I'd like to parse string | (string, int) and store it in a structure that defaults the int component to some value. The attribute of such a construction in X3 is a variant<string, tuple<string, int>>. I was thinking I could have a struct that takes either a string or a (string, int) to automagically be populated:
struct bar
{
bar (std::string x = "", int y = 0) : baz1 {x}, baz2 {y} {}
std::string baz1;
int baz2;
};
BOOST_FUSION_ADAPT_STRUCT (disj::ast::bar, baz1, baz2)
and then simply have:
const x3::rule<class bar, ast::bar> bar = "bar";
using x3::int_;
using x3::ascii::alnum;
auto const bar_def = (+(alnum) | ('(' >> +(alnum) >> ',' >> int_ >> ')')) >> ';';
BOOST_SPIRIT_DEFINE(bar);
However this does not work:
/usr/include/boost/spirit/home/x3/core/detail/parse_into_container.hpp:139:59: error: static assertion failed: Expecting a single element fusion sequence
139 | static_assert(traits::has_size<Attribute, 1>::value,
Setting baz2 to an optional does not help. One way to solve this is to have a variant field or inherit from that type:
struct string_int {
std::string s;
int i;
};
struct foo {
boost::variant<std::string, string_int> var;
};
BOOST_FUSION_ADAPT_STRUCT (disj::ast::string_int, s, i)
BOOST_FUSION_ADAPT_STRUCT (disj::ast::foo, var)
(For some reason, I have to use boost::variant instead of x3::variant for operator<< to work; also, using std::pair or tuple for string_int does not work, but boost::fusion::deque does.) One can then equip foo somehow to get the string and integer.
Question: What is the proper, clean way to do this in X3? Is there a more natural way than this second option and equipping foo with accessors?
Live On Coliru
Sadly the wording in the x3 section is exceedingly sparse and allows it (contrast the Qi section). A quick test confirms it:
Live On Coliru
#include <boost/spirit/home/x3.hpp>
namespace x3 = boost::spirit::x3;
template <typename Expr>
std::string inspect(Expr const& expr) {
using A = typename x3::traits::attribute_of<Expr, x3::unused_type>::type;
return boost::core::demangle(typeid(A).name());
}
int main()
{
std::cout << inspect(x3::double_ | x3::int_) << "\n"; // variant expected
std::cout << inspect(x3::int_ | "bla" >> x3::int_) << "\n"; // variant "understandable"
std::cout << inspect(x3::int_ | x3::int_) << "\n"; // variant suprising:
}
Prints
boost::variant<double, int>
boost::variant<int, int>
boost::variant<int, int>
All Hope Is Not Lost
In your specific case you could trick the system:
auto const bar_def = //
(+x3::alnum >> x3::attr(-1) //
| '(' >> +x3::alnum >> ',' >> x3::int_ >> ')' //
) >> ';';
Note how we "inject" an int value for the first branch. That satisfies the attribute propagation gods:
Live On Coliru
#include <boost/spirit/home/x3.hpp>
#include <boost/fusion/adapted/struct.hpp>
#include <boost/fusion/include/io.hpp>
#include <iomanip>
namespace x3 = boost::spirit::x3;
namespace disj::ast {
struct bar {
std::string x;
int y;
};
using boost::fusion::operator<<;
} // namespace disj::ast
BOOST_FUSION_ADAPT_STRUCT(disj::ast::bar, x, y)
namespace disj::parser {
const x3::rule<class bar, ast::bar> bar = "bar";
auto const bar_def = //
(+x3::alnum >> x3::attr(-1) //
| '(' >> +x3::alnum >> ',' >> x3::int_ >> ')' //
) >> ';';
BOOST_SPIRIT_DEFINE(bar)
}
namespace disj {
void run_tests() {
for (std::string const input : {
"",
";",
"bla;",
"bla, 42;",
"(bla, 42);",
}) {
ast::bar val;
auto f = begin(input), l = end(input);
std::cout << "\n" << quoted(input) << " -> ";
if (phrase_parse(f, l, parser::bar, x3::space, val)) {
std::cout << "Parsed: " << val << "\n";
} else {
std::cout << "Failed\n";
}
if (f!=l) {
std::cout << " -- Remaining " << quoted(std::string_view(f, l)) << "\n";
}
}
}
}
int main()
{
disj::run_tests();
}
Prints
"" -> Failed
";" -> Failed
-- Remaining ";"
"bla;" -> Parsed: (bla -1)
"bla, 42;" -> Failed
-- Remaining "bla, 42;"
"(bla, 42);" -> Parsed: (bla 42)
¹ just today
Related
I am trying to compose spirit rules but I cannot figure out what the attribute of this new rule would be.
The following code is working as I would expect it.
