Console/Parse.ih

   1 // $Id$
   2 //
   3 // Copyright (C) 2008
   4 // Fraunhofer Institute for Open Communication Systems (FOKUS)
   5 // Competence Center NETwork research (NET), St. Augustin, GERMANY
   6 //     Stefan Bund <g0dil@berlios.de>
   7 //
   8 // This program is free software; you can redistribute it and/or modify
   9 // it under the terms of the GNU General Public License as published by
  10 // the Free Software Foundation; either version 2 of the License, or
  11 // (at your option) any later version.
  12 //
  13 // This program is distributed in the hope that it will be useful,
  14 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 // GNU General Public License for more details.
  17 //
  18 // You should have received a copy of the GNU General Public License
  19 // along with this program; if not, write to the
  20 // Free Software Foundation, Inc.,
  21 // 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  22
  23 /** \file
  24     \brief Parse internal header */
  25
  26 #ifndef IH_Parse_
  27 #define IH_Parse_ 1
  28
  29 // Custom includes
  30 #include <vector>
  31 #include <boost/regex.hpp>
  32 #include <boost/spirit.hpp>
  33 #include <boost/spirit/utility/grammar_def.hpp>
  34 #include <boost/spirit/actor.hpp>
  35 #include <boost/bind.hpp>
  36 #include <boost/function.hpp>
  37 #include <boost/ref.hpp>
  38
  39 ///////////////////////////////ih.p////////////////////////////////////////
  40
  41 namespace senf {
  42 namespace console {
  43 namespace detail {
  44
  45     struct append_action
  46     {
  47         template <class T, class Value>
  48         void act(T & ref, Value const & value) const
  49             { ref += T(1, value); }
  50
  51         template <class T, class Iterator>
  52         void act(T & ref, Iterator const & f, Iterator const & l) const
  53             { ref += T(f,l); }
  54     };
  55
  56     template <class T>
  57     inline boost::spirit::ref_value_actor<T, append_action>
  58     append_a(T & ref)
  59     {
  60         return boost::spirit::ref_value_actor<T, append_action>(ref);
  61     }
  62
  63     template <class T, class Value>
  64     inline boost::spirit::ref_const_ref_actor<T, Value, append_action>
  65     append_a(T & ref, Value const & value)
  66     {
  67         return boost::spirit::ref_const_ref_actor<T, Value, append_action>(ref, value);
  68     }
  69
  70     template <class ParseDispatcher>
  71     struct CommandGrammar : boost::spirit::grammar<CommandGrammar<ParseDispatcher> >
  72     {
  73         ///////////////////////////////////////////////////////////////////////////
  74         // Start rules
  75
  76         enum { CommandParser, SkipParser };
  77
  78         ///////////////////////////////////////////////////////////////////////////
  79         // The parse context (variables needed while parsing)
  80
  81         struct Context {
  82             std::string str;
  83             std::vector<std::string> path;
  84             char ch;
  85         };
  86
  87         Context & context;
  88
  89         ///////////////////////////////////////////////////////////////////////////
  90         // Dispatching semantic actions
  91
  92         ParseDispatcher & dispatcher;
  93
  94         struct Dispatch_actor
  95         {
  96             Dispatch_actor(boost::function<void ()> fn_) : fn (fn_) {}
  97
  98             template <class Value>
  99             void operator()(Value const & value) const
 100                 { fn(); }
 101
 102             template <class Iterator>
 103             void operator()(Iterator const & f, Iterator const & l) const
 104                 { fn(); }
 105
 106             boost::function<void ()> fn;
 107         };
 108
 109         template <class Callback>
 110         Dispatch_actor dispatch(Callback cb) const
 111             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher))); }
 112
 113         template <class Callback, class Arg>
 114         Dispatch_actor dispatch(Callback cb, Arg const & arg) const
 115             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher), arg)); }
 116
 117         ///////////////////////////////////////////////////////////////////////////
 118
 119         CommandGrammar(ParseDispatcher & d, Context & c)
 120             : context(c), dispatcher(d) {}
 121
 122         template <class Scanner>
 123         struct definition
 124             : public boost::spirit::grammar_def< boost::spirit::rule<Scanner>,
 125                                                  boost::spirit::rule<Scanner> >
 126         {
 127             boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
 128                 punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
 129                 skip, commands, block, statement, relpath, abspath;
 130             boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
 131             boost::spirit::distinct_parser<> keyword_p;
 132
 133             definition(CommandGrammar const & self) :
 134
 135                 // Characters with a special meaning within the parser
 136                 special_p ("/(){};"),
 137
 138                 // Characters which are returned as punctuation tokens
 139                 punctuation_p (",="),
 140
 141                 // Whitespace characters
 142                 space_p (" \t\n\r"),
 143
 144                 // Invalid characters: All chars below \x20 (space) which are not space_p
 145                 // (don't put a \0 in the chset<> argument *string* ...)
