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 Token::TokenType TokenType;
  86
  87         struct Context {
  88             std::string str;
  89             std::vector<Token> path;
  90             char ch;
  91             Token token;
  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 Token EmptyToken;
  97         };
  98
  99         Context & context;
 100
 101         ///////////////////////////////////////////////////////////////////////////
 102         // Dispatching semantic actions
 103
 104         ParseDispatcher & dispatcher;
 105
 106         struct Dispatch_actor
 107         {
 108             Dispatch_actor(boost::function<void ()> fn_) : fn (fn_) {}
 109
 110             template <class Value>
 111             void operator()(Value const & value) const
 112                 { fn(); }
 113
 114             template <class Iterator>
 115             void operator()(Iterator const & f, Iterator const & l) const
 116                 { fn(); }
 117
 118             boost::function<void ()> fn;
 119         };
 120
 121         template <class Callback>
 122         Dispatch_actor dispatch(Callback cb) const
 123             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher))); }
 124
 125         template <class Callback, class Arg>
 126         Dispatch_actor dispatch(Callback cb, Arg const & arg) const
 127             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher), arg)); }
 128
 129         template <class Callback, class Arg1, class Arg2>
 130         Dispatch_actor dispatch(Callback cb, Arg1 const & arg1, Arg2 const & arg2) const
 131             { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher), arg1, arg2)); }
 132
 133         struct TokenSetter_actor
 134         {
 135             TokenSetter_actor(Context & c, TokenType t) : c_ (c), t_ (t) {}
 136
 137             void operator()(std::string const & value) const
 138                 { c_.token = Token(t_, value); }
 139             void operator()(char value) const
 140                 { c_.token = Token(t_, std::string(1,value)); }
 141             template <class Iterator> void operator()(Iterator const & f, Iterator const & l) const
 142                 { c_.token = Token(t_, std::string(f,l)); }
 143
 144             Context & c_;
 145             TokenType t_;
 146         };
 147
 148         struct TokenSetter_value_actor
 149             : public TokenSetter_actor
 150         {
 151             TokenSetter_value_actor(Context & c, TokenType t, std::string & v)
 152                 : TokenSetter_actor(c,t), v_ (v) {}
 153
 154             template <class Value> void operator()(Value const &) const
 155                 { TokenSetter_actor::operator()(v_); }
 156
 157             template <class Iterator> void operator()(Iterator const &, Iterator const &) const
 158                 { TokenSetter_actor::operator()(v_); }
 159
 160             std::string & v_;
 161         };
 162
 163         TokenSetter_actor set_token_a(TokenType t) const
 164             { return TokenSetter_actor(context, t); }
 165
 166         TokenSetter_value_actor set_token_a(TokenType t, std::string & arg) const
 167             { return TokenSetter_value_actor(context, t, arg); }
 168
 169         ///////////////////////////////////////////////////////////////////////////
 170
 171         CommandGrammar(ParseDispatcher & d, Context & c)
 172             : context(c), dispatcher(d) {}
 173
 174         template <class Scanner>
 175         struct definition
 176             : public boost::spirit::grammar_def< boost::spirit::rule<Scanner>,
 177                                                  boost::spirit::rule<Scanner> >
 178         {
 179             boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
 180                 punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
 181                 skip, commands, block, statement, relpath, abspath;
 182             boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
 183             boost::spirit::distinct_parser<> keyword_p;
 184
 185             definition(CommandGrammar const & self) :
 186
 187                 // Characters with a special meaning within the parser
 188                 special_p ("/(){};"),
 189
 190                 // Additional characters which are returned as punctuation tokens
 191                 // (only allowed within '()').
 192                 punctuation_p (",="),
 193
 194                 // Whitespace characters
 195                 space_p (" \t\n\r"),
 196
 197                 // Invalid characters: All chars below \x20 (space) which are not space_p
 198                 // (don't put a \0 in the chset<> argument *string* ...)
 199                 invalid_p (
 200                     boost::spirit::chset<>('\0') | boost::spirit::chset<>("\x01-\x20") - space_p ),
 201
 202                 // Valid word characters
 203                 word_p (
 204                     boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
 205
 206                 // Keywords must not be followed by a word char or '/'
 207                 keyword_p ( word_p | boost::spirit::ch_p('/') )
 208
 209             {
 210                 using namespace boost::spirit;
 211                 typedef ParseDispatcher PD;
 212                 typedef Token AT;
 213
 214                 ///////////////////////////////////////////////////////////////////
 215                 // Spirit grammar
 216                 //
 217                 // Syntax summary:
 218                 // This is EBNF with some minor tweaks to accommodate C++ syntax
 219                 //
 220                 //   * and +    like EBNF but they precede their argument
 221                 //   >>         is followed by
 222                 //   !          optional
 223                 //   a % b      match any number of a's separated by b
 224                 //   a - b      match a but not b
 225                 //
 226                 // Beside this, we use some special parsers (ch_p, eps_p, confix_p, lex_escape_ch_p,
 227                 // keyword_p, comment_p) and directives (lexeme_d), however, the parser should be
 228                 // quite readable.
