boost::xpressive : Get thousand separator as double

older
begin (cout) or boost::copy to cout

Jens Saathoff

8 Sep 2011 8 Sep '11

9:09 p.m.

Hi! I'm playing a bit with boost::xpressive and have no clue how to get a 100,000 (comma as separator) to a double as 1000000.00 or 250,000.10 as a double 250000.10 or 12,00 as double 12.00 or 1,200,500.00 as double 1200500.00. I'm working with xpressive since a day. What's the regex for that? And how can i use it in xpressive-grammar as a nested regex? Example subject: "I payed 100,200.50 Dollar" Example Regex on subject: 'I payed' >> (s1=getdoubleregex) >> ' Dollar' Is it possible to do it like that? Set return-value/type in a regex? Thank you very much!

Attachments:

attachment.html (text/html — 810 bytes)

Show replies by date

Eric Niebler

9 Sep 9 Sep

2:07 a.m.

New subject: boost::xpressive : Get thousand separator as double

On 9/8/2011 5:09 PM, Jens Saathoff wrote:

...

Hi!

I'm playing a bit with boost::xpressive and have no clue how to get a 100,000 (comma as separator) to a double as 1000000.00 or 250,000.10 as a double 250000.10 or 12,00 as double 12.00 or 1,200,500.00 as double 1200500.00. I'm working with xpressive since a day. What's the regex for that?

And how can i use it in xpressive-grammar as a nested regex?

Example subject: "I payed 100,200.50 Dollar"

Example Regex on subject: 'I payed' >> (s1=getdoubleregex) >> ' Dollar'

Is it possible to do it like that? Set return-value/type in a regex?

Try something like the following: #include <string> #include <iostream> #include <iomanip> #include #include namespace xp = boost::xpressive; int main() { using namespace xp; placeholder<double> result; sregex double_ = ~after(_d) >> // a number with no group separators, and either a period or comma // for fraction separator (+_d)[result = as<double>(_)] >> !((set=',','.') >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // A number with optional group separators | repeat<1,3>(_d)[result = as<int>(_)] >> ( // ... with comma group separators and period fraction separator *(',' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !('.' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // ... with period group separators and comma fraction separator | *('.' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !(',' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) ); std::string str; double d = 0.0; smatch what; what.let(result = d); std::cout << std::fixed << std::setprecision(2); while (std::getline(std::cin, str)) { if (regex_search(str, what, double_)) { std::cout << "Double = " << d << '\n'; } else { std::cout << "No match\n"; } } } -- Eric Niebler BoostPro Computing http://www.boostpro.com

Jens Saathoff

9:26 a.m.

New subject: boost::xpressive : Get thousand separator as double

Perfect! That works! I will compare it with my thousand-sep rule/grammar in boost::spirit, a little benchmark to find out which one is faster. Do have experiences/results with boost::spirit vs. boost::xpressive? Thank you very much. Have a nice weekend. 2011/9/9 Eric Niebler

...

#include <string> #include <iostream> #include <iomanip> #include #include namespace xp = boost::xpressive;

int main() { using namespace xp; placeholder<double> result; sregex double_ = ~after(_d) >> // a number with no group separators, and either a period or comma // for fraction separator (+_d)[result = as<double>(_)] >> !((set=',','.') >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // A number with optional group separators | repeat<1,3>(_d)[result = as<int>(_)] >> ( // ... with comma group separators and period fraction separator *(',' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !('.' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // ... with period group separators and comma fraction separator | *('.' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !(',' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) );

std::string str; double d = 0.0; smatch what; what.let(result = d); std::cout << std::fixed << std::setprecision(2); while (std::getline(std::cin, str)) { if (regex_search(str, what, double_)) { std::cout << "Double = " << d << '\n'; } else { std::cout << "No match\n"; } } }

Eric Niebler

2:09 p.m.

