The Spirit Parser Framework is an object oriented recursive descent parser generator implemented using template metaprogramming in C++. It leverages expression templates to approximate the syntax of extended Backus–Naur form (EBNF), enabling users to compose parser objects via operator overloading. The resulting backtracking LL(∞) parser can handle ambiguous grammars, and Spirit supports both lexing and parsing. This framework is included in the Boost libraries, making it a powerful tool for modern C++ parsing needs.
Operators
Because of limitations of the C++ language, the syntax of Spirit has been designed around the operator precedences of C++, while bearing resemblance to both EBNF and regular expressions.
| syntax | explanation |
|---|---|
| x >> y | Match x followed by y. |
| x > y | After matching x, expect y. |
| *x | Match x repeated zero or more times. This represents the Kleene star; C++ lacks an unary postfix operator *. |
| x | y | Match x. If x does not match, try to match y. |
| +x | Match a series of one or more occurrences of x. |
| -x | Match x zero or one time. |
| x & y | Match x and y. |
| x - y | Match x but not y. |
| x ^ y | Match x, or y, or both, in any order. |
| x || y | Match x, or y, or x followed by y. |
| x [ function_expression ] | Execute the function/functor returned by function_expression, if x matched. |
| ( x ) | Match x (can be used for priority grouping) |
| x % y | Match one or more occurrences of x, separated by occurrences of y. |
| ~x | Match anything but x (only with character classes such as ch_p or alnum_p) |
Example
This example shows how to use an inline parser expression with a semantic action.
#include <string> #include <iostream> #include <boost/spirit/include/qi.hpp> #include <boost/spirit/include/phoenix.hpp> int main() { namespace qi = boost::spirit::qi; std::string input; std::cout << "Input a line: \n"; getline(std::cin, input); std::cout << "Got '" << input << "'.\n"; unsigned count = 0; /* Next, parse the input (input.c_str()), using a parser constructed with the following semantics: Zero or more occurrences of ( literal string "cat" (when matched, increment the counter "count") or any character (which will be skipped) ) The parser is constructed by the compiler using operator overloading and template matching, so the actual work is done within qi::parse(), and the expression starting with * only initializes the rule object that the parse function uses. */ auto rule = *(qi::lit("cat") [ ++qi::_val ] | qi::omit[qi::char_]); qi::parse(input.begin(), input.end(), rule, count); // Finally, show results. std::cout << "The input contained " << count << " occurrences of 'cat'\n"; }External links
- Spirit parser framework github page
- Spirit parser framework
- Documentation in the Boost project
- Article on Spirit by designer Joel de Guzman in Dr. Dobb's Journal
References
Guzman, Joel de; Kaiser, Hartmut. "Spirit 2.59 - Introduction". Boost.org. Boost. Retrieved 22 March 2025. https://www.boost.org/doc/libs/1_87_0/libs/spirit/doc/html/spirit/introduction.html ↩