I need to parse following EBNF expression with Boost::Spirit.
period ::= date_part [time_part] , date_part [time_part]
time_part ::= hours:minutes[:seconds]
date_part ::= day.month.year
For example, 10.06.2014 10:00:15, 11.07.2014. I made my grammar in two ways, but can't exactly get working example.
1) First attempt
struct Parser: grammar<std::string::const_iterator, space_type>
{
Parser(): Parser::base_type(datetime_)
{
using boost::spirit::int_;
using boost::spirit::qi::_1;
using boost::spirit::qi::_2;
using boost::spirit::qi::_val;
datetime_ =
(date_ >> time_)
[
_val =
phoenix::construct<ptime>
(
date(_1[2]), _1[1], _1[0]),
hours(_2[0]) + minutes(_2[1]) + seconds[_2[0]]
)
|
_val =
phoenix::construct<ptime>
(
date(_1[2]), _1[1], _1[0]),
seconds(0)
)
];
date_ %= int_ % '.';
time_ %= int_ % ':';
BOOST_SPIRIT_DEBUG_NODE(datetime_);
BOOST_SPIRIT_DEBUG_NODE(date_);
BOOST_SPIRIT_DEBUG_NODE(time_);
}
rule<std::string::const_iterator, std::vector<int>(), space_type> date_, time_;
rule<std::string::const_iterator, ptime(), space_type> datetime_;
}
Parser parser;
std::string strTest("10.06.2014 10:00:15, 11.07.2014");
std::string::const_iterator it_begin(strTest.begin());
std::string::const_iterator it_end(strTest.end());
bool result = phrase_parse(it_begin, it_end, parser, space);
Errors:
/media/Data/Projects/Qt/Planner/parser.h:108: ошибка: no matching function for call to 'boost::gregorian::date::date(boost::phoenix::detail::make_index_composite<boost::phoenix::actor<boost::spirit::argument<0> >, int>::type)'
And so on. I can't cast boost::spirit::argument<0> to int or date::years_type. I tryed date((int)_1[2]), (int)_1[1], (int)_1[0])) and dynamic_cast<int>(_1[2]), but with no success (.
2) Second attempt
struct Parser: grammar<std::string::const_itearator, space_type>
{
Parser(ConditionTree& a_lTree):
Parser::base_type(time_period_),
m_lTree(a_lTree)
{
using boost::spirit::int_;
using boost::spirit::qi::_1;
using boost::spirit::qi::_2;
using boost::spirit::qi::_3;
using boost::spirit::qi::_4;
using boost::spirit::qi::_5;
using boost::spirit::qi::_val;
time_period_ = ( datetime_ > ',' > datetime_ ) [ _val = phoenix::construct<time_period>((int)_1, (int)_3) ];
datetime_ = (date_ >> time_duration_) [ _val = phoenix::construct<ptime>((int)_1, (int)_2) | _val = phoenix::construct<ptime>((int)_1, seconds(0)) ] ;
date_ = (int_ > '.' > int_ > '.' > int_) [ _val = phoenix::construct<date>((int)_5, (int)_3, (int)_1) ];
time_duration_ = (int_ > ':' > int_ > ':' > int_) [ _val = phoenix::construct<time_duration>((int)_1, (int)_3, (int)_5, 0)];
BOOST_SPIRIT_DEBUG_NODE(time_period_);
BOOST_SPIRIT_DEBUG_NODE(datetime_);
BOOST_SPIRIT_DEBUG_NODE(date_);
BOOST_SPIRIT_DEBUG_NODE(time_duration_);
}
rule<std::string::const_itarator, time_period(), space_type> time_period_;
rule<std::string::const_itarator, ptime(), space_type> datetime_;
rule<std::string::const_itarator, date(), space_type> date_;
rule<std::string::const_itarator, time_duration(), space_type> time_duration_;
ConditionTree& m_lTree;
};
Error:
/media/Data/Projects/Qt/Planner/parser.h:114: ошибка: invalid cast from type 'const _1_type {aka const boost::phoenix::actor<boost::spirit::argument<0> >}' to type 'int'...
Why I can't cast boost::spirit::argument<0> to int????
Better question, why would you be able to cast a placeholder type to a specific primitive type?
The place holder is a lazy actor only, so you should use Phoenix cast_ to cast it, if at all (hint: this should not be necessary): Live On Coliru
Output
<period_>
<try>10.06.2014 10:00:15,</try>
<date_>
<try>10.06.2014 10:00:15,</try>
<success> 10:00:15, 11.07.201</success>
<attributes>[[10, 6, 2014]]</attributes>
</date_>
<time_>
<try> 10:00:15, 11.07.201</try>
<success>, 11.07.2014</success>
<attributes>[[10, 0, 15]]</attributes>
</time_>
<date_>
<try> 11.07.2014</try>
<success></success>
<attributes>[[11, 7, 2014]]</attributes>
</date_>
<time_>
<try></try>
<fail/>
</time_>
<success></success>
<attributes>[[[[10, 6, 2014], [10, 0, 15]], [[11, 7, 2014], [empty]]]]</attributes>
</period_>
Parse success
Full Sample
#define BOOST_SPIRIT_DEBUG
#include <boost/fusion/adapted/struct.hpp>
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
namespace qi = boost::spirit::qi;
namespace phx = boost::phoenix;
namespace Ast {
using boost::optional;
struct date { unsigned day, month, year; };
struct time { unsigned hours, minutes, seconds; };
struct date_time { date date_part; optional<time> time_part; };
struct period { date_time start, end; };
}
BOOST_FUSION_ADAPT_STRUCT(Ast::date, (unsigned,day)(unsigned,month)(unsigned,year))
BOOST_FUSION_ADAPT_STRUCT(Ast::time, (unsigned,hours)(unsigned,minutes)(unsigned,seconds))
BOOST_FUSION_ADAPT_STRUCT(Ast::date_time, (Ast::date,date_part)(Ast::optional<Ast::time>, time_part))
BOOST_FUSION_ADAPT_STRUCT(Ast::period, (Ast::date_time,start)(Ast::date_time,end))
template <typename Iterator>
struct Parser : qi::grammar<Iterator, Ast::period(), qi::space_type>
{
int test;
Parser() : Parser::base_type(period_)
{
using namespace qi;
static const int_parser<unsigned, 10, 2, 2> _2digit = {};
static const int_parser<unsigned, 10, 4, 4> _4digit = {};
time_ = _2digit >> ":" >> _2digit >> ":" >> _2digit;
date_ = _2digit >> "." >> _2digit >> "." >> _4digit;
date_time_ = date_ >> -time_;
period_ = date_time_ >> "," >> date_time_;
BOOST_SPIRIT_DEBUG_NODES((period_)(time_)(date_))
}
private:
qi::rule<Iterator, Ast::period(), qi::space_type> period_;
qi::rule<Iterator, Ast::date(), qi::space_type> date_;
qi::rule<Iterator, Ast::time(), qi::space_type> time_;
qi::rule<Iterator, Ast::date_time(), qi::space_type> date_time_;
};
int main()
{
using It = std::string::const_iterator;
Parser<It> parser;
std::string input("10.06.2014 10:00:15, 11.07.2014");
It f(input.begin()), l(input.end());
Ast::period parsed;
bool ok = qi::phrase_parse(f, l, parser, qi::space, parsed);
if (ok)
{
std::cout << "Parse success\n";
}
else
{
std::cout << "Parse failed\n";
}
if (f!=l)
std::cout << "Remaining unparsed input: '" << std::string(f,l) << "'\n";
}
Related
I have this example code, which parses the string str correctly.
