Merge pull request #6 from ssciwr/from-csv

Add from_csv method that restores previously dumped distances

src/hamming.cc

@@ -2,6 +2,7 @@
 #include"hamming_impl.hh"
 
 #include<array>
+#include<algorithm>
 #include<fstream>
 #include<iostream>
 #include<sstream>
@@ -19,7 +20,28 @@ DataSet::DataSet(const std::vector<std::string>& data_)
 
 DataSet::DataSet(const std::string& filename)
 {
-  std::cout << "Not yet implemented" << std::endl;
+  // Determine correct dataset size
+  std::ifstream stream(filename);
+  std::string line;
+  nsamples = std::count(std::istreambuf_iterator<char>(stream),
+                        std::istreambuf_iterator<char>(), '\n');
+  result.resize(nsamples * (nsamples + 1) / 2);
+
+  // Read the data
+  stream = std::ifstream(filename);
+  std::size_t i{0};
+  std::size_t current{0};
+  while(std::getline(stream, line))
+  {
+    std::istringstream s(line);
+    std::string d;
+    for(std::size_t j=0; j<current; ++j)
+    {
+      std::getline(s, d, ',');
+      result[i++] = std::stoi(d);
+    }
+    ++current;
+  }
 }
 
 void DataSet::dump(const std::string& filename)

src/hamming_t.cc

@@ -161,3 +161,24 @@ TEST_CASE("invalid input data: inconsistent sequence lengths", "[invalid]") {
         REQUIRE_THROWS_WITH(from_stringlist(v), msg);
     }
 }
+
+TEST_CASE("from_csv reproduces correct data", "[hamming]") {
+    std::mt19937 gen(12345);
+    std::vector<std::string> data(10);
+    for(auto& d : data)
+        d = make_test_string(1000, gen);
+
+    DataSet ref(data);
+    char tmp_file_name[L_tmpnam];
+    REQUIRE(std::tmpnam(tmp_file_name) != nullptr);
+    ref.dump(std::string(tmp_file_name));
+
+    auto restore = from_csv(std::string(tmp_file_name));
+    REQUIRE(ref.nsamples == restore.nsamples);
+    for(std::size_t i=0; i<ref.nsamples; ++i) {
+        for(std::size_t j=0; j<ref.nsamples; ++j) {
+            REQUIRE(ref[{i, j}] == restore[{i, j}]);
+        }
+    }
+    std::remove(tmp_file_name);
+}