 146                 invalid_p (
 147                     boost::spirit::chset<>('\0') | boost::spirit::chset<>("\x01-\x20") - space_p ),
 148
 149                 // Valid word characters
 150                 word_p (
 151                     boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
 152
 153                 // Keywords must not be followed by a word char or '/'
 154                 keyword_p ( word_p | boost::spirit::ch_p('/') )
 155
 156             {
 157                 using namespace boost::spirit;
 158                 typedef ParseDispatcher PD;
 159
 160                 ///////////////////////////////////////////////////////////////////
 161                 // Spirit grammar
 162                 //
 163                 // Syntax summary:
 164                 // This is EBNF with some minor tweaks to accommodate C++ syntax
 165                 //
 166                 //   * and +    precede their argument
 167                 //   >>         is followed by
 168                 //   !          optional
 169                 //   a % b      match any number of a's separated by b
 170                 //   a - b      match a but not b
 171                 //
 172                 // Beside this, we use some special parsers (ch_p, eps_p, confix_p, lex_escape_ch_p,
 173                 // keyword_p, comment_p) and directives (lexeme_d), however, the parser should be
 174                 // quite readable.
 175                 //
 176                 //   ch_p             match character
 177                 //   eps_p            always matches nothing (to attach unconditional actions)
 178                 //   confix_p(a,b,c)  match b, preceded by a and terminated by c. Used to parse
 179                 //                    string literals and comments
 180                 //   lex_escape_ch_p  match a lex style escape char. This is like a C++ style
 181                 //                    literal string escape char, however \x will be replaced by 'x'
 182                 //                    for any char 'x' if it has no special meaning.
 183                 //   keyword_p        match a delimited keyword
 184                 //   comment_p(a,b)   match comment starting with a and terminated with b. b
 185                 //                    defaults to end-of-line
 186                 //
 187                 //   lexeme_d         don't skip whitespace (as defined by the skip parser)
 188                 //
 189                 // Aligned to the right at column 50 are semantic actions.
 190                 //
 191                 // For clarity, I have used 'ch_p' explicitly throughout even though it is auxiliary
 192                 // in most cases.
 193                 //
 194                 // More info is in the Boost.Spirit documentation
 195
 196                 commands
 197                     =  * command
 198                     ;
 199
 200                 command
 201                     =    builtin >> (ch_p(';') | end_p)
 202                     |    path  >> ( block | statement )
 203                     |    ch_p(';') // Ignore empty commands
 204                     ;
 205
 206                 builtin
 207                     =    keyword_p("cd")
 208                       >> path
 209                       >> eps_p                    [ self.dispatch(&PD::builtin_cd,
 210                                                                   boost::ref(self.context.path)) ]
 211                     |    keyword_p("ls")
 212                       >> ! path
 213                       >> eps_p                    [ self.dispatch(&PD::builtin_ls,
 214                                                                   boost::ref(self.context.path)) ]
 215                     |    keyword_p("exit")        [ self.dispatch(&PD::builtin_exit) ]
 216                     ;
 217
 218                 block
 219                     =    ch_p('{')                [ self.dispatch(&PD::pushDirectory,
 220                                                                   boost::ref(self.context.path)) ]
 221                       >> * command
 222                       >> ch_p('}')                [ self.dispatch(&PD::popDirectory) ]
 223                     ;
 224
 225                 statement
 226                     = eps_p                       [ self.dispatch(&PD::beginCommand,
 227                                                                   boost::ref(self.context.path)) ]
 228                       >> * argument
 229                       >> (ch_p(';') | end_p)
 230                       >> eps_p                    [ self.dispatch(&PD::endCommand) ]
 231                     ;
 232
 233                 argument
 234                     =    simple_argument          [ self.dispatch(&PD::pushArgument,
 235                                                                   boost::ref(self.context.str)) ]
 236                     |    complex_argument
 237                     ;
 238
 239                 simple_argument         // All these return their value in context.str
 240                     =    string
 241                     |    hexstring
 242                     |    word
 243                     ;
 244
 245                 complex_argument        // Argument consists of multiple tokens
 246                     =    ch_p('(')                [ self.dispatch(&PD::openGroup) ]
