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
C++ algoliadatastoreelasticsearchenterprise-searchfacetingfull-text-searchfuzzy-searchgeosearchin-memoryinstantsearchmerchandisingpineconesearchsearch-enginesemantic-searchsimilarity-searchsite-searchsynonymstypo-tolerancevector-search
Return vector distance when sort by vector query.
Kishore Nallan05b0faa955f2
diff --git a/src/collection.cpp b/src/collection.cppindex d40e4a48..7ea00e61 100644--- a/src/collection.cpp+++ b/src/collection.cpp@@ -2826,6 +2826,9 @@ Option<nlohmann::json> Collection::search(std::string raw_query,reference_lat_lng, sort_field.unit);} else if(sort_field.geopoint != 0) {geo_distances[sort_field.name] = std::abs(field_order_kv->scores[sort_field_index]);+ } else if(sort_field.name == sort_field_const::vector_query &&+ !sort_field.vector_query.query.field_name.empty()) {+ wrapper_doc["vector_distance"] = Index::int64_t_to_float(field_order_kv->scores[sort_field_index]);}}diff --git a/test/collection_vector_search_test.cpp b/test/collection_vector_search_test.cppindex af1c9516..cf1b6930 100644--- a/test/collection_vector_search_test.cpp+++ b/test/collection_vector_search_test.cpp@@ -1272,6 +1272,49 @@ TEST_F(CollectionVectorTest, EmbeddOptionalFieldNullValueUpsert) {ASSERT_EQ("Field `tags` must be an array of string.", add_op.error());}+TEST_F(CollectionVectorTest, SortKeywordSearchWithAutoEmbedVector) {+ nlohmann::json schema = R"({+ "name": "coll1",+ "fields": [+ {"name": "title", "type": "string"},+ {"name": "points", "type": "int32"},+ {"name": "embedding", "type":"float[]", "embed":{"from": ["title"],+ "model_config": {"model_name": "ts/e5-small"}}}+ ]+ })"_json;++ EmbedderManager::set_model_dir("/tmp/typesense_test/models");++ Collection* coll1 = collectionManager.create_collection(schema).get();++ nlohmann::json doc;+ doc["id"] = "0";+ doc["title"] = "The Lord of the Rings";+ doc["points"] = 100;++ auto add_op = coll1->add(doc.dump());+ ASSERT_TRUE(add_op.ok());++ std::vector<sort_by> sort_by_list = {sort_by("_vector_query(embedding:([]))", "asc")};++ auto results = coll1->search("lord", {"title"}, "", {}, sort_by_list, {0}, 10, 1, FREQUENCY, {true},+ Index::DROP_TOKENS_THRESHOLD,+ spp::sparse_hash_set<std::string>(),+ spp::sparse_hash_set<std::string>()).get();++ LOG(INFO) << results["hits"][0]["vector_distance"].get<float>();++ //ASSERT_EQ(1, results["found"].get<size_t>());+ //ASSERT_EQ(1.0f, results["hits"][0]["vector_distance"].get<float>());++ results = coll1->search("lord", {"title"}, "", {}, sort_by_list, {0}, 10, 1, FREQUENCY, {true},+ Index::DROP_TOKENS_THRESHOLD,+ spp::sparse_hash_set<std::string>(),+ spp::sparse_hash_set<std::string>()).get();++ LOG(INFO) << results["hits"][0]["vector_distance"].get<float>();+}+TEST_F(CollectionVectorTest, HybridSearchWithExplicitVector) {nlohmann::json schema = R"({"name": "objects",