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