How to I make it work if there any extra characters before and/or after the string? For example if I did str = std::string("AAA") + str + std::string("AAA")
frame.h
#define BOOST_SPIRIT_USE_PHOENIX_V3
//#define BOOST_SPIRIT_DEBUG
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/fusion/adapted/std_pair.hpp>
namespace qi = boost::spirit::qi;
namespace phx = boost::phoenix;
namespace ascii = boost::spirit::ascii;
struct frame
{
std::string addr;
std::string func;
std::string file;
std::string fullname;
std::string line;
std::map<std::string, std::string> kv;
};
template <typename Iterator>
struct argsArray : qi::grammar<Iterator, std::map<std::string, std::string>()>
{
argsArray() : argsArray::base_type(query)
{
query =
qi::lit("args=[") >> pair >> *(qi::lit(',') >> pair) >> qi::lit(']');
pair = qi::lit("{name=") >> quoted_string >> qi::lit(",value=") >>
quoted_string >> qi::lit("}");
key = qi::char_("a-zA-Z_") >> *qi::char_("a-zA-Z_0-9");
quoted_string %= boost::spirit::lexeme['"' >> +(qi::char_ - '"') >> '"'];
}
qi::rule<Iterator, std::map<std::string, std::string>()> query;
qi::rule<Iterator, std::pair<std::string, std::string>()> pair;
qi::rule<Iterator, std::string()> key;
qi::rule<Iterator, std::string()> quoted_string;
};
template <typename Iterator>
struct frameParser : qi::grammar<Iterator, frame(), ascii::space_type>
{
frameParser() : frameParser::base_type(frame_rule)
{
static const auto _addr = phx::bind(&frame::addr, qi::_r1);
static const auto _func = phx::bind(&frame::func, qi::_r1);
static const auto _file = phx::bind(&frame::file, qi::_r1);
static const auto _fullname = phx::bind(&frame::fullname, qi::_r1);
static const auto _line = phx::bind(&frame::line, qi::_r1);
static const auto _kv = phx::bind(&frame::kv, qi::_r1);
func = qi::lit("func=") >> quoted_string;
addr = qi::lit("addr=") >> quoted_string;
file = qi::lit("file=") >> quoted_string;
fullname = qi::lit("fullname=") >> quoted_string;
line = qi::lit("line=") >> quoted_string;
func_rule = func[_func = qi::_1];
addr_rule = addr[_addr = qi::_1];
file_rule = file[_file = qi::_1];
fullname_rule = fullname[_fullname = qi::_1];
line_rule = line[_line = qi::_1];
kv_rule = arrTest[_kv = qi::_1];
quoted_string %= boost::spirit::lexeme['"' >> +(qi::char_ - '"') >> '"'];
frame_rule = qi::lit("frame={") >>
(addr_rule(qi::_val) ^ qi::lit(',') ^ func_rule(qi::_val) ^
qi::lit(',') ^ file_rule(qi::_val) ^ qi::lit(',') ^
fullname_rule(qi::_val) ^ qi::lit(',') ^ line_rule(qi::_val) ^
qi::lit(',') ^ kv_rule(qi::_val)) >>
qi::lit('}');
BOOST_SPIRIT_DEBUG_NODES(
(frame_rule)(func_rule)(addr_rule)(fullname_rule)(line_rule))
}
qi::rule<Iterator, void(frame&), ascii::space_type> func_rule, addr_rule,
file_rule, fullname_rule, line_rule, kv_rule;
qi::rule<Iterator, frame(), ascii::space_type> frame_rule;
qi::rule<Iterator, std::string()> addr, func, file, fullname, line;
qi::rule<Iterator, std::string()> quoted_string;
argsArray<Iterator> arrTest;
};
test.cc
#include <iostream>
#include "gtest/gtest.h"
#include "parser/frame.h"
TEST(ParseFrameString, Test1)
{
std::string str = R"(frame={addr="0x0000000000414008",)"
R"(func="main",)"
R"(args=[{name="argc",value="1"},)"
R"({name="argv",value="0x7fffffffe1a8"}],)"
R"(file="/home/stiopa/development/gdbFront/main.cc",)"
R"(fullname="/home/stiopa/development/gdbFront/main.cc",)"
R"(line="90"}")";
typedef std::string::const_iterator It;
const frameParser<It> g;
It iter(str.begin()), end(str.end());
frame frame;
bool r = phrase_parse(iter, end, g, boost::spirit::ascii::space, frame);
EXPECT_EQ(r, true);
EXPECT_EQ(frame.addr, "0x0000000000414008");
EXPECT_EQ(frame.func, "main");
std::map<std::string, std::string> kv{{"argc", "1"},
{"argv", "0x7fffffffe1a8"}};
EXPECT_EQ(frame.kv, kv);
EXPECT_EQ(frame.file, "/home/stiopa/development/gdbFront/main.cc");
EXPECT_EQ(frame.fullname, "/home/stiopa/development/gdbFront/main.cc");
EXPECT_EQ(frame.line, "90");
}
The simple, low-tech solution would be to use qi::seek from the repository:
#include <boost/spirit/repository/include/qi_seek.hpp>
namespace qir = boost::spirit::repository::qi;
And then:
bool r = phrase_parse(iter, end, qir::seek[g], boost::spirit::ascii::space, frame);
DEMO
Live On Coliru
#define BOOST_SPIRIT_USE_PHOENIX_V3
//#define BOOST_SPIRIT_DEBUG
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/fusion/adapted/std_pair.hpp>
#include <boost/spirit/repository/include/qi_seek.hpp>
namespace qi = boost::spirit::qi;
namespace qir = boost::spirit::repository::qi;
namespace phx = boost::phoenix;
namespace ascii = boost::spirit::ascii;
struct frame
{
std::string addr;
std::string func;
std::string file;
std::string fullname;
std::string line;
std::map<std::string, std::string> kv;
};
template <typename Iterator>
struct argsArray : qi::grammar<Iterator, std::map<std::string, std::string>()>
{
argsArray() : argsArray::base_type(query)
{
query =
qi::lit("args=[") >> pair >> *(qi::lit(',') >> pair) >> qi::lit(']');
pair = qi::lit("{name=") >> quoted_string >> qi::lit(",value=") >>
quoted_string >> qi::lit("}");
key = qi::char_("a-zA-Z_") >> *qi::char_("a-zA-Z_0-9");
quoted_string %= boost::spirit::lexeme['"' >> +(qi::char_ - '"') >> '"'];
}
qi::rule<Iterator, std::map<std::string, std::string>()> query;
qi::rule<Iterator, std::pair<std::string, std::string>()> pair;
qi::rule<Iterator, std::string()> key;
qi::rule<Iterator, std::string()> quoted_string;
};
template <typename Iterator>
struct frameParser : qi::grammar<Iterator, frame(), ascii::space_type>
{
frameParser() : frameParser::base_type(frame_rule)
{
static const auto _addr = phx::bind(&frame::addr, qi::_r1);
static const auto _func = phx::bind(&frame::func, qi::_r1);
static const auto _file = phx::bind(&frame::file, qi::_r1);
static const auto _fullname = phx::bind(&frame::fullname, qi::_r1);
static const auto _line = phx::bind(&frame::line, qi::_r1);
