Scheduler/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 <boost/spirit.hpp>
  32 #include <boost/spirit/utility/grammar_def.hpp>
  33 #include <boost/spirit/dynamic.hpp>
  34 #include <boost/spirit/phoenix.hpp>
  35 #include "../../Utils/Phoenix.hh"
  36
  37 ///////////////////////////////ih.p////////////////////////////////////////
  38
  39 namespace senf {
  40 namespace console {
  41 namespace detail {
  42
  43 #ifndef DOXYGEN
  44
  45     ///////////////////////////////////////////////////////////////////////////
  46     // Grammar
  47
  48     template <class ParseDispatcher>
  49     struct CommandGrammar : boost::spirit::grammar<CommandGrammar<ParseDispatcher> >
  50     {
  51         ///////////////////////////////////////////////////////////////////////////
  52         // Start rules
  53
  54         enum { CommandParser, SkipParser, ArgumentsParser };
  55
  56         ///////////////////////////////////////////////////////////////////////////
  57         // The parse context (variables needed while parsing)
  58
  59         typedef Token::TokenType TokenType;
  60
  61         struct Context {
  62             std::string str;
  63             std::vector<Token> path;
  64             char ch;
  65             Token token;
  66         };
  67
  68         Context & context;
  69
  70         ///////////////////////////////////////////////////////////////////////////
  71         // Configuration
  72
  73         bool incremental;
  74
  75         ///////////////////////////////////////////////////////////////////////////
  76         // Dispatching semantic actions
  77
  78         ParseDispatcher & dispatcher;
  79
  80         ///////////////////////////////////////////////////////////////////////////
  81         // Errors
  82
  83         enum Errors {
  84             EndOfStatementExpected,
  85             GroupOrArgumentsExpected,
  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         {
 102             boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
 103                 punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
 104                 skip, statement, relpath, abspath, arguments, group_start, group_close,
 105                 statement_end;
 106             boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
 107             boost::spirit::distinct_parser<> keyword_p;
 108
 109             definition(CommandGrammar const & self) :
 110
 111                 // Characters with a special meaning within the parser
 112                 special_p ("/(){};\""),
 113
 114                 // Additional characters which are returned as punctuation tokens
 115                 // (only allowed within '()').
 116                 punctuation_p (",="),
 117
 118                 // Whitespace characters
 119                 space_p (" \t\n\r"),
 120
 121                 // Invalid characters: All chars below \x20 (space) which are not space_p
 122                 // (don't put a \0 in the chset<> argument *string* ...)
 123                 invalid_p ( (boost::spirit::chset<>('\0')
 124                              | boost::spirit::chset<>("\x01-\x20")) - space_p ),
 125
 126                 // Valid word characters
 127                 word_p (
 128                     boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
 129
 130                 // Keywords must not be followed by a word char or '/'
 131                 keyword_p ( word_p | boost::spirit::ch_p('/') )
 132
 133             {
 134                 using namespace boost::spirit;
 135                 using namespace ::phoenix;
 136                 using namespace senf::phoenix;
 137                 typedef ParseDispatcher PD;
 138
 139                 actor< variable< char > >               ch_    (self.context.ch);
 140                 actor< variable< std::string > >        str_   (self.context.str);
 141                 actor< variable< std::vector<Token> > > path_  (self.context.path);
 142                 actor< variable< Token > >              token_ (self.context.token);
 143                 actor< variable< ParseDispatcher > >    d_     (self.dispatcher);
 144
 145                 assertion<Errors> end_of_statement_expected   (EndOfStatementExpected);
 146                 assertion<Errors> group_or_arguments_expected (GroupOrArgumentsExpected);
 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                     |    path_expected(path)
 194                       >> group_or_arguments_expected( group_start | statement )
 195                     ;
 196
 197                 builtin
 198                     =    keyword_p("cd")
 199                       >> path_expected(path)
 200                       >> eps_p                    [ bind(&PD::builtin_cd)(d_, path_) ]
