#define IH_Parse_ 1
// Custom includes
-#include <boost/regex.hpp>
+#include <vector>
#include <boost/spirit.hpp>
-#include <boost/spirit/utility/regex.hpp>
-#include <boost/spirit/actor.hpp>
-#include <boost/bind.hpp>
-#include <boost/function.hpp>
-#include <boost/ref.hpp>
+#include <boost/spirit/utility/grammar_def.hpp>
+#include <boost/spirit/dynamic.hpp>
+#include <boost/spirit/phoenix.hpp>
+#include "../Utils/Phoenix.hh"
///////////////////////////////ih.p////////////////////////////////////////
namespace console {
namespace detail {
- struct append_action
- {
- template <class T, class Value>
- void act(T & ref, Value const & value) const
- { ref += T(1, value); }
-
- template <class T, class Iterator>
- void act(T & ref, Iterator const & f, Iterator const & l) const
- { ref += T(f,l); }
- };
+#ifndef DOXYGEN
- template <class T>
- inline boost::spirit::ref_value_actor<T, append_action>
- append_a(T & ref)
- {
- return boost::spirit::ref_value_actor<T, append_action>(ref);
- }
-
- template <class T, class Value>
- inline boost::spirit::ref_const_ref_actor<T, Value, append_action>
- append_a(T & ref, Value const & value)
- {
- return boost::spirit::ref_const_ref_actor<T, Value, append_action>(ref, value);
- }
+ ///////////////////////////////////////////////////////////////////////////
+ // Grammar
template <class ParseDispatcher>
struct CommandGrammar : boost::spirit::grammar<CommandGrammar<ParseDispatcher> >
{
///////////////////////////////////////////////////////////////////////////
+ // Start rules
+
+ enum { CommandParser, SkipParser, ArgumentsParser };
+
+ ///////////////////////////////////////////////////////////////////////////
// The parse context (variables needed while parsing)
+ typedef Token::TokenType TokenType;
+
struct Context {
std::string str;
+ std::vector<Token> path;
char ch;
+ Token token;
};
Context & context;
///////////////////////////////////////////////////////////////////////////
+ // Configuration
+
+ bool incremental;
+
+ ///////////////////////////////////////////////////////////////////////////
// Dispatching semantic actions
ParseDispatcher & dispatcher;
- struct Dispatch_actor
+ ///////////////////////////////////////////////////////////////////////////
+
+ CommandGrammar(ParseDispatcher & d, Context & c)
+ : context(c), incremental(false), dispatcher(d) {}
+
+ template <class Scanner>
+ struct definition
+ : public boost::spirit::grammar_def< boost::spirit::rule<Scanner>,
+ boost::spirit::rule<Scanner>,
+ boost::spirit::rule<Scanner> >
{
- Dispatch_actor(boost::function<void ()> fn_) : fn (fn_) {}
+ boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
+ punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
+ skip, statement, relpath, abspath, arguments, group_start, group_close,
+ statement_end;
+ boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
+ boost::spirit::distinct_parser<> keyword_p;
- template <class Value>
- void operator()(Value const & value) const
- { fn(); }
+ definition(CommandGrammar const & self) :
- template <class Iterator>
- void operator()(Iterator const & f, Iterator const & l) const
- { fn(); }
+ // Characters with a special meaning within the parser
+ special_p ("/(){};"),
- boost::function<void ()> fn;
- };
-
- template <class Callback>
- Dispatch_actor dispatch(Callback cb) const
- { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher))); }
+ // Additional characters which are returned as punctuation tokens
+ // (only allowed within '()').
+ punctuation_p (",="),
- template <class Callback, class Arg>
- Dispatch_actor dispatch(Callback cb, Arg const & arg) const
- { return Dispatch_actor(boost::bind(cb, boost::ref(dispatcher), arg)); }
+ // Whitespace characters
+ space_p (" \t\n\r"),
- ///////////////////////////////////////////////////////////////////////////
+ // Invalid characters: All chars below \x20 (space) which are not space_p
+ // (don't put a \0 in the chset<> argument *string* ...)
+ invalid_p (
+ boost::spirit::chset<>('\0') | boost::spirit::chset<>("\x01-\x20") - space_p ),
- CommandGrammar(ParseDispatcher & d, Context & c)
- : context(c), dispatcher(d) {}
+ // Valid word characters
+ word_p (
+ boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
- template <class Scanner>
- struct definition
- {
- boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token;
- boost::spirit::rule<Scanner> punctuation, hexbyte, balanced_tokens, simple_argument;
- boost::spirit::rule<Scanner> complex_argument;
+ // Keywords must not be followed by a word char or '/'
+ keyword_p ( word_p | boost::spirit::ch_p('/') )
- definition(CommandGrammar const & self) {
+ {
using namespace boost::spirit;
+ using namespace ::phoenix;
+ using namespace senf::phoenix;
typedef ParseDispatcher PD;
-
+
+ actor< variable< char > > ch_ (self.context.ch);
+ actor< variable< std::string > > str_ (self.context.str);
+ actor< variable< std::vector<Token> > > path_ (self.context.path);
+ actor< variable< Token > > token_ (self.context.token);
+ actor< variable< ParseDispatcher > > d_ (self.dispatcher);
+
+ ///////////////////////////////////////////////////////////////////
+ // Spirit grammar
+ //
+ // Syntax summary:
+ // This is EBNF with some minor tweaks to accommodate C++ syntax
+ //
+ // * a any number of a's
+ // + a at least one a
+ // ! a an optional a
+ // a >> b a followed by b
+ // a | b a or b
+ // a % b any number of a's separated by b's
+ // a - b a but not b
+ //
+ // Beside this, we use some special parsers (ch_p, eps_p, confix_p, lex_escape_ch_p,
+ // keyword_p, comment_p) and directives (lexeme_d), however, the parser should be
+ // quite readable.