static const auto _kv = phx::bind(&frame::kv, qi::_r1);
func = qi::lit("func=") >> quoted_string;
addr = qi::lit("addr=") >> quoted_string;
file = qi::lit("file=") >> quoted_string;
fullname = qi::lit("fullname=") >> quoted_string;
line = qi::lit("line=") >> quoted_string;
func_rule = func[_func = qi::_1];
addr_rule = addr[_addr = qi::_1];
file_rule = file[_file = qi::_1];
fullname_rule = fullname[_fullname = qi::_1];
line_rule = line[_line = qi::_1];
kv_rule = arrTest[_kv = qi::_1];
quoted_string %= boost::spirit::lexeme['"' >> +(qi::char_ - '"') >> '"'];
frame_rule = qi::lit("frame={") >>
(addr_rule(qi::_val) ^ qi::lit(',') ^ func_rule(qi::_val) ^
qi::lit(',') ^ file_rule(qi::_val) ^ qi::lit(',') ^
fullname_rule(qi::_val) ^ qi::lit(',') ^ line_rule(qi::_val) ^
qi::lit(',') ^ kv_rule(qi::_val)) >>
qi::lit('}');
BOOST_SPIRIT_DEBUG_NODES(
(frame_rule)(func_rule)(addr_rule)(fullname_rule)(line_rule))
}
qi::rule<Iterator, void(frame&), ascii::space_type> func_rule, addr_rule,
file_rule, fullname_rule, line_rule, kv_rule;
qi::rule<Iterator, frame(), ascii::space_type> frame_rule;
qi::rule<Iterator, std::string()> addr, func, file, fullname, line;
qi::rule<Iterator, std::string()> quoted_string;
argsArray<Iterator> arrTest;
};
#include <iostream>
//#include "parser/frame.h"
int main()
{
std::string str = R"(frame={addr="0x0000000000414008",)"
R"(func="main",)"
R"(args=[{name="argc",value="1"},)"
R"({name="argv",value="0x7fffffffe1a8"}],)"
R"(file="/home/stiopa/development/gdbFront/main.cc",)"
R"(fullname="/home/stiopa/development/gdbFront/main.cc",)"
R"(line="90"}")";
str = "AAA" + str + "AAA";
typedef std::string::const_iterator It;
const frameParser<It> g;
It iter(str.begin()), end(str.end());
frame frame;
bool r = phrase_parse(iter, end, qir::seek[g], boost::spirit::ascii::space, frame);
assert(r == true);
assert(frame.addr == "0x0000000000414008");
assert(frame.func == "main");
std::map<std::string, std::string> kv{{"argc", "1"},
{"argv", "0x7fffffffe1a8"}};
assert(frame.kv == kv);
assert(frame.file == "/home/stiopa/development/gdbFront/main.cc");
assert(frame.fullname == "/home/stiopa/development/gdbFront/main.cc");
assert(frame.line == "90");
}
Tests still pass.
And here's a free code review. Please see
Boost Spirit: "Semantic actions are evil"?
Boost spirit skipper issues
you didn't parse the delimiting ',' correctly at all. You must require it, unless end-of-frame ('}'). You cannot accept multiple in a row
Live On Coliru
#define BOOST_SPIRIT_USE_PHOENIX_V3
//#define BOOST_SPIRIT_DEBUG
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/fusion/adapted/std_pair.hpp>
#include <boost/fusion/adapted/struct.hpp>
#include <boost/spirit/repository/include/qi_seek.hpp>
namespace qi = boost::spirit::qi;
namespace qir = boost::spirit::repository::qi;
namespace phx = boost::phoenix;
namespace ascii = boost::spirit::ascii;
struct frame {
std::string addr;
std::string func;
std::string file;
std::string fullname;
std::string line;
std::map<std::string, std::string> kv;
};
BOOST_FUSION_ADAPT_STRUCT(frame, addr, func, file, fullname, line, kv)
template <typename Iterator>
struct argsArray : qi::grammar<Iterator, std::map<std::string, std::string>()>
{
argsArray() : argsArray::base_type(query)
{
query = "args=[" >> pair >> *(',' >> pair) >> ']';
pair = "{name=" >> quoted_string >> ",value=" >> quoted_string >> "}";
key = qi::char_("a-zA-Z_") >> *qi::char_("a-zA-Z_0-9");
quoted_string = '"' >> +(qi::char_ - '"') >> '"';
}
private:
qi::rule<Iterator, std::map<std::string, std::string>()> query;
qi::rule<Iterator, std::pair<std::string, std::string>()> pair;
qi::rule<Iterator, std::string()> key;
qi::rule<Iterator, std::string()> quoted_string;
};
template <typename Iterator>
struct frameParser : qi::grammar<Iterator, frame(), ascii::space_type>
{
frameParser() : frameParser::base_type(frame_rule)
{
quoted_string = '"' >> +(qi::char_ - '"') >> '"';
delim = (&qi::lit('}')) | ',';
field_rule = qi::lexeme [ qi::lit(qi::_r1) >> '=' ] >> quoted_string >> delim;
kv_rule = arrTest >> delim;
frame_rule = "frame={" >>
(field_rule(+"addr") ^
field_rule(+"func") ^
field_rule(+"file") ^
field_rule(+"fullname") ^
field_rule(+"line") ^
kv_rule
) >> '}';
BOOST_SPIRIT_DEBUG_NODES((frame_rule)(field_rule))
}
private:
qi::rule<Iterator> delim;
qi::rule<Iterator, std::string(char const*), ascii::space_type> field_rule;
qi::rule<Iterator, std::map<std::string, std::string>()> kv_rule;
qi::rule<Iterator, frame(), ascii::space_type> frame_rule;
qi::rule<Iterator, std::string()> quoted_string;
argsArray<Iterator> arrTest;
};
#include <iostream>
//#include "parser/frame.h"
int main()
{
std::string str = R"(frame={addr="0x0000000000414008",)"
R"(func="main",)"
R"(args=[{name="argc",value="1"},)"
R"({name="argv",value="0x7fffffffe1a8"}],)"
R"(file="/home/stiopa/development/gdbFront/main.cc",)"
R"(fullname="/home/stiopa/development/gdbFront/main.cc",)"
R"(line="90"}")";
str = "AAA" + str + "AAA";
typedef std::string::const_iterator It;
const frameParser<It> g;
It iter(str.begin()), end(str.end());
frame frame;
bool r = phrase_parse(iter, end, qir::seek[g], boost::spirit::ascii::space, frame);
if (iter != end)
std::cout << "Remaining unparsed: '" << std::string(iter,end) << "'\n";
assert(r == true);
assert(frame.addr == "0x0000000000414008");
assert(frame.func == "main");
std::map<std::string, std::string> kv{{"argc", "1"},
{"argv", "0x7fffffffe1a8"}};
assert(frame.kv == kv);
assert(frame.file == "/home/stiopa/development/gdbFront/main.cc");
assert(frame.fullname == "/home/stiopa/development/gdbFront/main.cc");
assert(frame.line == "90");
}
Prints:
Remaining unparsed: '"AAA'
Tests still pass.
Note your original sample input had a trailing ", which you simply ignored.
I managed to parse a pgn file thanks to the Boost Spirit library, but it fails as soon as there is some characters I did not "anticipated".
