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 #ifndef DOXYGEN
  46
  47     ///////////////////////////////////////////////////////////////////////////
  48     // append_a
  49
  50     struct append_action
  51     {
  52         template <class T, class Value>
  53         void act(T & ref, Value const & value) const
  54             { ref += T(1, value); }
  55
  56         template <class T, class Iterator>
  57         void act(T & ref, Iterator const & f, Iterator const & l) const
  58             { ref += T(f,l); }
  59     };
  60
  61     template <class T>
  62     inline boost::spirit::ref_value_actor<T, append_action>
  63     append_a(T & ref)
  64     { return boost::spirit::ref_value_actor<T, append_action>(ref); }
  65
  66     template <class T, class Value>
  67     inline boost::spirit::ref_const_ref_actor<T, Value, append_action>
  68     append_a(T & ref, Value const & value)
  69     { return boost::spirit::ref_const_ref_actor<T, Value, append_action>(ref, value); }
  70
  71     ///////////////////////////////////////////////////////////////////////////
  72     // Grammar
  73
  74     template <class ParseDispatcher>
  75     struct CommandGrammar : boost::spirit::grammar<CommandGrammar<ParseDispatcher> >
  76     {
  77         ///////////////////////////////////////////////////////////////////////////
  78         // Start rules
  79
  80         enum { CommandParser, SkipParser };
  81
  82         ///////////////////////////////////////////////////////////////////////////
  83         // The parse context (variables needed while parsing)
  84
  85         typedef ArgumentToken::TokenType TokenType;
  86
  87         struct Context {
  88             std::string str;
  89             std::vector<std::string> path;
  90             char ch;
  91             TokenType type;
  92
  93             // OUCH ... This is sooooo stupid .. push_back_a and assign_a take their
  94             // arguments by const-reference and STORE the REFERENCE ... they do NOT accept
  95             // literal values !!!!!!
  96             static const TokenType BasicString;
  97             static const TokenType HexString;
  98             static const TokenType Word;
  99             static const std::string EmptyString;
 100         };
 101
 102         Context & context;
 103
 104         ///////////////////////////////////////////////////////////////////////////
 105         // Dispatching semantic actions
 106
 107         ParseDispatcher & dispatcher;
 108
 109         struct Dispatch_actor
 110         {
 111             Dispatch_actor(boost::function<void ()> fn_) : fn (fn_) {}
 112
 113             template <class Value>
 114             void operator()(Value const & value) const
 115                 { fn(); }
 116
 117             template <class Iterator>
 118             void operator()(Iterator const & f, Iterator const & l) const
 119                 { fn(); }
 120
 121             boost::function<void ()> fn;
 122         };
 123
 124         template <class Callback>
 125         Dispatch_actor dispatch(Callback cb) const
 126             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher))); }
 127
 128         template <class Callback, class Arg>
 129         Dispatch_actor dispatch(Callback cb, Arg const & arg) const
 130             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher), arg)); }
 131
 132         template <class Callback, class Arg1, class Arg2>
 133         Dispatch_actor dispatch(Callback cb, Arg1 const & arg1, Arg2 const & arg2) const
 134             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher), arg1, arg2)); }
 135
 136         ///////////////////////////////////////////////////////////////////////////
 137
 138         CommandGrammar(ParseDispatcher & d, Context & c)
 139             : context(c), dispatcher(d) {}
 140
 141         template <class Scanner>
 142         struct definition
 143             : public boost::spirit::grammar_def< boost::spirit::rule<Scanner>,
 144                                                  boost::spirit::rule<Scanner> >
 145         {
 146             boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
 147                 punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
 148                 skip, commands, block, statement, relpath, abspath;
 149             boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
 150             boost::spirit::distinct_parser<> keyword_p;
 151
 152             definition(CommandGrammar const & self) :
 153
 154                 // Characters with a special meaning within the parser
 155                 special_p ("/(){};"),
 156
 157                 // Characters which are returned as punctuation tokens
 158                 // (only allowed within '()')
 159                 punctuation_p (",=/{};"),
 160
 161                 // Whitespace characters
 162                 space_p (" \t\n\r"),
 163
 164                 // Invalid characters: All chars below \x20 (space) which are not space_p
 165                 // (don't put a \0 in the chset<> argument *string* ...)
 166                 invalid_p (
 167                     boost::spirit::chset<>('\0') | boost::spirit::chset<>("\x01-\x20") - space_p ),
 168
 169                 // Valid word characters
 170                 word_p (
 171                     boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
 172
 173                 // Keywords must not be followed by a word char or '/'
 174                 keyword_p ( word_p | boost::spirit::ch_p('/') )
 175
 176             {
 177                 using namespace boost::spirit;
 178                 typedef ParseDispatcher PD;
 179
 180                 ///////////////////////////////////////////////////////////////////
 181                 // Spirit grammar
 182                 //
 183                 // Syntax summary:
 184                 // This is EBNF with some minor tweaks to accommodate C++ syntax
 185                 //
 186                 //   * and +    like EBNF but they precede their argument
 187                 //   >>         is followed by
 188                 //   !          optional
 189                 //   a % b      match any number of a's separated by b
 190                 //   a - b      match a but not b
 191                 //
 192                 // Beside this, we use some special parsers (ch_p, eps_p, confix_p, lex_escape_ch_p,
 193                 // keyword_p, comment_p) and directives (lexeme_d), however, the parser should be
 194                 // quite readable.
 195                 //
 196                 //   ch_p             match character
 197                 //   eps_p            always matches nothing (to attach unconditional actions)
 198                 //   confix_p(a,b,c)  match b, preceded by a and terminated by c. Used to parse
 199                 //                    string literals and comments
 200                 //   lex_escape_ch_p  match a lex style escape char. This is like a C++ style
 201                 //                    literal string escape char, however \x will be replaced by 'x'
 202                 //                    for any char 'x' if it has no special meaning.
 203                 //   keyword_p        match a delimited keyword
 204                 //   comment_p(a,b)   match comment starting with a and terminated with b. b
 205                 //                    defaults to end-of-line
 206                 //
 207                 //   lexeme_d         don't skip whitespace (as defined by the skip parser)
 208                 //
 209                 // Aligned to the right at column 50 are semantic actions.
 210                 //
 211                 // For clarity, I have used 'ch_p' explicitly throughout even though it is optional
 212                 // in most cases.
 213                 //
 214                 // More info is in the Boost.Spirit documentation
 215
 216                 commands
 217                     =  * command
 218                     ;
 219
 220                 command
 221                     =    builtin >> (ch_p(';') | end_p)
 222                     |    path  >> ( block | statement )
 223                     |    ch_p(';') // Ignore empty commands
 224                     ;
 225
 226                 builtin
 227                     =    keyword_p("cd")
 228                       >> path
 229                       >> eps_p                    [ self.dispatch(&PD::builtin_cd,
 230                                                                   boost::ref(self.context.path)) ]
 231                     |    keyword_p("ls")
 232                       >> ! path
 233                       >> eps_p                    [ self.dispatch(&PD::builtin_ls,
 234                                                                   boost::ref(self.context.path)) ]
 235                     |    keyword_p("exit")        [ self.dispatch(&PD::builtin_exit) ]
 236
 237                     |    keyword_p("help")
 238                       >> ! path
 239                       >> eps_p                    [ self.dispatch(&PD::builtin_help,
 240                                                                   boost::ref(self.context.path)) ]
 241                     ;
 242
 243                 block
 244                     =    ch_p('{')                [ self.dispatch(&PD::pushDirectory,
 245                                                                   boost::ref(self.context.path)) ]
 246                       >> * command
 247                       >> ch_p('}')                [ self.dispatch(&PD::popDirectory) ]
 248                     ;
 249
 250                 statement
 251                     = eps_p                       [ self.dispatch(&PD::beginCommand,
 252                                                                   boost::ref(self.context.path)) ]
 253                       >> * argument
 254                       >> (ch_p(';') | end_p)
 255                       >> eps_p                    [ self.dispatch(&PD::endCommand) ]
 256                     ;
 257
 258                 argument
 259                     =    simple_argument          [ self.dispatch(&PD::pushArgument,
 260                                                                   boost::ref(self.context.type),
 261                                                                   boost::ref(self.context.str)) ]
 262                     |    complex_argument
 263                     ;
 264
 265                 simple_argument         // All these return their value in context.str
 266                     =    string
 267                     |    hexstring
 268                     |    word
 269                     ;
 270
 271                 complex_argument        // Argument consists of multiple tokens
 272                     =    ch_p('(')                [ self.dispatch(&PD::openGroup) ]
 273                       >> * token
 274                       >> ch_p(')')                [ self.dispatch(&PD::closeGroup) ]
 275                     ;
 276
 277                 string                  // Returns value in context.str
 278                     =    eps_p                    [ clear_a(self.context.str) ]
 279                       >> eps_p                    [ assign_a(self.context.type,
 280                                                              self.context.BasicString) ]
 281                       >> lexeme_d
 282                          [
 283                              ch_p('"')
 284                           >> * ( ( lex_escape_ch_p[ assign_a(self.context.ch) ]
 285                                    - '"'
 286                                  )                [ append_a(self.context.str,
 287                                                              self.context.ch) ]
 288                                )
 289                           >> ch_p('"')
 290                          ]
 291                     ;
 292
 293                 hexstring               // Returns value in context.str
 294                     =    eps_p                    [ clear_a(self.context.str) ]
 295                       >> eps_p                    [ assign_a(self.context.type,
 296                                                              self.context.HexString) ]
 297                       >> confix_p( "x\"", * hexbyte, '"' )
 298                     ;
 299
 300                 path                    // Returns value in context.path
 301                     =    eps_p                    [ clear_a(self.context.path) ]
 302                       >> relpath | abspath
 303                     ;
 304
 305                 relpath
 306                     =    (   word                 [ push_back_a(self.context.path) ]
 307                            % ch_p('/') )
 308                       >> ( ! ch_p('/')            [ push_back_a(self.context.path,
 309                                                                 self.context.EmptyString) ] )
 310                     ;
 311
 312                 abspath
 313                     =    ch_p('/')                [ push_back_a(self.context.path,
 314                                                                 self.context.EmptyString) ]
 315                       >> ( relpath
 316                          | eps_p                  [ push_back_a(self.context.path, "") ] )
 317                     ;
 318
 319                 balanced_tokens
 320                     =    ch_p('(')                [ self.dispatch(&PD::pushPunctuation, "(") ]
 321                       >> * token
 322                       >> ch_p(')')                [ self.dispatch(&PD::pushPunctuation, ")") ]
 323                     ;
 324
 325                 token
 326                     =    simple_argument          [ self.dispatch(&PD::pushWord,
 327                                                                   boost::ref(self.context.type),
 328                                                                   boost::ref(self.context.str)) ]
 329                     |    punctuation              [ self.dispatch(&PD::pushPunctuation,
 330                                                                   boost::ref(self.context.str)) ]
 331                     |    balanced_tokens
 332                     ;
 333
 334                 punctuation             // Returns value in context.str
 335                     =    punctuation_p            [ assign_a(self.context.str) ]
 336                     ;
 337
 338                 word                    // Returns value in context.str
 339                     =    lexeme_d
 340                          [
 341                              eps_p                [ assign_a(self.context.type,
 342                                                              self.context.Word) ]
 343                           >> (+ word_p)           [ assign_a(self.context.str) ]
 344                          ]
 345                     ;
 346
 347                 hexbyte
 348                     =    uint_parser<char, 16, 2, 2>()
 349                                                   [ append_a(self.context.str) ]
 350                     ;
 351
 352                 skip
 353                     =    space_p | comment_p('#')
 354                     ;
 355
 356                 ///////////////////////////////////////////////////////////////////
 357
 358                 start_parsers(
 359                     commands,           // CommandParser
 360                     skip                // SkipParser
 361                 );
 362
 363                 BOOST_SPIRIT_DEBUG_TRACE_RULE(command,1);
 364                 BOOST_SPIRIT_DEBUG_TRACE_RULE(path,1);
 365                 BOOST_SPIRIT_DEBUG_TRACE_RULE(argument,1);
 366                 BOOST_SPIRIT_DEBUG_TRACE_RULE(word,1);
 367                 BOOST_SPIRIT_DEBUG_TRACE_RULE(string,1);
 368                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexstring,1);
 369                 BOOST_SPIRIT_DEBUG_TRACE_RULE(token,1);
 370                 BOOST_SPIRIT_DEBUG_TRACE_RULE(punctuation,1);
 371                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexbyte,1);
 372                 BOOST_SPIRIT_DEBUG_TRACE_RULE(balanced_tokens,1);
 373                 BOOST_SPIRIT_DEBUG_TRACE_RULE(simple_argument,1);
 374                 BOOST_SPIRIT_DEBUG_TRACE_RULE(complex_argument,1);
 375                 BOOST_SPIRIT_DEBUG_TRACE_RULE(builtin,1);
 376                 BOOST_SPIRIT_DEBUG_TRACE_RULE(commands,1);
 377                 BOOST_SPIRIT_DEBUG_TRACE_RULE(block,1);
 378                 BOOST_SPIRIT_DEBUG_TRACE_RULE(statement,1);
 379                 BOOST_SPIRIT_DEBUG_TRACE_RULE(relpath,1);
 380                 BOOST_SPIRIT_DEBUG_TRACE_RULE(abspath,1);
 381             }
 382         };
 383     };
 384
 385     template <class ParseDispatcher>
 386     ArgumentToken::TokenType const CommandGrammar<ParseDispatcher>::Context::BasicString (
 387         ArgumentToken::BasicString);
 388
 389     template <class ParseDispatcher>
 390     ArgumentToken::TokenType const CommandGrammar<ParseDispatcher>::Context::HexString(
 391         ArgumentToken::HexString);
 392
 393     template <class ParseDispatcher>
 394     ArgumentToken::TokenType const CommandGrammar<ParseDispatcher>::Context::Word(
 395         ArgumentToken::Word);
 396
 397     template <class ParseDispatcher>
 398     std::string const CommandGrammar<ParseDispatcher>::Context::EmptyString;
 399
 400 #endif
 401
 402 }}}
 403
 404 ///////////////////////////////ih.e////////////////////////////////////////
 405 #endif
 406
 407 \f
 408 // Local Variables:
 409 // mode: c++
 410 // fill-column: 100
 411 // comment-column: 40
 412 // c-file-style: "senf"
 413 // indent-tabs-mode: nil
 414 // ispell-local-dictionary: "american"
 415 // compile-command: "scons -u test"
 416 // End: