final?

include/config.hpp

@@ -4,14 +4,18 @@
 
 class Configuration {
 public:
+    // I/O settings
     std::string input_file = "1.in";
     std::string output_file = "1.out";
-    std::string tmp_dir = "build/run";
+    std::string tmp_directory = "run";
-    bool generate_data = false;
+
-    bool evaluate_file = false;
+    // Algorithm settings
-    int N = 100000; // number of records in a file
+    size_t buffer_rows = 101;
-    int n = 101; // number of buffers
+    size_t buffer_cols = 10;
-    int b = 10; // blocking factor
+
+    // Additional modes
+    size_t generate_data = 0;
+    bool read_file = false;
 
     Configuration(int argc, char** argv);
 };

include/merge_sorter.hpp

@@ -21,9 +21,9 @@ private:
 
 public:
     MergeSorter(size_t buffer_rows, size_t buffer_cols, const std::string& tmp_dir);
-    bool sort_file(const std::string& input_file, const std::string& output_file);
+    void sort_file(const std::string& input_file, const std::string& output_file);
 
-    size_t disk_reads() const;
+    size_t disk_reads();
-    size_t disk_writes() const;
+    size_t disk_writes();
-    size_t phases() const;
+    size_t phases();
 };

makefile

@@ -24,10 +24,10 @@ generate: build/run/1.in
 
 # actually should depend on build/main and build/run, but that will trigger it every recompile, and i don't need that
 build/run/1.in:
-	$(TARGET) -g -o build/run/1.in
+	$(TARGET) -g 100000 -o build/run/1.in
 
 run: $(TARGET) build/run/1.in
-	$(TARGET) -i build/run/1.in -d build/run/tmp
+	$(TARGET) -i build/run/1.in -d build/run/tmp -o build/run/1.out
 
 run_alt: $(TARGET) build/run/1.in
 	$(TARGET) -i build/run/1.in -d build/run/tmp -n 11 -b 100

src/config.cpp

@@ -6,13 +6,13 @@ Configuration::Configuration(int argc, char** argv)
 {
     if (argc <= 1) {
         std::cerr << "Please specify input arguments\n";
-        std::cerr << "Usage: " << argv[0] << " [-g] [-e] [-N count] [-i input] [-o output] [-d dir] [-n buffers] [-b block_size]\n";
+        std::cerr << "Usage: " << argv[0] << " [-g count] [-e]  [-i input] [-o output] [-d dir] [-n buffers] [-b block_size]\n";
         throw std::invalid_argument("No input arguments");
     }
     bool input_set = false;
     int opt;
 
-    while ((opt = getopt(argc, argv, "i:d:o:geN:n:b:")) != -1) {
+    while ((opt = getopt(argc, argv, "i:d:o:g:en:b:")) != -1) {
         switch (opt) {
         case 'i':
             input_file = optarg;
@@ -22,22 +22,19 @@ Configuration::Configuration(int argc, char** argv)
             output_file = optarg;
             break;
         case 'd':
-            tmp_dir = optarg;
+            tmp_directory = optarg;
             break;
         case 'g':
-            generate_data = true;
+            generate_data = std::stoi(optarg);
             break;
         case 'e':
-            evaluate_file = true;
+            read_file = true;
-            break;
-        case 'N':
-            N = std::stoi(optarg);
             break;
         case 'n':
-            n = std::stoi(optarg);
+            buffer_rows = std::stoi(optarg);
             break;
         case 'b':
-            b = std::stoi(optarg);
+            buffer_cols = std::stoi(optarg);
             break;
         case '?':
             throw std::invalid_argument("Invalid command line argument");
@@ -49,7 +46,7 @@ Configuration::Configuration(int argc, char** argv)
     if (generate_data && input_set) {
         throw std::invalid_argument("Cannot specify both -g and -i options");
     }
-    if (generate_data && evaluate_file) {
+    if (generate_data && read_file) {
         throw std::invalid_argument("Cannot specify both -g and -e options");
     }
 }

src/main.cpp

@@ -21,13 +21,13 @@ void generate_file(int N, const std::string& filename)
 void read_and_evaluate(std::string filename)
 {
     FileReader input(filename);
-    while (true) {
+    for (size_t i = 0;; i++) {
         auto rec = input.read();
         if (!rec.has_value()) {
             return;
         }
 
-        std::cout << "[Evaluate: " << rec.value().evaluate() << "] " << rec.value() << "\n";
+        std::cout << "record " << i << " [value: " << rec.value().evaluate() << "] " << rec.value() << "\n";
     }
 }
 
@@ -36,15 +36,18 @@ int main(int argc, char** argv)
     Configuration opts(argc, argv);
 
     if (opts.generate_data) {
-        generate_file(opts.N, opts.output_file);
+        generate_file(opts.generate_data, opts.output_file);
         return 0;
     }
 
-    if (opts.evaluate_file) {
+    if (opts.read_file) {
         read_and_evaluate(opts.input_file);
         return 0;
     }
 
-    MergeSorter sorter(opts.b, opts.n, opts.tmp_dir);
+    MergeSorter sorter(opts.buffer_cols, opts.buffer_rows, opts.tmp_directory);
     sorter.sort_file(opts.input_file, opts.output_file);
+    std::cout << "total reads: " << sorter.disk_reads() << '\n'
+              << "total writes: " << sorter.disk_writes() << '\n'
+              << "phases: " << sorter.phases() << "\n";
 }

src/merge_sorter.cpp

@@ -138,7 +138,7 @@ MergeSorter::MergeSorter(size_t buffer_rows, size_t buffer_cols, const std::stri
     }
 }
 
-bool MergeSorter::sort_file(const std::string& input_file, const std::string& output_file)
+void MergeSorter::sort_file(const std::string& input_file, const std::string& output_file)
 {
     std::string current_dir = tmp_dir_ + "/pass0/run_";
     std::string next_dir = tmp_dir_ + "/pass1/run_";
@@ -146,16 +146,29 @@ bool MergeSorter::sort_file(const std::string& input_file, const std::string& ou
 
     run_count = create_initial_runs(input_file, current_dir);
     phases_++;
-    std::cout << "created inital runs, disk reads: " << disk_reads_ << ", disk writes: " << disk_writes_ << '\n';
 
     while (run_count > 1) {
         run_count = merge_pass(current_dir, next_dir, run_count);
-        std::cout << "completed merge pass " << phases_ << ", disk reads: " << disk_reads_ << ", disk writes: " << disk_writes_ << '\n';
         phases_++;
         current_dir = next_dir;
         next_dir = tmp_dir_ + "/pass" + std::to_string(phases_) + "/run_";
     }
 
-    std::cout << output_file << '\n';
+    std::string final_run = current_dir + "0.dat";
-    return true;
+    std::filesystem::rename(final_run, output_file);
+}
+
+size_t MergeSorter::disk_reads()
+{
+    return disk_reads_;
+}
+
+size_t MergeSorter::disk_writes()
+{
+    return disk_writes_;
+}
+
+size_t MergeSorter::phases()
+{
+    return phases_;
 }