updated file check function and move it to MSite.*pp

Author: masarunakajima

src/analysis/roimethstat.cpp

@@ -362,59 +362,6 @@ process_with_cpgs_on_disk(const bool PRINT_NUMERIC_ONLY,
 ///
 ////////////////////////////////////////////////////////////////////////
 
-static inline bool
-is_float(const string &str) {
-  try {
-    size_t pos;
-    std::stof(str, &pos);
-    return pos == str.size(); // Check if entire string was consumed
-  } catch (const std::invalid_argument &) {
-    return false; // Conversion failed due to invalid argument
-  } catch (const std::out_of_range &) {
-    return false; // Conversion failed due to out of range
-  }
-}
-
-static inline bool 
-is_integer(const string &str) {
-	try {
-		size_t pos;
-		std::stoi(str, &pos);
-		return pos == str.size(); // Check if entire string was consumed
-	} catch (const std::invalid_argument &) {
-		return false; // Conversion failed due to invalid argument
-	} catch (const std::out_of_range &) {
-		return false; // Conversion failed due to out of range
-	}
-}
-
-
-static bool
-is_methylation_file(const string &file) {
-  ifstream in(file);
-  if (!in)
-    throw runtime_error("cannot open file: " + file);
-
-  string line;
-  getline(in, line);
-
-  std::istringstream iss(line);
-  string token;
-
-  vector<string> tokens;
-  while(iss >> token) {
-    tokens.push_back(token);
-  }
-
-  std::regex pattern("^C[pHWX][GH]$");
-
-  return tokens.size() == 6 && 
-         is_integer(tokens[1]) && 
-         (tokens[2] == "+" || tokens[2] == "-") && 
-         regex_match(tokens[3], pattern) && 
-         is_float(tokens[4]) && 
-         is_integer(tokens[5]);
-}
 
 static size_t
 check_bed_format(const string &regions_file) {
@@ -529,12 +476,12 @@ Columns (beyond the first 6) in the BED format output:
     // bed format
     if (n_columns != 3 && n_columns < 6)
       throw runtime_error("format must be 3 or 6+ column bed: " + regions_file);
-    if (is_methylation_file(regions_file)) {
+    if (is_msite_file(regions_file)) {
       cerr << opt_parse.help_message() << endl;
       throw runtime_error("The file seems to be a methylation file: " + 
           regions_file + "\nCheck the order of the input arguments");
     }
-    if (!is_methylation_file(cpgs_file)) {
+    if (!is_msite_file(cpgs_file)) {
       cerr << opt_parse.help_message() << endl;
       throw runtime_error("The file is not a methylation file: " + cpgs_file);
     }
@@ -584,3 +531,6 @@ Columns (beyond the first 6) in the BED format output:
   }
   return EXIT_SUCCESS;
 }
+
+
+

src/common/MSite.cpp

@@ -20,11 +20,13 @@
 #include <fstream>
 #include <sstream>
 #include <stdexcept>
+#include <regex>
 
 #include "smithlab_utils.hpp"
 
 using std::string;
 using std::runtime_error;
+using std::regex_match;
 
 MSite::MSite(const string &line) {
   /* GS: this is faster but seems to be genenerating issues when
@@ -152,3 +154,45 @@ find_offset_for_msite(const std::string &chr,
     move_to_start_of_line(site_in);
   }
 }
+
+
+
+bool
+is_msite_file(const string &file) {
+  ifstream in(file);
+  if (!in)
+    throw runtime_error("cannot open file: " + file);
+
+  string line;
+  if(!getline(in, line)) return false;
+
+  std::istringstream iss(line);
+
+  string chrom;
+  if (!(iss >> chrom)) return false;
+  
+  long int pos = 0;
+  if (!(iss >> pos)) return false;
+
+  string strand;
+  if (!(iss >> strand) || 
+      (strand.size() != 1) || 
+      ((strand != "+") && (strand != "-")) ) 
+    return false;
+
+  string context;
+  std::regex pattern("^C[pHWX][GH]$");
+  if (!(iss >> context) || !regex_match(context, pattern)) return false;
+
+  double level = 0.0;
+  if (!(iss >> level) || level < 0 || level > 1) return false;
+
+  long int n_reads = 0;
+  if (!(iss >> n_reads)) return false;
+
+  string temp;
+  if (iss >> temp) return false;
+  else return true;
+
+}
+

src/common/MSite.hpp

@@ -138,4 +138,9 @@ find_offset_for_msite(const std::string &chrom,
                       const size_t start_pos,
                       std::ifstream &site_in);
 
+
+bool
+is_msite_file(const std::string &file);
+
+
 #endif