#ifndef TOML11_PARSER_HPP #define TOML11_PARSER_HPP #include "result.hpp" #include "region.hpp" #include "combinator.hpp" #include "lexer.hpp" #include "types.hpp" #include "value.hpp" #include namespace toml { namespace detail { template result>, std::string> parse_boolean(location& loc) { const auto first = loc.iter(); if(const auto token = lex_boolean::invoke(loc)) { const auto reg = token.unwrap(); if (reg.str() == "true") {return ok(std::make_pair(true, reg));} else if(reg.str() == "false") {return ok(std::make_pair(false, reg));} else // internal error. { throw toml::internal_error(format_underline( "[error] toml::parse_boolean: internal error", reg, "invalid token")); } } loc.iter() = first; //rollback return err(format_underline("[error] toml::parse_boolean", loc, "token is not boolean", {"boolean is `true` or `false`"})); } template result>, std::string> parse_binary_integer(location& loc) { const auto first = loc.iter(); if(const auto token = lex_bin_int::invoke(loc)) { auto str = token.unwrap().str(); assert(str.size() > 2); // minimum -> 0b1 integer retval(0), base(1); for(auto i(str.rbegin()), e(str.rend() - 2); i!=e; ++i) { if (*i == '1'){retval += base; base *= 2;} else if(*i == '0'){base *= 2;} else if(*i == '_'){/* do nothing. */} else // internal error. { throw toml::internal_error(format_underline( "[error] toml::parse_integer: internal error", token.unwrap(), "invalid token")); } } return ok(std::make_pair(retval, token.unwrap())); } loc.iter() = first; return err(format_underline("[error] toml::parse_binary_integer", loc, "token is not binary integer", {"binary integer is like: 0b0011"})); } template result>, std::string> parse_octal_integer(location& loc) { const auto first = loc.iter(); if(const auto token = lex_oct_int::invoke(loc)) { auto str = token.unwrap().str(); str.erase(std::remove(str.begin(), str.end(), '_'), str.end()); str.erase(str.begin()); str.erase(str.begin()); // remove `0o` prefix std::istringstream iss(str); integer retval(0); iss >> std::oct >> retval; return ok(std::make_pair(retval, token.unwrap())); } loc.iter() = first; return err(format_underline("[error] toml::parse_octal_integer", loc, "token is not octal integer", {"octal integer is like: 0o775"})); } template result>, std::string> parse_hexadecimal_integer(location& loc) { const auto first = loc.iter(); if(const auto token = lex_hex_int::invoke(loc)) { auto str = token.unwrap().str(); str.erase(std::remove(str.begin(), str.end(), '_'), str.end()); str.erase(str.begin()); str.erase(str.begin()); // remove `0x` prefix std::istringstream iss(str); integer retval(0); iss >> std::hex >> retval; return ok(std::make_pair(retval, token.unwrap())); } loc.iter() = first; return err(format_underline("[error] toml::parse_hexadecimal_integer", loc, "token is not hex integer", {"hex integer is like: 0xC0FFEE"})); } template result>, std::string> parse_integer(location& loc) { const auto first = loc.iter(); if(first != loc.end() && *first == '0') { if(const auto bin = parse_binary_integer (loc)) {return bin;} if(const auto oct = parse_octal_integer (loc)) {return oct;} if(const auto hex = parse_hexadecimal_integer(loc)) {return hex;} // else, maybe just zero. } if(const auto token = lex_dec_int::invoke(loc)) { auto str = token.unwrap().str(); str.erase(std::remove(str.begin(), str.end(), '_'), str.end()); std::istringstream iss(str); integer retval(0); iss >> retval; return ok(std::make_pair(retval, token.unwrap())); } loc.iter() = first; return err(format_underline("[error] toml::parse_integer", loc, "token is not integer", {"integer is like: +42", "hex integer is like: 0xC0FFEE", "octal integer is like: 0o775", "binary integer is like: 0b0011"})); } template result>, std::string> parse_floating(location& loc) { const auto first = loc.iter(); if(const auto token = lex_float::invoke(loc)) { auto str = token.unwrap().str(); if(str == "inf" || str == "+inf") { if(std::numeric_limits::has_infinity) { return ok(std::make_pair( std::numeric_limits::infinity(), token.unwrap())); } else { throw std::domain_error("toml::parse_floating: inf value found" " but the current environment does not support inf. Please" " make sure that the floating-point implementation conforms" " IEEE 754/ISO 60559 international standard."); } } else if(str == "-inf") { if(std::numeric_limits::has_infinity) { return ok(std::make_pair( -std::numeric_limits::infinity(), token.unwrap())); } else { throw std::domain_error("toml::parse_floating: inf value found" " but the current environment does not support