+ //
+ // ch_p match character
+ // eps_p always matches nothing (to attach unconditional actions)
+ // confix_p(a,b,c) match b, preceded by a and terminated by c. Used to parse
+ // string literals and comments
+ // lex_escape_ch_p match a lex style escape char. This is like a C++ style
+ // literal string escape char, however \x will be replaced by 'x'
+ // for any char 'x' if it has no special meaning.
+ // keyword_p match a delimited keyword
+ // comment_p(a,b) match comment starting with a and terminated with b. b
+ // defaults to end-of-line
+ //
+ // lexeme_d don't skip whitespace (as defined by the skip parser)
+ //
+ // Aligned to the right at column 50 are semantic actions.
+ //
+ // For clarity, I have used 'ch_p' explicitly throughout even though it is optional
+ // in most cases.
+ //
+ // More info is in the Boost.Spirit documentation
+
command
- = path [ self.dispatch(&PD::beginCommand,
- boost::cref(self.context.str)) ]
- >> * argument
- >> ! ch_p(';')
- >> eps_p [ self.dispatch(&PD::endCommand) ]
+ = builtin >> statement_end
+ | path >> ( group_start | statement )
+ | group_close
+ | ch_p(';') // Ignore empty commands
+ ;
+
+ builtin
+ = keyword_p("cd")
+ >> path
+ >> eps_p [ bind(&PD::builtin_cd)(d_, path_) ]
+ | keyword_p("ls")
+ >> ! path
+ >> eps_p [ bind(&PD::builtin_ls)(d_, path_) ]
+ | keyword_p("exit") [ bind(&PD::builtin_exit)(d_) ]
+
+ | keyword_p("help")
+ >> ! path
+ >> eps_p [ bind(&PD::builtin_help)(d_, path_) ]
+ ;
+
+ group_start
+ = ch_p('{') [ bind(&PD::pushDirectory)(d_, path_) ]
+ ;
+
+ group_close
+ = ch_p('}') [ bind(&PD::popDirectory)(d_) ]
+ ;
+
+ statement
+ = eps_p [ bind(&PD::beginCommand)(d_, path_) ]
+ >> arguments
+ >> statement_end [ bind(&PD::endCommand)(d_) ]
+ ;
+
+ arguments
+ = * argument
;
argument
- = simple_argument [ self.dispatch(&PD::pushArgument,
- boost::cref(self.context.str)) ]
- | complex_argument
+ = simple_argument [ bind(&PD::pushToken)(d_, token_) ]
+ | balanced_tokens
;
- simple_argument // All these return their value in context.str
+ simple_argument // All these return their value in context.token
= string
| hexstring
- | path
+ | word
;
- complex_argument // Argument consists of multiple tokens
- = ch_p('(') [ self.dispatch(&PD::openGroup) ]
- >> * token
- >> ch_p(')') [ self.dispatch(&PD::closeGroup) ]
- ;
-
- string // Returns value in context.str
- = eps_p [ clear_a(self.context.str) ]
+ string // Returns value in context.token
+ = eps_p [ clear(str_) ]
>> lexeme_d
[
ch_p('"')
- >> * ( ( lex_escape_ch_p[ assign_a(self.context.ch) ]
+ >> * ( ( lex_escape_ch_p[ ch_ = arg1 ]
- '"'
- ) [ append_a(self.context.str,
- self.context.ch) ]
+ ) [ str_ += ch_ ]
)
- >> ch_p('"')
+ >> ch_p('"') [ token_ = construct_<Token>(Token::BasicString,
+ str_) ]
]
;
- hexstring // Returns value in context.str
- = eps_p [ clear_a(self.context.str) ]
+ hexstring // Returns value in context.token
+ = eps_p [ clear(str_) ]
>> confix_p( "x\"", * hexbyte, '"' )
+ [ token_ = construct_<Token>(Token::HexString,
+ str_) ]
;
- path // Returns value in context.str
- = eps_p [ clear_a(self.context.str) ]
- >> ( ! ch_p('/') [ append_a(self.context.str) ]
- )
- >> ( word [ append_a(self.context.str) ]
- % ch_p('/') [ append_a(self.context.str) ]
- )
+ path // Returns value in context.path
+ = eps_p [ clear(path_) ]
+ >> relpath | abspath
+ ;
+
+ relpath
+ = ( word [ push_back(path_, token_) ]
+ % ch_p('/') )
+ >> ( ! ch_p('/') [ push_back(path_, construct_<Token>()) ] )
+ ;
+
+ abspath
+ = ch_p('/') [ push_back(path_, construct_<Token>()) ]
+ >> ( relpath
+ | eps_p [ push_back(path_, construct_<Token>()) ] )
;
balanced_tokens
- = ch_p('(') [ self.dispatch(&PD::pushPunctuation, "(") ]
+ = ch_p('(') [ token_ = construct_<Token>(
+ Token::ArgumentGroupOpen,
+ "(") ]
+ [ bind(&PD::pushToken)(d_, token_) ]
>> * token
