C++ 精神语法解析问题
我有以下几点,C++ 精神语法解析问题,c++,boost-spirit,boost-spirit-qi,C++,Boost Spirit,Boost Spirit Qi,我有以下几点, class BATSTradeMsg : public BATSMessageBase { BATSTradeMsg(int timestamp, char msgtype, uint64_t orderId, char side, uint32_t shares, std::string const &symbol, uint64_t price, uint64_t execId) : BATSM
class BATSTradeMsg : public BATSMessageBase
{
BATSTradeMsg(int timestamp, char msgtype, uint64_t orderId, char side, uint32_t shares,
std::string const &symbol, uint64_t price, uint64_t execId) :
BATSMessageBase(timestamp, msgtype),
m_orderId(orderId),
m_side(side),
m_shares(shares),
m_symbol(symbol),
m_price(price),
m_execId(execId)
{
}
uint64_t m_orderId; // Base 36 Numeric values come over the wire in ascii
char m_side;
uint32_t m_shares;
std::string m_symbol;
uint64_t m_price;
uint64_t m_execId; // Base 36 Numeric values come over the wire in ascii
};
// order and execution ids are 12 characters base 36
qi::uint_parser< uint64_t, 36, 12, 12 > p_orderId;
qi::uint_parser< uint64_t, 36, 12, 12 > p_execId;
qi::uint_parser< uint32_t, 10, 6, 6 > p_shares;
qi::uint_parser< uint32_t, 10, 10, 10 > m_price;
qi::uint_parser< uint32_t, 10, 8, 8 > p_ts;
if (msgtype == BATSTradeMsg::longMsgCode)
m_wire_msg = ( p_ts >> qi::char_(msgtype)
>> p_orderId
>> qi::char_(BATSTradeMsg::sideFlag)
>> p_shares
>> qi::as_string[qi::repeat(8)[qi::char_]]
>> m_price
>> p_execId )
[qi::_val = phi::construct<BATSTradeMsg>(
qi::_1, qi::_2, qi::_3, qi::_4, qi::_5, qi::_6, qi::_7, qi::_8)];
else if ( msgtype == BATSTradeMsg::shortMsgCode )
m_wire_msg = ( p_ts >> qi::char_(msgtype)
>> p_orderId
>> qi::char_(BATSTradeMsg::sideFlag)
>> p_shares
>> qi::as_string[qi::repeat(6)[qi::char_]]
>> m_price
>> p_execId )
[qi::_val = phi::construct<BATSTradeMsg>(
qi::_1, qi::_2, qi::_3, qi::_4, qi::_5, qi::_6, qi::_7, qi::_8)];
以下是我的建议: 数据类型 然后使用简单的融合适应,而不是语义动作ª:
BOOST_FUSION_ADAPT_STRUCT(BATS::TradeMsg, timestamp, msgtype, orderId, side, shares, symbol, price, execId)
分析器
然后,解析器基本上变成:
我假设“side”可以是“B”或“S”(用于买入或卖出)
演示程序
下面是3个被剖析的测试用例:
int main() {
using It = std::string::const_iterator;
Parser<It> const parser;
for (std::string const input : {
"28800168P1K27GA00000YB000300AAPL 00018319001K27GA00000Z",
"28800168r1K27GA00000YB000300AAPLSPOT00018319001K27GA00000Z" })
{
std::cout << "Input: " << std::quoted(input) << "\n";
It f = begin(input), l = end(input);
BATS::TradeMsg msg;
if (parse(f, l, parser, msg)) {
std::cout << "Parsed\n";
} else {
std::cout << "Parse failed\n";
}
if (f!=l)
std::cout << "Remaining data: " << std::quoted(std::string(f,l), '\'') << "\n";
}
}
如果您提供一个完整(几乎?)的可编译示例(包含所需的输入和预期的结果),您将有更好的机会获得帮助。第一个示例消息中是否缺少字符?我们怎么知道
P
或r
是“短”还是“长”的代码?我们是否必须猜测“side”可以是“[B]uy”或“[S]ell”?很抱歉造成混淆,我不知道为什么,但上面带有“…AAPL00018…”的msg格式似乎只有一个空格,而不是正确的两个空格。让我把它改成代码吧,难怪大家都感到困惑。很抱歉。我已经搞定了。事实上完整的来源在这里。如果您查看测试用例test_parse_trade,您可以看到它检查BATS规范中的长MSG和短MSG是否正确解析。第一个测试用例parse(“2880016P1K27GA0000YB0000300AAPL 00018319001K27GA0000Z”)与第二个测试用例parse(“2880016R1K27GA0000YB0000300AAPLSPOT00018319001K27GA0000Z”)不同股票符号只需多出2个字符。因此,我的问题是,为什么我不能使用下面这样的OR构造(qi::as_string[qi::repeat(6)[qi::char_]]| qi::as_string[qi::repeat(8)[qi::char_]])来解析它们?让我们暂时撇开语义行为是否是邪恶的不谈……这不起作用,因为qi::repeat(6)[qi::char\uquo]
和qi::repeat(8)[qi::char\uquo]
通常都会成功。如果在a
和b
始终匹配时写入a | b
,则永远不会执行b
分支。
namespace BATS {
enum class MessageCode : char { Long = 'r', Short = 'P' };
struct MessageBase {
int timestamp;
MessageCode msgtype;
};
struct TradeMsg : MessageBase {
uint64_t orderId; // Base 36 Numeric values come over the wire in ascii
char side;
uint32_t shares;
std::string symbol;
uint64_t price;
uint64_t execId; // Base 36 Numeric values come over the wire in ascii
};
}
BOOST_FUSION_ADAPT_STRUCT(BATS::TradeMsg, timestamp, msgtype, orderId, side, shares, symbol, price, execId)
template <typename It>
struct Parser : qi::grammar<It, BATS::TradeMsg()> {
Parser() : Parser::base_type(r_wire_msg) {
// see below
r_wire_msg
= r_long_wire_msg
| r_short_wire_msg
;
BOOST_SPIRIT_DEBUG_NODES((r_wire_msg)(r_short_wire_msg)(r_long_wire_msg))
}
private:
// order and execution ids are 12 characters base 36
qi::uint_parser<uint64_t, 36, 12, 12> p_orderId;
qi::uint_parser<uint64_t, 36, 12, 12> p_execId;
qi::uint_parser<uint32_t, 10, 6, 6> p_shares;
qi::uint_parser<uint32_t, 10, 10, 10> p_price;
qi::uint_parser<uint32_t, 10, 8, 8> p_ts;
qi::rule<It, BATS::TradeMsg()> r_wire_msg, r_long_wire_msg, r_short_wire_msg;
};
r_long_wire_msg
= p_ts
>> qi::char_(BATS::MessageCode::Long)
>> p_orderId
>> qi::char_("BS")
>> p_shares
>> qi::as_string[qi::repeat(8)[qi::char_]]
>> p_price
>> p_execId
;
r_short_wire_msg
= p_ts
>> qi::char_(BATS::MessageCode::Short)
>> p_orderId
>> qi::char_("BS")
>> p_shares
>> qi::as_string[qi::repeat(6)[qi::char_]]
>> p_price
>> p_execId
;
int main() {
using It = std::string::const_iterator;
Parser<It> const parser;
for (std::string const input : {
"28800168P1K27GA00000YB000300AAPL 00018319001K27GA00000Z",
"28800168r1K27GA00000YB000300AAPLSPOT00018319001K27GA00000Z" })
{
std::cout << "Input: " << std::quoted(input) << "\n";
It f = begin(input), l = end(input);
BATS::TradeMsg msg;
if (parse(f, l, parser, msg)) {
std::cout << "Parsed\n";
} else {
std::cout << "Parse failed\n";
}
if (f!=l)
std::cout << "Remaining data: " << std::quoted(std::string(f,l), '\'') << "\n";
}
}
Input: "28800168P1K27GA00000YB000300AAPL 00018319001K27GA00000Z"
Parsed
Input: "28800168r1K27GA00000YB000300AAPLSPOT00018319001K27GA00000Z"
Parsed