New subject: boost::xpressive : Get thousand separator as double

If performance matters, you might want to avoid xpressive::as. It's just a thin wrapper over boost::lexical_cast, which may(?) do dynamic allocation while performing conversion. Try this, which uses a wrapper over std::atoi instead. (It's a bit sneaky in that it uses the fact that characters in a string are guaranteed contiguous, and that atoi stops parsing on the first non-digit.) Disclaimer: this code is not well-tested. struct asint_ : std::unary_function { int operator()(xp::ssub_match const &s) const { return (int)std::atoi(&*s.first); } }; xp::function::type const asint = {}; int main() { using namespace xp; placeholder<double> result; sregex double_ = ~after(_d) >> // a number with no group separators, and either a period or comma // for fraction separator (+_d)[result = asint(_)] >> !((set=',','.') >> (_d >> _d)[result += asint(_)/100.0]) >> ~before(set[_d|','|'.']) // A number with optional group separators | repeat<1,3>(_d)[result = asint(_)] >> ( // ... with comma group separators and period fraction separator *(',' >> repeat<3>(_d)[result *= 1000, result += asint(_)]) >> !('.' >> (_d >> _d)[result += asint(_)/100.0]) >> ~before(set[_d|','|'.']) // ... with period group separators and comma fraction separator | *('.' >> repeat<3>(_d)[result *= 1000, result += asint(_)]) >> !(',' >> (_d >> _d)[result += asint(_)/100.0]) >> ~before(set[_d|','|'.']) ); std::string str; double d = 0.0; smatch what; what.let(result = d); std::cout << std::fixed << std::setprecision(2); while (std::getline(std::cin, str)) { if (regex_search(str, what, double_)) { std::cout << "Double = " << d << '\n'; } else { std::cout << "No match\n"; } } } On 9/9/2011 5:26 AM, Jens Saathoff wrote:

...

Perfect! That works!

I will compare it with my thousand-sep rule/grammar in boost::spirit, a little benchmark to find out which one is faster. Do have experiences/results with boost::spirit vs. boost::xpressive?

Thank you very much. Have a nice weekend.

2011/9/9 Eric Niebler mailto:eric@boostpro.com>

#include <string> #include <iostream> #include <iomanip> #include #include namespace xp = boost::xpressive;

int main() { using namespace xp; placeholder<double> result; sregex double_ = ~after(_d) >> // a number with no group separators, and either a period or comma // for fraction separator (+_d)[result = as<double>(_)] >> !((set=',','.') >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // A number with optional group separators | repeat<1,3>(_d)[result = as<int>(_)] >> ( // ... with comma group separators and period fraction separator *(',' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !('.' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // ... with period group separators and comma fraction separator | *('.' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !(',' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) );

std::string str; double d = 0.0; smatch what; what.let(result = d); std::cout << std::fixed << std::setprecision(2); while (std::getline(std::cin, str)) { if (regex_search(str, what, double_)) { std::cout << "Double = " << d << '\n'; } else { std::cout << "No match\n"; } } }

_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

-- Eric Niebler BoostPro Computing http://www.boostpro.com

Jens Saathoff

10:30 p.m.

New subject: boost::xpressive : Get thousand separator as double

Hi! And again...thank you very much for your help. Very kind of you. There's one question left: What does smatch::let? In the doc there's a short comment on it: TODO document me Would you please explain it in short? Thank you! 2011/9/9 Eric Niebler

...

If performance matters, you might want to avoid xpressive::as. It's just a thin wrapper over boost::lexical_cast, which may(?) do dynamic allocation while performing conversion. Try this, which uses a wrapper over std::atoi instead. (It's a bit sneaky in that it uses the fact that characters in a string are guaranteed contiguous, and that atoi stops parsing on the first non-digit.)

Disclaimer: this code is not well-tested.

struct asint_ : std::unary_function { int operator()(xp::ssub_match const &s) const { return (int)std::atoi(&*s.first); } };

xp::function::type const asint = {};

int main() { using namespace xp; placeholder<double> result; sregex double_ = ~after(_d) >> // a number with no group separators, and either a period or comma // for fraction separator (+_d)[result = asint(_)] >> !((set=',','.') >> (_d >> _d)[result += asint(_)/100.0]) >> ~before(set[_d|','|'.']) // A number with optional group separators | repeat<1,3>(_d)[result = asint(_)] >> ( // ... with comma group separators and period fraction separator *(',' >> repeat<3>(_d)[result *= 1000, result += asint(_)]) >> !('.' >> (_d >> _d)[result += asint(_)/100.0]) >> ~before(set[_d|','|'.']) // ... with period group separators and comma fraction separator | *('.' >> repeat<3>(_d)[result *= 1000, result += asint(_)]) >> !(',' >> (_d >> _d)[result += asint(_)/100.0]) >> ~before(set[_d|','|'.']) );

std::string str; double d = 0.0; smatch what; what.let(result = d); std::cout << std::fixed << std::setprecision(2); while (std::getline(std::cin, str)) { if (regex_search(str, what, double_)) { std::cout << "Double = " << d << '\n'; } else { std::cout << "No match\n"; } } }

On 9/9/2011 5:26 AM, Jens Saathoff wrote:

...
Perfect! That works!

I will compare it with my thousand-sep rule/grammar in boost::spirit, a little benchmark to find out which one is faster. Do have experiences/results with boost::spirit vs. boost::xpressive?

Thank you very much. Have a nice weekend.

2011/9/9 Eric Niebler mailto:eric@boostpro.com>

#include <string> #include <iostream> #include <iomanip> #include #include namespace xp = boost::xpressive;

int main() { using namespace xp; placeholder<double> result; sregex double_ = ~after(_d) >> // a number with no group separators, and either a period or comma // for fraction separator (+_d)[result = as<double>(_)] >> !((set=',','.') >> (_d >> _d)[result += as<int>(_)/100.0])

...
~before(set[_d|','|'.']) // A number with optional group separators | repeat<1,3>(_d)[result = as<int>(_)] >> ( // ... with comma group separators and period fraction separator *(',' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !('.' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) // ... with period group separators and comma fraction separator | *('.' >> repeat<3>(_d)[result *= 1000, result += as<int>(_)]) >> !(',' >> (_d >> _d)[result += as<int>(_)/100.0]) >> ~before(set[_d|','|'.']) );

std::string str; double d = 0.0; smatch what; what.let(result = d); std::cout << std::fixed << std::setprecision(2); while (std::getline(std::cin, str)) { if (regex_search(str, what, double_)) { std::cout << "Double = " << d << '\n'; } else { std::cout << "No match\n"; } } }

_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

-- Eric Niebler BoostPro Computing http://www.boostpro.com _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

Eric Niebler

10 Sep 10 Sep

9:02 a.m.

New subject: boost::xpressive : Get thousand separator as double

On 9/9/2011 6:30 PM, Jens Saathoff wrote:

...

There's one question left: What does smatch::let? In the doc there's a short comment on it: TODO document me

Would you please explain it in short?

Ah, the reference section needs updating, but this is described in the users' guide: http://www.boost.org/doc/libs/1_47_0/doc/html/xpressive/user_s_guide.html#bo... HTH, -- Eric Niebler BoostPro Computing http://www.boostpro.com

Jens Saathoff

13 Sep 13 Sep

10:54 p.m.

New subject: boost::xpressive : Get thousand separator as double

Thank you very much! Is it possible to mix static and dynamic xpressive as nested xpressions like: staticregex = double_; (the code you provided) dynregex.compile("i paid:" >> staticregex); That's just a very short example, but i think you understand what i mean here. Thankx a lot! 2011/9/10 Eric Niebler

...

On 9/9/2011 6:30 PM, Jens Saathoff wrote:

...
There's one question left: What does smatch::let? In the doc there's a short comment on it: TODO document me

Would you please explain it in short?

Ah, the reference section needs updating, but this is described in the users' guide:

http://www.boost.org/doc/libs/1_47_0/doc/html/xpressive/user_s_guide.html#bo...

HTH,

-- Eric Niebler BoostPro Computing http://www.boostpro.com _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

Eric Niebler

14 Sep 14 Sep

4:26 a.m.

New subject: boost::xpressive : Get thousand separator as double

(Please don't top-post) On 9/13/2011 6:54 PM, Jens Saathoff wrote:

...

Is it possible to mix static and dynamic xpressive as nested xpressions like:

staticregex = double_; (the code you provided) dynregex.compile("i paid:" >> staticregex);

That's just a very short example, but i think you understand what i mean here.

Yes, dynamic regexes can embedded in static regexes, and static regexes can be embedded in dynamic regexes via named regexes. It's all described in the users' guide. -- Eric Niebler BoostPro Computing http://www.boostpro.com

Jens Saathoff

12:50 p.m.

New subject: boost::xpressive : Get thousand separator as double

Yes, you are right. I overlooked it and used the wrong search term. Thanks 2011/9/14 Eric Niebler

...

(Please don't top-post)

On 9/13/2011 6:54 PM, Jens Saathoff wrote:

...
Is it possible to mix static and dynamic xpressive as nested xpressions like:

staticregex = double_; (the code you provided) dynregex.compile("i paid:" >> staticregex);

That's just a very short example, but i think you understand what i mean here.

Yes, dynamic regexes can embedded in static regexes, and static regexes can be embedded in dynamic regexes via named regexes. It's all described in the users' guide.

-- Eric Niebler BoostPro Computing http://www.boostpro.com _______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

4966

Age (days ago)

4972

Last active (days ago)

List overview

Download

8 comments

2 participants

participants (2)

Eric Niebler
Jens Saathoff