#include <iostream>
#include <boost/spirit/home/x3.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
#include <boost/fusion/include/io.hpp>
#include <boost/fusion/tuple.hpp>
namespace ast{
struct Record{
int id;
std::string name;
};
struct Document{
Record rec;
Record rec2;
//std::vector<Record> rec;
std::string name;
};
using boost::fusion::operator<<;
}
BOOST_FUSION_ADAPT_STRUCT(ast::Record,
name, id
)
BOOST_FUSION_ADAPT_STRUCT(ast::Document,
rec, rec2,
//rec,
name
)
namespace parser{
namespace x3 = boost::spirit::x3;
namespace ascii = boost::spirit::x3::ascii;
using x3::lit;
using x3::int_;
using ascii::char_;
const auto identifier = +char_("a-z");
const x3::rule<class record, ast::Record> record = "record";
const auto record_def = lit("record") >> identifier >> lit("{") >> int_ >> lit("}");
const x3::rule<class document, ast::Document> document = "document";
const auto document_def =
record >> record
//+record // This should generate a sequence
>> identifier
;
BOOST_SPIRIT_DEFINE(document, record);
}
namespace{
constexpr char g_input[] = R"input(
record foo{42}
record bar{73}
foobar
)input";
}
int main(){
using boost::spirit::x3::ascii::space;
std::string str = g_input;
ast::Document unit;
bool r = phrase_parse(str.begin(), str.end(), parser::document, space, unit);
std::cout << "Got: " << unit << "\n";
return 0;
}
But when I change the rule to parse multiple records(instead of exactly 2) I would expect it to have a std::vector<Record> as an attribute. But all I get is a long compiler error that does not help me very much.
Can someone point me to what I am doing wrong in order to compose the attributes correctly?
I think the whole reason it didn't compile is because you tried to print the result... and std::vector<Record> doesn't know how to be streamed:
namespace ast {
using boost::fusion::operator<<;
static inline std::ostream& operator<<(std::ostream& os, std::vector<Record> const& rs) {
os << "{ ";
for (auto& r : rs) os << r << " ";
return os << "}";
}
}
Some more notes:
adding lexemes where absolutely required (!)
simplifying (no need to BOOST_SPIRIT_DEFINE unless recursive rules/separate TUs)
dropping redundant lit
I arrived at
Live On Coliru
#include <iostream>
#include <boost/spirit/home/x3.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
#include <boost/fusion/include/io.hpp>
namespace ast {
struct Record{
int id;
std::string name;
};
struct Document{
std::vector<Record> rec;
std::string name;
};
}
BOOST_FUSION_ADAPT_STRUCT(ast::Record, name, id)
BOOST_FUSION_ADAPT_STRUCT(ast::Document, rec, name)
namespace ast {
using boost::fusion::operator<<;
static inline std::ostream& operator<<(std::ostream& os, std::vector<Record> const& rs) {
os << "{ ";
for (auto& r : rs) os << r << " ";
return os << "}";
}
}
namespace parser {
namespace x3 = boost::spirit::x3;
namespace ascii = x3::ascii;
const auto identifier = x3::lexeme[+x3::char_("a-z")];
const auto record = x3::rule<class record, ast::Record> {"record"}
= x3::lexeme["record"] >> identifier >> "{" >> x3::int_ >> "}";
const auto document = x3::rule<class document, ast::Document> {"document"}
= +record
>> identifier
;
}
int main(){
std::string const str = "record foo{42} record bar{73} foobar";
auto f = str.begin(), l = str.end();
ast::Document unit;
if (phrase_parse(f, l, parser::document, parser::ascii::space, unit)) {
std::cout << "Got: " << unit << "\n";
} else {
std::cout << "Parse failed\n";
}
if (f != l) {
std::cout << "Remaining unparsed input: '" << std::string(f,l) << "'\n";
}
}
Prints
Got: ({ (foo 42) (bar 73) } foobar)
I need a comma delimited output from a struct with optionals. For example, if I have this struct:
MyStruct
{
boost::optional<std::string> one;
boost::optional<int> two;
boost::optional<float> three;
};
An output like: { "string", 1, 3.0 } or { "string" } or { 1, 3.0 } and so on.
Now, I have code like this:
struct MyStruct
{
boost::optional<std::string> one;
boost::optional<int> two;
boost::optional<float> three;
};
BOOST_FUSION_ADAPT_STRUCT
(MyStruct,
one,
two,
three)
template<typename Iterator>
struct MyKarmaGrammar : boost::spirit::karma::grammar<Iterator, MyStruct()>
{
MyKarmaGrammar() : MyKarmaGrammar::base_type(request_)
{
using namespace std::literals::string_literals;
namespace karma = boost::spirit::karma;
using karma::int_;
using karma::double_;
using karma::string;
using karma::lit;
using karma::_r1;
key_ = '"' << string(_r1) << '"';
str_prop_ = key_(_r1) << ':'
<< string
;
int_prop_ = key_(_r1) << ':'
<< int_
;
dbl_prop_ = key_(_r1) << ':'
<< double_
;
//REQUEST
request_ = '{'
<< -str_prop_("one"s) <<
-int_prop_("two"s) <<
-dbl_prop_("three"s)
<< '}'
;
}
private:
//GENERAL RULES
boost::spirit::karma::rule<Iterator, void(std::string)> key_;
boost::spirit::karma::rule<Iterator, double(std::string)> dbl_prop_;
boost::spirit::karma::rule<Iterator, int(std::string)> int_prop_;
boost::spirit::karma::rule<Iterator, std::string(std::string)> str_prop_;
//REQUEST
boost::spirit::karma::rule<Iterator, MyStruct()> request_;
};
int main()
{
using namespace std::literals::string_literals;
MyStruct request = {std::string("one"), 2, 3.1};
std::string generated;
std::back_insert_iterator<std::string> sink(generated);
MyKarmaGrammar<std::back_insert_iterator<std::string>> serializer;
boost::spirit::karma::generate(sink, serializer, request);
std::cout << generated << std::endl;
}
This works but I need a comma delimited output. I tried with a grammar like:
request_ = '{'
<< (str_prop_("one"s) |
int_prop_("two"s) |
dbl_prop_("three"s)) % ','
<< '}'
;
But I receive this compile error:
/usr/include/boost/spirit/home/support/container.hpp:194:52: error: no type named ‘const_iterator’ in ‘struct MyStruct’
typedef typename Container::const_iterator type;
thanks!
Your struct is not a container, therefore list-operator% will not work. The documentation states it expects the attribute to be a container type.
So, just like in the Qi counterpart I showed you to create a conditional delim production:
delim = (&qi::lit('}')) | ',';
You'd need something similar here. However, everything about it is reversed. Instead of "detecting" the end of the input sequence from the presence of a {, we need to track the absense of preceding field from "not having output a field since opening brace yet".
That's a bit trickier since the required state cannot come from the same source as the input. We'll use a parser-member for simplicity here¹:
private:
bool _is_first_field;
Now, when we generate the opening brace, we want to initialize that to true:
auto _f = px::ref(_is_first_field); // short-hand
request_ %= lit('{') [ _f = true ]
Note: Use of %= instead of = tells Spirit that we want automatic attribute propagation to happen, in spite of the presence of a Semantic Action ([ _f = true ]).
Now, we need to generate the delimiter:
delim = eps(_f) | ", ";
Simple. Usage is also simple, except we'll want to conditionally reset the _f:
auto reset = boost::proto::deep_copy(eps [ _f = false ]);
str_prop_ %= (delim << key_(_r1) << string << reset) | "";
int_prop_ %= (delim << key_(_r1) << int_ << reset) | "";
dbl_prop_ %= (delim << key_(_r1) << double_ << reset) | "";
A very subtle point here is that I changed to the declared rule attribute types from T to optional<T>. This allows Karma to do the magic to fail the value generator if it's empty (boost::none), and skipping the reset!
ka::rule<Iterator, boost::optional<double>(std::string)> dbl_prop_;
ka::rule<Iterator, boost::optional<int>(std::string)> int_prop_;
ka::rule<Iterator, boost::optional<std::string>(std::string)> str_prop_;
Now, let's put together some testcases:
Test Cases
Live On Coliru
#include "iostream"
#include <boost/optional/optional_io.hpp>
#include <boost/fusion/include/io.hpp>
#include <boost/spirit/include/karma.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <string>
struct MyStruct {
boost::optional<std::string> one;
boost::optional<int> two;
boost::optional<double> three;
};
BOOST_FUSION_ADAPT_STRUCT(MyStruct, one, two, three)
namespace ka = boost::spirit::karma;
namespace px = boost::phoenix;
template<typename Iterator>
struct MyKarmaGrammar : ka::grammar<Iterator, MyStruct()> {
MyKarmaGrammar() : MyKarmaGrammar::base_type(request_) {
using namespace std::literals::string_literals;
using ka::int_;
using ka::double_;
using ka::string;
using ka::lit;
using ka::eps;
using ka::_r1;
auto _f = px::ref(_is_first_field);
auto reset = boost::proto::deep_copy(eps [ _f = false ]);
key_ = '"' << string(_r1) << "\":";
delim = eps(_f) | ", ";
str_prop_ %= (delim << key_(_r1) << string << reset) | "";
int_prop_ %= (delim << key_(_r1) << int_ << reset) | "";
dbl_prop_ %= (delim << key_(_r1) << double_ << reset) | "";
//REQUEST
request_ %= lit('{') [ _f = true ]
<< str_prop_("one"s) <<
int_prop_("two"s) <<
dbl_prop_("three"s)
<< '}';
}
private:
bool _is_first_field = true;
//GENERAL RULES
ka::rule<Iterator, void(std::string)> key_;
ka::rule<Iterator, boost::optional<double>(std::string)> dbl_prop_;
ka::rule<Iterator, boost::optional<int>(std::string)> int_prop_;
ka::rule<Iterator, boost::optional<std::string>(std::string)> str_prop_;
ka::rule<Iterator> delim;
//REQUEST
ka::rule<Iterator, MyStruct()> request_;
};
template <typename T> std::array<boost::optional<T>, 2> option(T const& v) {
return { { v, boost::none } };
}
int main() {
using namespace std::literals::string_literals;
for (auto a : option("one"s))
for (auto b : option(2))
for (auto c : option(3.1))
for (auto request : { MyStruct { a, b, c } }) {
std::string generated;
std::back_insert_iterator<std::string> sink(generated);
MyKarmaGrammar<std::back_insert_iterator<std::string>> serializer;
ka::generate(sink, serializer, request);
std::cout << boost::fusion::as_vector(request) << ":\t" << generated << "\n";
}
}
Printing:
( one 2 3.1): {"one":one, "two":2, "three":3.1}
( one 2 --): {"one":one, "two":2}
( one -- 3.1): {"one":one, "three":3.1}
( one -- --): {"one":one}
(-- 2 3.1): {"two":2, "three":3.1}
(-- 2 --): {"two":2}
(-- -- 3.1): {"three":3.1}
(-- -- --): {}
¹ Note this limits re-entrant use of the parser, as well as making it non-const etc. karma::locals are the true answer to that, adding a little more complexity
I am currentl adding expectation points to my grammar in X3.
Now I came accross an rule, which looks like this.
auto const id_string = +x3::char("A-Za-z0-9_);
auto const nested_identifier_def =
x3::lexeme[
*(id_string >> "::")
>> *(id_string >> ".")
>> id_string
];
I am wondering how I can add conditional expectation points to this rule.
Like "if there is a "::" then there musst follow an id_string" or "when there is a . then there musst follow an id_string"
and so on.
How can I achieve such a behaviour for such a rule?
I'd write it exactly the way you intend it:
auto const identifier
= lexeme [+char_("A-Za-z0-9_")];
auto const qualified_id
= identifier >> *("::" > identifier);
auto const simple_expression // only member expressions supported now
= qualified_id >> *('.' > identifier);
With a corresponding AST:
namespace AST {
using identifier = std::string;
struct qualified_id : std::vector<identifier> { using std::vector<identifier>::vector; };
struct simple_expression {
qualified_id lhs;
std::vector<identifier> rhs;
};
}
LIVE DEMO
Live On Coliru
#include <iostream>
#include <string>
#include <vector>
namespace AST {
using identifier = std::string;
struct qualified_id : std::vector<identifier> { using std::vector<identifier>::vector; };
struct simple_expression {
qualified_id lhs;
std::vector<identifier> rhs;
};
}
#include <boost/fusion/adapted.hpp>
BOOST_FUSION_ADAPT_STRUCT(AST::simple_expression, lhs, rhs)
#include <boost/spirit/home/x3.hpp>
namespace Parser {
using namespace boost::spirit::x3;
auto const identifier
= rule<struct identifier_, AST::identifier> {}
= lexeme [+char_("A-Za-z0-9_")];
auto const qualified_id
= rule<struct qualified_id_, AST::qualified_id> {}
= identifier >> *("::" > identifier);
auto const simple_expression // only member expressions supported now
= rule<struct simple_expression_, AST::simple_expression> {}
= qualified_id >> *('.' > identifier);
}
int main() {
using It = std::string::const_iterator;
for (std::string const input : { "foo", "foo::bar", "foo.member", "foo::bar.member.subobject" }) {
It f = input.begin(), l = input.end();
AST::simple_expression data;
bool ok = phrase_parse(f, l, Parser::simple_expression, boost::spirit::x3::space, data);
if (ok) {
std::cout << "Parse success: ";
for (auto& el : data.lhs) std::cout << "::" << el;
for (auto& el : data.rhs) std::cout << "." << el;
std::cout << "\n";
}
else {
std::cout << "Parse failure ('" << input << "')\n";
}
if (f != l)
std::cout << "Remaining unparsed input: '" << std::string(f, l) << "'\n";
}
}
Prints
Parse success: ::foo
Parse success: ::foo::bar
Parse success: ::foo.member
Parse success: ::foo::bar.member.subobject
Can you help me understand the difference between the a % b parser and its expanded a >> *(b >> a) form in Boost.Spirit? Even though the reference manual states that they are equivalent,
The list operator, a % b, is a binary operator that matches a list of one or more repetitions of a separated by occurrences of b. This is equivalent to a >> *(b >> a).
the following program produces different results depending on which is used:
#include <iostream>
#include <string>
#include <vector>
#include <boost/fusion/include/adapt_struct.hpp>
#include <boost/spirit/include/qi.hpp>
struct Record {
int id;
std::vector<int> values;
};
BOOST_FUSION_ADAPT_STRUCT(Record,
(int, id)
(std::vector<int>, values)
)
int main() {
namespace qi = boost::spirit::qi;
const auto str = std::string{"1: 2, 3, 4"};
const auto rule1 = qi::int_ >> ':' >> (qi::int_ % ',') >> qi::eoi;
const auto rule2 = qi::int_ >> ':' >> (qi::int_ >> *(',' >> qi::int_)) >> qi::eoi;
Record record1;
if (qi::phrase_parse(str.begin(), str.end(), rule1, qi::space, record1)) {
std::cout << record1.id << ": ";
for (const auto& value : record1.values) { std::cout << value << ", "; }
std::cout << '\n';
} else {
std::cerr << "syntax error\n";
}
Record record2;
if (qi::phrase_parse(str.begin(), str.end(), rule2, qi::space, record2)) {
std::cout << record2.id << ": ";
for (const auto& value : record2.values) { std::cout << value << ", "; }
std::cout << '\n';
} else {
std::cerr << "syntax error\n";
}
}
Live on Coliru
1: 2, 3, 4,
1: 2,
rule1 and rule2 are different only in that rule1 uses the list operator ((qi::int_ % ',')) and rule2 uses its expanded form ((qi::int_ >> *(',' >> qi::int_))). However, rule1 produced 1: 2, 3, 4, (as expected) and rule2 produced 1: 2,. I cannot understand the result of rule2: 1) why is it different from that of rule1 and 2) why were 3 and 4 not included in record2.values even though phrase_parse returned true somehow?
Update X3 version added
First off, you fallen into a deep trap here:
Qi rules don't work with auto. Use qi::copy or just used qi::rule<>. Your program has undefined behaviour and indeed it crashed for me (valgrind pointed out where the dangling references originated).
So, first off:
const auto rule = qi::copy(qi::int_ >> ':' >> (qi::int_ % ',') >> qi::eoi);
Now, when you delete the redundancy in the program, you get:
Reproducing the problem
Live On Coliru
int main() {
test(qi::copy(qi::int_ >> ':' >> (qi::int_ % ',')));
test(qi::copy(qi::int_ >> ':' >> (qi::int_ >> *(',' >> qi::int_))));
}
Printing
1: 2, 3, 4,
1: 2,
The cause and the fix
What happened to 3, 4 which was successfully parsed?
Well, the attribute propagation rules indicate that qi::int_ >> *(',' >> qi::int_) exposes a tuple<int, vector<int> >. In a bid to magically DoTheRightThing(TM) Spirit accidentally misfires and "assigngs" the int into the attribute reference, ignoring the remaining vector<int>.
If you want to make container attributes parse as "an atomic group", use qi::as<>:
test(qi::copy(qi::int_ >> ':' >> qi::as<Record::values_t>() [ qi::int_ >> *(',' >> qi::int_)]));
Here as<> acts as a barrier for the attribute compatibility heuristics and the grammar knows what you meant:
Live On Coliru
#include <iostream>
#include <string>
#include <vector>
#include <boost/fusion/include/adapt_struct.hpp>
#include <boost/spirit/include/qi.hpp>
struct Record {
int id;
using values_t = std::vector<int>;
values_t values;
};
BOOST_FUSION_ADAPT_STRUCT(Record, id, values)
namespace qi = boost::spirit::qi;
template <typename T>
void test(T const& rule) {
const std::string str = "1: 2, 3, 4";
Record record;
if (qi::phrase_parse(str.begin(), str.end(), rule >> qi::eoi, qi::space, record)) {
std::cout << record.id << ": ";
for (const auto& value : record.values) { std::cout << value << ", "; }
std::cout << '\n';
} else {
std::cerr << "syntax error\n";
}
}
int main() {
test(qi::copy(qi::int_ >> ':' >> (qi::int_ % ',')));
test(qi::copy(qi::int_ >> ':' >> (qi::int_ >> *(',' >> qi::int_))));
test(qi::copy(qi::int_ >> ':' >> qi::as<Record::values_t>() [ qi::int_ >> *(',' >> qi::int_)]));
}
Prints
1: 2, 3, 4,
1: 2,
1: 2, 3, 4,
Because it's time to get people started with X3 (the new version of Spirit), and because I like to challenge msyelf to do the corresponding tasks in Spirit X3, here is the Spirit X3 version.
There's no problem with auto in X3.
The "broken" case also behaves much better, triggering this static assertion:
// If you got an error here, then you are trying to pass
// a fusion sequence with the wrong number of elements
// as that expected by the (sequence) parser.
static_assert(
fusion::result_of::size<Attribute>::value == (l_size + r_size)
, "Attribute does not have the expected size."
);
That's nice, right?
