update fasta_format::do_scan() to read sequence portion of fasta file

include/boost/bioseq/io_formats/fasta.hpp

@@ -10,6 +10,8 @@
 #ifndef BOOST_BIOSEQ_IO_FORMATS_FASTA_HPP
 #define BOOST_BIOSEQ_IO_FORMATS_FASTA_HPP
 
+#include <iostream>
+
 namespace boost {
 namespace bioseq {
 
@@ -35,8 +37,31 @@ private:
 
     virtual void do_scan(std::basic_istream<CharT,Traits>& in, SeqT& s)
     {
+        auto c = in.peek();
+        if (c != '>')
+        {
+            in.setstate(std::ios::failbit);
+            return;
+        }
+
+        std::basic_string<CharT,Traits> line;
+        line.reserve(71);
+        std::getline(in, line);
+        std::list<std::basic_string<CharT,Traits>> lines;
+        c = in.peek();
+        while (c != '>' && std::getline(in, line))
+        {
+            lines.push_back(std::move(line));
+            c = in.peek();
+        }
+
+        std::cout << "Number of lines: " << lines.size() << std::endl;
+        for (auto line : lines)
+            s.insert(s.end(), line.begin(), line.end());
+
+        std::cout << "Sequence Size: " << s.size() << std::endl;
         // s.insert(0, "ACGT");
-        s.insert(s.begin(), 'A');
+        // s.insert(s.begin(), 'A');
     }
 };
 

test/test_io_fasta.cpp

@@ -26,16 +26,19 @@
 
 namespace bioseq = boost::bioseq;
 
+// preparing str_seq type
 template<class CharT, class AllocatorT> using bs_string
 =   std::basic_string<CharT, std::char_traits<CharT>, AllocatorT>;
 typedef bioseq::sequence_host<char, std::allocator, bs_string> str_seq;
 typedef bioseq::ioformat<char,std::char_traits<char>,str_seq,bioseq::fasta_format> frmt1;
 template<> struct bioseq::io_format_traits<str_seq> { typedef frmt1 type; };
 
+// preparing vec_seq type
 typedef bioseq::sequence_host<char, std::allocator, std::vector> vec_seq;
 typedef bioseq::ioformat<char,std::char_traits<char>,vec_seq,bioseq::fasta_format> frmt2;
 template<> struct bioseq::io_format_traits<vec_seq> { typedef frmt2 type; };
 
+// adding sequence types to the test list
 typedef boost::mpl::list<vec_seq, str_seq> seq_types;
 
 BOOST_AUTO_TEST_CASE_TEMPLATE(read_single_fasta, T, seq_types)
@@ -44,7 +47,13 @@ BOOST_AUTO_TEST_CASE_TEMPLATE(read_single_fasta, T, seq_types)
     T mt;
     in >> mt;
 
-    BOOST_CHECK_EQUAL(mt[0], 'A');
+    BOOST_CHECK_EQUAL(mt[0], 'G');
+    BOOST_CHECK_EQUAL(mt[1], 'A');
+    BOOST_CHECK_EQUAL(mt[69], 'G');
+    BOOST_CHECK_EQUAL(mt[139], 'C');
+    BOOST_CHECK_EQUAL(mt[16566], 'A');
+    BOOST_CHECK_EQUAL(mt[16567], 'T');
+    BOOST_CHECK_EQUAL(mt[16568], 'G');
 
     // BOOST_CHECK( mt == "ACGT" );