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

Exhaustive search should not always be enabled during token drop search.

Kishore Nallan27b392cee95b
diff --git a/src/index.cpp b/src/index.cpp
index 28ab879f..73b6e7e6 100644
--- a/src/index.cpp
+++ b/src/index.cpp
@@ -2269,6 +2269,7 @@ void Index::search_field(const uint8_t & field_id,
costs[i] = token_to_costs[i][q.rem];
}
+ unique_tokens.clear();
token_candidates_vec.clear();
size_t token_index = 0;
@@ -2388,7 +2389,7 @@ void Index::search_field(const uint8_t & field_id,
sort_fields, num_typos,searched_queries, topster, groups_processed, all_result_ids,
all_result_ids_len, field_num_results, group_limit, group_by_fields,
prioritize_exact_match, concurrency,
- token_order, prefix, drop_tokens_threshold, typo_tokens_threshold, combination_limit);
+ token_order, prefix, drop_tokens_threshold, typo_tokens_threshold, exhaustive_search);
}
}
diff --git a/test/collection_test.cpp b/test/collection_test.cpp
index 5bef8283..bac2bf8c 100644
--- a/test/collection_test.cpp
+++ b/test/collection_test.cpp
@@ -317,7 +317,7 @@ TEST_F(CollectionTest, SkipUnindexedTokensDuringPhraseSearch) {
spp::sparse_hash_set<std::string>(),
spp::sparse_hash_set<std::string>(), 10, "", 30, 5,
"", 10).get();
- ASSERT_EQ(9, results["hits"].size());
+ ASSERT_EQ(7, results["hits"].size());
results.clear();
results = collection->search("the a", query_fields, "", facets, sort_fields, {0}, 10, 1, FREQUENCY, {false}, 0).get();
@@ -347,10 +347,11 @@ TEST_F(CollectionTest, SkipUnindexedTokensDuringPhraseSearch) {
TEST_F(CollectionTest, PartialPhraseSearch) {
std::vector<std::string> facets;
- nlohmann::json results = collection->search("rocket research", query_fields, "", facets, sort_fields, {0}, 10).get();
- ASSERT_EQ(6, results["hits"].size());
+ nlohmann::json results = collection->search("rocket research", query_fields, "", facets,
+ sort_fields, {0}, 10, 1, FREQUENCY, {false}, 10).get();
+ ASSERT_EQ(4, results["hits"].size());
- std::vector<std::string> ids = {"19", "1", "10", "8", "16", "17"};
+ std::vector<std::string> ids = {"1", "8", "16", "17"};
for(size_t i = 0; i < results["hits"].size(); i++) {
nlohmann::json result = results["hits"].at(i);
@@ -648,7 +649,7 @@ TEST_F(CollectionTest, PrefixSearching) {
}
// only the last token in the query should be used for prefix search - so, "math" should not match "mathematics"
- results = collection->search("math fx", query_fields, "", facets, sort_fields, {0}, 1, 1, FREQUENCY, {true}).get();
+ results = collection->search("math fx", query_fields, "", facets, sort_fields, {0}, 1, 1, FREQUENCY, {true}, 0).get();
ASSERT_EQ(0, results["hits"].size());
// single and double char prefixes should set a ceiling on the num_typos possible