUCT(
  54. client::mini_xml,
  55. (std::string, name)
  56. (std::vector<client::mini_xml_node>, children)
  57. )
  58. //]
  59. namespace client
  60. {
  61. ///////////////////////////////////////////////////////////////////////////
  62. // Print out the mini xml tree
  63. ///////////////////////////////////////////////////////////////////////////
  64. int const tabsize = 4;
  65. void tab(int indent)
  66. {
  67. for (int i = 0; i < indent; ++i)
  68. std::cout << ' ';
  69. }
  70. struct mini_xml_printer
  71. {
  72. mini_xml_printer(int indent = 0)
  73. : indent(indent)
  74. {
  75. }
  76. void operator()(mini_xml const& xml) const;
  77. int indent;
  78. };
  79. struct mini_xml_node_printer : boost::static_visitor<>
  80. {
  81. mini_xml_node_printer(int indent = 0)
  82. : indent(indent)
  83. {
  84. }
  85. void operator()(mini_xml const& xml) const
  86. {
  87. mini_xml_printer(indent+tabsize)(xml);
  88. }
  89. void operator()(std::string const& text) const
  90. {
  91. tab(indent+tabsize);
  92. std::cout << "text: \"" << text << '"' << std::endl;
  93. }
  94. int indent;
  95. };
  96. void mini_xml_printer::operator()(mini_xml const& xml) const
  97. {
  98. tab(indent);
  99. std::cout << "tag: " << xml.name << std::endl;
  100. tab(indent);
  101. std::cout << '{' << std::endl;
  102. BOOST_FOREACH(mini_xml_node const& node, xml.children)
  103. {
  104. boost::apply_visitor(mini_xml_node_printer(indent), node);
  105. }
  106. tab(indent);
  107. std::cout << '}' << std::endl;
  108. }
  109. ///////////////////////////////////////////////////////////////////////////
  110. // Our mini XML grammar definition
  111. ///////////////////////////////////////////////////////////////////////////
  112. //[tutorial_xml1_grammar
  113. template <typename Iterator>
  114. struct mini_xml_grammar : qi::grammar<Iterator, mini_xml(), ascii::space_type>
  115. {
  116. mini_xml_grammar() : mini_xml_grammar::base_type(xml)
  117. {
  118. using qi::lit;
  119. using qi::lexeme;
  120. using ascii::char_;
  121. using ascii::string;
  122. using namespace qi::labels;
  123. using phoenix::at_c;
  124. using phoenix::push_back;
  125. text = lexeme[+(char_ - '<') [_val += _1]];
  126. node = (xml | text) [_val = _1];
  127. start_tag =
  128. '<'
  129. >> !lit('/')
  130. >> lexeme[+(char_ - '>') [_val += _1]]
  131. >> '>'
  132. ;
  133. end_tag =
  134. "</"
  135. >> lit(_r1)
  136. >> '>'
  137. ;
  138. xml =
  139. start_tag [at_c<0>(_val) = _1]
  140. >> *node [push_back(at_c<1>(_val), _1)]
  141. >> end_tag(at_c<0>(_val))
  142. ;
  143. }
  144. qi::rule<Iterator, mini_xml(), ascii::space_type> xml;
  145. qi::rule<Iterator, mini_xml_node(), ascii::space_type> node;
  146. qi::rule<Iterator, std::string(), ascii::space_type> text;
  147. qi::rule<Iterator, std::string(), ascii::space_type> start_tag;
  148. qi::rule<Iterator, void(std::string), ascii::space_type> end_tag;
  149. };
  150. //]
  151. }
  152. ///////////////////////////////////////////////////////////////////////////////
  153. // Main program
  154. ///////////////////////////////////////////////////////////////////////////////
  155. int main(int argc, char **argv)
  156. {
  157. char const* filename;
  158. if (argc > 1)
  159. {
  160. filename = argv[1];
  161. }
  162. else
  163. {
  164. std::cerr << "Error: No input file provided." << std::endl;
  165. return 1;
  166. }
  167. std::ifstream in(filename, std::ios_base::in);
  168. if (!in)
  169. {
  170. std::cerr << "Error: Could not open input file: "
  171. << filename << std::endl;
  172. return 1;
  173. }
  174. std::string storage; // We will read the contents here.
  175. in.unsetf(std::ios::skipws); // No white space skipping!
  176. std::copy(
  177. std::istream_iterator<char>(in),
  178. std::istream_iterator<char>(),
  179. std::back_inserter(storage));
  180. typedef client::mini_xml_grammar<std::string::const_iterator> mini_xml_grammar;
  181. mini_xml_grammar xml; // Our grammar
  182. client::mini_xml ast; // Our tree
  183. using boost::spirit::ascii::space;
  184. std::string::const_iterator iter = storage.begin();
  185. std::string::const_iterator end = storage.end();
  186. bool r = phrase_parse(iter, end, xml, space, ast);
  187. if (r && iter == end)
  188. {
  189. std::cout << "-------------------------\n";
  190. std::cout << "Parsing succeeded\n";
  191. std::cout << "-------------------------\n";
  192. client::mini_xml_printer printer;
  193. printer(ast);
  194. return 0;
  195. }
  196. else
  197. {
  198. std::string::const_iterator some = iter + std::min(30, int(end - iter));
  199. std::string context(iter, (some>end)?end:some);
  200. std::cout << "-------------------------\n";
  201. std::cout << "Parsing failed\n";
  202. std::cout << "stopped at: \"" << context << "...\"\n";
  203. std::cout << "-------------------------\n";
  204. return 1;
  205. }
  206. }