Resolved. I have to apologise - I just worked out that I needed to use anychar - blank_p because blank_p was matching elsewhere in the grammar. -John On 1/24/06, John Ky <newhoggy@gmail.com> wrote:

Hi Hartmut,

Thanks for the help.

I got it to work with an extra ::context_t in:

typedef rule<strlit_closure::context_t> rule_type;

Is that the correct way to do it?

Also, I have another problem:

The rule:

rule_string mystringrule; mystringrule = * ( confix_p('"', (*c_escape_ch_p)[mystringrule.val += construct_<string>(arg1, arg2)], "') );

Will match:

"hello""world"

as:

helloworld

But the rule:

rule_string mystringrule; mystringrule = * ( confix_p('"', (*c_escape_ch_p)[mystringrule.val += construct_<string>(arg1, arg2)], "') | (anychar_p - '"') /* this line added */ );

matched returning an empty string.

-John

On 1/24/06, Hartmut Kaiser <hartmut.kaiser@gmail.com> wrote:

...

John Ky wrote:

...

Is it possible to define a rule:

rule<> mystringrule = confix_p('"', *c_escape_ch_p, '"')

differently such that I can use

std::string value;

bool result = parse("\"hello world\"", mystringrule[assign_a(value)]).full;

and get the text = "hello world" assigned to value?

Currently, I get text = "\"hello world\"" instead. ie with double quotes at either end.

I'ld suggest to use closures for that (not tested):

struct strlit_closure : public boost::spirit::closure<strlit_closure, std::string> { member1 val; };

typedef rule<strlit_closure> rule_type;

rule_type mystringrule = confix_p('"', *c_escape_ch_p[assign_a(mystringrule.val)], '"');

bool result = parse("\"hello world\"", mystringrule[assign_a(value)]).full;

This is based on the fact, that the first closure member of a rule is used as the rule's value.

HTH Regards Hartmut