fast_parser.cpp

#include <pybind11/pybind11.h>
#include <pybind11/numpy.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <unistd.h>

namespace py = pybind11;

struct MarketTick {
    uint64_t timestamp; 
    char symbol[8];     
    double price;       
};

class FastParser {
public:
    void* mapped_data;
    size_t filesize;
    int fd;

    FastParser(const std::string& filepath) {
        fd = open(filepath.c_str(), O_RDONLY);
        struct stat st;
        fstat(fd, &st);
        filesize = st.st_size;
        mapped_data = mmap(NULL, filesize, PROT_READ, MAP_PRIVATE, fd, 0);
    }

    ~FastParser() {
        munmap(mapped_data, filesize);
        close(fd);
    }

    py::array_t<double> get_prices() {
        size_t num_ticks = filesize / sizeof(MarketTick);
        
        return py::array_t<double>(
            { (ssize_t)num_ticks },            
            { (ssize_t)sizeof(MarketTick) },  
            (double*)((char*)mapped_data + 16),
            py::cast(this)                    
        );
    }
};

PYBIND11_MODULE(fast_parser, m) {
    py::class_<FastParser>(m, "FastParser")
        .def(py::init<const std::string&>())
        .def("get_prices", &FastParser::get_prices);
}