Skip to content

Open Source alternative to Algolia + Pinecone and an Easier-to-Use alternative to ElasticSearch ⚡ 🔍 ✨ Fast, typo tolerant, in-memory fuzzy Search Engine for building delightful search experiences

Fix block intersection contination.

Kishore Nallanbe44d92539e9
diff --git a/src/posting.cpp b/src/posting.cpp
index 293b9b53..ff7412f8 100644
--- a/src/posting.cpp
+++ b/src/posting.cpp
@@ -408,15 +408,15 @@ bool posting_t::block_intersect(const std::vector<void*>& raw_posting_lists, siz
std::vector<uint32_t> expanded_plist_indices;
to_expanded_plists(raw_posting_lists, plists, expanded_plist_indices);
- bool done = posting_list_t::block_intersect(plists, batch_size, its, iter_state);
+ bool has_more = posting_list_t::block_intersect(plists, batch_size, its, iter_state);
- if(done) {
+ if(!has_more) {
for(uint32_t expanded_plist_index: expanded_plist_indices) {
delete plists[expanded_plist_index];
}
}
- return done;
+ return has_more;
}
void posting_t::destroy_list(void*& obj) {
diff --git a/src/posting_list.cpp b/src/posting_list.cpp
index bc63c6ed..e95fd283 100644
--- a/src/posting_list.cpp
+++ b/src/posting_list.cpp
@@ -651,7 +651,7 @@ bool posting_list_t::block_intersect(const std::vector<posting_list_t*>& posting
}
if(iter_state.ids.size() == batch_size) {
- return true;
+ return !at_end2(its);
}
}
break;
@@ -678,7 +678,7 @@ bool posting_list_t::block_intersect(const std::vector<posting_list_t*>& posting
}
if(iter_state.ids.size() == batch_size) {
- return true;
+ return !at_end(its);
}
}
}
diff --git a/test/posting_list_test.cpp b/test/posting_list_test.cpp
index 71a66670..5a646abb 100644
--- a/test/posting_list_test.cpp
+++ b/test/posting_list_test.cpp
@@ -605,6 +605,34 @@ TEST(PostingListTest, IntersectionBasics) {
ASSERT_EQ(2, result_ids.size());
ASSERT_EQ(3, result_ids[0]);
ASSERT_EQ(20, result_ids[1]);
+
+ std::vector<posting_list_t::iterator_t> its;
+ posting_list_t::result_iter_state_t iter_state;
+ bool has_more = posting_list_t::block_intersect(lists, 2, its, iter_state);
+ ASSERT_FALSE(has_more);
+ ASSERT_EQ(2, iter_state.ids.size());
+ ASSERT_EQ(3, iter_state.ids[0]);
+ ASSERT_EQ(20, iter_state.ids[1]);
+
+ ASSERT_EQ(2, iter_state.blocks.size());
+ ASSERT_EQ(3, iter_state.blocks[0].size());
+ ASSERT_EQ(3, iter_state.blocks[1].size());
+
+ ASSERT_EQ(2, iter_state.indices.size());
+
+ // try with smaller batch size
+
+ std::vector<posting_list_t::iterator_t> its2;
+ posting_list_t::result_iter_state_t iter_state2;
+ has_more = posting_list_t::block_intersect(lists, 1, its2, iter_state2);
+ ASSERT_TRUE(has_more);
+ ASSERT_EQ(1, iter_state2.ids.size());
+ ASSERT_EQ(3, iter_state2.ids[0]);
+
+ has_more = posting_list_t::block_intersect(lists, 1, its2, iter_state2);
+ ASSERT_FALSE(has_more);
+ ASSERT_EQ(1, iter_state2.ids.size());
+ ASSERT_EQ(20, iter_state2.ids[0]);
}
TEST(PostingListTest, ResultsAndOffsetsBasics) {
@@ -660,19 +688,9 @@ TEST(PostingListTest, ResultsAndOffsetsBasics) {
lists.push_back(&p3);
std::vector<posting_list_t::iterator_t> its;
-
posting_list_t::result_iter_state_t iter_state;
- posting_list_t::block_intersect(lists, 2, its, iter_state);
-
- ASSERT_EQ(2, iter_state.ids.size());
- ASSERT_EQ(3, iter_state.ids[0]);
- ASSERT_EQ(20, iter_state.ids[1]);
-
- ASSERT_EQ(2, iter_state.blocks.size());
- ASSERT_EQ(3, iter_state.blocks[0].size());
- ASSERT_EQ(3, iter_state.blocks[1].size());
-
- ASSERT_EQ(2, iter_state.indices.size());
+ bool has_more = posting_list_t::block_intersect(lists, 2, its, iter_state);
+ ASSERT_FALSE(has_more);
std::vector<std::unordered_map<size_t, std::vector<token_positions_t>>> array_token_positions_vec;
posting_list_t::get_offsets(iter_state, array_token_positions_vec);