Here is my Spirit grammar :
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
BOOST_FUSION_ADAPT_STRUCT(
loloof64::pgn_tag,
(std::string, key),
(std::string, value)
)
BOOST_FUSION_ADAPT_STRUCT(
loloof64::game_move,
(unsigned, move_number),
(std::string, move_turn),
(std::string, white_move),
(std::string, black_move),
(std::string, result)
)
BOOST_FUSION_ADAPT_STRUCT(
loloof64::pgn_game,
(std::vector<loloof64::pgn_tag>, header),
(std::vector<loloof64::game_move>, moves)
)
namespace loloof64 {
namespace qi = boost::spirit::qi;
namespace ascii = boost::spirit::ascii;
namespace phoenix = boost::phoenix;
template <typename Iterator>
struct pgn_parser : qi::grammar<Iterator, std::vector<pgn_game>, qi::unused_type>
{
pgn_parser() : pgn_parser::base_type(games)
{
using qi::lexeme;
using ascii::char_;
using qi::uint_;
using qi::alnum;
using qi::space;
using qi::omit;
using qi::eol;
using qi::lit;
quoted_string %= lexeme[lit('"') >> *(char_ - '"') >> lit('"')];
tag %=
'['
>> +alnum
>> omit[+space]
>> quoted_string
>> ']'
>> omit[+eol]
;
header %= +tag;
move_turn %= qi::string("...") | qi::string(".");
regular_move %=
+char_("a-hNBRQK")
>> +char_("a-h1-8x=NBRQK")
>> -qi::string("e.p.")
;
castle_move %= qi::string("O-O-O") | qi::string("O-O");
single_move %=
(regular_move | castle_move) >> -(char_('+') | char_('#'))
;
result %= qi::string("1-0") | qi::string("0-1") | qi::string("1/2-1/2") | qi::string("*");
full_move %=
uint_
>> move_turn
>> omit[*space]
>> single_move
>> -(omit[+space] >> single_move)
>> -(omit[+space] >> result)
;
game_description %= full_move
>> *(omit[*space] >> full_move);
single_game %=
-header
>> game_description
;
games %=
single_game
>> *(omit[*(space|eol)] >> single_game)
;
}
qi::rule<Iterator, pgn_tag(), qi::unused_type> tag;
qi::rule<Iterator, std::vector<pgn_tag>, qi::unused_type> header;
qi::rule<Iterator, std::string(), qi::unused_type> quoted_string;
qi::rule<Iterator, std::string(), qi::unused_type> result;
qi::rule<Iterator, std::string(), qi::unused_type> regular_move;
qi::rule<Iterator, std::string(), qi::unused_type> castle_move;
qi::rule<Iterator, std::string(), qi::unused_type> single_move;
qi::rule<Iterator, std::string(), qi::unused_type> move_turn;
qi::rule<Iterator, game_move(), qi::unused_type> full_move;
qi::rule<Iterator, std::vector<game_move>, qi::unused_type> game_description;
qi::rule<Iterator, pgn_game, qi::unused_type> single_game;
qi::rule<Iterator, std::vector<pgn_game>, qi::unused_type> games;
};
}
How could I simply consume any character I could not "anticipate" ? I mean, how could I ignore any character that I don't want in none of my grammar rule ?
As for testing purposes :
here my parser header (pgn_games_extractor.hpp)
#ifndef PGNGAMESEXTRACTOR_HPP
#define PGNGAMESEXTRACTOR_HPP
#include <string>
#include <vector>
#include <fstream>
#include <stdexcept>
namespace loloof64 {
struct pgn_tag {
std::string key;
std::string value;
};
struct game_move {
unsigned move_number;
std::string move_turn;
std::string white_move;
std::string black_move;
std::string result;
};
struct pgn_game {
std::vector<pgn_tag> header;
std::vector<game_move> moves;
};
class PgnGamesExtractor
{
public:
PgnGamesExtractor(std::string inputFilePath);
PgnGamesExtractor(std::ifstream &inputFile);
/*
Both constructos may throw PgnParsingException (if bad pgn format) and InputFileException (if missing file)
*/
std::vector<pgn_game> getGames() const { return games; }
virtual ~PgnGamesExtractor();
protected:
private:
std::vector<pgn_game> games;
void parseInput(std::ifstream &inputFile);
};
class PgnParsingException : public std::runtime_error
{
public:
PgnParsingException(std::string message): std::runtime_error(message){}
};
class InputFileException : public std::runtime_error
{
public:
InputFileException(std::string message) : std::runtime_error(message){}
};
}
#endif // PGNGAMESEXTRACTOR_HPP
Here is my parser source (pgn_games_extractor.cpp) :
#include "pgn_games_extractor.hpp"
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
BOOST_FUSION_ADAPT_STRUCT(
loloof64::pgn_tag,
(std::string, key),
(std::string, value)
)
BOOST_FUSION_ADAPT_STRUCT(
loloof64::game_move,
(unsigned, move_number),
(std::string, move_turn),
(std::string, white_move),
(std::string, black_move),
(std::string, result)
)
BOOST_FUSION_ADAPT_STRUCT(
loloof64::pgn_game,
(std::vector<loloof64::pgn_tag>, header),
(std::vector<loloof64::game_move>, moves)
)
namespace loloof64 {
namespace qi = boost::spirit::qi;
namespace ascii = boost::spirit::ascii;
namespace phoenix = boost::phoenix;
template <typename Iterator>
struct pgn_parser : qi::grammar<Iterator, std::vector<pgn_game>, qi::unused_type>
{
pgn_parser() : pgn_parser::base_type(games)
{
using qi::lexeme;
using ascii::char_;
using qi::uint_;
using qi::alnum;
using qi::space;
using qi::omit;
using qi::eol;
using qi::lit;
quoted_string %= lexeme[lit('"') >> *(char_ - '"') >> lit('"')];
tag %=
'['
>> +alnum
>> omit[+space]
>> quoted_string
>> ']'
>> omit[+eol]
;
header %= +tag;
move_turn %= qi::string("...") | qi::string(".");
regular_move %=
+char_("a-hNBRQK")
>> +char_("a-h1-8x=NBRQK")
>> -qi::string("e.p.")
;
castle_move %= qi::string("O-O-O") | qi::string("O-O");
single_move %=
(regular_move | castle_move) >> -(char_('+') | char_('#'))
;
result %= qi::string("1-0") | qi::string("0-1") | qi::string("1/2-1/2") | qi::string("*");
full_move %=
uint_
>> move_turn
>> omit[*space]
>> single_move
>> -(omit[+space] >> single_move)
>> -(omit[+space] >> result)
;
game_description %= full_move
>> *(omit[*space] >> full_move);
single_game %=
-header
>> game_description
;
games %=
single_game
>> *(omit[*(space|eol)] >> single_game)
;
}
qi::rule<Iterator, pgn_tag(), qi::unused_type> tag;
qi::rule<Iterator, std::vector<pgn_tag>, qi::unused_type> header;
qi::rule<Iterator, std::string(), qi::unused_type> quoted_string;
qi::rule<Iterator, std::string(), qi::unused_type> result;
qi::rule<Iterator, std::string(), qi::unused_type> regular_move;
qi::rule<Iterator, std::string(), qi::unused_type> castle_move;
qi::rule<Iterator, std::string(), qi::unused_type> single_move;
qi::rule<Iterator, std::string(), qi::unused_type> move_turn;
qi::rule<Iterator, game_move(), qi::unused_type> full_move;
qi::rule<Iterator, std::vector<game_move>, qi::unused_type> game_description;
qi::rule<Iterator, pgn_game, qi::unused_type> single_game;
qi::rule<Iterator, std::vector<pgn_game>, qi::unused_type> games;
};
}
loloof64::PgnGamesExtractor::PgnGamesExtractor(std::string inputFilePath)
{
std::ifstream inputFile(inputFilePath);
parseInput(inputFile);
}
loloof64::PgnGamesExtractor::PgnGamesExtractor(std::ifstream &inputFile)
{
parseInput(inputFile);
}
loloof64::PgnGamesExtractor::~PgnGamesExtractor()
{
//dtor
}
void loloof64::PgnGamesExtractor::parseInput(std::ifstream &inputFile)
{
using namespace std;
if (! inputFile) throw InputFileException("File does not exist !");
string content("");
getline(inputFile, content, (char) inputFile.eof());
if (inputFile.fail() || inputFile.bad()) throw new InputFileException("Could not read the input file !");
loloof64::pgn_parser<string::const_iterator> parser;
std::vector<loloof64::pgn_game> temp_games;
string::const_iterator iter = content.begin();
string::const_iterator end = content.end();
bool success = boost::spirit::qi::phrase_parse(iter, end, parser, boost::spirit::qi::eol, temp_games);
if (success && iter == end)
{
games = temp_games;
}
else
{
string error_fragment(iter, end);
string error_message("");
error_message = "Failed to parse the input at :'" + error_fragment + "' !";
throw PgnParsingException(error_message);
}
}
I am asking this question because I could not parse the following pgn : ScotchGambitPgn.zip. I think it is because of an encoding issue with this file.
