#define IH_Parse_ 1
// Custom includes
+#include <vector>
#include <boost/regex.hpp>
#include <boost/spirit.hpp>
-#include <boost/spirit/utility/regex.hpp>
+#include <boost/spirit/utility/grammar_def.hpp>
#include <boost/spirit/actor.hpp>
#include <boost/bind.hpp>
#include <boost/function.hpp>
struct CommandGrammar : boost::spirit::grammar<CommandGrammar<ParseDispatcher> >
{
///////////////////////////////////////////////////////////////////////////
+ // Start rules
+
+ enum { CommandParser, SkipParser };
+
+ ///////////////////////////////////////////////////////////////////////////
// The parse context (variables needed while parsing)
struct Context {
std::string str;
+ std::vector<std::string> path;
char ch;
};
: context(c), dispatcher(d) {}
template <class Scanner>
- struct definition
+ struct definition
+ : public boost::spirit::grammar_def< boost::spirit::rule<Scanner>,
+ boost::spirit::rule<Scanner> >
{
- 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;
+ boost::spirit::rule<Scanner> command, path, argument, word, string, hexstring, token,
+ punctuation, hexbyte, balanced_tokens, simple_argument, complex_argument, builtin,
+ skip, commands, block, statement, relpath, abspath;
+ boost::spirit::chset<> special_p, punctuation_p, space_p, invalid_p, word_p;
+ boost::spirit::distinct_parser<> keyword_p;
+
+ definition(CommandGrammar const & self) :
+
+ // Characters with a special meaning within the parser
+ special_p ("/(){};"),
+
+ // Characters which are returned as punctuation tokens
+ punctuation_p (",="),
+
+ // 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 ),
+
+ // Valid word characters
+ word_p (
+ boost::spirit::anychar_p - special_p - punctuation_p - space_p - invalid_p),
+
+ // 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;
typedef ParseDispatcher PD;
-
+
+ ///////////////////////////////////////////////////////////////////
+ // Spirit grammar
+ //
+ // Syntax summary:
+ // This is EBNF with some minor tweaks to accommodate C++ syntax
+ //
+ // * and + precede their argument
+ // >> is followed by
+ // ! optional
+ // a % b match any number of a's separated by b
+ // a - b match 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 auxiliary
+ // in most cases.
+ //
+ // More info is in the Boost.Spirit documentation
+
+ commands
+ = * command
+ ;
+
command
- = path [ self.dispatch(&PD::beginCommand,
- boost::cref(self.context.str)) ]
+ = builtin >> (ch_p(';') | end_p)
+ | path >> ( block | statement )
+ | ch_p(';') // Ignore empty commands
+ ;
+
+ builtin
+ = keyword_p("cd")
+ >> path
+ >> eps_p [ self.dispatch(&PD::builtin_cd,
+ boost::ref(self.context.path)) ]
+ | keyword_p("ls")
+ >> ! path
+ >> eps_p [ self.dispatch(&PD::builtin_ls,
+ boost::ref(self.context.path)) ]
+ | keyword_p("exit") [ self.dispatch(&PD::builtin_exit) ]
+ ;
+
+ block
+ = ch_p('{') [ self.dispatch(&PD::pushDirectory,
+ boost::ref(self.context.path)) ]
+ >> * command
+ >> ch_p('}') [ self.dispatch(&PD::popDirectory) ]
+ ;
+
+ statement
+ = eps_p [ self.dispatch(&PD::beginCommand,
+ boost::ref(self.context.path)) ]
>> * argument
- >> ! ch_p(';')
+ >> (ch_p(';') | end_p)
>> eps_p [ self.dispatch(&PD::endCommand) ]
;
argument
= simple_argument [ self.dispatch(&PD::pushArgument,
- boost::cref(self.context.str)) ]
+ boost::ref(self.context.str)) ]
| complex_argument
;
simple_argument // All these return their value in context.str
= string
| hexstring
- | path
+ | word
;
complex_argument // Argument consists of multiple tokens
= ch_p('(') [ self.dispatch(&PD::openGroup) ]
- >> * token
- >> ch_p(')') [ self.dispatch(&PD::closeGroup) ]
+ >> * token
+ >> ch_p(')') [ self.dispatch(&PD::closeGroup) ]
;
string // Returns value in context.str
>> confix_p( "x\"", * hexbyte, '"' )
;
- 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_a(self.context.path) ]
+ >> relpath | abspath
+ ;
+
+ relpath
+ = ( word [ push_back_a(self.context.path) ]
+ % ch_p('/') )
+ >> ( ! ch_p('/') [ push_back_a(self.context.path,"") ] )
+ ;
+
+ abspath
+ = ch_p('/') [ push_back_a(self.context.path, "") ]
+ >> ! ( ( word [ push_back_a(self.context.path) ]
+ % ch_p('/') )
+ >> ( ! ch_p('/') [ push_back_a(self.context.path,"") ] ) )
;
balanced_tokens
token
= simple_argument [ self.dispatch(&PD::pushWord,
- boost::cref(self.context.str)) ]
+ boost::ref(self.context.str)) ]
| punctuation [ self.dispatch(&PD::pushPunctuation,
- boost::cref(self.context.str)) ]
+ boost::ref(self.context.str)) ]
| balanced_tokens
;
punctuation // Returns value in context.str
- = regex_p("[,=]") [ assign_a(self.context.str) ]
+ = punctuation_p [ assign_a(self.context.str) ]
;
- word
- = regex_p("[^ \t\n\r;,=(){}/\"]+")
+ word // Returns value in context.str
+ = lexeme_d[ + word_p ] [ assign_a(self.context.str) ]
;
hexbyte
[ append_a(self.context.str) ]
;
- 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;
-
- 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(
+ commands, // CommandParser
+ skip // SkipParser
+ );
+
+ 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);
+ }
};
};
projects / senf.git / blobdiff