The workaround seems a bit less readable:
test(int_ >> ':' >> (rule<struct _, Record::values_t>{} = (int_ >> *(',' >> int_))));
But it would be trivial to write your own as<> "directive" (or just a function), if you wanted:
namespace {
template <typename T>
struct as_type {
template <typename Expr>
auto operator[](Expr&& expr) const {
return x3::rule<struct _, T>{"as"} = x3::as_parser(std::forward<Expr>(expr));
}
};
template <typename T> static const as_type<T> as = {};
}
DEMO
Live On Coliru
#include <iostream>
#include <string>
#include <vector>
#include <boost/fusion/adapted/std_tuple.hpp>
#include <boost/spirit/home/x3.hpp>
struct Record {
int id;
using values_t = std::vector<int>;
values_t values;
};
namespace x3 = boost::spirit::x3;
template <typename T>
void test(T const& rule) {
const std::string str = "1: 2, 3, 4";
Record record;
auto attr = std::tie(record.id, record.values);
if (x3::phrase_parse(str.begin(), str.end(), rule >> x3::eoi, x3::space, attr)) {
std::cout << record.id << ": ";
for (const auto& value : record.values) { std::cout << value << ", "; }
std::cout << '\n';
} else {
std::cerr << "syntax error\n";
}
}
namespace {
template <typename T>
struct as_type {
template <typename Expr>
auto operator[](Expr&& expr) const {
return x3::rule<struct _, T>{"as"} = x3::as_parser(std::forward<Expr>(expr));
}
};
template <typename T> static const as_type<T> as = {};
}
int main() {
using namespace x3;
test(int_ >> ':' >> (int_ % ','));
//test(int_ >> ':' >> (int_ >> *(',' >> int_))); // COMPILER asserts "Attribute does not have the expected size."
// "clumsy" x3 style workaround
test(int_ >> ':' >> (rule<struct _, Record::values_t>{} = (int_ >> *(',' >> int_))));
// using an ad-hoc `as<>` implementation:
test(int_ >> ':' >> as<Record::values_t>[int_ >> *(',' >> int_)]);
}
Prints
1: 2, 3, 4,
1: 2, 3, 4,
1: 2, 3, 4,
I want to use boost spirit to parse an expression like
function1(arg1, arg2, function2(arg1, arg2, arg3),
function3(arg1,arg2))
and call corresponding c++ functions. What should be the grammar to parse above expression and call the corresponding c++ function by phoneix::bind()?
I have 2 types of functions to call
1) string functions;
wstring GetSubString(wstring stringToCut, int position, int length);
wstring GetStringToken(wstring stringToTokenize, wstring seperators,
int tokenNumber );
2) Functions that return integer;
int GetCount();
int GetId(wstring srcId, wstring srcType);
Second Answer (more pragmatic)
Here's a second take, for comparison:
Just in case you really didn't want to parse into an abstract syntax tree representation, but rather evaluate the functions on-the-fly during parsing, you can simplify the grammar.
It comes in at 92 lines as opposed to 209 lines in the first answer. It really depends on what you're implementing which approach is more suitable.
This shorter approach has some downsides:
less flexible (not reusable)
less robust (if functions have side effects, they will happen even if parsing fails halfway)
less extensible (the supported functions are hardwired into the grammar1)
Full code:
//#define BOOST_SPIRIT_DEBUG
#define BOOST_SPIRIT_USE_PHOENIX_V3
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/phoenix/function.hpp>
namespace qi = boost::spirit::qi;
namespace phx = boost::phoenix;
typedef boost::variant<int, std::string> value;
//////////////////////////////////////////////////
// Demo functions:
value AnswerToLTUAE() {
return 42;
}
value ReverseString(value const& input) {
auto& as_string = boost::get<std::string>(input);
return std::string(as_string.rbegin(), as_string.rend());
}
value Concatenate(value const& a, value const& b) {
std::ostringstream oss;
oss << a << b;
return oss.str();
}
BOOST_PHOENIX_ADAPT_FUNCTION_NULLARY(value, AnswerToLTUAE_, AnswerToLTUAE)
BOOST_PHOENIX_ADAPT_FUNCTION(value, ReverseString_, ReverseString, 1)
BOOST_PHOENIX_ADAPT_FUNCTION(value, Concatenate_, Concatenate, 2)
//////////////////////////////////////////////////
// Parser grammar
template <typename It, typename Skipper = qi::space_type>
struct parser : qi::grammar<It, value(), Skipper>
{
parser() : parser::base_type(expr_)
{
using namespace qi;
function_call_ =
(lit("AnswerToLTUAE") > '(' > ')')
[ _val = AnswerToLTUAE_() ]
| (lit("ReverseString") > '(' > expr_ > ')')
[ _val = ReverseString_(_1) ]
| (lit("Concatenate") > '(' > expr_ > ',' > expr_ > ')')
[ _val = Concatenate_(_1, _2) ]
;
string_ = as_string [
lexeme [ "'" >> *~char_("'") >> "'" ]
];
value_ = int_ | string_;
expr_ = function_call_ | value_;
on_error<fail> ( expr_, std::cout
<< phx::val("Error! Expecting ") << _4 << phx::val(" here: \"")
<< phx::construct<std::string>(_3, _2) << phx::val("\"\n"));
BOOST_SPIRIT_DEBUG_NODES((expr_)(function_call_)(value_)(string_))
}
private:
qi::rule<It, value(), Skipper> value_, function_call_, expr_, string_;
};
int main()
{
for (const std::string input: std::vector<std::string> {
"-99",
"'string'",
"AnswerToLTUAE()",
"ReverseString('string')",
"Concatenate('string', 987)",
"Concatenate('The Answer Is ', AnswerToLTUAE())",
})
{
auto f(std::begin(input)), l(std::end(input));
const static parser<decltype(f)> p;
value direct_eval;
bool ok = qi::phrase_parse(f,l,p,qi::space,direct_eval);
if (!ok)
std::cout << "invalid input\n";
else
{
std::cout << "input:\t" << input << "\n";
std::cout << "eval:\t" << direct_eval << "\n\n";
}
if (f!=l) std::cout << "unparsed: '" << std::string(f,l) << "'\n";
}
}
Note how, instead of using BOOST_PHOENIX_ADAPT_FUNCTION* we could have directly used boost::phoenix::bind.
The output is still the same:
input: -99
eval: -99
input: 'string'
eval: string
input: AnswerToLTUAE()
eval: 42
input: ReverseString('string')
eval: gnirts
input: Concatenate('string', 987)
eval: string987
input: Concatenate('The Answer Is ', AnswerToLTUAE())
eval: The Answer Is 42
1 This last downside is easily remedied by using the 'Nabialek Trick'
First Answer (complete)
I've gone and implemented a simple recursive expression grammar for functions having up-to-three parameters:
for (const std::string input: std::vector<std::string> {
"-99",
"'string'",
"AnswerToLTUAE()",
"ReverseString('string')",
"Concatenate('string', 987)",
"Concatenate('The Answer Is ', AnswerToLTUAE())",
})
{
auto f(std::begin(input)), l(std::end(input));
const static parser<decltype(f)> p;
expr parsed_script;
bool ok = qi::phrase_parse(f,l,p,qi::space,parsed_script);
if (!ok)
std::cout << "invalid input\n";
else
{
const static generator<boost::spirit::ostream_iterator> g;
std::cout << "input:\t" << input << "\n";
std::cout << "tree:\t" << karma::format(g, parsed_script) << "\n";
std::cout << "eval:\t" << evaluate(parsed_script) << "\n";
}
if (f!=l) std::cout << "unparsed: '" << std::string(f,l) << "'\n";
}
Which prints:
input: -99
tree: -99
eval: -99
input: 'string'
tree: 'string'
eval: string
input: AnswerToLTUAE()
tree: nullary_function_call()
eval: 42
input: ReverseString('string')
tree: unary_function_call('string')
eval: gnirts
input: Concatenate('string', 987)
tree: binary_function_call('string',987)
eval: string987
input: Concatenate('The Answer Is ', AnswerToLTUAE())
tree: binary_function_call('The Answer Is ',nullary_function_call())
eval: The Answer Is 42
Some notes:
I separated parsing from execution (which is always a good idea IMO)
I implemented function evaluation for zero, one or two parameters (this should be easy to extend)
Values are assumed to be integers or strings (should be easy to extend)
I added a karma generator to display the parsed expression (with a TODO marked in the comment)
I hope this helps:
//#define BOOST_SPIRIT_DEBUG
#include <boost/fusion/adapted/struct.hpp>
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/spirit/include/karma.hpp>
#include <boost/variant/recursive_wrapper.hpp>
namespace qi = boost::spirit::qi;
namespace karma = boost::spirit::karma;
namespace phx = boost::phoenix;
typedef boost::variant<int, std::string> value;
typedef boost::variant<value, boost::recursive_wrapper<struct function_call> > expr;
typedef std::function<value() > nullary_function_impl;
typedef std::function<value(value const&) > unary_function_impl;
typedef std::function<value(value const&, value const&)> binary_function_impl;
typedef boost::variant<nullary_function_impl, unary_function_impl, binary_function_impl> function_impl;
typedef qi::symbols<char, function_impl> function_table;
struct function_call
{
typedef std::vector<expr> arguments_t;
function_call() = default;
function_call(function_impl f, arguments_t const& arguments)
: f(f), arguments(arguments) { }
function_impl f;
arguments_t arguments;
};
BOOST_FUSION_ADAPT_STRUCT(function_call, (function_impl, f)(function_call::arguments_t, arguments))
#ifdef BOOST_SPIRIT_DEBUG
namespace std
{
static inline std::ostream& operator<<(std::ostream& os, nullary_function_impl const& f) { return os << "<nullary_function_impl>"; }
static inline std::ostream& operator<<(std::ostream& os, unary_function_impl const& f) { return os << "<unary_function_impl>"; }
static inline std::ostream& operator<<(std::ostream& os, binary_function_impl const& f) { return os << "<binary_function_impl>"; }
}
static inline std::ostream& operator<<(std::ostream& os, function_call const& call) { return os << call.f << "(" << call.arguments.size() << ")"; }
#endif
//////////////////////////////////////////////////
// Evaluation
value evaluate(const expr& e);
struct eval : boost::static_visitor<value>
{
eval() {}
value operator()(const value& v) const
{
return v;
}
value operator()(const function_call& call) const
{
return boost::apply_visitor(invoke(call.arguments), call.f);
}
private:
struct invoke : boost::static_visitor<value>
{
function_call::arguments_t const& _args;
invoke(function_call::arguments_t const& args) : _args(args) {}
value operator()(nullary_function_impl const& f) const {
return f();
}
value operator()(unary_function_impl const& f) const {
auto a = evaluate(_args.at(0));
return f(a);
}
value operator()(binary_function_impl const& f) const {
auto a = evaluate(_args.at(0));
auto b = evaluate(_args.at(1));
return f(a, b);
}
};
};
value evaluate(const expr& e)
{
return boost::apply_visitor(eval(), e);
}
//////////////////////////////////////////////////
// Demo functions:
value AnswerToLTUAE() {
return 42;
}
value ReverseString(value const& input) {
auto& as_string = boost::get<std::string>(input);
return std::string(as_string.rbegin(), as_string.rend());
}
value Concatenate(value const& a, value const& b) {
std::ostringstream oss;
oss << a << b;
return oss.str();
}
//////////////////////////////////////////////////
// Parser grammar
template <typename It, typename Skipper = qi::space_type>
struct parser : qi::grammar<It, expr(), Skipper>
{
parser() : parser::base_type(expr_)
{
using namespace qi;
n_ary_ops.add
("AnswerToLTUAE", nullary_function_impl{ &::AnswerToLTUAE })
("ReverseString", unary_function_impl { &::ReverseString })
("Concatenate" , binary_function_impl { &::Concatenate });
function_call_ = n_ary_ops > '(' > expr_list > ')';
string_ = qi::lexeme [ "'" >> *~qi::char_("'") >> "'" ];
value_ = qi::int_ | string_;
expr_list = -expr_ % ',';
expr_ = function_call_ | value_;
on_error<fail> ( expr_, std::cout
<< phx::val("Error! Expecting ") << _4 << phx::val(" here: \"")
<< phx::construct<std::string>(_3, _2) << phx::val("\"\n"));
BOOST_SPIRIT_DEBUG_NODES((expr_)(expr_list)(function_call_)(value_)(string_))
}
private:
function_table n_ary_ops;
template <typename Attr> using Rule = qi::rule<It, Attr(), Skipper>;
Rule<std::string> string_;
Rule<value> value_;
Rule<function_call> function_call_;
Rule<std::vector<expr>> expr_list;
Rule<expr> expr_;
};
//////////////////////////////////////////////////
// Output generator
template <typename It>
struct generator : karma::grammar<It, expr()>
{
generator() : generator::base_type(expr_)
{
using namespace karma;
nullary_ = eps << "nullary_function_call"; // TODO reverse lookup :)
unary_ = eps << "unary_function_call";
binary_ = eps << "binary_function_call";
function_ = nullary_ | unary_ | binary_;
function_call_ = function_ << expr_list;
expr_list = '(' << -(expr_ % ',') << ')';
value_ = karma::int_ | ("'" << karma::string << "'");
expr_ = function_call_ | value_;
}
private:
template <typename Attr> using Rule = karma::rule<It, Attr()>;
Rule<nullary_function_impl> nullary_;
Rule<unary_function_impl> unary_;
Rule<binary_function_impl> binary_;
Rule<function_impl> function_;
Rule<function_call> function_call_;
Rule<value> value_;
Rule<std::vector<expr>> expr_list;
Rule<expr> expr_;
};
int main()
{
for (const std::string input: std::vector<std::string> {
"-99",
"'string'",
"AnswerToLTUAE()",
"ReverseString('string')",
"Concatenate('string', 987)",
"Concatenate('The Answer Is ', AnswerToLTUAE())",
})
{
auto f(std::begin(input)), l(std::end(input));
const static parser<decltype(f)> p;
expr parsed_script;
bool ok = qi::phrase_parse(f,l,p,qi::space,parsed_script);
if (!ok)
std::cout << "invalid input\n";
else
{
const static generator<boost::spirit::ostream_iterator> g;
std::cout << "input:\t" << input << "\n";
std::cout << "tree:\t" << karma::format(g, parsed_script) << "\n";
std::cout << "eval:\t" << evaluate(parsed_script) << "\n\n";
}
if (f!=l) std::cout << "unparsed: '" << std::string(f,l) << "'\n";
}
}