I am using Spirit 2 and C++ 11 (Gnu)
As requested the simple X3 translation.
fewer lines of code (10 lines)
compilation time down from 7.4s to 3.6s (clang)
compilation time down from 11.4s to 6.0s (gcc5)
runtime down from 0.80s to 0.55s (clang and gcc)
The outputs are identical (exactly).
Live On Coliru
//#define BOOST_SPIRIT_DEBUG
#ifndef PGNGAMESEXTRACTOR_HPP
#define PGNGAMESEXTRACTOR_HPP
#include <string>
#include <vector>
#include <fstream>
#include <stdexcept>
namespace loloof64 {
struct pgn_tag {
std::string key;
std::string value;
};
struct game_move {
unsigned move_number;
std::string white_move;
std::string black_move;
enum result_t { white_won, black_won, draw, undecided } result;
};
struct pgn_game {
std::vector<pgn_tag> header;
std::vector<game_move> moves;
};
class PgnGamesExtractor {
public:
PgnGamesExtractor(std::string inputFilePath);
PgnGamesExtractor(std::istream &inputFile);
/*
Both constructos may throw PgnParsingException (if bad pgn format) and InputFileException (if missing file)
*/
std::vector<pgn_game> getGames() const { return games; }
virtual ~PgnGamesExtractor();
protected:
private:
std::vector<pgn_game> games;
void parseInput(std::istream &inputFile);
};
class PgnParsingException : public virtual std::runtime_error {
public:
PgnParsingException(std::string message) : std::runtime_error(message) {}
};
class InputFileException : public virtual std::runtime_error {
public:
InputFileException(std::string message) : std::runtime_error(message) {}
};
}
#endif // PGNGAMESEXTRACTOR_HPP
#include <boost/spirit/home/x3.hpp>
#include <boost/spirit/include/support_istream_iterator.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
BOOST_FUSION_ADAPT_STRUCT(loloof64::pgn_tag, key, value)
BOOST_FUSION_ADAPT_STRUCT(loloof64::game_move, move_number, white_move, black_move, result)
BOOST_FUSION_ADAPT_STRUCT(loloof64::pgn_game, header, moves)
namespace loloof64 {
namespace pgn_parser {
using namespace boost::spirit::x3;
static std::string const no_move;
static auto const result = []{
symbols<game_move::result_t> table;
table.add
("1-0", game_move::white_won)
("0-1", game_move::black_won)
("1/2-1/2", game_move::draw)
("*", game_move::undecided);
return table;
}();
static auto const quoted_string = lexeme['"' >> *~char_('"') >> '"'];
static auto const tag = '[' >> +alnum >> quoted_string >> ']';
static auto const header = +tag;
static auto const regular_move = as_parser("O-O-O") | "O-O" | (+char_("a-hNBRQK") >> +char_("a-h1-8x=NBRQK") >> -lit("e.p."));
static auto const single_move = rule<struct single_move_, std::string> { "single_move" }
= raw [ lexeme [ regular_move >> -char_("+#")] ];
static auto const full_move = rule<struct full_move_, game_move> { "full_move" }
= uint_
>> (lexeme["..." >> attr(no_move)] | "." >> single_move)
>> (single_move | attr(no_move))
>> -result;
static auto const game_description = +full_move;
static auto const single_game = rule<struct single_game_, pgn_game> { "single_game" }
= -header >> game_description;
static auto const games = *single_game;
}
}
loloof64::PgnGamesExtractor::PgnGamesExtractor(std::string inputFilePath) {
std::ifstream inputFile(inputFilePath);
parseInput(inputFile);
}
loloof64::PgnGamesExtractor::PgnGamesExtractor(std::istream &inputFile) { parseInput(inputFile); }
loloof64::PgnGamesExtractor::~PgnGamesExtractor() {
// dtor
}
void loloof64::PgnGamesExtractor::parseInput(std::istream &inputFile) {
if (inputFile.fail() || inputFile.bad())
throw new InputFileException("Could not read the input file !");
typedef boost::spirit::istream_iterator It;
std::vector<loloof64::pgn_game> temp_games;
It iter(inputFile >> std::noskipws), end;
bool success = boost::spirit::x3::phrase_parse(iter, end, pgn_parser::games, boost::spirit::x3::space, temp_games);
if (success && iter == end) {
games.swap(temp_games);
} else {
std::string error_fragment(iter, end);
throw PgnParsingException("Failed to parse the input at :'" + error_fragment + "' !");
}
}
#include <iostream>
int main() {
loloof64::PgnGamesExtractor pge("ScotchGambit.pgn");
std::cout << "Parsed " << pge.getGames().size() << " games\n";
for (auto& g : pge.getGames())
for (auto& m : g.moves)
std::cout << m.move_number << ".\t" << m.white_move << "\t" << m.black_move << "\n";
}
For what it's worth, here's significantly simplified:
Live On Coliru
//#define BOOST_SPIRIT_DEBUG
#ifndef PGNGAMESEXTRACTOR_HPP
#define PGNGAMESEXTRACTOR_HPP
#include <string>
#include <vector>
#include <fstream>
#include <stdexcept>
namespace loloof64 {
struct pgn_tag {
std::string key;
std::string value;
};
struct game_move {
unsigned move_number;
std::string white_move;
std::string black_move;
enum result_t { white_won, black_won, draw, undecided } result;
};
struct pgn_game {
std::vector<pgn_tag> header;
std::vector<game_move> moves;
};
class PgnGamesExtractor {
public:
PgnGamesExtractor(std::string inputFilePath);
PgnGamesExtractor(std::istream &inputFile);
/*
Both constructos may throw PgnParsingException (if bad pgn format) and InputFileException (if missing file)
*/
std::vector<pgn_game> getGames() const { return games; }
virtual ~PgnGamesExtractor();
protected:
private:
std::vector<pgn_game> games;
void parseInput(std::istream &inputFile);
};
class PgnParsingException : public virtual std::runtime_error {
public:
PgnParsingException(std::string message) : std::runtime_error(message) {}
};
class InputFileException : public virtual std::runtime_error {
public:
InputFileException(std::string message) : std::runtime_error(message) {}
};
}
#endif // PGNGAMESEXTRACTOR_HPP
#include <boost/spirit/include/qi.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
BOOST_FUSION_ADAPT_STRUCT(loloof64::pgn_tag, key, value)
BOOST_FUSION_ADAPT_STRUCT(loloof64::game_move, move_number, white_move, black_move, result)
BOOST_FUSION_ADAPT_STRUCT(loloof64::pgn_game, header, moves)
namespace loloof64 {
namespace qi = boost::spirit::qi;
template <typename Iterator> struct pgn_parser : qi::grammar<Iterator, std::vector<pgn_game>, qi::space_type> {
