make mupa file loading eol-style insensitive

ticket:260
I refactored the load_mupa_file function into one that takes a file,
opens it, and then passes it to a different function that takes a stream.
AKA split a function in two, so I could more easily unit test the
mupa loading code.

Once it was unit testable, I moved multiplatform_getline out of sequence,
into its own new cpp file and changed the mupa loading code to use it instead.

diff --git a/alg/CMakeLists.txt b/alg/CMakeLists.txt
index 0234d943f14f8dbd761248239c59c4b5b2292459..9963d64542b3f4a549e3a1d6a249e6ef5a3a2d11 100644 (file)
--- a/alg/CMakeLists.txt
+++ b/alg/CMakeLists.txt
@@ -19,6 +19,7 @@ SET(SOURCES alphabet.cpp
             flp_seqcomp.cpp 
             glseqbrowser.cpp 
             glsequence.cpp 
             flp_seqcomp.cpp 
             glseqbrowser.cpp 
             glsequence.cpp 

+            io.cpp

             mussa.cpp 
             motif_parser.cpp 
             nway_entropy.cpp
             mussa.cpp 
             motif_parser.cpp 
             nway_entropy.cpp

diff --git a/alg/io.cpp b/alg/io.cpp
new file mode 100644 (file)
index 0000000..4d7624b
--- /dev/null
+++ b/alg/io.cpp
@@ -0,0 +1,17 @@
+#include "io.hpp"
+
+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();
+  }
+}
\ No newline at end of file

diff --git a/alg/io.hpp b/alg/io.hpp
new file mode 100644 (file)
index 0000000..29e1105
--- /dev/null
+++ b/alg/io.hpp
@@ -0,0 +1,10 @@
+#ifndef IO_HPP_
+#define IO_HPP_
+
+#include <iostream>
+#include <string>
+
+//! useful function for ignoring the various end of line conventions
+void multiplatform_getline(std::istream& in, std::string& line);
+
+#endif /*IO_HPP_*/

diff --git a/alg/mussa.cpp b/alg/mussa.cpp
index 82b1d4a6585e49c1a946e9093413b0be357ab570..9a5935e19947cbd68e28e7c7a3f88b267816030f 100644 (file)
--- a/alg/mussa.cpp
+++ b/alg/mussa.cpp
@@ -21,9 +21,11 @@ namespace fs = boost::filesystem;
 #include <sstream>
 
 #include "mussa_exceptions.hpp"
 #include <sstream>
 
 #include "mussa_exceptions.hpp"

-#include "alg/flp.hpp"
-#include "alg/mussa.hpp"
-#include "alg/motif_parser.hpp"
+
+#include "flp.hpp"
+#include "io.hpp"
+#include "mussa.hpp"
+#include "motif_parser.hpp"

 
 using namespace std;
 
 
 using namespace std;
 


@@ -350,10 +352,35 @@ void Mussa::load_sequence(fs::path seq_file, fs::path annot_file,
   set_dirty(true);
 }
 
   set_dirty(true);
 }
 

+void Mussa::load_mupa_file(std::string para_file_path) {
+  load_mupa_file(boost::filesystem::path(para_file_path));
+}
+

 void
 Mussa::load_mupa_file(fs::path para_file_path)
 {
 void
 Mussa::load_mupa_file(fs::path para_file_path)
 {

-  fs::ifstream para_file;
+  if (not fs::exists(para_file_path))
+  {
+    throw mussa_load_error("Config File: " + para_file_path.string() + " not found");
+  } else if (fs::is_directory(para_file_path)) {
+    throw mussa_load_error("Config File: " + para_file_path.string() + " is a directory.");
+  } else if (fs::is_empty(para_file_path)) {
+    throw mussa_load_error("Config File: " + para_file_path.string() + " is empty");
+  } else  {
+    // what directory is the mupa file in?
+    fs::path file_path_base( para_file_path.branch_path()) ;
+
+    fs::ifstream para_file;
+    para_file.open(para_file_path, ios::in);
+    
+    load_mupa_stream(para_file, file_path_base);
+    para_file.close();   
+  }
+}
+
+void
+Mussa::load_mupa_stream(std::istream& para_file, fs::path& file_path_base)
+{

   string file_data_line;
   string param, value; 
   fs::path annot_file;
   string file_data_line;
   string param, value; 
   fs::path annot_file;


@@ -367,101 +394,84 @@ Mussa::load_mupa_file(fs::path para_file_path)
   // initialize values
   clear();
 
   // initialize values
   clear();
 

-  // if file was opened, read the parameter values
-  if (not fs::exists(para_file_path))
+  // setup loop by getting file's first line
+  getline(para_file, file_data_line);
+  split_index = file_data_line.find(" ");
+  param = file_data_line.substr(0,split_index);
+  value = file_data_line.substr(split_index+1);
+  
+  while (para_file)

