make regex static and initliaze once

Author: Gaura

include/SingleCellProtocols.hpp

@@ -223,7 +223,7 @@ namespace alevin{
 
     // Custom geometry specification using regex for extraction
     struct CustomGeo {
-      // store regex for reads 1 and 2
+      // store regex string for reads 1 and 2
       static std::string reg[2];
       // store positions of matches for bc and umi
       static std::vector<int> b[2], u[2];
@@ -237,6 +237,8 @@ namespace alevin{
       boost::smatch match[2];
       // store the success of regex search
       bool rgx_search[2];
+      // store the regex
+      static boost::regex rgx[2];
       // required
       static uint32_t barcodeLength, umiLength;
       BarcodeEnd end;

src/AlevinUtils.cpp

@@ -445,9 +445,8 @@ namespace alevin {
                                      std::string& bc){
       std::string barcode="";
       for(int r=0; r < 2; r++) {
-        boost::regex rgx(pt.reg[r]);
-        pt.rgx_search[r] = (r == 0) ? boost::regex_search(read1,pt.match[r],rgx) :
-          boost::regex_search(read2,pt.match[r],rgx); // using std::string instead of read1/2 results in blank umi. strange!
+        pt.rgx_search[r] = (r == 0) ? boost::regex_search(read1,pt.match[r],pt.rgx[r]) :
+          boost::regex_search(read2,pt.match[r],pt.rgx[r]); // using std::string instead of read1/2 results in blank umi. strange!
         if(!pt.b[r].empty()) { // if read r has barcode
           if(pt.rgx_search[r]){ // if rgx search was successful
             for(int i : pt.b[r]) {
@@ -1473,7 +1472,8 @@ namespace alevin {
         }
         // std::cout << "reg[0]: " << customGeo.reg[0] << std::endl;
         // std::cout << "reg[1]: " << customGeo.reg[1] << std::endl;
-
+        customGeo.rgx[0] = customGeo.reg[0];
+        customGeo.rgx[1] = customGeo.reg[1];
         // validate that BC and UMI lengths are OK
         uint32_t maxBC{31};
         uint32_t maxUMI{31};

src/SingleCellProtocols.cpp

@@ -38,6 +38,6 @@ namespace alevin{
   extern uint32_t CustomGeo::barcodeLength, CustomGeo::umiLength;
   // bool alevin::protocols::CustomGeo::bioReadFound;
   extern bool CustomGeo::bioReadFound;
-
+  extern boost::regex CustomGeo::rgx[2];
   }// protocols
 }//alevin