make the mupa parser more robust

ticket:268
improved mupa parser

diff --git a/alg/mussa.cpp b/alg/mussa.cpp
index 0912cb428be56d6ea4f1e270cf4b0acb2b9517a9..cda06ffe3ab9001d78bfa85323b7ee3b12d98435 100644 (file)
--- a/alg/mussa.cpp
+++ b/alg/mussa.cpp
@@ -17,6 +17,8 @@
 #include <boost/filesystem/fstream.hpp>
 namespace fs = boost::filesystem;
 
+#include <boost/algorithm/string.hpp>
+
 #include <iostream>
 #include <sstream>
 
@@ -209,6 +211,7 @@ int Mussa::get_soft_threshold() const
   return soft_thres;
 }
 
+
 void
 Mussa::set_analysis_mode(enum analysis_modes new_ana_mode)
 {
@@ -401,104 +404,131 @@ Mussa::load_mupa_file(fs::path para_file_path)
 void
 Mussa::load_mupa_stream(std::istream& para_file, fs::path& file_path_base)
 {
-  bool error_occured = false;
-  string file_data_line;
-  string param, value; 
+  std::string line;
+  std::vector< std::string > tokens;
+  int line_count = 0;
+
+  enum parsing_state_enum { START, INSEQUENCE };
+  parsing_state_enum parsing_state = START;
+  
+  // sequence file parameters
+  fs::path seq_file;
   fs::path annot_file;
   int split_index, fasta_index;
   int sub_seq_start, sub_seq_end;
   bool seq_params, did_seq;
-  string err_msg;
-  bool parsing_path;
-  string::size_type new_index, dir_index;
+  std::string err_msg;
 
   // initialize values
   clear();
 
-  // 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)
+  while (para_file.good())
   {
-    did_seq = false;
-    if (param == "ANA_NAME")
-      analysis_name = value;
-    else if (param == "APPEND_WIN")
+    ++line_count;
+    // setup loop by getting file's first line
+    multiplatform_getline(para_file, line);
+    // strip leading/trailing whitespace
+    boost::trim(line);
+    // ignore commented out or blank lines
+    if ( line.size() == 0 or line[0] == '#' ) {
+      continue;
+    } 
+
+    // split the line on white spance
+    boost::split(tokens, line, boost::is_space());
+    // do we have a name/value pair?
+    if (tokens.size() != 2) { 
+      std::stringstream errmsg;
+      errmsg << "Error parsing MUPA file line: "
+             << line_count << std::endl
+             << line;
+      throw mussa_load_error(errmsg.str());
+    }
+  
+    boost::to_upper(tokens[0]);
+    // Parameters only useful after a sequence block
+    if (parsing_state == INSEQUENCE) {
+      // in the following if blocks, if we do 
+      // successfully match a token we should continue
+      // on to the next token
+      // but if we don't match a token we want to 
+      // fall through to the top level parsing
+
+      if (tokens[0] == "FASTA_INDEX") {
+        fasta_index = atoi(tokens[1].c_str());
+        continue;
+      } else if (tokens[0] == "ANNOTATION") {
+        annot_file = file_path_base / tokens[1];
+        continue;
+      } else if (tokens[0] == "SEQ_START") {
+        sub_seq_start = atoi(tokens[1].c_str());
+        continue;
+      } else if (tokens[0] == "SEQ_END") {
+        sub_seq_end = atoi(tokens[1].c_str());
+        continue;
+      } else {
+        // any other token means we're done with this
+        // sequence so we should load it 
+        // (and let the "unknown" token fall through into the 
+        // top level token parser)
+        load_sequence(seq_file, annot_file, fasta_index, sub_seq_start, 
+                      sub_seq_end);
+        parsing_state = START;
+      }
+    }
+    // if we didn't consume a token from the previous if block
+    // try 
+    // top level token parsing
+    if (tokens[0] == "ANA_NAME")  {
+      analysis_name = tokens[1];
+    } else if (tokens[0] == "APPEND_WIN") {
       win_append = true;
-    else if (param == "APPEND_THRES")
+    } else if (tokens[0] == "APPEND_THRES") {
       thres_append = true;
-    else if (param == "SEQUENCE_NUM")
+    } else if (tokens[0] == "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;
+    } else if (tokens[0] == "WINDOW") {
+      window = atoi(tokens[1].c_str());
+    } else if (tokens[0] == "THRESHOLD") {
+      threshold = atoi(tokens[1].c_str());
+    } else if (tokens[0] == "SEQUENCE") {
+      if (parsing_state == INSEQUENCE) {
+        cout << "seq_file_name call2" << seq_file << endl;
+        load_sequence(seq_file, annot_file, fasta_index, sub_seq_start, 
+                      sub_seq_end);
+        parsing_state = START;
+      }
+      // reset sequence parameters
+      seq_file = file_path_base / tokens[1];
       fasta_index = 1;
       annot_file = "";
       sub_seq_start = 0;
       sub_seq_end = 0;
       seq_params = true;
-
-      while (para_file && seq_params)
-      {
-        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);
-
-        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.size() == 0) || (param[0] == '#')) 
-    {} // pass
-    else
-    {
+      parsing_state = INSEQUENCE;
+    } else {
       clog << "Illegal/misplaced mussa parameter in file\n";
-      clog << param << "\n";
-      error_occured = true;
-    }
-
-    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;
+      clog << tokens[0] << "\n";
+      std::stringstream errmsg;
+      errmsg << "Invalid mussa paaramater '" 
+             << tokens[0] 
+             << "' on line: "
+             << line_count << std::endl
+             << line;
+      throw mussa_load_error(errmsg.str());
+      throw mussa_load_error("Error parsing MUPA file");
     }
   }
 
-  if (error_occured) {
-    throw mussa_load_error("Error parsing MUPA file");
+  // if we hit the end of the file and there's a sequence
+  // pending, go ahead and load it
+  if (parsing_state == INSEQUENCE) {
+    load_sequence(seq_file, annot_file, fasta_index, sub_seq_start, 
+                  sub_seq_end);
   }
+
   soft_thres = threshold;
-  // no file was loaded, signal error
   set_dirty(true);
 }