Make RandstrobeIterator/-Generator yield the same results

src/index.cpp

@@ -27,18 +27,15 @@ static const uint32_t STI_FILE_FORMAT_VERSION = 3;
 namespace {
 
 uint64_t count_randstrobes(const std::string& seq, const IndexParameters& parameters) {
+    // To reduce runtime, this function actually counts *syncmers* and
+    // assumes that one randstrobe is generated per syncmer
     uint64_t n_syncmers = 0;
     SyncmerIterator syncmer_iterator(seq, parameters.syncmer);
     Syncmer syncmer;
     while (!(syncmer = syncmer_iterator.next()).is_end()) {
         n_syncmers++;
     }
-    // The last w_min syncmers do not result in a randstrobe
-    if (n_syncmers < parameters.randstrobe.w_min) {
-        return 0;
-    } else {
-        return n_syncmers - parameters.randstrobe.w_min;
-    }
+    return n_syncmers;
 }
 
 std::vector<uint64_t> count_all_randstrobes(const References& references, const IndexParameters& parameters, size_t n_threads) {

src/randstrobes.cpp

@@ -202,10 +202,21 @@ Randstrobe RandstrobeGenerator::next() {
         }
         syncmers.push_back(syncmer);
     }
-    if (syncmers.size() <= w_min) {
+    if (syncmers.empty()) {
         return RandstrobeGenerator::end();
     }
     auto strobe1 = syncmers[0];
+
+    if (syncmers.size() < w_min) {
+        auto randstrobe = Randstrobe{
+            randstrobe_hash(strobe1.hash, 0, aux_len),
+            static_cast<uint32_t>(strobe1.position),
+            static_cast<uint32_t>(strobe1.position),
+            true
+        };
+        syncmers.pop_front();
+        return randstrobe;
+    }
     auto max_position = strobe1.position + max_dist;
     uint64_t min_val = std::numeric_limits<uint64_t>::max();
     Syncmer strobe2 = strobe1; // Default if no nearby syncmer

tests/tests.cpp

@@ -138,7 +138,7 @@ TEST_CASE("both randstrobes iterator implementations give same results") {
 }
 
 // This tests an assumption that we need to hold for count_randstrobes()
-TEST_CASE("syncmer and randstrobe iterators return (nearly) same no. of items") {
+TEST_CASE("syncmer and randstrobe iterators return same no. of items") {
     auto references = References::from_fasta("tests/phix.fasta");
     auto& seq = references.sequences[0];
     auto parameters = IndexParameters::from_read_length(100);
@@ -157,7 +157,7 @@ TEST_CASE("syncmer and randstrobe iterators return (nearly) same no. of items")
         syncmer_count++;
     }
 
-    CHECK(randstrobe_count + parameters.randstrobe.w_min == syncmer_count);
+    CHECK(randstrobe_count == syncmer_count);
 }
 
 TEST_CASE("reverse complement") {