 201                     |    keyword_p("ls")
 202                       >> ! path
 203                       >> eps_p                    [ bind(&PD::builtin_ls)(d_, path_) ]
 204                     |    keyword_p("exit")        [ bind(&PD::builtin_exit)(d_) ]
 205                     |    keyword_p("help")
 206                       >> ! path
 207                       >> eps_p                    [ bind(&PD::builtin_help)(d_, path_) ]
 208                     ;
 209
 210                 group_start
 211                     =    ch_p('{')                [ bind(&PD::pushDirectory)(d_, path_) ]
 212                     ;
 213
 214                 group_close
 215                     =    ch_p('}')                [ bind(&PD::popDirectory)(d_) ]
 216                     ;
 217
 218                 statement
 219                     =    eps_p                    [ bind(&PD::beginCommand)(d_, path_) ]
 220                       >> arguments
 221                       >> end_of_statement_expected(statement_end)
 222                                                   [ bind(&PD::endCommand)(d_) ]
 223                     ;
 224
 225                 arguments
 226                     =    * argument
 227                     ;
 228
 229                 argument
 230                     =    simple_argument          [ bind(&PD::pushToken)(d_, token_) ]
 231                     |    balanced_tokens
 232                     ;
 233
 234                 simple_argument         // All these return their value in context.token
 235                     =    string
 236                     |    hexstring
 237                     |    word
 238                     ;
 239
 240                 string                  // Returns value in context.token
 241                     =    eps_p                    [ clear(str_) ]
 242                       >> lexeme_d
 243                          [
 244                              ch_p('"')
 245                           >> * ( ( lex_escape_ch_p[ ch_ = arg1 ]
 246                                    - '"'
 247                                  )                [ str_ += ch_ ]
 248                                )
 249                           >> quote_expected(ch_p('"'))
 250                                                   [ token_ = construct_<Token>(Token::BasicString,
 251                                                                                str_) ]
 252                          ]
 253                     ;
 254
 255                 hexstring               // Returns value in context.token
 256                     =    eps_p                    [ clear(str_) ]
 257                       >>  "x\""
 258                       >> * ( hexbyte - ch_p('"') )
 259                       >> quote_expected(ch_p('"'))
 260                                                   [ token_ = construct_<Token>(Token::HexString,
 261                                                                                str_) ]
 262                     ;
 263
 264                 path                    // Returns value in context.path
 265                     =    eps_p                    [ clear(path_) ]
 266                       >> relpath | abspath
 267                     ;
 268
 269                 relpath
 270                     =    (   word                 [ push_back(path_, token_) ]
 271                            % ch_p('/') )
 272                       >> ( ! ch_p('/')            [ push_back(path_, construct_<Token>()) ] )
 273                     ;
 274
 275                 abspath
 276                     =    ch_p('/')                [ push_back(path_, construct_<Token>()) ]
 277                       >> ( relpath
 278                          | eps_p                  [ push_back(path_, construct_<Token>()) ] )
 279                     ;
 280
 281                 balanced_tokens
 282                     =    ch_p('(')                [ token_ = construct_<Token>(
 283                                                         Token::ArgumentGroupOpen,
 284                                                         "(") ]
 285                                                   [ bind(&PD::pushToken)(d_, token_) ]
 286                       >> * token
 287                       >> closing_paren_expected(ch_p(')'))
 288                                                   [ token_ = construct_<Token>(
 289                                                         Token::ArgumentGroupClose,
 290                                                         ")") ]
 291                                                   [ bind(&PD::pushToken)(d_, token_) ]
 292                     ;
 293
 294                 token
 295                     =    simple_argument          [ bind(&PD::pushToken)(d_, token_) ]
 296                     |    punctuation              [ bind(&PD::pushToken)(d_, token_) ]
 297                     |    balanced_tokens
 298                     ;
 299
 300                 punctuation             // Returns value in context.str
 301                     =    ch_p('/')                [ token_ = construct_<Token>(
 302                                                         Token::PathSeparator,
 303                                                         "/") ]
 304                     |    ch_p('{')                [ token_ = construct_<Token>(
 305                                                         Token::DirectoryGroupOpen,
 306                                                         "{") ]
 307                     |    ch_p('}')                [ token_ = construct_<Token>(
 308                                                         Token::DirectoryGroupClose,
 309                                                         "}") ]
 310                     |    ch_p(';')                [ token_ = construct_<Token>(
 311                                                         Token::CommandTerminator,
 312                                                         ";") ]
 313                     |    punctuation_p            [ token_ = construct_<Token>(
 314                                                         Token::OtherPunctuation,
 315                                                         construct_<std::string>(1u, arg1)) ]
 316                     ;
 317
 318                 word                    // Returns value in context.token
 319                     =    lexeme_d
 320                          [
 321                              (+ word_p)           [ str_ = construct_<std::string>(arg1, arg2) ]
 322                          ]
 323                       >> eps_p                    [ token_ = construct_<Token>(
 324                                                         Token::Word,
 325                                                         str_) ]
 326                     ;
 327
 328                 hexbyte
 329                     =    uint_parser<char, 16, 2, 2>()
 330                                                   [ push_back(str_, arg1) ]
 331                     ;
 332
 333                 statement_end
 334                     =    if_p(var(self.incremental)) [
 335                                ch_p(';')
 336                          ]
 337                          .else_p [
 338                                ch_p(';')
 339                              | end_p
 340                          ]
 341                     ;
 342
 343                 skip
 344                     =    space_p | comment_p('#')
 345                     ;
 346
 347                 ///////////////////////////////////////////////////////////////////
 348
 349                 start_parsers(
 350                     command,            // CommandParser
 351                     skip,               // SkipParser
 352                     arguments           // ArgumentsParser
 353                 );
 354
 355                 BOOST_SPIRIT_DEBUG_TRACE_RULE(command,1);
 356                 BOOST_SPIRIT_DEBUG_TRACE_RULE(path,1);
 357                 BOOST_SPIRIT_DEBUG_TRACE_RULE(argument,1);
 358                 BOOST_SPIRIT_DEBUG_TRACE_RULE(word,1);
 359                 BOOST_SPIRIT_DEBUG_TRACE_RULE(string,1);
 360                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexstring,1);
 361                 BOOST_SPIRIT_DEBUG_TRACE_RULE(token,1);
 362                 BOOST_SPIRIT_DEBUG_TRACE_RULE(punctuation,1);
 363                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexbyte,1);
 364                 BOOST_SPIRIT_DEBUG_TRACE_RULE(balanced_tokens,1);
 365                 BOOST_SPIRIT_DEBUG_TRACE_RULE(simple_argument,1);
 366                 BOOST_SPIRIT_DEBUG_TRACE_RULE(complex_argument,1);
 367                 BOOST_SPIRIT_DEBUG_TRACE_RULE(builtin,1);
 368                 BOOST_SPIRIT_DEBUG_TRACE_RULE(commands,1);
 369                 BOOST_SPIRIT_DEBUG_TRACE_RULE(block,1);
 370                 BOOST_SPIRIT_DEBUG_TRACE_RULE(statement,1);
 371                 BOOST_SPIRIT_DEBUG_TRACE_RULE(relpath,1);
 372                 BOOST_SPIRIT_DEBUG_TRACE_RULE(abspath,1);
 373             }
 374         };
 375     };
 376
 377 #endif
 378
 379 }}}
 380
 381 ///////////////////////////////ih.e////////////////////////////////////////
 382 #endif
 383
 384 \f
 385 // Local Variables:
 386 // mode: c++
 387 // fill-column: 100
 388 // comment-column: 40
 389 // c-file-style: "senf"
 390 // indent-tabs-mode: nil
 391 // ispell-local-dictionary: "american"
 392 // compile-command: "scons -u test"
 393 // End: