Utils/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_SENF_Scheduler_Console_Parse_
  27 #define IH_SENF_Scheduler_Console_Parse_ 1
  28
  29 // Custom includes
  30 #include <vector>
  31 #include "../../config.hh"
  32 #include <boost/spirit.hpp>
  33 #include <boost/spirit/utility/grammar_def.hpp>
  34 #include <boost/spirit/dynamic.hpp>
  35 #include <boost/spirit/phoenix.hpp>
  36 #include "../../Utils/Phoenix.hh"
  37
  38 ///////////////////////////////ih.p////////////////////////////////////////
  39
  40 namespace senf {
  41 namespace console {
  42 namespace detail {
  43
  44 #ifndef DOXYGEN
  45
  46     ///////////////////////////////////////////////////////////////////////////
  47     // Grammar
  48
  49     template <class ParseDispatcher>
  50     struct CommandGrammar : boost::spirit::grammar<CommandGrammar<ParseDispatcher> >
  51     {
  52         ///////////////////////////////////////////////////////////////////////////
  53         // Start rules
  54
  55         enum { CommandParser, SkipParser, ArgumentsParser, PathParser };
  56
  57         ///////////////////////////////////////////////////////////////////////////
  58         // The parse context (variables needed while parsing)
  59
  60         typedef Token::TokenType TokenType;
  61
  62         struct Context {
  63             std::string str;
  64             std::vector<Token> path;
  65             char ch;
  66             Token token;
  67         };
  68
  69         Context & context;
  70
  71         ///////////////////////////////////////////////////////////////////////////
  72         // Configuration
  73
  74         bool incremental;
  75
  76         ///////////////////////////////////////////////////////////////////////////
  77         // Dispatching semantic actions
  78
  79         ParseDispatcher & dispatcher;
  80
  81         ///////////////////////////////////////////////////////////////////////////
  82         // Errors
  83
  84         enum Errors {
  85             EndOfStatementExpected,
  86             PathExpected,
  87             ClosingParenExpected,
  88             QuoteExpected
  89         };
  90
  91         ///////////////////////////////////////////////////////////////////////////
  92
  93         CommandGrammar(ParseDispatcher & d, Context & c)
  94             : context(c), incremental(false), dispatcher(d) {}
  95
  96         template <class Scanner>
  97         struct definition
  98             : public boost::spirit::grammar_def< boost::spirit::rule<Scanner>,
  99                                                  boost::spirit::rule<Scanner>,
 100                                                  boost::spirit::rule<Scanner>,
 101                                                  boost::spirit::rule<Scanner> >
 102         {
 103             boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
 104                 punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
 105                 skip, statement, relpath, abspath, arguments, group_start, group_close,
 106                 statement_end, opt_path;
 107             boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
 108             boost::spirit::distinct_parser<> keyword_p;
 109
 110             definition(CommandGrammar const & self) :
 111
 112                 // Characters with a special meaning within the parser
 113                 special_p ("/(){};\""),
 114
 115                 // Additional characters which are returned as punctuation tokens
 116                 // (only allowed within '()').
 117                 punctuation_p (",="),
 118
 119                 // Whitespace characters
 120                 space_p (" \t\n\r"),
 121
 122                 // Invalid characters: All chars below \x20 (space) which are not space_p
 123                 // (don't put a \0 in the chset<> argument *string* ...)
 124                 invalid_p ( (boost::spirit::chset<>('\0')
 125                              | boost::spirit::chset<>("\x01-\x20")) - space_p ),
 126
 127                 // Valid word characters
 128                 word_p (
 129                     boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
 130
 131                 // Keywords must not be followed by a word char or '/'
 132                 keyword_p ( word_p | boost::spirit::ch_p('/') )
 133
 134             {
 135                 using namespace boost::spirit;
 136                 using namespace ::phoenix;
 137                 using namespace senf::phoenix;
 138                 typedef ParseDispatcher PD;
 139
 140                 actor< variable< char > >               ch_    (self.context.ch);
 141                 actor< variable< std::string > >        str_   (self.context.str);
 142                 actor< variable< std::vector<Token> > > path_  (self.context.path);
 143                 actor< variable< Token > >              token_ (self.context.token);
 144                 actor< variable< ParseDispatcher > >    d_     (self.dispatcher);
 145
 146                 assertion<Errors> end_of_statement_expected   (EndOfStatementExpected);
 147                 assertion<Errors> path_expected               (PathExpected);
 148                 assertion<Errors> closing_paren_expected      (ClosingParenExpected);
 149                 assertion<Errors> quote_expected              (QuoteExpected);
 150
 151                 ///////////////////////////////////////////////////////////////////
 152                 // Spirit grammar
 153                 //
 154                 // Syntax summary:
 155                 // This is EBNF with some minor tweaks to accommodate C++ syntax
 156                 //
 157                 //   * a        any number of a's
 158                 //   + a        at least one a
 159                 //   ! a        an optional a
 160                 //   a >> b     a followed by b
 161                 //   a | b      a or b
 162                 //   a % b      any number of a's separated by b's
 163                 //   a - b      a but not b
 164                 //
 165                 // Beside this, we use some special parsers (ch_p, eps_p, confix_p, lex_escape_ch_p,
 166                 // keyword_p, comment_p) and directives (lexeme_d), however, the parser should be
 167                 // quite readable.