- >> ch_p(')') [ self.dispatch(&PD::pushPunctuation, ")") ]
+ >> ch_p(')') [ token_ = construct_<Token>(
+ Token::ArgumentGroupClose,
+ ")") ]
+ [ bind(&PD::pushToken)(d_, token_) ]
;
token
- = simple_argument [ self.dispatch(&PD::pushWord,
- boost::cref(self.context.str)) ]
- | punctuation [ self.dispatch(&PD::pushPunctuation,
- boost::cref(self.context.str)) ]
+ = simple_argument [ bind(&PD::pushToken)(d_, token_) ]
+ | punctuation [ bind(&PD::pushToken)(d_, token_) ]
| balanced_tokens
;
punctuation // Returns value in context.str
- = regex_p("[,=]") [ assign_a(self.context.str) ]
+ = ch_p('/') [ token_ = construct_<Token>(
+ Token::PathSeparator,
+ "/") ]
+ | ch_p('{') [ token_ = construct_<Token>(
+ Token::DirectoryGroupOpen,
+ "{") ]
+ | ch_p('}') [ token_ = construct_<Token>(
+ Token::DirectoryGroupClose,
+ "}") ]
+ | ch_p(';') [ token_ = construct_<Token>(
+ Token::CommandTerminator,
+ ";") ]
+ | punctuation_p [ token_ = construct_<Token>(
+ Token::OtherPunctuation,
+ construct_<std::string>(1u, arg1)) ]
;
- word
- = regex_p("[^ \t\n\r;,=(){}/\"]+")
+ word // Returns value in context.token
+ = lexeme_d
+ [
+ (+ word_p) [ str_ = construct_<std::string>(arg1, arg2) ]
+ ]
+ >> eps_p [ token_ = construct_<Token>(
+ Token::Word,
+ str_) ]
;
hexbyte
= uint_parser<char, 16, 2, 2>()
- [ append_a(self.context.str) ]
+ [ push_back(str_, arg1) ]
;
- BOOST_SPIRIT_DEBUG_RULE(command);
- BOOST_SPIRIT_DEBUG_RULE(path);
- BOOST_SPIRIT_DEBUG_RULE(argument);
- BOOST_SPIRIT_DEBUG_RULE(word);
- BOOST_SPIRIT_DEBUG_RULE(string);
- BOOST_SPIRIT_DEBUG_RULE(hexstring);
- BOOST_SPIRIT_DEBUG_RULE(token);
- BOOST_SPIRIT_DEBUG_RULE(punctuation);
- BOOST_SPIRIT_DEBUG_RULE(hexbyte);
- BOOST_SPIRIT_DEBUG_RULE(balanced_tokens);
- BOOST_SPIRIT_DEBUG_RULE(simple_argument);
- BOOST_SPIRIT_DEBUG_RULE(complex_argument);
- }
-
- boost::spirit::rule<Scanner> const & start() const { return command; }
- };
- };
-
- struct SkipGrammar
- : public boost::spirit::grammar<SkipGrammar>
- {
- template <class Scanner>
- struct definition
- {
- boost::spirit::rule<Scanner> rule;
+ statement_end
+ = if_p(var(self.incremental)) [
+ ch_p(';')
+ ]
+ .else_p [
+ ch_p(';')
+ | end_p
+ ]
+ ;
- definition(SkipGrammar const & self) {
- rule
- = boost::spirit::regex_p("[ \t]+")
- | boost::spirit::comment_p('#')
+ skip
+ = space_p | comment_p('#')
;
- }
- boost::spirit::rule<Scanner> const & start() const { return rule; }
+ ///////////////////////////////////////////////////////////////////
+
+ start_parsers(
+ command, // CommandParser
+ skip, // SkipParser
+ arguments // ArgumentsParser
+ );
+
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(command,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(path,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(argument,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(word,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(string,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(hexstring,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(token,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(punctuation,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(hexbyte,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(balanced_tokens,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(simple_argument,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(complex_argument,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(builtin,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(commands,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(block,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(statement,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(relpath,1);
+ BOOST_SPIRIT_DEBUG_TRACE_RULE(abspath,1);
+ }
};
};
+#endif
+
}}}
///////////////////////////////ih.e////////////////////////////////////////