Parse string with boost::spirit phrase_parse

Question

I would like to parse the string

 std::string entry = "127.0.0.1 - [16/Aug/2012:01:50:02 +0000] \"GET /check.htm HTTP/1.1\" 200 17 \"AgentName/0.1 libwww-perl/5.833\""

with the followings:

 ip_rule %= lexeme[(+char_("0-9."))[ref(ip) = _1]]; timestamp_rule %= lexeme[('[' >> +(char_ - ']') >> ']')[ref(timestamp) = _1]]; user_rule %= lexeme[(+char_)[ref(user) = _1]]; request_rule %= lexeme[('"' >> +(char_ - '"') >> '"')[ref(req) = _1]]; referer_rule %= lexeme[('"' >> +(char_ - '"') >> '"')[ref(referer) = _1]]; bool r = phrase_parse(first, last, ip_rule >> user_rule >> timestamp_rule >> request_rule >> uint_[ref(status) = _1] >> uint_[ref(transferred_bytes) = _1] >> referer_rule, space);

but it does not match. If I remove the "-" from the string, and the rule "user_rule" of course, than it matches. Could you please advise how to match the string with the "-"?

The rules have the following type: rule<Iterator, std::string(), space_type> ip_rule, timestamp_rule, user_rule, request_rule, referer_rule; — bayerb
– bayerb, Commented Aug 23, 2012 at 9:14

Igor R. · Accepted Answer · 2012-08-23 10:10:49Z

3

Your user_rule "eats" the rest of the text. Define it like this: +~qi::char_("[")), so that it would stop at '[' character. The following code works as expected:

#include <boost/spirit/include/qi.hpp> using namespace boost::spirit::qi; int main() { std::string ip, user, timestamp, req, referer; unsigned status, transferred_bytes; std::string entry = "127.0.0.1 - [16/Aug/2012:01:50:02 +0000] \"GET /check.htm HTTP/1.1\" 200 17 \"AgentName/0.1 libwww-perl/5.833\""; bool r = phrase_parse(entry.begin(), entry.end(), lexeme[+char_("0-9.")] >> +~char_("[") >> lexeme[('[' >> +~char_("]") >> ']')] >> lexeme[('"' >> +~char_("\"") >> '"')] >> uint_ >> uint_ >> lexeme[('"' >> +~char_("\"") >> '"')], space, ip, user, timestamp, req, status, transferred_bytes, referer); }

edited Aug 23, 2012 at 10:10

answered Aug 23, 2012 at 10:03

Igor R.

15.1k2 gold badges52 silver badges89 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

bayerb Over a year ago

Actually my goal is to be able to change the sequence of the different rules. This is an entry from an access log, and thus the sequence can change depending on the logformat directive of the webservers config. How is it possible to solve it with boost spirit?

Igor R. Over a year ago

@user777377 I'm not sure I got your question and how it's related to the original one - could you provide an example? If it's just another one, please write a separate question.

bayerb Over a year ago

e.g.: std::string1 = '127.0.0.1 - [16/Aug/2012:01:50:02 +0000]'; std::string2 = '127.0.0.1 [16/Aug/2012:01:50:02 +0000] -';

Igor R. Over a year ago

@user777377 well, you should define how the ip is separated from the timestamp. For instance, the definition may be: "any characters, but "[", or nothing". In this case, just change +~char_("[") in my code above to *~char_("[").

bayerb Over a year ago

Please ignore my previous comment. So e.g.: std::string s1 = '127.0.0.1 - [16/Aug/2012:01:50:02 +0000]'; std::string s2 = '127.0.0.1 [16/Aug/2012:01:50:02 +0000] -'; std::vector<rules> input1{ip, user, timestamp} configurable_smart_parser(input1, s1); std::vector<rules> input2{ip, timestamp, user} configurable_smart_parser(input2, user, s2); I want to build something similar to this configurable_smart_parser

|

Collectives™ on Stack Overflow

Parse string with boost::spirit phrase_parse

1 Answer 1

6 Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

6 Comments

Related