 168                 //
 169                 //   ch_p             match character
 170                 //   eps_p            always matches nothing (to attach unconditional actions)
 171                 //   confix_p(a,b,c)  match b, preceded by a and terminated by c. Used to parse
 172                 //                    string literals and comments
 173                 //   lex_escape_ch_p  match a lex style escape char. This is like a C++ style
 174                 //                    literal string escape char, however \x will be replaced by 'x'
 175                 //                    for any char 'x' if it has no special meaning.
 176                 //   keyword_p        match a delimited keyword
 177                 //   comment_p(a,b)   match comment starting with a and terminated with b. b
 178                 //                    defaults to end-of-line
 179                 //
 180                 //   lexeme_d         don't skip whitespace (as defined by the skip parser)
 181                 //
 182                 // Aligned to the right at column 50 are semantic actions.
 183                 //
 184                 // For clarity, I have used 'ch_p' explicitly throughout even though it is optional
 185                 // in most cases.
 186                 //
 187                 // More info is in the Boost.Spirit documentation
 188
 189                 command
 190                     =    builtin >> end_of_statement_expected(statement_end)
 191                     |    group_close
 192                     |    ch_p(';') // Ignore empty commands
 193                     |    statement
 194                     ;
 195
 196                 statement
 197                     =    path_expected(path)      [ bind(&PD::beginCommand)(d_, path_) ]
 198                       >> arguments
 199                       >> end_of_statement_expected(
 200                            ( group_start | statement_end )
 201                                                   [ bind(&PD::endCommand)(d_) ]
 202                          )
 203                     ;
 204
 205                 builtin
 206                     =    keyword_p("cd")
 207                       >> path_expected(path)
 208                       >> eps_p                    [ bind(&PD::builtin_cd)(d_, path_) ]
 209                     |    keyword_p("ls")
 210                       >> ! path
 211                       >> eps_p                    [ bind(&PD::builtin_ls)(d_, path_) ]
 212                     |    keyword_p("exit")        [ bind(&PD::builtin_exit)(d_) ]
 213                     |    keyword_p("help")
 214                       >> ! path
 215                       >> eps_p                    [ bind(&PD::builtin_help)(d_, path_) ]
 216                     ;
 217
 218                 group_start
 219                     =    ch_p('{')                [ bind(&PD::pushDirectory)(d_) ]
 220                     ;
 221
 222                 group_close
 223                     =    ch_p('}')                [ bind(&PD::popDirectory)(d_) ]
 224                     ;
 225
 226                 arguments
 227                     =    * argument
 228                     ;
 229
 230                 argument
 231                     =    simple_argument          [ bind(&PD::pushToken)(d_, token_) ]
 232                     |    balanced_tokens
 233                     ;
 234
 235                 simple_argument         // All these return their value in context.token
 236                     =    string
 237                     |    hexstring
 238                     |    word
 239                     ;
 240
 241                 string                  // Returns value in context.token
 242                     =    eps_p                    [ clear(str_) ]
 243                       >> lexeme_d
 244                          [
 245                              ch_p('"')
 246                           >> * ( ( lex_escape_ch_p[ ch_ = arg1 ]
 247                                    - '"'
 248                                  )                [ str_ += ch_ ]
 249                                )
 250                           >> quote_expected(ch_p('"'))
 251                                                   [ token_ = construct_<Token>(Token::BasicString,
 252                                                                                str_) ]
 253                          ]
 254                     ;
 255
 256                 hexstring               // Returns value in context.token
 257                     =    eps_p                    [ clear(str_) ]
 258                       >>  "x\""
 259                       >> * ( hexbyte - ch_p('"') )
 260                       >> quote_expected(ch_p('"'))
 261                                                   [ token_ = construct_<Token>(Token::HexString,
 262                                                                                str_) ]
 263                     ;
 264
 265                 opt_path
 266                     = ! path                      [ bind(&PD::beginCommand)(d_, path_) ]
 267                                                   [ bind(&PD::endCommand)(d_) ]
 268                     ;
 269
 270                 path                    // Returns value in context.path
 271                     =    eps_p                    [ clear(path_) ]
 272                       >> relpath | abspath
 273                     ;
 274
 275                 relpath
 276                     =    (   word                 [ push_back(path_, token_) ]
 277                            % ch_p('/') )
 278                       >> ( ! ch_p('/')            [ push_back(path_, construct_<Token>()) ] )
 279                     ;
 280
 281                 abspath
 282                     =    ch_p('/')                [ push_back(path_, construct_<Token>()) ]
 283                       >> ( relpath
 284                          | eps_p                  [ push_back(path_, construct_<Token>()) ] )
 285                     ;
 286
 287                 balanced_tokens
 288                     =    ch_p('(')                [ token_ = construct_<Token>(
 289                                                         Token::ArgumentGroupOpen,
 290                                                         "(") ]
 291                                                   [ bind(&PD::pushToken)(d_, token_) ]
 292                       >> * token
 293                       >> closing_paren_expected(ch_p(')'))
 294                                                   [ token_ = construct_<Token>(
 295                                                         Token::ArgumentGroupClose,
 296                                                         ")") ]
 297                                                   [ bind(&PD::pushToken)(d_, token_) ]
 298                     ;
 299
 300                 token
 301                     =    simple_argument          [ bind(&PD::pushToken)(d_, token_) ]
 302                     |    punctuation              [ bind(&PD::pushToken)(d_, token_) ]
 303                     |    balanced_tokens
 304                     ;
 305
 306                 punctuation             // Returns value in context.str
 307                     =    ch_p('/')                [ token_ = construct_<Token>(
 308                                                         Token::PathSeparator,
 309                                                         "/") ]
 310                     |    ch_p('{')                [ token_ = construct_<Token>(
 311                                                         Token::DirectoryGroupOpen,
 312                                                         "{") ]
 313                     |    ch_p('}')                [ token_ = construct_<Token>(
 314                                                         Token::DirectoryGroupClose,
 315                                                         "}") ]
 316                     |    ch_p(';')                [ token_ = construct_<Token>(
 317                                                         Token::CommandTerminator,
 318                                                         ";") ]
 319                     |    punctuation_p            [ token_ = construct_<Token>(
 320                                                         Token::OtherPunctuation,
 321                                                         construct_<std::string>(1u, arg1)) ]
 322                     ;
 323
 324                 word                    // Returns value in context.token
 325                     =    lexeme_d
 326                          [
 327                              (+ word_p)           [ str_ = construct_<std::string>(arg1, arg2) ]
 328                          ]
 329                       >> eps_p                    [ token_ = construct_<Token>(
 330                                                         Token::Word,
 331                                                         str_) ]
 332                     ;
 333
 334                 hexbyte
 335                     =    uint_parser<char, 16, 2, 2>()
 336                                                   [ push_back(str_, arg1) ]
 337                     ;
 338
 339                 statement_end
 340                     =    if_p(var(self.incremental)) [
 341                                ch_p(';')
 342                          ]
 343                          .else_p [
 344                                ch_p(';')
 345                              | end_p
 346                          ]
 347                     ;
 348
 349                 skip
 350                     =    space_p | comment_p('#')
 351                     ;
 352
 353                 ///////////////////////////////////////////////////////////////////
 354
 355                 start_parsers(
 356                     command,            // CommandParser
 357                     skip,               // SkipParser
 358                     arguments,          // ArgumentsParser
 359                     opt_path            // PathParser
 360                 );
 361
 362                 BOOST_SPIRIT_DEBUG_TRACE_RULE(command,1);
 363                 BOOST_SPIRIT_DEBUG_TRACE_RULE(path,1);
 364                 BOOST_SPIRIT_DEBUG_TRACE_RULE(argument,1);
 365                 BOOST_SPIRIT_DEBUG_TRACE_RULE(word,1);
 366                 BOOST_SPIRIT_DEBUG_TRACE_RULE(string,1);
 367                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexstring,1);
 368                 BOOST_SPIRIT_DEBUG_TRACE_RULE(token,1);
 369                 BOOST_SPIRIT_DEBUG_TRACE_RULE(punctuation,1);
 370                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexbyte,1);
 371                 BOOST_SPIRIT_DEBUG_TRACE_RULE(balanced_tokens,1);
 372                 BOOST_SPIRIT_DEBUG_TRACE_RULE(simple_argument,1);
 373                 BOOST_SPIRIT_DEBUG_TRACE_RULE(complex_argument,1);
 374                 BOOST_SPIRIT_DEBUG_TRACE_RULE(builtin,1);
 375                 BOOST_SPIRIT_DEBUG_TRACE_RULE(commands,1);
 376                 BOOST_SPIRIT_DEBUG_TRACE_RULE(block,1);
 377                 BOOST_SPIRIT_DEBUG_TRACE_RULE(statement,1);
 378                 BOOST_SPIRIT_DEBUG_TRACE_RULE(relpath,1);
 379                 BOOST_SPIRIT_DEBUG_TRACE_RULE(abspath,1);
 380             }
 381         };
 382     };
 383
 384 #endif
 385
 386 }}}
 387
 388 ///////////////////////////////ih.e////////////////////////////////////////
 389 #endif
 390
 391 \f
 392 // Local Variables:
 393 // mode: c++
 394 // fill-column: 100
 395 // comment-column: 40
 396 // c-file-style: "senf"
 397 // indent-tabs-mode: nil
 398 // ispell-local-dictionary: "american"
 399 // compile-command: "scons -u test"
 400 // End: