Add test for BucketIndexStore and the iterator.

3 files changed, 44 insertions, 10 deletions

diff --git a/searchlib/src/tests/docstore/store_by_bucket/store_by_bucket_test.cpp b/searchlib/src/tests/docstore/store_by_bucket/store_by_bucket_test.cpp
index 50e99b15fb2..5684f4a4070 100644
--- a/searchlib/src/tests/docstore/store_by_bucket/store_by_bucket_test.cpp
+++ b/searchlib/src/tests/docstore/store_by_bucket/store_by_bucket_test.cpp
@@ -4,7 +4,7 @@
 
 #include <vespa/document/bucket/bucketid.h>
 #include <vespa/document/base/documentid.h>
-#include <vespa/searchlib/docstore/storebybucket.h>
+#include <vespa/searchlib/docstore/compacter.h>
 #include <vespa/vespalib/stllike/asciistream.h>
 #include <vespa/vespalib/stllike/hash_set.h>
 #include <vespa/vespalib/util/threadstackexecutor.h>
@@ -25,8 +25,8 @@ createPayload(BucketId b) {
 }
 uint32_t userId(size_t i) { return i%100; }
 
-void
-add(StoreByBucket & sbb, size_t i) {
+BucketId
+createBucketId(size_t i) {
     constexpr size_t USED_BITS=5;
     vespalib::asciistream os;
     os << "id:a:b:n=" << userId(i) << ":" << i;
@@ -34,6 +34,11 @@ add(StoreByBucket & sbb, size_t i) {
     BucketId b = docId.getGlobalId().convertToBucketId();
     EXPECT_EQUAL(userId(i), docId.getGlobalId().getLocationSpecificBits());
     b.setUsedBits(USED_BITS);
+    return b;
+}
+void
+add(StoreByBucket & sbb, size_t i) {
+    BucketId b = createBucketId(i);
     vespalib::string s = createPayload(b);
     sbb.add(b, i%10, i, {s.c_str(), s.size()});
 }
@@ -78,7 +83,7 @@ struct StoreIndex : public StoreByBucket::StoreIndex {
 StoreIndex::~StoreIndex() = default;
 
 struct Iterator : public StoreByBucket::IndexIterator {
-    Iterator(const std::vector<StoreByBucket::Index> & where) : _where(where), _current(0) {}
+    explicit Iterator(const std::vector<StoreByBucket::Index> & where) : _where(where), _current(0) {}
 
     bool has_next() noexcept override {
         return _current < _where.size();
@@ -99,19 +104,48 @@ TEST("require that StoreByBucket gives bucket by bucket and ordered within")
     vespalib::ThreadStackExecutor executor(8);
     StoreIndex storeIndex;
     StoreByBucket sbb(storeIndex, backing, executor, CompressionConfig::LZ4);
-    for (size_t i(1); i <=500; i++) {
+    for (size_t i(1); i <= 500u; i++) {
         add(sbb, i);
     }
-    for (size_t i(1000); i > 500; i--) {
+    for (size_t i(1000); i > 500u; i--) {
         add(sbb, i);
     }
     sbb.close();
     std::sort(storeIndex._where.begin(), storeIndex._where.end());
-    //EXPECT_EQUAL(32u, sbb.getBucketCount());
     EXPECT_EQUAL(1000u, storeIndex._where.size());
     VerifyBucketOrder vbo;
     Iterator all(storeIndex._where);
     sbb.drain(vbo, all);
 }
 
+constexpr uint32_t NUM_PARTS = 3;
+
+void
+verifyIter(BucketIndexStore &store, uint32_t partId, uint32_t expected_count) {
+    auto iter = store.createIterator(partId);
+    uint32_t count(0);
+    while (iter->has_next()) {
+        StoreByBucket::Index idx = iter->next();
+        EXPECT_EQUAL(store.toPartitionId(idx._bucketId), partId);
+        count++;
+    }
+    EXPECT_EQUAL(expected_count, count);
+}
+
+TEST("test that iterators cover the whole corpus and maps to correct partid") {
+
+    BucketIndexStore bucketIndexStore(32, NUM_PARTS);
+    for (size_t i(1); i <= 500u; i++) {
+        bucketIndexStore.store(StoreByBucket::Index(createBucketId(i), 1, 2, i));
+    }
+    bucketIndexStore.prepareForIterate();
+    EXPECT_EQUAL(500u, bucketIndexStore.getLidCount());
+    EXPECT_EQUAL(32u, bucketIndexStore.getBucketCount());
+    constexpr uint32_t COUNT_0 = 175, COUNT_1 = 155, COUNT_2 = 170;
+    verifyIter(bucketIndexStore, 0, COUNT_0);
+    verifyIter(bucketIndexStore, 1, COUNT_1);
+    verifyIter(bucketIndexStore, 2, COUNT_2);
+    EXPECT_EQUAL(500u, COUNT_0 + COUNT_1 + COUNT_2);
+}
+
 TEST_MAIN() { TEST_RUN_ALL(); }
diff --git a/searchlib/src/vespa/searchlib/docstore/compacter.cpp b/searchlib/src/vespa/searchlib/docstore/compacter.cpp
index 44fd88e7994..ebef684479f 100644
--- a/searchlib/src/vespa/searchlib/docstore/compacter.cpp
+++ b/searchlib/src/vespa/searchlib/docstore/compacter.cpp
@@ -25,7 +25,7 @@ Compacter::write(LockGuard guard, uint32_t chunkId, uint32_t lid, ConstBufferRef
 }
 
 BucketIndexStore::BucketIndexStore(size_t maxSignificantBucketBits, uint32_t numPartitions) noexcept
-    : _unSignificantBucketBits((maxSignificantBucketBits > 8) ? (maxSignificantBucketBits - 8) : 0),
+    : _inSignificantBucketBits((maxSignificantBucketBits > 8) ? (maxSignificantBucketBits - 8) : 0),
       _where(),
       _numPartitions(numPartitions),
       _readyForIterate(true)
diff --git a/searchlib/src/vespa/searchlib/docstore/compacter.h b/searchlib/src/vespa/searchlib/docstore/compacter.h
index 7dff421c1ba..57f65b249f5 100644
--- a/searchlib/src/vespa/searchlib/docstore/compacter.h
+++ b/searchlib/src/vespa/searchlib/docstore/compacter.h
@@ -29,7 +29,7 @@ public:
     ~BucketIndexStore() override;
     size_t toPartitionId(document::BucketId bucketId) const noexcept {
         uint64_t sortableBucketId = bucketId.toKey();
-        return (sortableBucketId >> _unSignificantBucketBits) % _numPartitions;
+        return (sortableBucketId >> _inSignificantBucketBits) % _numPartitions;
     }
     void store(const StoreByBucket::Index & index) override;
     size_t getBucketCount() const noexcept;
@@ -48,7 +48,7 @@ private:
         size_t                        _partitionId;
         IndexVector::const_iterator   _current;
     };
-    size_t       _unSignificantBucketBits;
+    size_t       _inSignificantBucketBits;
     IndexVector  _where;
     uint32_t     _numPartitions;
     bool         _readyForIterate;