Re: Parse succeeds, but doesn't propagate to first attribute with extra actions.
Henri Menke <[email protected]> Thu, 21 Nov 2019 11:32:35 +1300
| Newsgroups | gmane.comp.parsers.spirit.general |
|---|---|
| Message-ID | <[email protected]> |
Dear Dan,
Sorry for the confusion, I now see why you need the rule to be
'=' >> bool_ | attr(true)
However, I think your code is not as concise as it could be. I suggest
you keep using the `set' lambda in all the places and complement it with
a `set_or' lambda which either sets the member to the attribute of the
parser or a default value.
I also don't see why you reverted to `words[face]' instead of
`set(&font::face_name, words)' which I find much cleaner.
Probably you could also use the expectation operator `>' in more places
to minimize backtracking for faster parsing and earlier error reporting.
Cheers, Henri
Live example also on Wandbox:
https://wandbox.org/permlink/yRexnqNB8JzT5Ota
---
#include <boost/spirit/home/x3.hpp>
#include <iostream>
struct font {
std::string face_name;
size_t point = 0;
bool bold = false;
bool italic = false;
bool strike = false;
size_t id = 0;
};
namespace parsers {
// Thanks to the ingenious Seth!
// https://stackoverflow.com/a/49269239
auto const font_parser = boost::spirit::x3::rule<class _, font>{"font"} = [] {
using namespace boost::spirit::x3;
auto const set = [](auto member, auto p) {
auto propagate = [member](auto &ctx) {
traits::move_to(_attr(ctx), _val(ctx).*(member));
};
return as_parser(p)[propagate];
};
auto const set_or = [&set](auto member, auto p, auto def) {
return set(member, p) | set(member, attr(def));
};
auto const words = lexeme[+char_("a-zA-Z ")];
return lit("font(") > set(&font::face_name, words)
>> *(
';'
>> (
lit("size") >> set(&font::point, '=' > int_)
| lit("bold") >> set_or(&font::bold, '=' > bool_, true)
| lit("italic") >> set_or(&font::italic, '=' > bool_, true)
| lit("strike") >> set_or(&font::strike, '=' > bool_, true)
)
);
}();
} // namespace parsers
int main() {
font afont;
std::string target;
std::string test_str(
"font(Times New Roman;italic;size=12;bold=false;strike=true);circle");
auto begin = test_str.begin();
phrase_parse(begin, test_str.end(), parsers::font_parser,
boost::spirit::x3::space, afont);
std::cout << std::boolalpha << "face_name = " << afont.face_name << '\n'
<< "point = " << afont.point << '\n'
<< "bold = " << afont.bold << '\n'
<< "italic = " << afont.italic << '\n'
<< "strike = " << afont.strike << '\n';
return 0;
}
On 11/20/19 9:56 AM, Dan Bloomquist wrote:
> Henri Menke wrote:
> ...
> Hi Henri,
> Thank you very much for setting me on the way. I'm now much more
> comfortable with 'as_parser' and putting a parser in a lambda. Note that
> 'true_parser' just parses. And yes, Seth is brilliant, I've learned very
> much from his posts.
>
> //https://stackoverflow.com/a/49269239
> It seems I had already voted that one up in the past. I just missed the treasure there on this endeavor. I just don't use spirit often enough to keep it all in my head. At that, I'm 68 so don't know how much mind is left. :)
>
> The objective is that the user, as not a programmer, should get as much latitude as possible. Like in the case of fonts, the defaults are false and they can:
> bold;
> bold=true;
> bold=false; //this is redundant as is the case for no bold defined.
> size=12; //a value is required.
>
> And, of course, they can put them in any order they would like. They just must lead with a font face or 'Default' for the globally used font.
> This:
> https://github.com/lakeweb/stuff/blob/master/void_method2.cpp
>
> I've already plugged into my app and it is working nicely. This kind of parser is part of the support so a user has control over how their reports look without my hard-coding.
>
> Best, Dan.
>
> New code: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> #include <iostream>
> #include <boost/fusion/adapted/struct.hpp>
> #include <boost/spirit/home/x3.hpp>
>
> struct font {
> std::string face_name;
> size_t point = 0;
> bool bold = false;
> bool italic = false;
> bool strike = false;
> size_t id = 0;
> };
>
> namespace parsers {
> using namespace boost::spirit::x3;
>
> auto true_parser = [](auto at) {
> auto propagate = [at](auto ctx) {
> traits::move_to(_attr(ctx), _val(ctx).*(at)); };
> return (('=' >> bool_[propagate]) | attr(true)[propagate]);
> };
> auto set_val = [](auto member, auto p) {
> auto propagate = [member](auto& ctx) {
> traits::move_to(_attr(ctx), _val(ctx).*(member));
> };
> return as_parser('=' >> p)[propagate];
> };
> auto const words = lexeme[+char_("a-zA-Z ")];
> auto const face = [](auto& attr) { _val(attr).face_name = _attr(attr); };
>
> auto const font_parser = boost::spirit::x3::rule<class _, font>{ "font" } = [] {
> return lit("font(") >> words[face]
> >> *(
> ';'
> >> (
> lit("size") >> set_val(&font::point, int_)
> | lit("bold") >> true_parser(&font::bold)
> | lit("italic") >> true_parser(&font::italic)
> | lit("strike") >> true_parser(&font::strike)
> )
> );
> }();
> }
>
> int main() {
> font afont;
> std::string target;
> std::string test_str("font(Times New Roman;italic;size=12;bold=false;strike=true);circle");
> auto begin = test_str.begin();
> phrase_parse(begin, test_str.end(), parsers::font_parser, boost::spirit::x3::space, afont);
> std::cout << std::boolalpha
> << "face_name = " << afont.face_name << '\n'
> << "point = " << afont.point << '\n'
> << "bold = " << afont.bold << '\n'
> << "italic = " << afont.italic << '\n'
> << "strike = " << afont.strike << '\n';
> return 0;
> }
>
>
>
> _______________________________________________
> Spirit-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/spirit-general
>