Added ability to detect file format from sequence file without known extension.

Author: Daniel Mapleson

lib/include/kat/input_handler.hpp

@@ -67,6 +67,8 @@ namespace kat {
 
         static shared_ptr<vector<path>> globFiles(const string& input);
         static shared_ptr<vector<path>> globFiles(const vector<path>& input);
+        
+        static string determineSequenceFileType(const path& file);
 
     private:
         static int globerr(const char *path, int eerrno);

lib/src/input_handler.cc

@@ -20,7 +20,9 @@
 #endif
 
 #include <iostream>
+#include <fstream>
 #include <glob.h>
+using std::fstream;
 using std::stringstream;
 
 #include <boost/filesystem.hpp>
@@ -259,3 +261,37 @@ shared_ptr<vector<path>> kat::InputHandler::globFiles(const vector<path>& input)
 
     return globbed;
 }
+
+string kat::InputHandler::determineSequenceFileType(const path& filename) {
+    
+    string ext = filename.extension().string();
+
+    // Check extension first
+    if (boost::iequals(ext, ".fastq") || boost::iequals(ext, ".fq")) {
+        return "fastq";
+    }
+    else if (   boost::iequals(ext, ".fasta") ||
+                boost::iequals(ext, ".fa") ||
+                boost::iequals(ext, ".fna") ||
+                boost::iequals(ext, ".fas") ||
+                boost::iequals(ext, ".scafSeq")) {
+        return "fasta";
+    }
+    else {
+        // Now check first character of the file
+        char ch;
+        fstream fin(filename.string(), fstream::in);
+        fin >> ch;
+        fin.close();
+
+        if (ch == '>') {
+            return "fasta";
+        }
+        else if (ch == '@') {
+            return "fastq";
+        }
+    }
+
+    // If we've got this far then it's not obviously a sequence file we recognise.
+    BOOST_THROW_EXCEPTION(InputFileException() << InputFileErrorInfo("Unknown file type"));
+}

tests/check_jellyfish.cc

@@ -28,7 +28,9 @@ template<typename DtnType>
 inline double as_seconds(DtnType dtn) { return duration_cast<duration<double>>(dtn).count(); }
 
 #include <kat/jellyfish_helper.hpp>
+#include <kat/input_handler.hpp>
 using kat::JellyfishHelper;
+using kat::InputHandler;
 using kat::HashLoader;
 
 namespace kat {
@@ -209,4 +211,12 @@ TEST(jellyfish, unknownexttest) {
     EXPECT_EQ( res, true );
 }
 
+TEST(jellyfish, determineexttest) {
+    path unknownpath = path(DATADIR "/unknown.dat");
+    
+    string ft = InputHandler::determineSequenceFileType(unknownpath);
+    
+    EXPECT_EQ( ft, "fasta" );
+}
+
 }