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

Only the last facet query token must be treated as prefix.

kishorenc991bf03bcba7
diff --git a/include/collection.h b/include/collection.h
index 971527eb..907693a2 100644
--- a/include/collection.h
+++ b/include/collection.h
@@ -169,7 +169,7 @@ public:
Collection(const std::string name, const uint32_t collection_id, const uint64_t created_at,
const uint32_t next_seq_id, Store *store, const std::vector<field> & fields,
- const std::string & default_sorting_field, const size_t num_indices=4);
+ const std::string & default_sorting_field, const size_t num_indices);
~Collection();
diff --git a/src/index.cpp b/src/index.cpp
index 4dc469d4..3d3717c5 100644
--- a/src/index.cpp
+++ b/src/index.cpp
@@ -555,10 +555,12 @@ void Index::do_facets(std::vector<facet> & facets, facet_query_t & facet_query,
auto & q = query_tokens[qtoken_index];
string_utils.unicode_normalize(q);
int bounded_cost = (q.size() < 3) ? 0 : 1;
+ bool prefix_search = (qtoken_index == (query_tokens.size()-1)); // only last token must be used as prefix
+
std::vector<art_leaf*> leaves;
art_fuzzy_search(search_index.at(a_facet.field_name), (const unsigned char *) q.c_str(),
- q.size(),0, bounded_cost, 10000,
- token_ordering::MAX_SCORE, true, leaves);
+ q.size(), 0, bounded_cost, 10000,
+ token_ordering::MAX_SCORE, prefix_search, leaves);
for(size_t i = 0; i < leaves.size(); i++) {
const auto & leaf = leaves[i];
// calculate hash without terminating null char
diff --git a/test/collection_test.cpp b/test/collection_test.cpp
index 49ad46ba..b2e47020 100644
--- a/test/collection_test.cpp
+++ b/test/collection_test.cpp
@@ -2082,6 +2082,22 @@ TEST_F(CollectionTest, FacetCountsHighlighting) {
ASSERT_STREQ("Cell Phone Accessories", results["facet_counts"][0]["counts"][0]["value"].get<std::string>().c_str());
ASSERT_STREQ("Cell Phone <mark>Acces</mark>sories", results["facet_counts"][0]["counts"][0]["highlighted"].get<std::string>().c_str());
+ // ensure that only the last token is treated as prefix search
+
+ coll1->remove("100");
+ doc["categories"] = {"Cell Phones", "Cell Phone Accessories", "Cellophanes"};
+ coll1->add(doc.dump());
+
+ results = coll1->search("phone", {"categories"}, "", facets, sort_fields, 0, 10, 1,
+ token_ordering::FREQUENCY, true, 10, spp::sparse_hash_set<std::string>(),
+ spp::sparse_hash_set<std::string>(), 10, 500, "categories:cell ph").get();
+
+ ASSERT_EQ(1, results["facet_counts"].size());
+ ASSERT_EQ(2, results["facet_counts"][0]["counts"].size());
+
+ ASSERT_STREQ("Cell Phones", results["facet_counts"][0]["counts"][0]["value"].get<std::string>().c_str());
+ ASSERT_STREQ("Cell Phone Accessories", results["facet_counts"][0]["counts"][1]["value"].get<std::string>().c_str());
+
collectionManager.drop_collection("coll1");
}
@@ -2673,4 +2689,6 @@ TEST_F(CollectionTest, StringArrayFieldShouldNotAllowPlainString) {
auto add_op = coll1->add(doc.dump());
ASSERT_FALSE(add_op.ok());
ASSERT_STREQ("Field `categories` must be a string array.", add_op.error().c_str());
+
+ collectionManager.drop_collection("coll1");
}
\ No newline at end of file