 247                       >> * token
 248                       >> ch_p(')')                [ self.dispatch(&PD::closeGroup) ]
 249                     ;
 250
 251                 string                  // Returns value in context.str
 252                     =    eps_p                    [ clear_a(self.context.str) ]
 253                       >> lexeme_d
 254                          [
 255                              ch_p('"')
 256                           >> * ( ( lex_escape_ch_p[ assign_a(self.context.ch) ]
 257                                    - '"'
 258                                  )                [ append_a(self.context.str,
 259                                                              self.context.ch) ]
 260                                )
 261                           >> ch_p('"')
 262                          ]
 263                     ;
 264
 265                 hexstring               // Returns value in context.str
 266                     =    eps_p                    [ clear_a(self.context.str) ]
 267                       >> confix_p( "x\"", * hexbyte, '"' )
 268                     ;
 269
 270                 path                    // Returns value in context.path
 271                     =    eps_p                    [ clear_a(self.context.path) ]
 272                       >> relpath | abspath
 273                     ;
 274
 275                 relpath
 276                     =    (   word                 [ push_back_a(self.context.path) ]
 277                            % ch_p('/') )
 278                       >> ( ! ch_p('/')            [ push_back_a(self.context.path,"") ] )
 279                     ;
 280
 281                 abspath
 282                     =    ch_p('/')                [ push_back_a(self.context.path, "") ]
 283                       >> ( relpath
 284                          | eps_p                  [ push_back_a(self.context.path, "") ] )
 285                     ;
 286
 287                 balanced_tokens
 288                     =    ch_p('(')                [ self.dispatch(&PD::pushPunctuation, "(") ]
 289                       >> * token
 290                       >> ch_p(')')                [ self.dispatch(&PD::pushPunctuation, ")") ]
 291                     ;
 292
 293                 token
 294                     =    simple_argument          [ self.dispatch(&PD::pushWord,
 295                                                                   boost::ref(self.context.str)) ]
 296                     |    punctuation              [ self.dispatch(&PD::pushPunctuation,
 297                                                                   boost::ref(self.context.str)) ]
 298                     |    balanced_tokens
 299                     ;
 300
 301                 punctuation             // Returns value in context.str
 302                     =    punctuation_p            [ assign_a(self.context.str) ]
 303                     ;
 304
 305                 word                    // Returns value in context.str
 306                     =    lexeme_d[ + word_p ]     [ assign_a(self.context.str) ]
 307                     ;
 308
 309                 hexbyte
 310                     =    uint_parser<char, 16, 2, 2>()
 311                                                   [ append_a(self.context.str) ]
 312                     ;
 313
 314                 skip
 315                     =    space_p | comment_p('#')
 316                     ;
 317
 318                 ///////////////////////////////////////////////////////////////////
 319
 320                 start_parsers(
 321                     commands,           // CommandParser
 322                     skip                // SkipParser
 323                 );
 324
 325                 BOOST_SPIRIT_DEBUG_TRACE_RULE(command,1);
 326                 BOOST_SPIRIT_DEBUG_TRACE_RULE(path,1);
 327                 BOOST_SPIRIT_DEBUG_TRACE_RULE(argument,1);
 328                 BOOST_SPIRIT_DEBUG_TRACE_RULE(word,1);
 329                 BOOST_SPIRIT_DEBUG_TRACE_RULE(string,1);
 330                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexstring,1);
 331                 BOOST_SPIRIT_DEBUG_TRACE_RULE(token,1);
 332                 BOOST_SPIRIT_DEBUG_TRACE_RULE(punctuation,1);
 333                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexbyte,1);
 334                 BOOST_SPIRIT_DEBUG_TRACE_RULE(balanced_tokens,1);
 335                 BOOST_SPIRIT_DEBUG_TRACE_RULE(simple_argument,1);
 336                 BOOST_SPIRIT_DEBUG_TRACE_RULE(complex_argument,1);
 337                 BOOST_SPIRIT_DEBUG_TRACE_RULE(builtin,1);
 338                 BOOST_SPIRIT_DEBUG_TRACE_RULE(commands,1);
 339                 BOOST_SPIRIT_DEBUG_TRACE_RULE(block,1);
 340                 BOOST_SPIRIT_DEBUG_TRACE_RULE(statement,1);
 341                 BOOST_SPIRIT_DEBUG_TRACE_RULE(relpath,1);
 342                 BOOST_SPIRIT_DEBUG_TRACE_RULE(abspath,1);
 343             }
 344         };
 345     };
 346
 347 }}}
 348
 349 ///////////////////////////////ih.e////////////////////////////////////////
 350 #endif
 351
 352 \f
 353 // Local Variables:
 354 // mode: c++
 355 // fill-column: 100
 356 // comment-column: 40
 357 // c-file-style: "senf"
 358 // indent-tabs-mode: nil
 359 // ispell-local-dictionary: "american"
 360 // compile-command: "scons -u test"
 361 // End: