update

CMakeLists.txt

@@ -26,7 +26,7 @@ foreach(SRC_CPP IN LISTS SRC_CPPS)
     message("found cpp : ${SRC_CPP}")
 endforeach()
 
-add_executable(${PROJECT_NAME}_bin ${SRC_HPPS} ${SRC_CPPS})
+# add_executable(${PROJECT_NAME}_bin ${SRC_HPPS} ${SRC_CPPS})
 
 add_subdirectory(pybind11)
 

dna.pyi

@@ -0,0 +1,11 @@
+# void convert_from_file(
+#     std::string source_file=std::string("filteredReads.txt"),
+#     std::string destination_file=std::string("reversedSequence.txt"),
+#     size_t buffer_size_gb=4
+# )
+
+def convert_from_file(
+        source_file:str="filteredReads.txt",
+        destination_file:str="reversedSequence.txt",
+        buffer_size_gb:int=4
+    )->None:...

src/dna.hpp

@@ -1 +1,37 @@
-#pragma once
+#pragma once
+#include <iostream>
+#include <sstream>
+#include <string>
+#include <utility>
+#include <chrono>
+
+template<class ...Args>
+std::string fmt(Args&&...args){
+    std::ostringstream oss;
+    (oss<<...<<std::forward<Args>(args));
+    return std::move(std::string(oss.str()));
+}
+
+template<class ...Args>
+void print(Args...args){
+    std::ostringstream oss;
+    (oss<<...<<std::forward<Args>(args));
+    std::cout<<oss.str();
+}
+
+#define NVC(v)#v," : ",(v)
+
+class Spent{ // 使用RAII原理的自动计时器，计算主函数运行时间，析构时自动输出
+    private:
+        const decltype(std::chrono::system_clock::now()) start;
+        const std::string name;
+    public:
+        Spent(const std::string name)noexcept:start(std::chrono::system_clock::now()),name(name){
+            print("[Timer: ",name,"]"," Start timing","\n");
+        }
+        ~Spent()noexcept{
+            const auto end = std::chrono::system_clock::now();
+            const auto dur = std::chrono::duration_cast<std::chrono::milliseconds> (end-start);
+            print("[Timer: ",name,"]"," Stop timing , used ", dur.count(),"ms\n");
+        }
+    };

src/main.cpp

@@ -1,8 +1,92 @@
 #include "dna.hpp"
+#include <array>
+#include <cstddef>
+#include <cstring>
+#include <fstream>
 #include <iostream>
-#include 
+#include <istream>
+#include <pybind11/cast.h>
+#include <pybind11/pybind11.h>
+#include <pybind11/stl.h>
+#include <stdexcept>
+#include <string>
 
-int main(){
+namespace py = pybind11;
-    
+static constexpr size_t default_buffer_size{512*1024*1024};
+
+static const std::unordered_map<char, char> complement = { //这里使用查表的方式大大提高CPU速度，因为if分支CPU不容易命中缓存，需要使用查表加速
+    {'A', 'T'}, {'a', 'T'},
+    {'T', 'A'}, {'t', 'A'},
+    {'C', 'G'}, {'c', 'G'},
+    {'G', 'C'}, {'g', 'C'}
+};
+
+void reverseComplement(char *begin, char *end) 
+{
+    //注意end是开区间，不能访问end
+    std::reverse(begin, end); //翻转DNA序列
+
+    for (ptrdiff_t i = 0; i < (end - begin); ++i) {
+        // static int _ = (zt::print(NAME_VALUE(omp_get_num_threads()),"\n"),0); // 打印线程数量
+        auto it = complement.find(begin[i]);
+        if (it != complement.end()) {
+            begin[i] = it->second;
+        }
+    }
 }
 
+void convert_from_file(
+    std::string source_file=std::string("filteredReads.txt"),
+    std::string destination_file=std::string("reversedSequence.txt"),
+    size_t buffer_size=default_buffer_size
+)
+{
+    // std::iostream::sync_with_stdio(false);
+
+    const size_t max_size_pre_dna{(size_t)5e4+5};
+    const size_t all_buf_size = {buffer_size};
+    const size_t read_bufsize{all_buf_size/2},write_bufsize{all_buf_size/2}; 
+
+    print(NVC(max_size_pre_dna),'\n',
+            NVC(all_buf_size),'\n',
+            NVC(read_bufsize),'\n',
+            NVC(write_bufsize),'\n');   
+    std::ifstream ifs(source_file);
+    if(ifs.is_open()==false)throw std::runtime_error(fmt("Cannot open input file stream\nfilename: ",source_file,'\n'));
+
+    std::ofstream ofs(destination_file);
+    if(ofs.is_open()==false)throw std::runtime_error(fmt("Cannot open output file stream\nfilename: ",destination_file,'\n'));
+    std::cout<<"Open file ok ,getting memory\n";
+    std::vector<char> read_buf(read_bufsize), write_buf(write_bufsize);
+    ifs.rdbuf()->pubsetbuf(read_buf.data(), read_buf.size());
+    ofs.rdbuf()->pubsetbuf(write_buf.data(), write_buf.size());
+
+    std::array<char, max_size_pre_dna> dna_buf;
+    bool is_dna_line{false};
+    std::cout<<"computing\n";
+    Spent all_spent("all_spent_time");
+    while(ifs.getline(dna_buf.data(),dna_buf.size())){
+        const size_t new_buflen{strlen(dna_buf.data())};
+        if(is_dna_line){
+            reverseComplement(dna_buf.data(), dna_buf.data()+new_buflen);
+            // std::cout<<"complete one ok\n";
+        }
+        // print(NVC(new_buflen),'\n');
+        // print(NVC(dna_buf.data()));
+        dna_buf[new_buflen]='\n';
+        ofs.write(dna_buf.data(),new_buflen+1);
+        is_dna_line=!is_dna_line;
+    }
+    ofs.flush();
+    print("done\n");
+}
+
+PYBIND11_MODULE(dna, m) {
+    m.doc() = "A dna base conversion library"; // optional module docstring
+
+    m.def("convert_from_file", &convert_from_file, "dna base switching from file.",
+           py::arg("source_file")= std::string("filteredReads.txt"),
+           py::arg("destination_file")=std::string("reversedSequence.txt"),
+           py::arg("buffer_size")=default_buffer_size
+        );
+}