inf. Please" " make sure that the floating-point implementation conforms" " IEEE 754/ISO 60559 international standard."); } } else if(str == "nan" || str == "+nan") { if(std::numeric_limits::has_quiet_NaN) { return ok(std::make_pair( std::numeric_limits::quiet_NaN(), token.unwrap())); } else if(std::numeric_limits::has_signaling_NaN) { return ok(std::make_pair( std::numeric_limits::signaling_NaN(), token.unwrap())); } else { throw std::domain_error("toml::parse_floating: NaN value found" " but the current environment does not support NaN. Please" " make sure that the floating-point implementation conforms" " IEEE 754/ISO 60559 international standard."); } } else if(str == "-nan") { if(std::numeric_limits::has_quiet_NaN) { return ok(std::make_pair( -std::numeric_limits::quiet_NaN(), token.unwrap())); } else if(std::numeric_limits::has_signaling_NaN) { return ok(std::make_pair( -std::numeric_limits::signaling_NaN(), token.unwrap())); } else { throw std::domain_error("toml::parse_floating: NaN value found" " but the current environment does not support NaN. Please" " make sure that the floating-point implementation conforms" " IEEE 754/ISO 60559 international standard."); } } str.erase(std::remove(str.begin(), str.end(), '_'), str.end()); std::istringstream iss(str); floating v(0.0); iss >> v; return ok(std::make_pair(v, token.unwrap())); } loc.iter() = first; return err(format_underline("[error] toml::parse_floating: ", loc, "token is not a float", {"floating point is like: -3.14e+1"})); } template std::string read_utf8_codepoint(const region& reg) { const auto str = reg.str().substr(1); std::uint_least32_t codepoint; std::istringstream iss(str); iss >> std::hex >> codepoint; std::string character; if(codepoint < 0x80) // U+0000 ... U+0079 ; just an ASCII. { character += static_cast(codepoint); } else if(codepoint < 0x800) //U+0080 ... U+07FF { // 110yyyyx 10xxxxxx; 0x3f == 0b0011'1111 character += static_cast(0xC0| codepoint >> 6); character += static_cast(0x80|(codepoint & 0x3F)); } else if(codepoint < 0x10000) // U+0800...U+FFFF { // 1110yyyy 10yxxxxx 10xxxxxx character += static_cast(0xE0| codepoint >> 12); character += static_cast(0x80|(codepoint >> 6 & 0x3F)); character += static_cast(0x80|(codepoint & 0x3F)); } else if(codepoint < 0x200000) // U+10000 ... U+1FFFFF { if(0x10FFFF < codepoint) // out of Unicode region { std::cerr << format_underline(concat_to_string("[warning] " "input codepoint (", str, ") is too large to decode as " "a unicode character. The result may not be able to render " "to your screen."), reg, "should be in [0x00..0x10FFFF]") << std::endl; } // 11110yyy 10yyxxxx 10xxxxxx 10xxxxxx character += static_cast(0xF0| codepoint >> 18); character += static_cast(0x80|(codepoint >> 12 & 0x3F)); character += static_cast(0x80|(codepoint >> 6 & 0x3F)); character += static_cast(0x80|(codepoint & 0x3F)); } else // out of UTF-8 region { throw std::range_error(format_underline(concat_to_string("[error] " "input codepoint (", str, ") is too large to encode as utf-8."), reg, "should be in [0x00..0x10FFFF]")); } return character; } template result parse_escape_sequence(location& loc) { const auto first = loc.iter(); if(first == loc.end() || *first != '\\') { return err(format_underline("[error]: " "toml::parse_escape_sequence: location does not points \"\\\"", loc, "should be \"\\\"")); } ++loc.iter(); switch(*loc.iter()) { case '\\':{++loc.iter(); return ok(std::string("\\"));} case '"' :{++loc.iter(); return ok(std::string("\""));} case 'b' :{++loc.iter(); return ok(std::string("\b"));} case 't' :{++loc.iter(); return ok(std::string("\t"));} case 'n' :{++loc.iter(); return ok(std::string("\n"));} case 'f' :{++loc.iter(); return ok(std::string("\f"));} case 'r' :{++loc.iter(); return ok(std::string("\r"));} case 'u' : { if(const auto token = lex_escape_unicode_short::invoke(loc)) { return ok(read_utf8_codepoint(token.unwrap())); } else { return err(format_underline("[error] parse_escape_sequence: " "invalid token found in UTF-8 codepoint uXXXX.", loc, token.unwrap_err())); } } case 'U': { if(const auto token = lex_escape_unicode_long::invoke(loc)) { return ok(read_utf8_codepoint(token.unwrap())); } else { return err(format_underline("[error] parse_escape_sequence: " "invalid token found in UTF-8 codepoint Uxxxxxxxx", loc, token.unwrap_err())); } } } const auto msg = format_underline("[error] parse_escape_sequence: " "unknown escape sequence appeared.", loc, "escape sequence is one of" " \\, \", b, t, n, f, r, uxxxx, Uxxxxxxxx", {"if you want to write " "backslash as just