Merge branch 'master' into docs

src/process_reads.cpp

@@ -26,7 +26,7 @@ void process_reads(Parameters& parameters) {
     for (int i=0; i<parameters.get_value_from_name<int>(par); ++i) {
         threads.push_back(std::thread(file_processor, std::ref(input_files), std::ref(results), std::ref(results_mutex), std::ref(files_mutex)));
     }
-
+    
     for (auto &t : threads) t.join();
 
     par = "output_file_path";
@@ -36,6 +36,7 @@ void process_reads(Parameters& parameters) {
     std::vector<std::string> individuals;
     for (auto i: input_files) individuals.push_back(i.individual_name);
 
+
     // Generate the output file
     output_process_reads(output_file_path, individuals, results, min_cov);
 
@@ -50,6 +51,7 @@ inline void file_processor(std::vector<InputFile>& input_files, std::unordered_m
      * A mutex protects the list of input files and another mutex protects the results structure
      */
 
+
     bool remaining_files = true;
 
     while (remaining_files) {
@@ -89,6 +91,7 @@ inline void process_file(InputFile& input_file, std::unordered_map<std::string,
     // Read through the file and store the results
     while ((line_n = kseq_read(sequence)) >= 0) {
         ++temp_results[sequence->seq.s];
+        if (line_n % 1000 == 0) std::cout << line_n / 1000 << std::endl;
     }
 
     kseq_destroy(sequence); // Destroy the seq object

src/sequence_file.cpp

@@ -23,7 +23,7 @@ std::vector<InputFile> get_input_files(const std::string& input_dir_path) {
     std::vector<std::string> split_name;
     InputFile temp;
 
-    while ((dir_content=readdir(dir))){
+    while ((dir_content = readdir(dir))) {
 
         current_file = dir_content->d_name;
         split_name = split(current_file, ".");
@@ -32,9 +32,15 @@ std::vector<InputFile> get_input_files(const std::string& input_dir_path) {
 
         // Careful: this will fail if the file name contains "." ... TODO: solve this problem
         if (s > 1) {
-            for (uint i=1; i<split_name.size(); ++i) extension += "." + split_name[i];
+            if (split_name[s - 1] == "gz" and s > 2) {
+                extension = "." + split_name[s - 2] + "." + split_name[s - 1];
+            } else {
+                extension = "." + split_name[s - 1];
+            }
         }
 
+        std::cout << current_file << " : " << extension << std::endl;
+
         if(std::find(extensions.begin(), extensions.end(), extension) != extensions.end()) {
             temp.individual_name = split_name[0];
             temp.path = input_dir_path + current_file;
@@ -44,5 +50,11 @@ std::vector<InputFile> get_input_files(const std::string& input_dir_path) {
         }
     }
 
+    if (files.size() == 0) {
+
+        std::cout << " ** Error: no valid input file found in input directory \"" << input_dir_path <<"\"." << std::endl;
+        exit(1);
+    }
+
     return files;
 }