 229                 //
 230                 //   ch_p             match character
 231                 //   eps_p            always matches nothing (to attach unconditional actions)
 232                 //   confix_p(a,b,c)  match b, preceded by a and terminated by c. Used to parse
 233                 //                    string literals and comments
 234                 //   lex_escape_ch_p  match a lex style escape char. This is like a C++ style
 235                 //                    literal string escape char, however \x will be replaced by 'x'
 236                 //                    for any char 'x' if it has no special meaning.
 237                 //   keyword_p        match a delimited keyword
 238                 //   comment_p(a,b)   match comment starting with a and terminated with b. b
 239                 //                    defaults to end-of-line
 240                 //
 241                 //   lexeme_d         don't skip whitespace (as defined by the skip parser)
 242                 //
 243                 // Aligned to the right at column 50 are semantic actions.
 244                 //
 245                 // For clarity, I have used 'ch_p' explicitly throughout even though it is optional
 246                 // in most cases.
 247                 //
 248                 // More info is in the Boost.Spirit documentation
 249
 250                 commands
 251                     =  * command
 252                     ;
 253
 254                 command
 255                     =    builtin >> (ch_p(';') | end_p)
 256                     |    path  >> ( block | statement )
 257                     |    ch_p(';') // Ignore empty commands
 258                     ;
 259
 260                 builtin
 261                     =    keyword_p("cd")
 262                       >> path
 263                       >> eps_p                    [ self.dispatch(&PD::builtin_cd,
 264                                                                   boost::ref(self.context.path)) ]
 265                     |    keyword_p("ls")
 266                       >> ! path
 267                       >> eps_p                    [ self.dispatch(&PD::builtin_ls,
 268                                                                   boost::ref(self.context.path)) ]
 269                     |    keyword_p("exit")        [ self.dispatch(&PD::builtin_exit) ]
 270
 271                     |    keyword_p("help")
 272                       >> ! path
 273                       >> eps_p                    [ self.dispatch(&PD::builtin_help,
 274                                                                   boost::ref(self.context.path)) ]
 275                     ;
 276
 277                 block
 278                     =    ch_p('{')                [ self.dispatch(&PD::pushDirectory,
 279                                                                   boost::ref(self.context.path)) ]
 280                       >> * command
 281                       >> ch_p('}')                [ self.dispatch(&PD::popDirectory) ]
 282                     ;
 283
 284                 statement
 285                     = eps_p                       [ self.dispatch(&PD::beginCommand,
 286                                                                   boost::ref(self.context.path)) ]
 287                       >> * argument
 288                       >> (ch_p(';') | end_p)
 289                       >> eps_p                    [ self.dispatch(&PD::endCommand) ]
 290                     ;
 291
 292                 argument
 293                     =    simple_argument          [ self.dispatch(&PD::pushToken,
 294                                                                   boost::ref(self.context.token)) ]
 295                     |    balanced_tokens
 296                     ;
 297
 298                 simple_argument         // All these return their value in context.token
 299                     =    string
 300                     |    hexstring
 301                     |    word
 302                     ;
 303
 304                 string                  // Returns value in context.token
 305                     =    eps_p                    [ clear_a(self.context.str) ]
 306                       >> lexeme_d
 307                          [
 308                              ch_p('"')
 309                           >> * ( ( lex_escape_ch_p[ assign_a(self.context.ch) ]
 310                                    - '"'
 311                                  )                [ append_a(self.context.str,
 312                                                              self.context.ch) ]
 313                                )
 314                           >> ch_p('"')            [ self.set_token_a(AT::BasicString,
 315                                                                      self.context.str) ]
 316                          ]
 317                     ;
 318
 319                 hexstring               // Returns value in context.token
 320                     =    eps_p                    [ clear_a(self.context.str) ]
 321                       >> confix_p( "x\"", * hexbyte, '"' )
 322                                                   [ self.set_token_a(AT::HexString,
 323                                                                      self.context.str) ]
 324                     ;
 325
 326                 path                    // Returns value in context.path
 327                     =    eps_p                    [ clear_a(self.context.path) ]
