More robust eol handling for Sequence::load_fasta

[mussa.git] / alg / sequence.cpp

diff --git a/alg/sequence.cpp b/alg/sequence.cpp
index 3ff96fc08b9cba1e73a2c4aa810a359e2512c853..90b48052d14fe617c49248a7292e02e575f83828 100644 (file)
--- a/alg/sequence.cpp
+++ b/alg/sequence.cpp
@@ -112,6 +112,23 @@ std::ostream& operator<<(std::ostream& out, const Sequence& seq)
   return out;
 }
 
+static void multiplatform_getline(std::istream& in, std::string& line)
+{
+  line.clear();
+  char c;
+  in.get(c);
+  while(in.good() and !(c == '\012' or c == '\015') ) {
+    line.push_back(c);
+    in.get(c);
+  }
+  // if we have cr-lf eat it
+  c = in.peek();
+  if (c=='\012' or c == '\015') {
+    in.get();
+  }
+ 
+}
+
 //! load a fasta file into a sequence
 /*! 
  * \param file_path the location of the fasta file in the filesystem
@@ -120,11 +137,39 @@ std::ostream& operator<<(std::ostream& out, const Sequence& seq)
  * \param end_index ending position in the fasta sequence, 0 for end
  * \return error message, empty string if no error. (gag!)
  */
+void Sequence::load_fasta(fs::path file_path, int seq_num,
+                          int start_index, int end_index)
+{
+  fs::fstream data_file;
+  data_file.open(file_path, std::ios::in);
+
+  if (!data_file.good())
+  {    
+    throw mussa_load_error("Sequence File: "+file_path.string()+" not found"); 
+  } else {
+    try {
+      load_fasta(data_file, seq_num, start_index, end_index);
+    } catch(sequence_empty_error e) {
+      // there doesn't appear to be any sequence
+      // catch and rethrow to include the filename
+      std::stringstream msg;
+      msg << "The selected sequence in " 
+          << file_path.native_file_string()
+          << " appears to be empty"; 
+      throw sequence_empty_error(msg.str());
+    } catch(sequence_empty_file_error e) {
+      std::stringstream errormsg;
+      errormsg << file_path.native_file_string()
+               << " did not have any fasta sequences" << std::endl;
+      throw sequence_empty_file_error(errormsg.str());
+    }
+  }
+}
+
 void
-Sequence::load_fasta(fs::path file_path, int seq_num, 
+Sequence::load_fasta(std::iostream& data_file, int seq_num, 
                      int start_index, int end_index)
 {
-  fs::fstream data_file;
   std::string file_data_line;
   int header_counter = 0;
   bool read_seq = true;
@@ -132,61 +177,45 @@ Sequence::load_fasta(fs::path file_path, int seq_num,
   std::string sequence_raw;
   std::string seq_tmp;             // holds sequence during basic filtering
 
-  data_file.open(file_path, std::ios::in);
-
   if (seq_num == 0) {
     throw mussa_load_error("fasta sequence number is 1 based (can't be 0)");
   }
-  if (!data_file)
-  {    
-    throw mussa_load_error("Sequence File: " + file_path.string() + " not found"); 
-  }
-  // if file opened okay, read it
-  else
+
+  // search for the header of the fasta sequence we want
+  while ( (!data_file.eof()) && (header_counter < seq_num) )
   {
-    // search for the header of the fasta sequence we want
-    while ( (!data_file.eof()) && (header_counter < seq_num) )
-    {
-      getline(data_file,file_data_line);
-      if (file_data_line.substr(0,1) == ">")
-        header_counter++;
-    }
+    multiplatform_getline(data_file, file_data_line);
+    if (file_data_line.substr(0,1) == ">")
+      header_counter++;
+  }
 
-    if (header_counter > 0) {
-      header = file_data_line.substr(1);
+  if (header_counter > 0) {
+    header = file_data_line.substr(1);
 
-      sequence_raw = "";
+    sequence_raw = "";
 
-      while ( !data_file.eof() && read_seq ) {
-        getline(data_file,file_data_line);
-        if (file_data_line.substr(0,1) == ">")
-          read_seq = false;
-        else sequence_raw += file_data_line;
-      }
+    while ( !data_file.eof() && read_seq ) {
+      multiplatform_getline(data_file,file_data_line);
+      if (file_data_line.substr(0,1) == ">")
+        read_seq = false;
+      else sequence_raw += file_data_line;
+    }
 
-      // Lastly, if subselection of the sequence was specified we keep cut out
-      // and only keep that part
-      // end_index = 0 means no end was specified, so cut to the end 
-      if (end_index == 0)
-        end_index = sequence_raw.size();
-
-      // sequence filtering for upcasing agctn and convert non AGCTN to N
-      if (end_index-start_index <= 0) {
-        // there doesn't appear to be any sequence
-        std::stringstream msg;
-        msg << "The selected sequence in " 
-            << file_path.native_file_string()
-            << " appears to be empty"; 
-        throw mussa_load_error(msg.str());
-      }
-      set_filtered_sequence(sequence_raw, start_index, end_index-start_index);
-    } else {
-      std::stringstream errormsg;
-      errormsg << file_path.native_file_string()
-               << " did not have any fasta sequences" << std::endl;
-      throw mussa_load_error(errormsg.str());
+    // Lastly, if subselection of the sequence was specified we keep cut out
+    // and only keep that part
+    // end_index = 0 means no end was specified, so cut to the end 
+    if (end_index == 0)
+      end_index = sequence_raw.size();
+
+    // sequence filtering for upcasing agctn and convert non AGCTN to N
+    if (end_index-start_index <= 0) {
+      std::string msg("The selected sequence appears to be empty"); 
+      throw sequence_empty_error(msg);
     }
-    data_file.close();
+    set_filtered_sequence(sequence_raw, start_index, end_index-start_index);
+  } else {
+    std::string errormsg("There were no fasta sequences");
+    throw sequence_empty_file_error(errormsg);
   }
 }