adding write_sam_header and faster to_string

Author: guilhermesena1

sam_record.cpp

@@ -34,6 +34,7 @@ using std::ostream;
 using std::begin;
 using std::end;
 using std::ostringstream;
+using std::to_string;
 
 // ADS: this is for debugging purposes
 string
@@ -73,42 +74,36 @@ format_sam_flags(const uint16_t the_flags) {
 //   return regex_match(qual, regex("[!-~]+"));
 // }
 
+size_t
+sam_rec::estimate_line_size() const {
+  static const size_t all_field_estimates = 100;
+  return qname.size() + rname.size() + qual.size() + all_field_estimates;
+}
+
 string
 sam_rec::tostring() const {
-  ostringstream oss;
-  oss << qname << '\t'
-      << flags << '\t'
-      << rname << '\t'
-      << pos << '\t'
-      << static_cast<unsigned>(mapq) << '\t'
-      << cigar << '\t'
-      << rnext << '\t'
-      << pnext << '\t'
-      << tlen << '\t'
-      << seq << '\t'
-      << qual;
-
+  string out;
+  out.reserve(estimate_line_size());
+  out.append(qname + "\t" +
+               to_string(flags) + "\t" +
+               rname + "\t" +
+               to_string(pos) + "\t" +
+               to_string(static_cast<unsigned>(mapq)) + "\t" +
+               cigar + "\t" +
+               rnext + "\t" +
+               to_string(pnext) + "\t" +
+               to_string(tlen) + "\t" +
+               seq + "\t" +
+               qual);
   for (auto it(begin(tags)); it != end(tags); ++it)
-    oss << '\t' << *it;
-  return oss.str() + "\n";
+    out.append("\t" + *it);
+
+  return out;
 }
 
 ostream &
 operator<<(std::ostream &the_stream, const sam_rec &r) {
-  the_stream << r.qname << '\t'
-             << r.flags << '\t'
-             << r.rname << '\t'
-             << r.pos << '\t'
-             << static_cast<unsigned>(r.mapq) << '\t'
-             << r.cigar << '\t'
-             << r.rnext << '\t'
-             << r.pnext << '\t'
-             << r.tlen << '\t'
-             << r.seq << '\t'
-             << r.qual;
-
-  for (auto it(begin(r.tags)); it != end(r.tags); ++it)
-    the_stream << '\t' << *it;
+  the_stream << r.tostring();
   return the_stream;
 }
 

sam_record.hpp

@@ -22,6 +22,8 @@
 #include <string>
 #include <vector>
 #include <iostream>
+#include <sstream>
+#include <iterator>
 
 // from 30 April 2020 SAM documentation
 // 1    0x1   template having multiple segments in sequencing
@@ -117,6 +119,7 @@ class sam_rec {
     seq(_seq),
     qual(_qual) {}
   void add_tag(const std::string &the_tag) {tags.push_back(the_tag);}
+  size_t estimate_line_size() const;
   std::string tostring() const;
 };
 
@@ -145,4 +148,36 @@ void
 inflate_with_cigar(const sam_rec &sr, std::string &to_inflate,
                    const char inflation_symbol = 'N');
 
+template<typename T>
+static void
+write_sam_header(const std::vector<std::string> &chrom_names,
+                 const std::vector<T> &chrom_starts,
+                 const std::string program_name,
+                 const std::string program_version,
+                 const int argc, const char **argv,
+                 std::ostream &out) {
+  static const std::string SAM_VERSION = "1.0";
+
+  // sam version
+  out <<"@HD" << '\t' << "VN:" << SAM_VERSION << '\n'; // sam version
+
+  // chromosome sizes
+  const size_t n_chroms = chrom_names.size() - 1;
+  for (size_t i = 1; i < n_chroms; ++i) {
+    out << "@SQ" << '\t'
+        << "SN:" << chrom_names[i] << '\t'
+        << "LN:" << chrom_starts[i+1] - chrom_starts[i] << '\n';
+  }
+
+  // program details
+  out << "@PG" << '\t'
+      << "ID:" << program_name << '\t'
+      << "VN:" << program_version << '\t';
+
+  // how the program was run
+  std::ostringstream the_command;
+  copy(argv, argv + argc, std::ostream_iterator<const char*>(the_command, " "));
+  out << "CL:\"" << the_command.str() << "\"" << std::endl;
+}
+
 #endif