one backslash, use literal string like:", "regex = '<\\i\\c*\\s*>'"}); loc.iter() = first; return err(msg); } template result>, std::string> parse_ml_basic_string(location& loc) { const auto first = loc.iter(); if(const auto token = lex_ml_basic_string::invoke(loc)) { auto inner_loc = loc; inner_loc.iter() = first; std::string retval; retval.reserve(token.unwrap().size()); auto delim = lex_ml_basic_string_delim::invoke(inner_loc); if(!delim) { throw internal_error(format_underline("[error] " "parse_ml_basic_string: invalid token", inner_loc, "should be \"\"\"")); } // immediate newline is ignored (if exists) /* discard return value */ lex_newline::invoke(inner_loc); delim = err("tmp"); while(!delim) { using lex_unescaped_seq = repeat< either, unlimited>; if(auto unescaped = lex_unescaped_seq::invoke(inner_loc)) { retval += unescaped.unwrap().str(); } if(auto escaped = parse_escape_sequence(inner_loc)) { retval += escaped.unwrap(); } if(auto esc_nl = lex_ml_basic_escaped_newline::invoke(inner_loc)) { // ignore newline after escape until next non-ws char } if(inner_loc.iter() == inner_loc.end()) { throw internal_error(format_underline("[error] " "parse_ml_basic_string: unexpected end of region", inner_loc, "not sufficient token")); } delim = lex_ml_basic_string_delim::invoke(inner_loc); } return ok(std::make_pair(toml::string(retval), token.unwrap())); } else { loc.iter() = first; return err(token.unwrap_err()); } } template result>, std::string> parse_basic_string(location& loc) { const auto first = loc.iter(); if(const auto token = lex_basic_string::invoke(loc)) { auto inner_loc = loc; inner_loc.iter() = first; auto quot = lex_quotation_mark::invoke(inner_loc); if(!quot) { throw internal_error(format_underline("[error] parse_basic_string: " "invalid token", inner_loc, "should be \"")); } std::string retval; retval.reserve(token.unwrap().size()); quot = err("tmp"); while(!quot) { using lex_unescaped_seq = repeat; if(auto unescaped = lex_unescaped_seq::invoke(inner_loc)) { retval += unescaped.unwrap().str(); } if(auto escaped = parse_escape_sequence(inner_loc)) { retval += escaped.unwrap(); } if(inner_loc.iter() == inner_loc.end()) { throw internal_error(format_underline("[error] " "parse_ml_basic_string: unexpected end of region", inner_loc, "not sufficient token")); } quot = lex_quotation_mark::invoke(inner_loc); } return ok(std::make_pair(toml::string(retval), token.unwrap())); } else { loc.iter() = first; // rollback return err(token.unwrap_err()); } } template result>, std::string> parse_ml_literal_string(location& loc) { const auto first = loc.iter(); if(const auto token = lex_ml_literal_string::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto open = lex_ml_literal_string_delim::invoke(inner_loc); if(!open) { throw internal_error(format_underline("[error] " "parse_ml_literal_string: invalid token", inner_loc, "should be '''")); } // immediate newline is ignored (if exists) /* discard return value */ lex_newline::invoke(inner_loc); const auto body = lex_ml_literal_body::invoke(inner_loc); const auto close = lex_ml_literal_string_delim::invoke(inner_loc); if(!close) { throw internal_error(format_underline("[error] " "parse_ml_literal_string: invalid token", inner_loc, "should be '''")); } return ok(std::make_pair( toml::string(body.unwrap().str(), toml::string_t::literal), token.unwrap())); } else { loc.iter() = first; // rollback return err(token.unwrap_err()); } } template result>, std::string> parse_literal_string(location& loc) { const auto first = loc.iter(); if(const auto token = lex_literal_string::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto open = lex_apostrophe::invoke(inner_loc); if(!open) { throw internal_error(format_underline("[error] " "parse_literal_string: invalid token", inner_loc, "should be '")); } const auto body = repeat::invoke(inner_loc); const auto close = lex_apostrophe::invoke(inner_loc); if(!close) { throw internal_error(format_underline("[error] " "parse_literal_string: invalid token", inner_loc, "should be '")); } return ok(std::make_pair( toml::string(body.unwrap().str(), toml::string_t::literal), token.unwrap())); } else { loc.iter() = first; // rollback return err(token.unwrap_err()); } } template result>, std::string> parse_string(location& loc) { if(const auto rslt = parse_ml_basic_string(loc)) {return rslt;} if(const auto rslt = parse_ml_literal_string(loc)) {return rslt;} if(const auto rslt = parse_basic_string(loc)) {return rslt;} if(const auto rslt = parse_literal_string(loc)) {return rslt;} return err(format_underline("[error] toml::parse_string: not a string", loc, "not a string")); } template result>, std::string> parse_local_date(location& loc) { const auto first = loc.iter(); if(const auto token = lex_local_date::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto y = lex_date_fullyear::invoke(inner_loc); if(!y || inner_loc.iter() == inner_loc.end() || *inner_loc.iter() != '-') { throw internal_error(format_underline("[error]: " "toml::parse_inner_local_date: invalid year format", inner_loc, y.map_err_or_else([](const std::string& msg) { return msg; }, "should be `-`"))); } ++inner_loc.iter(); const auto m = lex_date_month::invoke(inner_loc); if(!m || inner_loc.iter() == inner_loc.end() || *inner_loc.iter() != '-') { throw internal_error(format_underline("[error]: " "toml::parse_local_date: invalid month format", inner_loc, m.map_err_or_else([](const std::string& msg) { return msg; }, "should be `-`"))); } ++inner_loc.iter(); const auto d = lex_date_mday::invoke(inner_loc); if(!d) { throw internal_error(format_underline("[error]: " "toml::parse_local_date: invalid day format", inner_loc, d.unwrap_err())); } return ok(std::make_pair(local_date( static_cast(from_string(y.unwrap().str(), 0)), static_cast( static_cast(from_string(m.unwrap().str(), 0)-1)), static_cast(from_string(d.unwrap().str(), 0))), token.unwrap())); } else { auto msg = format_underline("[error]: toml::parse_local_date: " "invalid format", loc, token.unwrap_err(), {"local date is like: 1979-05-27"}); loc.iter() = first; return err(std::move(msg)); } } template result>, std::string> parse_local_time(location& loc) { const auto first = loc.iter(); if(const auto token = lex_local_time::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto h = lex_time_hour::invoke(inner_loc); if(!h || inner_loc.iter() == inner_loc.end() || *inner_loc.iter() != ':') { throw internal_error(format_underline("[error]: " "toml::parse_local_time: invalid year format", inner_loc, h.map_err_or_else([](const std::string& msg) { return msg; }, "should be `:`"))); } ++inner_loc.iter(); const auto m = lex_time_minute::invoke(inner_loc); if(!m || inner_loc.iter() == inner_loc.end() || *inner_loc.iter() != ':') { throw internal_error(format_underline("[error]: " "toml::parse_local_time: invalid month format", inner_loc, m.map_err_or_else([](const std::string& msg) { return msg; }, "should be `:`"))); } ++inner_loc.iter(); const auto s = lex_time_second::invoke(inner_loc); if(!s) { throw internal_error(format_underline("[error]: " "toml::parse_local_time: invalid second format", inner_loc, s.unwrap_err())); } local_time time( static_cast(from_string(h.unwrap().str(), 0)), static_cast(from_string(m.unwrap().str(), 0)), static_cast(from_string(s.unwrap().str(), 0)), 0, 0); const auto before_secfrac = inner_loc.iter(); if(const auto secfrac = lex_time_secfrac::invoke(inner_loc)) { auto sf = secfrac.unwrap().str(); sf.erase(sf.begin()); // sf.front() == '.' switch(sf.size() % 3) { case 2: sf += '0'; break; case 1: sf += "00"; break; case 0: break; default: break; } if(sf.size() >= 6) { time.millisecond = from_string(sf.substr(0, 3), 0); time.microsecond = from_string(sf.substr(3, 3), 0); } else if(sf.size() >= 3) { time.millisecond = from_string(sf, 0); time.microsecond = 0; } } else { if(before_secfrac != inner_loc.iter()) { throw internal_error(format_underline("[error]: " "toml::parse_local_time: invalid subsecond format", inner_loc, secfrac.unwrap_err())); } } return ok(std::make_pair(time, token.unwrap())); } else { auto msg = format_underline("[error]: toml::parse_local_time: " "invalid format", loc, token.unwrap_err(), {"local time is like: 00:32:00.999999"}); loc.iter() = first; return err(std::move(msg)); } } template result>, std::string> parse_local_datetime(location& loc) { const auto first = loc.iter(); if(const auto token = lex_local_date_time::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto date = parse_local_date(inner_loc); if(!date || inner_loc.iter() == inner_loc.end()) { throw internal_error(format_underline("[error]: " "toml::parse_local_datetime: invalid datetime format", inner_loc, date.map_err_or_else([](const std::string& msg){ return msg; }, "date, not datetime"))); } const char delim = *(inner_loc.iter()++); if(delim != 'T' && delim != 't' && delim != ' ') { throw internal_error(format_underline("[error]: " "toml::parse_local_datetime: invalid datetime format", inner_loc, "should be `T` or ` ` (space)")); } const auto time = parse_local_time(inner_loc); if(!time) { throw internal_error(format_underline("[error]: " "toml::parse_local_datetime: invalid datetime format", inner_loc, "invalid time fomrat")); } return ok(std::make_pair( local_datetime(date.unwrap().first, time.unwrap().first), token.unwrap())); } else { auto msg = format_underline("[error]: toml::parse_local_datetime: " "invalid format", loc, token.unwrap_err(), {"local datetime is like: 1979-05-27T00:32:00.999999"}); loc.iter() = first; return err(std::move(msg)); } } template result>, std::string> parse_offset_datetime(location& loc) { const auto first = loc.iter(); if(const auto token = lex_offset_date_time::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto datetime = parse_local_datetime(inner_loc); if(!datetime || inner_loc.iter() == inner_loc.end()) { throw internal_error(format_underline("[error]: " "toml::parse_offset_datetime: invalid datetime format", inner_loc, datetime.map_err_or_else([](const std::string& msg){ return msg; }, "date, not datetime"))); } time_offset offset(0, 0); if(const auto ofs = lex_time_numoffset::invoke(inner_loc)) { const auto str = ofs.unwrap().str(); if(str.front() == '+') { offset.hour = static_cast(from_string(str.substr(1,2), 0)); offset.minute = static_cast(from_string(str.substr(4,2), 0)); } else { offset.hour = -static_cast(from_string(str.substr(1,2), 0)); offset.minute = -static_cast(from_string(str.substr(4,2), 0)); } } else if(*inner_loc.iter() != 'Z' && *inner_loc.iter() != 'z') { throw internal_error(format_underline("[error]: " "toml::parse_offset_datetime: invalid datetime format", inner_loc, "should be `Z` or `+HH:MM`")); } return ok(std::make_pair(offset_datetime(datetime.unwrap().first, offset), token.unwrap())); } else { auto msg = format_underline("[error]: toml::parse_offset_datetime: " "invalid format", loc, token.unwrap_err(), {"offset datetime is like: 1979-05-27T00:32:00-07:00", "or in UTC (w/o offset) : 1979-05-27T00:32:00Z"}); loc.iter() = first; return err(std::move(msg)); } } template result parse_simple_key(location& loc) { if(const auto bstr = parse_basic_string(loc)) { return ok(bstr.unwrap().first.str); } if(const auto lstr = parse_literal_string(loc)) { return ok(lstr.unwrap().first.str); } if(const auto bare = lex_unquoted_key::invoke(loc)) { return ok(bare.unwrap().str()); } return err(format_underline("[error] toml::parse_simple_key: " "the next token is not a simple key", loc, "not a key")); } // dotted key become vector of keys template result, std::string> parse_key(location& loc) { const auto first = loc.iter(); // dotted key -> foo.bar.baz whitespaces are allowed if(const auto token = lex_dotted_key::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); std::vector keys; while(inner_loc.iter() != inner_loc.end()) { lex_ws::invoke(inner_loc); if(const auto k = parse_simple_key(inner_loc)) { keys.push_back(k.unwrap()); } else { throw internal_error(format_underline("[error] " "toml::detail::parse_key: dotted key contains invalid key", inner_loc, k.unwrap_err())); } lex_ws::invoke(inner_loc); if(inner_loc.iter() == inner_loc.end()) { break; } else if(*inner_loc.iter() == '.') { ++inner_loc.iter(); // to skip `.` } else { throw internal_error(format_underline("[error] toml::parse_key: " "dotted key contains invalid key ", inner_loc, "should be `.`")); } } return ok(keys); } loc.iter() = first; // simple key -> foo if(const auto smpl = parse_simple_key(loc)) { return ok(std::vector(1, smpl.unwrap())); } return err(format_underline("toml::parse_key: the next token is not a key", loc, "not a key")); } // forward-decl to implement parse_array and parse_table template result parse_value(location&); template result>, std::string> parse_array(location& loc) { const auto first = loc.iter(); if(loc.iter() == loc.end()) { return err("[error] toml::parse_array: input is empty"); } if(*loc.iter() != '[') { return err(format_underline("[error] toml::parse_array: " "token is not an array", loc, "should be [")); } ++loc.iter(); using lex_ws_comment_newline = repeat< either, unlimited>; array retval; while(loc.iter() != loc.end()) { lex_ws_comment_newline::invoke(loc); // skip if(loc.iter() != loc.end() && *loc.iter() == ']') { ++loc.iter(); // skip ']' return ok(std::make_pair(retval, region(loc, first, loc.iter()))); } if(auto val = parse_value(loc)) { if(!retval.empty() && retval.front().type() != val.as_ok().type()) { throw syntax_error(format_underline( "[error] toml::parse_array: type of elements should be the " "same each other.", region(loc, first, loc.iter()), "inhomogenous types")); } retval.push_back(std::move(val.unwrap())); } else { return err(val.unwrap_err()); } using lex_array_separator = sequence, character<','>>; const auto sp = lex_array_separator::invoke(loc); if(!sp) { lex_ws_comment_newline::invoke(loc); if(loc.iter() != loc.end() && *loc.iter() == ']') { ++loc.iter(); // skip ']' return ok(std::make_pair(retval, region(loc, first, loc.iter()))); } else { return err(format_underline("[error] toml::parse_array: " "missing array separator `,`", loc, "should be `,`")); } } } loc.iter() = first; return err(format_underline("[error] toml::parse_array: " "array did not closed by `]`", loc, "should be closed")); } template result, value>, std::string> parse_key_value_pair(location& loc) { const auto first = loc.iter(); auto key = parse_key(loc); if(!key) { std::string msg = std::move(key.unwrap_err()); // if the next token is keyvalue-separator, it means that there are no // key. then we need to show error as "empty key is not allowed". if(const auto keyval_sep = lex_keyval_sep::invoke(loc)) { loc.iter() = first; msg = format_underline("[error] toml::parse_key_value_pair: " "empty key is not allowed.", loc, "key expected before '='"); } return err(std::move(msg)); } const auto kvsp = lex_keyval_sep::invoke(loc); if(!kvsp) { std::string msg; // if the line contains '=' after the invalid sequence, possibly the // error is in the key (like, invalid character in bare key). const auto line_end = std::find(loc.iter(), loc.end(), '\n'); if(std::find(loc.iter(), line_end, '=') != line_end) { msg = format_underline("[error] toml::parse_key_value_pair: " "invalid format for key", loc, "invalid character in key", { "Did you forget '.' to separate dotted-key?", "Allowed characters for bare key are [0-9a-zA-Z_-]."}); } else // if not, the error is lack of key-value separator. { msg = format_underline("[error] toml::parse_key_value_pair: " "missing key-value separator `=`", loc, "should be `=`"); } loc.iter() = first; return err(std::move(msg)); } auto val = parse_value(loc); if(!val) { loc.iter() = first; return err(val.unwrap_err()); } return ok(std::make_pair(std::move(key.unwrap()), std::move(val.unwrap()))); } // for error messages. template std::string format_dotted_keys(InputIterator first, const InputIterator last) { static_assert(std::is_same::value_type>::value,""); std::string retval(*first++); for(; first != last; ++first) { retval += '.'; retval += *first; } return retval; } template result insert_nested_key(table& root, const toml::value& v, InputIterator iter, const InputIterator last, const bool is_array_of_table = false) { static_assert(std::is_same::value_type>::value,""); const auto first = iter; assert(iter != last); table* tab = std::addressof(root); for(; iter != last; ++iter) // search recursively { const key& k = *iter; if(std::next(iter) == last) // k is the last key { // XXX if the value is array-of-tables, there can be several // tables that are in the same array. in that case, we need to // find the last element and insert it to there. if(is_array_of_table) { if(tab->count(k) == 1) // there is already an array of table { if(!(tab->at(k).is(value_t::Array))) { throw syntax_error("toml::detail::insert_nested_key: " "target is not an array of table: " + format_dotted_keys(first, last)); } array& a = tab->at(k).template cast(); if(!(a.front().is(value_t::Table))) { throw syntax_error("toml::detail::insert_nested_key: " "target is not an array of table: " + format_dotted_keys(first, last)); } a.push_back(v); return ok(true); } else // if not, we need to create the array of table { array aot(1, v); // array having one table tab->insert(std::make_pair(k, value(aot))); return ok(true); } } if(tab->count(k) == 1) { throw syntax_error("[error] toml::detail::insert_nested_key: " "while inserting value to table: value already exists. " + format_dotted_keys(first, last)); } tab->insert(std::make_pair(k, v)); return ok(true); } else { // if there is no corresponding value, insert it first. if(tab->count(k) == 0) {(*tab)[k] = table{};} // type checking... if(tab->at(k).is(value_t::Table)) { tab = std::addressof((*tab)[k].template cast()); } else if(tab->at(k).is(value_t::Array)) // array-of-table case { array& a = (*tab)[k].template cast(); if(!a.back().is(value_t::Table)) { throw syntax_error("toml::detail::insert_nested_key: value " "is not a table but an array: " + format_dotted_keys(first, last)); } tab = std::addressof(a.back().template cast()); } else { throw syntax_error("toml::detail::insert_nested_key: value " "is not a table but an array: " + format_dotted_keys(first, last)); } } } return err(std::string("toml::detail::insert_nested_key: never reach here")); } template result>, std::string> parse_inline_table(location& loc) { const auto first = loc.iter(); table retval; if(!(loc.iter() != loc.end() && *loc.iter() == '{')) { return err(format_underline("[error] toml::parse_inline_table: " "the next token is not an inline table", loc, "not `{`.")); } ++loc.iter(); while(loc.iter() != loc.end()) { maybe::invoke(loc); if(loc.iter() != loc.end() && *loc.iter() == '}') { ++loc.iter(); // skip `}` return ok(std::make_pair( retval, region(loc, first, loc.iter()))); } const auto kv_r = parse_key_value_pair(loc); if(!kv_r) { return err(kv_r.unwrap_err()); } const std::vector& keys = kv_r.unwrap().first; const value& val = kv_r.unwrap().second; const auto inserted = insert_nested_key(retval, val, keys.begin(), keys.end()); if(!inserted) { throw internal_error("[error] toml::parse_inline_table: " "failed to insert value into table: " + inserted.unwrap_err()); } using lex_table_separator = sequence, character<','>>; const auto sp = lex_table_separator::invoke(loc); if(!sp) { maybe::invoke(loc); if(loc.iter() != loc.end() && *loc.iter() == '}') { ++loc.iter(); // skip `}` return ok(std::make_pair( retval, region(loc, first, loc.iter()))); } else { return err(format_underline("[error] toml:::parse_inline_table:" " missing table separator `,` ", loc, "should be `,`")); } } } loc.iter() = first; return err(format_underline("[error] toml::parse_inline_table: " "inline table did not closed by `}`", loc, "should be closed")); } template result parse_value(location& loc) { const auto first = loc.iter(); if(first == loc.end()) { return err(std::string("[error] toml::parse_value: input is empty")); } if(auto r = parse_string (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_array (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_inline_table (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_boolean (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_offset_datetime(loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_local_datetime (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_local_date (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_local_time (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_floating (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} if(auto r = parse_integer (loc)) {return ok(value(std::move(r.unwrap().first), std::move(r.unwrap().second)));} const auto msg = format_underline("[error] toml::parse_value: " "unknown token appeared", loc, "unknown"); loc.iter() = first; return err(msg); } template result, region>, std::string> parse_table_key(location& loc) { if(auto token = lex_std_table::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto open = lex_std_table_open::invoke(inner_loc); if(!open || inner_loc.iter() == inner_loc.end()) { throw internal_error(format_underline("[error] " "toml::parse_table_key: no `[`", inner_loc, "should be `[`")); } // to skip [ a . b . c ] // ^----------- this whitespace lex_ws::invoke(inner_loc); const auto keys = parse_key(inner_loc); if(!keys) { throw internal_error(format_underline("[error] " "toml::parse_table_key: invalid key", inner_loc, "not key")); } // to skip [ a . b . c ] // ^-- this whitespace lex_ws::invoke(inner_loc); const auto close = lex_std_table_close::invoke(inner_loc); if(!close) { throw internal_error(format_underline("[error] " "toml::parse_table_key: no `]`", inner_loc, "should be `]`")); } return ok(std::make_pair(keys.unwrap(), token.unwrap())); } else { return err(token.unwrap_err()); } } template result, region>, std::string> parse_array_table_key(location& loc) { if(auto token = lex_array_table::invoke(loc)) { location inner_loc(loc.name(), token.unwrap().str()); const auto open = lex_array_table_open::invoke(inner_loc); if(!open || inner_loc.iter() == inner_loc.end()) { throw internal_error(format_underline("[error] " "toml::parse_array_table_key: no `[[`", inner_loc, "should be `[[`")); } lex_ws::invoke(inner_loc); const auto keys = parse_key(inner_loc); if(!keys) { throw internal_error(format_underline("[error] " "toml::parse_array_table_key: invalid key", inner_loc, "not key")); } lex_ws::invoke(inner_loc); const auto close = lex_array_table_close::invoke(inner_loc); if(!close) { throw internal_error(format_underline("[error] " "toml::parse_table_key: no `]]`", inner_loc, "should be `]]`")); } return ok(std::make_pair(keys.unwrap(), token.unwrap())); } else { return err(token.unwrap_err()); } } // parse table body (key-value pairs until the iter hits the next [tablekey]) template result parse_ml_table(location& loc) { const auto first = loc.iter(); if(first == loc.end()) { return err(std::string("toml::parse_ml_table: input is empty")); } // XXX at lest one newline is needed using skip_line = repeat< sequence, maybe, lex_newline>, at_least<1>>; skip_line::invoke(loc); table tab; while(loc.iter() != loc.end()) { lex_ws::invoke(loc); const auto before = loc.iter(); if(const auto tmp = parse_array_table_key(loc)) // next table found { loc.iter() = before; return ok(tab); } if(const auto tmp = parse_table_key(loc)) // next table found { loc.iter() = before; return ok(tab); } if(const auto kv = parse_key_value_pair(loc)) { const std::vector& keys = kv.unwrap().first; const value& val = kv.unwrap().second; const auto inserted = insert_nested_key(tab, val, keys.begin(), keys.end()); if(!inserted) { return err(inserted.unwrap_err()); } } else { return err(kv.unwrap_err()); } const auto newline = skip_line::invoke(loc); if(!newline && loc.iter() != loc.end()) { const auto before = loc.iter(); lex_ws::invoke(loc); // skip whitespace const auto msg = format_underline("[error] toml::parse_table: " "invalid line format", loc, concat_to_string( "expected newline, but got '", show_char(*loc.iter()), "'.")); loc.iter() = before; return err(msg); } // comment lines are skipped by the above function call. // However, if the file ends with comment without newline, // it might cause parsing error because skip_line matches // `comment + newline`, not `comment` itself. to skip the // last comment, call lex_comment one more time. lex_comment::invoke(loc); } return ok(tab); } template result parse_toml_file(location& loc) { const auto first = loc.iter(); if(first == loc.end()) { return err(std::string("toml::detail::parse_toml_file: input is empty")); } table data; /* root object is also table, but without [tablename] */ if(auto tab = parse_ml_table(loc)) { data = std::move(tab.unwrap()); } else // failed (empty table is regarded as success in parse_ml_table) { return err(tab.unwrap_err()); } while(loc.iter() != loc.end()) { // here, the region of [table] is regarded as the table-key because // the table body is normally too big and it is not so informative // if the first key-value pair of the table is shown in the error // message. if(const auto tabkey = parse_array_table_key(loc)) { const auto tab = parse_ml_table(loc); if(!tab){return err(tab.unwrap_err());} const auto& keys = tabkey.unwrap().first; const auto& reg = tabkey.unwrap().second; const auto inserted = insert_nested_key(data, toml::value(tab.unwrap(), reg), keys.begin(), keys.end(), /*is_array_of_table=*/ true); if(!inserted) {return err(inserted.unwrap_err());} continue; } if(const auto tabkey = parse_table_key(loc)) { const auto tab = parse_ml_table(loc); if(!tab){return err(tab.unwrap_err());} const auto& keys = tabkey.unwrap().first; const auto& reg = tabkey.unwrap().second; const auto inserted = insert_nested_key(data, toml::value(tab.unwrap(), reg), keys.begin(), keys.end()); if(!inserted) {return err(inserted.unwrap_err());} continue; } return err(format_underline("[error]: toml::parse_toml_file: " "unknown line appeared", loc, "unknown format")); } return ok(data); } } // detail inline table parse(std::istream& is, std::string fname = "unknown file") { const auto beg = is.tellg(); is.seekg(0, std::ios::end); const auto end = is.tellg(); const auto fsize = end - beg; is.seekg(beg); // read whole file as a sequence of char std::vector letters(fsize); is.read(letters.data(), fsize); detail::location> loc(std::move(fname), std::move(letters)); // skip BOM if exists. // XXX component of BOM (like 0xEF) exceeds the representable range of // signed char, so on some (actually, most) of the environment, these cannot // be compared to char. However, since we are always out of luck, we need to // check our chars are equivalent to BOM. To do this, first we need to // convert char to unsigned char to guarantee the comparability. if(letters.size() >= 3) { std::array BOM; std::memcpy(BOM.data(), letters.data(), 3); if(BOM[0] == 0xEF && BOM[1] == 0xBB && BOM[2] == 0xBF) { loc.iter() += 3; // BOM found. skip. } } const auto data = detail::parse_toml_file(loc); if(!data) { throw syntax_error(data.unwrap_err()); } return data.unwrap(); } inline table parse(const std::string& fname) { std::ifstream ifs(fname.c_str()); if(!ifs.good()) { throw std::runtime_error("toml::parse: file open error -> " + fname); } return parse(ifs, fname); } } // toml #endif// TOML11_PARSER_HPP