From glenn.ehrlich2 at boeing.com  Wed Oct 29 12:33:22 2014
From: glenn.ehrlich2 at boeing.com (EXT-Ehrlich, Glenn)
Date: Wed Oct 29 13:14:36 2014
Subject: [studxml-users] libstudxml: attributes with spaces are not fully
 extracted with parser::attribute<T> when T is std::string
Message-ID: <C2E6AB55108BE5409FC8C5D8623CC59C3D593D2C@XCH-PHX-205.sw.nos.boeing.com>

Hi,

First off, I'd like to say how great libstudxml is.  It's made our xml parsing quite a bit better than using the alternatives.  Anyway....

There's a minor "defect" with the way std::string is handled with parser::attribute<T> or parser::element<T> when T is std::string.  Since the default implementation of these uses operator>> to pull the value out of the text, if the text has multiple spaces in it, the operator>> will stop at the first space.  Thus when parsing something like <foo bar="some text"/>, if bar is going into a std::string variable, it's text will be "some" instead of "some text".

I've fixed this in our version of the code by making this mod to value-traits:

  template <>
  struct LIBSTUDXML_EXPORT default_value_traits<std::string>
  {
    static std::string
    parse (std::string s, const parser&)
    {
      return s;
    }
  };

I don't implement serialize() because we don't use it and I'd like to think that the default implementation of serialize() would work, so this may not be a complete fix.

By the way, I work for Jeff Garland and he said to mention him, as you guys have met several times.  He's the one who suggested using libstudxml.

Glenn Ehrlich
glenn.ehrlich2@boeing.com<mailto:glenn.ehrlich2@boeing.com>



From boris at codesynthesis.com  Wed Oct 29 13:52:07 2014
From: boris at codesynthesis.com (Boris Kolpackov)
Date: Wed Oct 29 13:59:18 2014
Subject: [studxml-users] Re: libstudxml: attributes with spaces are not
 fully extracted with parser::attribute<T> when T is std::string
In-Reply-To: <C2E6AB55108BE5409FC8C5D8623CC59C3D593D2C@XCH-PHX-205.sw.nos.boeing.com>
References: <C2E6AB55108BE5409FC8C5D8623CC59C3D593D2C@XCH-PHX-205.sw.nos.boeing.com>
Message-ID: <boris.20141029194646@codesynthesis.com>

Hi Glenn,

EXT-Ehrlich, Glenn <glenn.ehrlich2@boeing.com> writes:

> First off, I'd like to say how great libstudxml is. It's made our xml
> parsing quite a bit better than using the alternatives.

Glad to hear that, thanks!


> There's a minor "defect" with the way std::string is handled with
> parser::attribute<T> or parser::element<T> when T is std::string.
> Since the default implementation of these uses operator>> to pull
> the value out of the text, if the text has multiple spaces in it,
> the operator>> will stop at the first space.  Thus when parsing
> something like <foo bar="some text"/>, if bar is going into a
> std::string variable, it's text will be "some" instead of "some text".

Yes, you are right. Thanks for the bug report!


> I've fixed this in our version of the code by making this mod to
> value-traits:
> 
>   template <>
>   struct LIBSTUDXML_EXPORT default_value_traits<std::string>
>   {
>     static std::string
>     parse (std::string s, const parser&)
>     {
>       return s;
>     }
>   };
> 
> I don't implement serialize() because we don't use it and I'd like
> to think that the default implementation of serialize() would work,
> so this may not be a complete fix.

I think serialize() that simply returns the passed string will be
at least significantly faster. I've added the fix that specializes
both parse() and serialize(), as well as the test:

http://scm.codesynthesis.com/?p=libstudxml/libstudxml.git;a=commit;h=7275dc4d919b0cb6f8713d0b4a9a2b507051cded


> By the way, I work for Jeff Garland and he said to mention him, as you
> guys have met several times.  He's the one who suggested using libstudxml.

Yes, Jeff was at the C++Now 2014 talk where I introduced libstudxml.
Please pass my regards to him when you have a chance.

Thanks,
Boris