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