   {
   {

-    throw mussa_load_error("Config File: " + para_file_path.string() + " not found");
-  } else if (fs::is_directory(para_file_path)) {
-    throw mussa_load_error("Config File: " + para_file_path.string() + " is a directory.");
-  } else if (fs::is_empty(para_file_path)) {
-    throw mussa_load_error("Config File: " + para_file_path.string() + " is empty");
-  } else  {
-    para_file.open(para_file_path, ios::in);
-
-    // what directory is the mupa file in?
-    fs::path file_path_base = para_file_path.branch_path();
-
-    // setup loop by getting file's first line
-    getline(para_file,file_data_line);
-    split_index = file_data_line.find(" ");
-    param = file_data_line.substr(0,split_index);
-    value = file_data_line.substr(split_index+1);
-    
-    while (para_file)
+    did_seq = false;
+    if (param == "ANA_NAME")
+      analysis_name = value;
+    else if (param == "APPEND_WIN")
+      win_append = true;
+    else if (param == "APPEND_THRES")
+      thres_append = true;
+    else if (param == "SEQUENCE_NUM")
+      ; // ignore sequence_num now
+    else if (param == "WINDOW")
+      window = atoi(value.c_str());
+    else if (param == "THRESHOLD")
+      threshold = atoi(value.c_str());
+    else if (param == "SEQUENCE")

     {
     {

-      did_seq = false;
-      if (param == "ANA_NAME")
-        analysis_name = value;
-      else if (param == "APPEND_WIN")
-        win_append = true;
-      else if (param == "APPEND_THRES")
-        thres_append = true;
-      else if (param == "SEQUENCE_NUM")
-        ; // ignore sequence_num now
-      else if (param == "WINDOW")
-        window = atoi(value.c_str());
-      else if (param == "THRESHOLD")
-        threshold = atoi(value.c_str());
-      else if (param == "SEQUENCE")
-      {
-        fs::path seq_file = file_path_base / value;
-        //cout << "seq_file_name " << seq_files.back() << endl;
-        fasta_index = 1;
-        annot_file = "";
-        sub_seq_start = 0;
-        sub_seq_end = 0;
-        seq_params = true;
-
-        while (para_file && seq_params)
-        {
-          getline(para_file,file_data_line);
-          split_index = file_data_line.find(" ");
-          param = file_data_line.substr(0,split_index);
-          value = file_data_line.substr(split_index+1);
-
-          if (param == "FASTA_INDEX")
-            fasta_index = atoi(value.c_str());
-          else if (param == "ANNOTATION")
-            annot_file = file_path_base / value;
-          else if (param == "SEQ_START")
-            sub_seq_start = atoi(value.c_str());
-          else if (param == "SEQ_END")
-          {
-            sub_seq_end = atoi(value.c_str());
-          }
-          //ignore empty lines or that start with '#'
-          else if ((param == "") || (param == "#")) {}
-          else seq_params = false;
-        }
-        load_sequence(seq_file, annot_file, fasta_index, sub_seq_start, 
-                      sub_seq_end);
-        did_seq = true;
-      }
-      //ignore empty lines or that start with '#'
-      else if ((param == "") || (param == "#")) {}
-      else
-      {
-        clog << "Illegal/misplaced mussa parameter in file\n";
-        clog << param << "\n";
-      }
-
-      if (!did_seq)
+      fs::path seq_file = file_path_base / value;
+      //cout << "seq_file_name " << seq_files.back() << endl;
+      fasta_index = 1;
+      annot_file = "";
+      sub_seq_start = 0;
+      sub_seq_end = 0;
+      seq_params = true;
+
+      while (para_file && seq_params)

       {
       {

-        getline(para_file,file_data_line);
+        multiplatform_getline(para_file,file_data_line);

         split_index = file_data_line.find(" ");
         param = file_data_line.substr(0,split_index);
         value = file_data_line.substr(split_index+1);
         split_index = file_data_line.find(" ");
         param = file_data_line.substr(0,split_index);
         value = file_data_line.substr(split_index+1);

-        did_seq = false;
+
+        if (param == "FASTA_INDEX")
+          fasta_index = atoi(value.c_str());
+        else if (param == "ANNOTATION")
+          annot_file = file_path_base / value;
+        else if (param == "SEQ_START")
+          sub_seq_start = atoi(value.c_str());
+        else if (param == "SEQ_END")
+        {
+          sub_seq_end = atoi(value.c_str());
+        }
+        //ignore empty lines or that start with '#'
+        else if ((param == "") || (param == "#")) {
+          // pass 
+        } else {
+          seq_params = false;
+        }

       }
       }

+      load_sequence(seq_file, annot_file, fasta_index, sub_seq_start, 
+                    sub_seq_end);
+      did_seq = true;
+    }
+    //ignore empty lines or that start with '#'
+    else if ((param == "") || (param == "#")) {}
+    else
+    {
+      clog << "Illegal/misplaced mussa parameter in file\n";
+      clog << param << "\n";

     }
 
     }
 

-    para_file.close();
-
-    soft_thres = threshold;
-    //cout << "nway mupa: analysis_name = " << analysis_name 
-    //     << " window = " << window 
-    //     << " threshold = " << threshold << endl;
+    if (!did_seq)
+    {
+      multiplatform_getline(para_file,file_data_line);
+      split_index = file_data_line.find(" ");
+      param = file_data_line.substr(0,split_index);
+      value = file_data_line.substr(split_index+1);
+      did_seq = false;
+    }

   }
   }

+
+  soft_thres = threshold;

   // no file was loaded, signal error
   set_dirty(true);
 }
   // no file was loaded, signal error
   set_dirty(true);
 }

diff --git a/alg/mussa.hpp b/alg/mussa.hpp
index 83e08cba7f7e42e8a32c9d092d2a516020d6a700..152d8217349568755913e67c1a77fed6aaa6eaa6 100644 (file)
--- a/alg/mussa.hpp
+++ b/alg/mussa.hpp
@@ -72,8 +72,14 @@ public:
     void clear();
 
     //! set parameters from a file - 'mupa' ~ mussa parameters
     void clear();
 
     //! set parameters from a file - 'mupa' ~ mussa parameters

-    void load_mupa_file(std::string para_file_path) { load_mupa_file(boost::filesystem::path(para_file_path));}
+    void load_mupa_file(std::string para_file_path);

     void load_mupa_file(boost::filesystem::path para_file_path);
     void load_mupa_file(boost::filesystem::path para_file_path);

+    //! load mussa parameters from a stream, specifing output location
+    void load_mupa_stream(
+           std::istream & para_file, 
+           boost::filesystem::path& file_path_base
+         );
+    

 
     // set parameters individually (eg from user input into gui classes)
     //! set analysis name
 
     // set parameters individually (eg from user input into gui classes)
     //! set analysis name

diff --git a/alg/sequence.cpp b/alg/sequence.cpp
index 2e845c67bdfaa84d5843e0b05a5e3a766e363eaf..1f21bd5edc83e8e66894ffc7ee473c31bfbd170c 100644 (file)
--- a/alg/sequence.cpp
+++ b/alg/sequence.cpp
@@ -32,6 +32,7 @@ namespace fs = boost::filesystem;
 namespace spirit = boost::spirit;
 
 #include "alg/sequence.hpp"
 namespace spirit = boost::spirit;
 
 #include "alg/sequence.hpp"

+#include "io.hpp"

 #include "mussa_exceptions.hpp"
 
 #include <string>
 #include "mussa_exceptions.hpp"
 
 #include <string>


@@ -162,22 +163,6 @@ Sequence &Sequence::operator=(const Sequence& s)
   return *this;
 }
 
   return *this;
 }
 

-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();
-  }
-}
-

 void Sequence::load_fasta(fs::path file_path, int seq_num, int start_index, int end_index)
 {
   load_fasta(file_path, reduced_nucleic_alphabet, seq_num, start_index, end_index);
 void Sequence::load_fasta(fs::path file_path, int seq_num, int start_index, int end_index)
 {
   load_fasta(file_path, reduced_nucleic_alphabet, seq_num, start_index, end_index);

diff --git a/alg/test/test_mussa.cpp b/alg/test/test_mussa.cpp
index 55188eab8052b9e128d4470d2dd33ef1911644ec..e1cef22fb6cda9e9213e04e0642bd6306e05a8ab 100644 (file)
--- a/alg/test/test_mussa.cpp
+++ b/alg/test/test_mussa.cpp
@@ -110,6 +110,25 @@ BOOST_AUTO_TEST_CASE ( empty_mussa_set_threshold )
   m.nway();
 }
 
   m.nway();
 }
 

+BOOST_AUTO_TEST_CASE( mussa_load_mupa_crlf )
+{
+  fs::path example_path(EXAMPLE_DIR, fs::native);
+  fs::path seq_path(example_path / "seq" / "mouse_mck_pro.fa");
+  fs::path annot_path(example_path / "mm_mck3test.annot");
+
+  std::string mupa(
+    "# hello\015\012"
+    "ANA_NAME load_mupa_crlf\015\012");
+  mupa += "SEQUENCE " + seq_path.native_file_string() + "\015\012";
+  mupa += "ANNOTATION " + annot_path.native_file_string() + "\015\012";
+  
+  istringstream mupa_stream(mupa);
+  Mussa m;
+  fs::path base;
+  m.load_mupa_stream( mupa_stream, base );
+  // Should run with no exceptions
+}
+

 BOOST_AUTO_TEST_CASE( mussa_load_mupa )
 {
   fs::path mupa_path(EXAMPLE_DIR, fs::native);
 BOOST_AUTO_TEST_CASE( mussa_load_mupa )
 {
   fs::path mupa_path(EXAMPLE_DIR, fs::native);