cleanup

main.cpp

@@ -15,9 +15,9 @@ using namespace indicators;
 // first letter of an ecoded sequence is the prepend symbol, last is the append symbol
 // there's no need for actually prepending/appending all k symbols
 // we also skip sequences shorter than k
-void computeMultiKMap(const std::vector<encoded_seq_t>& seqs, size_t nbDatasets,
- size_t datasetId, int k, Alphabet alpha, size_t lowerBound,
- size_t upperBound, multikmap_t& res) {
+void computeMultiKMap(const std::vector<encoded_seq_t>& seqs, size_t datasetId, int k,
+ Alphabet alpha, size_t lowerBound, size_t upperBound,
+ multikmap_t& res) {
  if (k <= 0) throw std::invalid_argument("k must be > 0 ");
  const auto ALPHABET_SIZE = alphaMap.alphabetSizes[alpha];
  for (size_t s = lowerBound; s < upperBound; ++s) {
@@ -57,8 +57,7 @@ void computeDatasetMultiKmap(const std::vector<dataset_t>& datasets, size_t data
  auto& dataset = datasets[datasetId];
  const size_t lower = i;
  const size_t upper = std::min(i + nbSeqPerTask, dataset.size());
- computeMultiKMap(dataset, nbDatasets, datasetId, K, alpha, lower, upper,
- allmaps[procId]);
+ computeMultiKMap(dataset, datasetId, K, alpha, lower, upper, allmaps[procId]);
  int ticksize = 0;
  for (size_t j = lower; j < upper; ++j) ticksize += dataset[j].size();
  mainbar.tick(ticksize);
@@ -67,7 +66,7 @@ void computeDatasetMultiKmap(const std::vector<dataset_t>& datasets, size_t data
  tp.waitAll();
  mergeMultiMaps(outmap, allmaps);
  } else {
- computeMultiKMap(datasets[datasetId], datasets.size(), datasetId, K, alpha, 0,
+ computeMultiKMap(datasets[datasetId], datasetId, K, alpha, 0,
  datasets[datasetId].size(), outmap);
  for (const auto& d : datasets[datasetId]) mainbar.tick(d.size());
  }
@@ -196,7 +195,6 @@ int main(int argc, char** argv) {
  mainbar.set_option(option::PostfixText{" "});
  mainbar.mark_as_completed();
 
- // multikmap_t result = mergeMaps(allkmaps);
  mergeMultiMaps(allkmaps[0], allkmaps, true, 1);
  auto& result = allkmaps[0];
 
@@ -213,6 +211,7 @@ int main(int argc, char** argv) {
  break;
  }
 
+ std::cerr << "Done. Cleaning up memory..." << std::endl;
  show_console_cursor(true);
  return 0;
 }

utils.hpp

@@ -7,6 +7,7 @@
 #include <memory>
 #include <random>
 #include <robinhood.hpp>
+#include <string_view>
 #include <unordered_map>
 #include <xhash.hpp>
 
@@ -20,13 +21,9 @@ namespace fs = std::filesystem;
 
 using seqview_t = std::basic_string_view<int8_t>;
 
-// template <typename... T> using umap_t = robin_hood::unordered_flat_map<T...>;
-template <typename... T> using umap_t = tsl::robin_map<T...>;
-
 template <typename V> using seqmap_t = robin_hood::unordered_map<seqview_t, V>;
 
 using kmap_t = seqmap_t<std::vector<size_t>>; // SeqView -> counts
-
 using multikmap_t = seqmap_t<robin_hood::unordered_map<
  size_t, std::vector<size_t>>>; // SeqView -> counts per dataset
 
@@ -244,7 +241,7 @@ inline void mergeMultiMaps(multikmap_t& res, std::vector<multikmap_t>& maps,
  for (size_t m = startIndex; m < N; ++m) {
  for (auto& [kmer, map] : maps[m]) {
  if (!res.count(kmer))
- res[kmer] = map;
+ res[kmer] = decltype(map)(std::move(map));
  else {
  for (auto& [dataset, counts] : map) {
  if (res[kmer].count(dataset)) {
@@ -261,9 +258,9 @@ inline void mergeMultiMaps(multikmap_t& res, std::vector<multikmap_t>& maps,
  ticks = 0;
  }
  }
- maps[m].clear(); // avoid wasting memory
+ // maps[m].clear(); // avoid wasting memory
  }
-
+ // maps.clear();
  if (mainbar) { // spinner update
  mainbar->set_option(option::ForegroundColor{Color::green});
  mainbar->set_option(option::PrefixText{"✔ " + std::to_string(N) + " kmaps merged (" +
@@ -405,10 +402,11 @@ inline void dumpMultiMap_sparseMatrix(const multikmap_t& m, int k, Alphabet alph
  for (const auto& c : allcounts) buff += std::to_string(c) + ",";
  buff.pop_back(); // removes trailing comma
  buff += "]\n";
- if (++c % CHUNKSIZE == 0) {
+ if (++c > CHUNKSIZE) {
  std::fwrite(buff.c_str(), 1, buff.size(), file);
  buff.clear();
- mainbar.tick(CHUNKSIZE);
+ mainbar.tick(c);
+ c = 0;
  }
  }
  std::fwrite(buff.c_str(), 1, buff.size(), file);