Re: [Boost-users] spirit::, token_node_d, xml-tree

26 Jan 2006

      My original problem is the following code:
I'd like to parse a C++ function definition. I'm using file_iterator
to read a c++ file.
The parser that parses the entire c++ code is called default_parser.
The parser that matches only C++ method is called method_parser. I
know it needs some tuning but I use that problem to improve my
knowledge of spirit.

"Supress_code" is parser that every time precedes C++ function
definition(;, }, #define, */, //).
 "residue_parser" is everything but C++ function definition or supress_code.

e.g.
I'd like to parse:
template  void function(ddd::ssss<ddd> ss, dfdf dfdf) {
}

::::::::::::::::part of the
grammar::::::::::::::::::::::::::::::::::::::::::::::::
           method_parser =
                   !template_parser
               >>
                   !(*space_p >> str_p("inline") >> *space_p)
               >>
                   argument_parser
               >>
                   identifier_name
               >>
                   ch_p('(')
               >>
                   !list_p(argument_parser, ch_p(','))
               >>
                   ch_p(')')
               >>
                   !member_initialization_p
               >>
                   !(*space_p >> (str_p("const")|str_p("throw()")))>> *space_p
               >>
                   cpp_code_block_p
               ;

           residue_parser = *(anychar_p - (supress_code >> method_parser));

           default_parser =
                   token_node_d[residue_parser]
               %
                   (token_node_d[supress_code] >> token_node_d[method_parser])
               ;

I also deifned
        rule, parser_tag >
supress_code;
        rule, parser_tag
...
residue_parser;
        rule, parser_tag >
method_parser;
        rule, parser_tag
default_parser;
        rule, parser_tag > const&
        start() const { return default_parser; } //the default parser
in the grammar.

Plus invocation of the function xml-tree generation the tree. The
output is the following xml code:

<parsetree version="1.0">
    <parsenode rule="default_parser">
        <parsenode rule="residue_parser">
        </parsenode>
        <parsenode rule="supress_code">
        </parsenode>
        <parsenode rule="method_parser">
        </parsenode>
        <parsenode rule="residue_parser">
        </parsenode>
        <parsenode rule="supress_code">
        </parsenode>
        <parsenode rule="method_parser">
        </parsenode>
        ...........
        .........
        .....
        ...
        <parsenode rule="residue_parser">
        </parsenode>
        <parsenode rule="supress_code">
        </parsenode>
        <parsenode rule="method_parser">
        </parsenode>
        <parsenode rule="residue_parser">
        </parsenode>
    </parsenode>
</parsetree>
Parse succeeded!

With additional message proving that the file was parsed successfully.
The structure of the tree is correct, that's excatly what I want, but
I'm missing any string included between entities <value> and </value>.

I'm expecting something like this:

        <parsenode rule="method_parser">
            <value> "template  void function(ddd::ssss<ddd> ss,
dfdf dfdf) {
}"
            </value>
        </parsenode>

On 26/01/06, Hartmut Kaiser  wrote:
...
Elviin wrote:
...
I'd like to use non-ast tree. I'm also using the tree_to_xml
function to generate xml code.
Plus my own grammar. The problem is with the token_node_d
directive. I assumed that this directive  merge the parsed
code to an one node. At this time my parser works for me very
fine except the tree because there are missing value objects
in xml structure. So it seems that the whole tree is empty:/
I can reproduce that state with the example file parse_tree_calc1.cpp:
IIRC, the token_node_d or leaf_node_d cannot be used with a rule inside the
[]. But I'm not sure anymore about the rationale behind this. I'm CC'ing to
Dan Nuffer, he's the original author of this code, perhaps he has some
additional information.
Regards Hartmut
...
The correct output from the tree_to_xml function is following
when parsing "1":
<?xml version="1.0" encoding="ISO-8859-1"?> <!DOCTYPE
parsetree SYSTEM "parsetree.dtd">
<!-- 1 -->
<parsetree version="1.0">
    <parsenode rule="expression">
        <parsenode rule="term">
            <parsenode rule="factor">
                <parsenode rule="integer">
                    <parsenode rule="integer">
                        <value>1</value> <<<<======== it's OK here
                    </parsenode>
                </parsenode>
            </parsenode>
        </parsenode>
    </parsenode>
</parsetree>
parsing succeeded
... and the corresponfing code/rules:
integer     =   lexeme_d[ token_node_d[ (!ch_p('-') >>
+digit_p) ] ];
    factor      =   integer
                |   '(' >> expression >> ')'
                |   ('-' >> factor);
    term        =   factor >>
                    *(  ('*' >> factor)
                      | ('/' >> factor)
                    );
    expression  =   term >>
                    *(  ('+' >> term)
                      | ('-' >> term)
                    );
But if I move the token_node_d directive to another position,
so that is similar to the problem with my grammar:
integer     =   lexeme_d[  (!ch_p('-') >> +digit_p) ];
    factor      =   token_node_d[integer]
                |   '(' >> expression >> ')'
                |   ('-' >> factor);
    term        =   factor >>
                    *(  ('*' >> factor)
                      | ('/' >> factor)
                    );
    expression  =   term >>
                    *(  ('+' >> term)
                      | ('-' >> term)
                    );
... then the XML structure looks like this. The "value"
object are missing:
<?xml version="1.0" encoding="ISO-8859-1"?> <!DOCTYPE
parsetree SYSTEM "parsetree.dtd">
<!-- 1 -->
<parsetree version="1.0">
    <parsenode rule="expression">
        <parsenode rule="term">
            <parsenode rule="factor">
                <parsenode rule="integer">
<<<<======== it is not OK here
                </parsenode>
            </parsenode>
        </parsenode>
    </parsenode>
</parsetree>
parsing succeeded
Can anyone explain me why it is not possible to use  the
directive token_node_d like this token_node_d[integer]? I
just want to group all the parsed text (integer) in one node.
May I have to change the policy or to create proper typedefs.
The problem is with hierarchy, that I want to group the
parsed text only in the higher levels of parsing process.
Thank you.
Elviin
_______________________________________________
Boost-users mailing list
Boost-users@lists.boost.org
http://lists.boost.org/mailman/listinfo.cgi/boost-users
_______________________________________________
Boost-users mailing list
Boost-users@lists.boost.org
http://lists.boost.org/mailman/listinfo.cgi/boost-users

Re: [Boost-users] spirit::, token_node_d, xml-tree

elviin