 328                       >> relpath | abspath
 329                     ;
 330
 331                 relpath
 332                     =    (   word                 [ push_back_a(self.context.path,
 333                                                                 self.context.token) ]
 334                            % ch_p('/') )
 335                       >> ( ! ch_p('/')            [ push_back_a(self.context.path,
 336                                                                 self.context.EmptyToken) ] )
 337                     ;
 338
 339                 abspath
 340                     =    ch_p('/')                [ push_back_a(self.context.path,
 341                                                                 self.context.EmptyToken) ]
 342                       >> ( relpath
 343                          | eps_p                  [ push_back_a(self.context.path,
 344                                                                 self.context.EmptyToken) ] )
 345                     ;
 346
 347                 balanced_tokens
 348                     =    ch_p('(')                [ self.set_token_a(AT::ArgumentGroupOpen) ]
 349                                                   [ self.dispatch(&PD::pushToken,
 350                                                                   boost::ref(self.context.token)) ]
 351                       >> * token
 352                       >> ch_p(')')                [ self.set_token_a(AT::ArgumentGroupClose) ]
 353                                                   [ self.dispatch(&PD::pushToken,
 354                                                                   boost::ref(self.context.token)) ]
 355                     ;
 356
 357                 token
 358                     =    simple_argument          [ self.dispatch(&PD::pushToken,
 359                                                                   boost::ref(self.context.token)) ]
 360                     |    punctuation              [ self.dispatch(&PD::pushToken,
 361                                                                   boost::ref(self.context.token)) ]
 362                     |    balanced_tokens
 363                     ;
 364
 365                 punctuation             // Returns value in context.str
 366                     =    ch_p('/')                [ self.set_token_a(AT::PathSeparator) ]
 367                     |    ch_p('{')                [ self.set_token_a(AT::DirectoryGroupOpen) ]
 368                     |    ch_p('}')                [ self.set_token_a(AT::DirectoryGroupClose) ]
 369                     |    ch_p(';')                [ self.set_token_a(AT::CommandTerminator) ]
 370                     |    punctuation_p            [ self.set_token_a(AT::OtherPunctuation) ]
 371                     ;
 372
 373                 word                    // Returns value in context.token
 374                     =    lexeme_d
 375                          [
 376                              eps_p
 377                           >> (+ word_p)           [ assign_a(self.context.str) ]
 378                          ]
 379                       >> eps_p                    [ self.set_token_a(AT::Word, self.context.str) ]
 380                     ;
 381
 382                 hexbyte
 383                     =    uint_parser<char, 16, 2, 2>()
 384                                                   [ append_a(self.context.str) ]
 385                     ;
 386
 387                 skip
 388                     =    space_p | comment_p('#')
 389                     ;
 390
 391                 ///////////////////////////////////////////////////////////////////
 392
 393                 start_parsers(
 394                     commands,           // CommandParser
 395                     skip                // SkipParser
 396                 );
 397
 398                 BOOST_SPIRIT_DEBUG_TRACE_RULE(command,1);
 399                 BOOST_SPIRIT_DEBUG_TRACE_RULE(path,1);
 400                 BOOST_SPIRIT_DEBUG_TRACE_RULE(argument,1);
 401                 BOOST_SPIRIT_DEBUG_TRACE_RULE(word,1);
 402                 BOOST_SPIRIT_DEBUG_TRACE_RULE(string,1);
 403                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexstring,1);
 404                 BOOST_SPIRIT_DEBUG_TRACE_RULE(token,1);
 405                 BOOST_SPIRIT_DEBUG_TRACE_RULE(punctuation,1);
 406                 BOOST_SPIRIT_DEBUG_TRACE_RULE(hexbyte,1);
 407                 BOOST_SPIRIT_DEBUG_TRACE_RULE(balanced_tokens,1);
 408                 BOOST_SPIRIT_DEBUG_TRACE_RULE(simple_argument,1);
 409                 BOOST_SPIRIT_DEBUG_TRACE_RULE(complex_argument,1);
 410                 BOOST_SPIRIT_DEBUG_TRACE_RULE(builtin,1);
 411                 BOOST_SPIRIT_DEBUG_TRACE_RULE(commands,1);
 412                 BOOST_SPIRIT_DEBUG_TRACE_RULE(block,1);
 413                 BOOST_SPIRIT_DEBUG_TRACE_RULE(statement,1);
 414                 BOOST_SPIRIT_DEBUG_TRACE_RULE(relpath,1);
 415                 BOOST_SPIRIT_DEBUG_TRACE_RULE(abspath,1);
 416             }
 417         };
 418     };
 419
 420     template <class ParseDispatcher>
 421     Token const CommandGrammar<ParseDispatcher>::Context::EmptyToken;
 422
 423 #endif
 424
 425 }}}
 426
 427 ///////////////////////////////ih.e////////////////////////////////////////
 428 #endif
 429
 430 \f
 431 // Local Variables:
 432 // mode: c++
 433 // fill-column: 100
 434 // comment-column: 40
 435 // c-file-style: "senf"
 436 // indent-tabs-mode: nil
 437 // ispell-local-dictionary: "american"
 438 // compile-command: "scons -u test"
 439 // End: