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

Consider cross field matching during token splitting.

Kishore Nallan623a65c5385d
diff --git a/src/index.cpp b/src/index.cpp
index 3be32394..bd36cec6 100644
--- a/src/index.cpp
+++ b/src/index.cpp
@@ -5514,7 +5514,7 @@ void Index::resolve_space_as_typos(std::vector<std::string>& qtokens, const stri
leaves.push_back(leaf);
}
- if(candidate_tokens.size() == leaves.size() && common_results_exist(leaves, false)) {
+ if(common_results_exist(leaves, false)) {
resolved_queries.push_back(candidate_tokens);
return;
}
diff --git a/test/collection_specific_more_test.cpp b/test/collection_specific_more_test.cpp
index 64ad5be6..be76ec9c 100644
--- a/test/collection_specific_more_test.cpp
+++ b/test/collection_specific_more_test.cpp
@@ -609,3 +609,25 @@ TEST_F(CollectionSpecificMoreTest, ExactFilteringOnArray) {
ASSERT_EQ(0, results["hits"].size());
}
+
+TEST_F(CollectionSpecificMoreTest, SplitTokensCrossFieldMatching) {
+ 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();
+
+ nlohmann::json doc1;
+ doc1["id"] = "0";
+ doc1["name"] = "Vitamin C1";
+ doc1["brand"] = "Paulas Choice";
+
+ ASSERT_TRUE(coll1->add(doc1.dump()).ok());
+
+ auto results = coll1->search("paulaschoice c1", {"name", "brand"},
+ "", {}, {}, {2}, 10,
+ 1, FREQUENCY, {true},
+ 0).get();
+
+ ASSERT_EQ(1, results["hits"].size());
+ collectionManager.drop_collection("coll1");
+}
diff --git a/test/collection_specific_test.cpp b/test/collection_specific_test.cpp
index 528442b8..eaaebb83 100644
--- a/test/collection_specific_test.cpp
+++ b/test/collection_specific_test.cpp
@@ -2076,6 +2076,9 @@ TEST_F(CollectionSpecificTest, SearchShouldSplitAndJoinTokens) {
results = coll1->search("the pressurecooker", {"title"}, "", {}, {}, {0}, 10, 1, FREQUENCY, {false}, 0).get();
ASSERT_EQ(1, results["hits"].size());
+ results = coll1->search("pressurecooker great", {"title"}, "", {}, {}, {0}, 10, 1, FREQUENCY, {false}, 0).get();
+ ASSERT_EQ(1, results["hits"].size());
+
// splitting requires tokens to co-occur as a phrase in the dataset
results = coll1->search("the pressureis", {"title"}, "", {}, {}, {0}, 10, 1, FREQUENCY, {false}, 0).get();