pgn_parser() : pgn_parser::base_type(games) {
using namespace qi;
const std::string no_move;
result.add
("1-0", game_move::white_won)
("0-1", game_move::black_won)
("1/2-1/2", game_move::draw)
("*", game_move::undecided);
quoted_string = '"' >> *~char_('"') >> '"';
tag = '[' >> +alnum >> quoted_string >> ']';
header = +tag;
regular_move = lit("O-O-O") | "O-O" | (+char_("a-hNBRQK") >> +char_("a-h1-8x=NBRQK") >> -lit("e.p."));
single_move = raw [ regular_move >> -char_("+#") ];
full_move = uint_
>> (lexeme["..." >> attr(no_move)] | "." >> single_move)
>> (single_move | attr(no_move))
>> -result;
game_description = +full_move;
single_game = -header >> game_description;
games = *single_game;
BOOST_SPIRIT_DEBUG_NODES(
(tag)(header)(quoted_string)(regular_move)(single_move)
(full_move)(game_description)(single_game)(games)
)
}
private:
qi::rule<Iterator, pgn_tag(), qi::space_type> tag;
qi::rule<Iterator, std::vector<pgn_tag>, qi::space_type> header;
qi::rule<Iterator, game_move(), qi::space_type> full_move;
qi::rule<Iterator, std::vector<game_move>, qi::space_type> game_description;
qi::rule<Iterator, pgn_game, qi::space_type> single_game;
qi::rule<Iterator, std::vector<pgn_game>, qi::space_type> games;
// lexemes
qi::symbols<char, game_move::result_t> result;
qi::rule<Iterator, std::string()> quoted_string;
qi::rule<Iterator> regular_move;
qi::rule<Iterator, std::string()> single_move;
};
}
loloof64::PgnGamesExtractor::PgnGamesExtractor(std::string inputFilePath) {
std::ifstream inputFile(inputFilePath);
parseInput(inputFile);
}
loloof64::PgnGamesExtractor::PgnGamesExtractor(std::istream &inputFile) { parseInput(inputFile); }
loloof64::PgnGamesExtractor::~PgnGamesExtractor() {
// dtor
}
void loloof64::PgnGamesExtractor::parseInput(std::istream &inputFile) {
if (inputFile.fail() || inputFile.bad())
throw new InputFileException("Could not read the input file !");
typedef boost::spirit::istream_iterator It;
loloof64::pgn_parser<It> parser;
std::vector<loloof64::pgn_game> temp_games;
It iter(inputFile >> std::noskipws), end;
bool success = boost::spirit::qi::phrase_parse(iter, end, parser, boost::spirit::qi::space, temp_games);
if (success && iter == end) {
games.swap(temp_games);
} else {
std::string error_fragment(iter, end);
throw PgnParsingException("Failed to parse the input at :'" + error_fragment + "' !");
}
}
int main() {
loloof64::PgnGamesExtractor pge(std::cin); // "ScotchGambit.pgn"
std::cout << "Parsed " << pge.getGames().size() << " games\n";
for (auto& g : pge.getGames())
for (auto& m : g.moves)
std::cout << m.move_number << ".\t" << m.white_move << "\t" << m.black_move << "\n";
}
Notes:
don't read full file in memory (boost::spirit::istream_iterator)
don't manually skip (use skippers)
don't explicitly lexeme (Boost spirit skipper issues)
don't use %= if not needed
don't synthesize unneeded attributes (use raw[])
treat optional parts of move as optional, don't store assymetric magic flags like "..." (look for no_move)
don't be overly specific (use istream& instead of ifstream&)
Probably some other things I forgot. Output is e.g.
Parsed 6166 games
1. e4 e5
2. Nf3 Nc6
3. d4 exd4
4. Bc4 Qf6
5. O-O d6
6. Ng5 Nh6
7. f4 Be7
8. e5 Qg6
9. exd6 cxd6
10. c3 dxc3
11. Nxc3 O-O
12. Nd5 Bd7
13. Rf3 Bg4
14. Bd3 Bxf3
15. Qxf3 f5
16. Bc4 Kh8
17. Nxe7 Nxe7
18. Qxb7 Qf6
19. Be3 Rfb8
20. Qd7 Rd8
21. Qb7 d5
22. Bb3 Nc6
23. Bxd5 Nd4
24. Rd1 Ne2+
25. Kf1 Rab8
26. Qxa7 Rxb2
27. Ne6 Qxe6
28. Bxe6 Rxd1+
29. Kf2
1. e4 e5
2. Nf3 Nc6
3. d4 exd4
4. Bc4 Bc5
5. Ng5 Ne5
6. Bxf7+ Nxf7
7. Nxf7 Bb4+
8. c3 dxc3
9. bxc3 Bxc3+
10. Nxc3 Kxf7
11. Qd5+ Kf8
12. Ba3+ d6
13. e5 Qg5
14. exd6 Qxd5
Indeed the problem is with Veronica. Or, actually, it's with Ver?nica. Where ? is the code unit <93> - which, lacking codepage/encoding information could mean anything really.
You're using ascii::char and this requires 7-bit only characters.
Easily fix it by changing
using ascii::char_;
into
using qi::char_;
^ No it is not. This was part of the problem, but if review the code as is right now, it already does what the pointed out question/answer shows ... and the errors are still not triggered.
I have this boost spirit parser for string literal. It works. Now I would like to start handle errors when it fail. I copied the on_error handle 1-1 from the mini xml example and it compiles, but it is never triggered (no errors are outputted).
This is the parser:
#define BOOST_SPIRIT_USE_PHOENIX_V3
#define BOOST_SPIRIT_DEBUG
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <boost/spirit/home/support/iterators/line_pos_iterator.hpp>
#include <boost/spirit/include/phoenix_fusion.hpp>
#include <boost/spirit/include/phoenix_stl.hpp>
#include <boost/fusion/include/adapt_struct.hpp>
namespace qi = boost::spirit::qi;
struct my_handler_f
{
template <typename...> struct result { typedef void type; };
template <typename... T>
void operator()(T&&...) const {
std::cout << "\nmy_handler_f() invoked with " << sizeof...(T) << " arguments\n";
}
};
struct append_utf8_f
{
template <typename, typename>
struct result { typedef void type; };
template <typename INT>
void operator()(INT in, std::string& to) const
{
auto out = std::back_inserter(to);
boost::utf8_output_iterator<decltype(out)> convert(out);
*convert++ = in;
}
};
struct get_line_f
{
template <typename> struct result { typedef size_t type; };
template <typename It> size_t operator()(It const& pos_iter) const
{
return get_line(pos_iter);
}
};
struct RangePosition { size_t beginLine, endLine; };
struct String : public RangePosition
{
String()
: RangePosition()
, value()
, source()
{
}
std::string value;
std::string source;
};
BOOST_FUSION_ADAPT_STRUCT(String,
(std::string, value)
(std::string, source)
(size_t, beginLine)
(size_t, endLine)
)
template <typename Iterator>
struct source_string : qi::grammar<Iterator, String(), qi::space_type>
{
struct escape_symbols : qi::symbols<char, char>
{
escape_symbols()
{
add
("\'" , '\'')
("\"" , '\"')
("\?" , '\?')
