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
Fix field-wise prefix searching edge case.
Kishore Nallan77efb5c0ed51
diff --git a/src/index.cpp b/src/index.cppindex 804147d2..7a584d45 100644--- a/src/index.cpp+++ b/src/index.cpp@@ -1171,7 +1171,7 @@ void Index::search_all_candidates(const size_t num_search_fields,LOG(INFO) << "token_candidates_vec.size(): " << token_candidates_vec.size();}*/- std::unordered_set<std::string> trimmed_candidates;+ std::set<std::string> trimmed_candidates;if(token_candidates_vec.size() > 1 && token_candidates_vec.back().candidates.size() > max_candidates) {std::vector<uint32_t> temp_ids;@@ -1187,7 +1187,7 @@ void Index::search_all_candidates(const size_t num_search_fields,const bool prefix_search = query_tokens.back().is_prefix_searched;const uint32_t token_num_typos = query_tokens.back().num_typos;- const bool token_prefix = query_tokens.back().is_prefix_searched;+ const bool token_prefix = (token_str.size() > token_candidates_vec.back().token.value.size());auto token_c_str = (const unsigned char*) token_str.c_str();const size_t token_len = token_str.size() + 1;diff --git a/test/collection_specific_more_test.cpp b/test/collection_specific_more_test.cppindex ecb0d50d..fb916320 100644--- a/test/collection_specific_more_test.cpp+++ b/test/collection_specific_more_test.cpp@@ -631,3 +631,50 @@ TEST_F(CollectionSpecificMoreTest, SplitTokensCrossFieldMatching) {ASSERT_EQ(1, results["hits"].size());collectionManager.drop_collection("coll1");}++TEST_F(CollectionSpecificMoreTest, PrefixSearchOnSpecificFields) {+ std::vector<field> fields = {field("name", field_types::STRING, false),+ field("brand", field_types::STRING, false),};++ Collection* coll1 = collectionManager.create_collection("coll1", 1, fields).get();++ // atleast 4 tokens that begin with "girl" to trigger this regression+ std::vector<std::string> names = {+ "Jungle Girl", "Jungle Girlz", "Jam Foo1", "Jam Foo2", "Jam Foo3", "Jam Foo4", "Jam Foo"+ };++ std::vector<std::string> brands = {+ "Foobar", "Foobar2", "Girlx", "Girly", "Girlz", "Girlz", "Girlzz"+ };++ for(size_t i = 0; i < names.size(); i++) {+ nlohmann::json doc1;+ doc1["name"] = names[i];+ doc1["brand"] = brands[i];+ ASSERT_TRUE(coll1->add(doc1.dump()).ok());+ }++ auto results = coll1->search("jungle girl", {"name", "brand"},+ "", {}, {}, {0}, 10,+ 1, FREQUENCY, {false, true},+ 0).get();++ ASSERT_EQ(1, results["hits"].size());++ results = coll1->search("jam foo", {"name"},+ "", {}, {}, {0}, 10,+ 1, FREQUENCY, {true},+ 0).get();++ ASSERT_EQ(4, results["hits"].size());+ ASSERT_EQ("6", results["hits"][0]["document"]["id"].get<std::string>());++ results = coll1->search("jam foo", {"name"},+ "", {}, {}, {0}, 10,+ 1, FREQUENCY, {false},+ 0).get();++ ASSERT_EQ(1, results["hits"].size());++ collectionManager.drop_collection("coll1");+}