("\\" , '\\')
("0" , '\0')
("a" , '\a')
("b" , '\b')
("f" , '\f')
("n" , '\n')
("r" , '\r')
("t" , '\t')
("v" , '\v')
;
}
} escape_symbol;
source_string() : source_string::base_type(start)
{
using qi::raw;
using qi::_val;
using qi::_1;
using qi::_2;
using qi::_3;
using qi::_4;
using qi::space;
using qi::omit;
using qi::no_case;
using qi::print;
using qi::eps;
using qi::on_error;
using qi::fail;
using qi::lit;
namespace phx = boost::phoenix;
using phx::at_c;
using phx::begin;
using phx::end;
using phx::construct;
using phx::ref;
using phx::val;
escape %= escape_symbol;
character %= (no_case["\\x"] > hex12)
| ("\\" > (oct123 | escape))
| (print - (lit('"') | '\\'));
unicode = ("\\u" > hex4[append_utf8(_1, _val)])
| ("\\U" > hex8[append_utf8(_1, _val)]);
string_section %= '"' > *(unicode | character) > '"';
string %= string_section % omit[*space];
main = raw [
string[at_c<0>(_val) = _1]
]
[
at_c<1>(_val) = construct<std::string>(begin(_1), end(_1)),
at_c<2>(_val) = get_line_(begin(_1)),
at_c<3>(_val) = get_line_(end(_1))
];
start %= eps > main;
on_error<fail>(start, my_handler);
}
boost::phoenix::function<my_handler_f> my_handler;
qi::rule<Iterator, std::string()> escape;
qi::uint_parser<char, 16, 1, 2> hex12;
qi::uint_parser<char, 8, 1, 3> oct123;
qi::rule<Iterator, std::string()> character;
qi::uint_parser<uint16_t, 16, 4, 4> hex4;
qi::uint_parser<uint32_t, 16, 8, 8> hex8;
boost::phoenix::function<append_utf8_f> append_utf8;
qi::rule<Iterator, std::string()> unicode;
qi::rule<Iterator, std::string()> string_section;
qi::rule<Iterator, std::string()> string;
boost::phoenix::function<get_line_f> get_line_;
qi::rule<Iterator, String(), qi::space_type> main;
qi::rule<Iterator, String(), qi::space_type> start;
};
and this is the test code
int main()
{
std::string str[] =
{
"\"\\u1234\\U0002345\"",
//"\"te\"\"st\"",
//"\"te\" \"st\"",
//"\"te\" \n \"st\"",
//"\"\"",
//"\"\\\"\"",
//"\"test\"",
//"\"test\" something",
//"\"\\\'\\\"\\\?\\\\\\a\\b\\f\\n\\r\\t\\v\"",
//"\"\\x61cd\\X3012\\x7z\"",
//"\"\\141cd\\06012\\78\\778\"",
"\"te",
//"\"te\nst\"",
//"\"test\\\"",
//"\"te\\st\"",
//
};
typedef boost::spirit::line_pos_iterator<std::string::const_iterator> Iterator;
for (size_t i = 0; i < sizeof(str) / sizeof(str[0]); ++i)
{
source_string<Iterator> g;
Iterator iter(str[i].begin());
Iterator end(str[i].end());
String string;
bool r = phrase_parse(iter, end, g, qi::space, string);
if (r)
std::cout << string.beginLine << "-" << string.endLine << ": " << string.value << " === " << string.source << "\n";
else
std::cout << "Parsing failed\n";
}
}
I use boost spirit to parse a color. That worked quite well,
but after I changed the the iterator type, the skipper stopped working.
"rgb(1.0,1.0,0.5)" // this works
" rgb(0.2,0.2,0.2)" // this fails
Here is the header:
struct ColorGrammar : public qi::grammar<StringIterator, Color(), chs::space_type>
{
//! Iterator type for this grammar
typedef StringIterator ItType;
//! Skipper type used in this grammar
typedef chs::space_type Skipper;
//! Rule to parse a number with up to 3 digits
qi::uint_parser<uint8, 10, 1, 3> number;
//! Rule to parse a hex digit
qi::uint_parser<uint8, 16, 1, 1> hexdigit;
ColorGrammar();
//! Rule for rgb(...)
qi::rule<ItType, Color(), qi::locals<float, float>, Skipper> rule_rgb;
//! Rule for rgba(...)
qi::rule<ItType, Color(), qi::locals<float, float, float>, Skipper> rule_rgba;
//! Mainrule
qi::rule<ItType, Color(), Skipper> rule_color;
};
Here is the cpp
ColorGrammar::ColorGrammar()
: ColorGrammar::base_type(rule_color, "color-grammar")
{
using namespace qi::labels;
using boost::phoenix::construct;
auto& _1 = qi::_1;
rule_rgb = '(' >> qi::float_[_a = _1] >> ',' >> qi::float_[_b = _1] >> ',' >> qi::float_[_val = phx::construct<Color>(_a, _b, _1)] >> ')';
rule_rgba = '(' >> qi::float_[_a = _1] >> ',' >> qi::float_[_b = _1] >> ',' >> qi::float_[_c = _1] >> ',' >> qi::float_[_val = phx::construct<Color>(_a, _b, _c, _1)] >> ')';
rule_color = (qi::lit("rgb") >> rule_rgb)
| (qi::lit("rgba") >> rule_rgba);
}
And the call:
Color out;
StringIterator begin = str.cbegin();
StringIterator end = str.cend();
bool result = qi::phrase_parse(begin, end, color_, chs::space, out);
I'm sure, it is only a little misstake, but I am not able to see it.
Maybe i watched too long at the source... can you see a misstake?
I can't see what's wrong: I've taken the effort to reconstruct your SSCCE.
http://liveworkspace.org/code/1pDtmn$1
In the process, it seems I must have removed the problem. I suggest you do the same.
Oh, and this is how I'd write this:
no more phoenix
no more constructors
no more qi::locals
no more needless copying
using expectation points
In short: no more fuss.
#include <boost/spirit/include/qi.hpp>
#include <boost/fusion/adapted/struct.hpp>
#include <cstdint>
namespace qi = boost::spirit::qi;
namespace chs = boost::spirit::ascii; //qi;
typedef std::string::const_iterator StringIterator;
struct Color
{
float r,g,b,a;
};
BOOST_FUSION_ADAPT_STRUCT(Color, (float, r)(float, g)(float, b)(float, a))
template <typename ItType, typename Skipper>
struct ColorGrammar : public qi::grammar<StringIterator, Color(), Skipper>
{
ColorGrammar()
: ColorGrammar::base_type(rule_color, "color-grammar")
{
using namespace qi;
rule_rgb = lit("rgb") >> '(' > float_ > ',' > float_ > ',' > float_ > attr(1.0f) > ')';
rule_rgba = lit("rgba") >> '(' > float_ > ',' > float_ > ',' > float_ > ',' > float_ > ')';
rule_color = rule_rgb | rule_rgba;
}
private:
qi::uint_parser<uint8_t, 10, 1, 3> number; // unused
qi::uint_parser<uint8_t, 16, 1, 1> hexdigit; // unused
qi::rule<ItType, Color(), Skipper> rule_rgb, rule_rgba, rule_color;
};
int main()
{
Color out;
std::string str = " rgb ( 0.3 , .4 , 0.5 )";
StringIterator begin = str.cbegin();
StringIterator end = str.cend();
ColorGrammar<StringIterator, chs::space_type> color_;
bool result = qi::phrase_parse(begin, end, color_, chs::space, out);
std::cout << std::boolalpha << result << '\n';
std::cout << "remains: '" << std::string(begin, end) << "'\n";
}
Live on http://liveworkspace.org/code/35htD$3
I have a map of string-rule pairs and I would like to create a "joint rule"(rule_t joint_rule;) of them somehow. If I do this this way:
joint_rule = convert_logformat["%h"] >> convert_logformat["%t"];
than the joint rule with the parse_phrase matches the string
std::string entry = "127.0.0.1 [16/Aug/2012:01:50:02 +0000]";
But if I create the joint rule this way:
for (it = convert_logformat.begin(); it != convert_logformat.end(); it++)
{
joint_rule = joint_rule.copy() >> (*it).second.copy();
}
It does not match the same string. Why? How could I achieve something similar to the latter?
Relevant code:
template <typename Iterator>
bool parse_logentry(Iterator first, Iterator last, std::vector<char>& ip, std::vector<char>& timestamp, std::vector<char>& req, unsigned int& status, unsigned int& transferred_bytes, std::vector<char>& referer, std::vector<char>& ua)
{
using boost::spirit::qi::char_;
using boost::spirit::qi::int_;
using boost::spirit::qi::uint_;
using boost::spirit::qi::phrase_parse;
using boost::spirit::ascii::space;
using boost::spirit::ascii::space_type;
using boost::phoenix::ref;
using boost::phoenix::push_back;
using boost::spirit::qi::_1;
using boost::spirit::qi::lexeme;
using boost::spirit::qi::rule;
typedef boost::spirit::qi::rule<Iterator, std::string(), space_type> rule_t;
rule_t ip_rule, timestamp_rule, user_rule, req_rule, ref_rule, ua_rule, bytes_rule, status_rule;
ip_rule %= lexeme[(+char_("0-9."))[ref(ip) = _1]];
timestamp_rule %= lexeme[('[' >> +(~char_(']')) >> ']')[ref(timestamp) = _1]];
user_rule %= lexeme[(+~char_(" "))];
req_rule %= lexeme[('"' >> +(~char_('"')) >> '"')[ref(req) = _1]];
ref_rule %= lexeme[('"' >> +(~char_('"')) >> '"')[ref(referer) = _1]];
ua_rule %= lexeme[('"' >> +(~char_('"')) >> '"')[ref(ua) = _1]];
bytes_rule %= uint_[ref(transferred_bytes) = _1];
status_rule %= uint_[ref(status) = _1];
std::map<std::string, rule_t> convert_logformat;
typename std::map<std::string, rule_t>::iterator it;
convert_logformat.insert(std::pair<std::string, rule_t>("%h", ip_rule));
convert_logformat.insert(std::pair<std::string, rule_t>("%t", timestamp_rule));
//convert_logformat.insert(std::pair<std::string, rule_t>("%r", req_rule));
//convert_logformat.insert(std::pair<std::string, rule_t>("%>s", status_rule));
//convert_logformat.insert(std::pair<std::string, rule_t>("%b", bytes_rule));
//convert_logformat.insert(std::pair<std::string, rule_t>("%u", user_rule));
//convert_logformat.insert(std::pair<std::string, rule_t>("%{User-agent}i", ua_rule));
//convert_logformat.insert(std::pair<std::string, rule_t>("%{Referer}i", ref_rule));
rule_t joint_rule;
//joint_rule = convert_logformat["%h"] >> convert_logformat["%t"];
for (it = convert_logformat.begin(); it != convert_logformat.end(); it++)
{
joint_rule = joint_rule.copy() >> (*it).second.copy();
std::cout << (*it).first << ": " << typeid((*it).second).name() << "\n";
}
std::cout << "convert_logformath: " << typeid(convert_logformat["%h"]).name() << "\n";
bool r = phrase_parse(first, last, joint_rule, space);
if (first != last)
return false;
return r;
}
Ahem. It is really quite simple. You should initialize your variables :)
rule_t joint_rule; // what is it initialized to?
for (auto it = convert_logformat.begin(); it != convert_logformat.end(); it++)
{
joint_rule = joint_rule.copy() >> (*it).second.copy();
}
Change the first line to
rule_t joint_rule = qi::eps;
And it works:
sehe#mint12:/tmp$ ./test
127.0.0.1
16/Aug/2012:01:50:02 +0000
Your parser lacks some (good) common practice. See below for tidied up source (C++11).
Note that using a map to store the rules looks odd, because maps iteration will order by the key, not insertion order.
See the code live at http://liveworkspace.org/code/a7f2f94840d63fce43d8c3f56236330e
// #define BOOST_SPIRIT_DEBUG
#include <boost/spirit/include/qi.hpp>
#include <boost/spirit/include/phoenix.hpp>
#include <typeinfo>
namespace qi = boost::spirit::qi;
namespace phx = boost::phoenix;
template <typename Iterator>
struct Grammar : qi::grammar<Iterator, std::string(), qi::space_type>
{
Grammar() : Grammar::base_type(joint_rule)
{
using namespace qi;
ip_rule %= lexeme[ (+char_("0-9."))[phx::ref(ip) = _1] ];
timestamp_rule %= lexeme[ ('[' >> +(~char_(']')) >> ']')[phx::ref(timestamp) = _1] ];
user_rule %= lexeme[ (+~char_(" ")) ];
req_rule %= lexeme[ ('"' >> +(~char_('"')) >> '"')[phx::ref(req) = _1] ];
ref_rule %= lexeme[ ('"' >> +(~char_('"')) >> '"')[phx::ref(referer) = _1] ];
ua_rule %= lexeme[ ('"' >> +(~char_('"')) >> '"')[phx::ref(ua) = _1] ];
bytes_rule %= uint_[phx::ref(transferred_bytes) = _1 ];
status_rule %= uint_[phx::ref(status) = _1 ];
auto convert_logformat = std::map<std::string, rule_t> {
{ "%h" , ip_rule } ,
{ "%t" , timestamp_rule },
// { "%r" , req_rule } ,
// { "%>s" , status_rule } ,
// { "%b" , bytes_rule } ,
// { "%u" , user_rule } ,
// { "%{User-agent}i", ua_rule } ,
// { "%{Referer}i" , ref_rule }
};
joint_rule = eps;
for (auto const& p: convert_logformat)
{
joint_rule = joint_rule.copy() >> p.second.copy();
}
BOOST_SPIRIT_DEBUG_NODE(ip_rule);
BOOST_SPIRIT_DEBUG_NODE(timestamp_rule);
BOOST_SPIRIT_DEBUG_NODE(user_rule);
BOOST_SPIRIT_DEBUG_NODE(req_rule);
BOOST_SPIRIT_DEBUG_NODE(ref_rule);
BOOST_SPIRIT_DEBUG_NODE(ua_rule);
BOOST_SPIRIT_DEBUG_NODE(bytes_rule);
BOOST_SPIRIT_DEBUG_NODE(status_rule);
}
typedef qi::rule<Iterator, std::string(), qi::space_type> rule_t;
rule_t ip_rule, timestamp_rule, user_rule, req_rule, ref_rule, ua_rule, bytes_rule, status_rule;
rule_t joint_rule;
std::vector<char> ip;
std::vector<char> timestamp;
std::vector<char> req;
unsigned int status;
unsigned int transferred_bytes;
std::vector<char> referer;
std::vector<char> ua;
};
template <typename Iterator>
bool parse_logentry(Iterator first, Iterator last,
Grammar<Iterator>& parser)
{
bool r = phrase_parse(first, last, parser, qi::space);
return (r && (first == last));
}
int main(void)
{
std::string entry = "127.0.0.1 [16/Aug/2012:01:50:02 +0000]";
//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\"";
Grammar<std::string::iterator> parser;
if (parse_logentry(entry.begin(), entry.end(), parser))
{
for (auto i : parser.ip)
std::cout << i;
std::cout << "\n";
for (auto ts: parser.timestamp)
std::cout << ts;
std::cout << "\n";
}
else
{
std::cout << "not ok\n";
}
return 0;
}
Note that, among other things, this setup allows you to enable debugging of your grammar by simply defining BOOST_SPIRIT_DEBUG at the start.