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

Handle reference to an id field from an object array field. (#1721)

Harpreet Sangar8d7def3eb995
diff --git a/src/collection.cpp b/src/collection.cpp
index f2d61c13..3b36af6f 100644
--- a/src/collection.cpp
+++ b/src/collection.cpp
@@ -154,9 +154,57 @@ Option<bool> Collection::add_reference_helper_fields(nlohmann::json& document, c
auto const reference_helper_field = field_name + fields::REFERENCE_HELPER_FIELD_SUFFIX;
+ bool is_object_reference_field = flat_fields.count(field_name) != 0;
+ std::string object_key;
+ bool is_object_array = false;
+ if (is_object_reference_field) {
+ object_reference_helper_fields.insert(reference_helper_field);
+
+ std::vector<std::string> tokens;
+ StringUtils::split(field_name, tokens, ".");
+ if (schema.count(tokens[0]) == 0) {
+ return Option<bool>(400, "Could not find `" + tokens[0] + "` object/object[] field in the schema.");
+ }
+ object_key = tokens[0];
+ is_object_array = schema.at(object_key).is_array();
+ }
+
if (reference_field_name == "id") {
auto id_field_type_error_op = Option<bool>(400, "Field `" + field_name + "` must have string value.");
- if (document[field_name].is_array()) {
+ if (is_object_array) {
+ if (!document[field_name].is_array()) {
+ return Option<bool>(400, "Expected `" + field_name + "` to be an array.");
+ }
+
+ document[reference_helper_field] = nlohmann::json::array();
+ document[fields::reference_helper_fields] += reference_helper_field;
+
+ std::vector<std::string> keys;
+ StringUtils::split(field_name, keys, ".");
+ auto const& object_array = document[keys[0]];
+
+ for (uint32_t i = 0; i < object_array.size(); i++) {
+ if (optional && object_array[i].count(keys[1]) == 0) {
+ continue;
+ } else if (object_array[i].count(keys[1]) == 0) {
+ return Option<bool>(400, "Object at index `" + std::to_string(i) + "` is missing `" + field_name + "`.");
+ } else if (!object_array[i].at(keys[1]).is_string()) {
+ return id_field_type_error_op;
+ }
+
+ auto id = object_array[i].at(keys[1]).get<std::string>();
+ auto ref_doc_id_op = ref_collection->doc_id_to_seq_id_with_lock(id);
+ if (!ref_doc_id_op.ok()) {
+ return Option<bool>(400, "Referenced document having `id: " + id +
+ "` not found in the collection `" +
+ reference_collection_name + "`." );
+ }
+
+ // Adding the index of the object along with referenced doc id to account for the scenario where a
+ // reference field of an object array might be optional and missing.
+ document[reference_helper_field] += nlohmann::json::array({i, ref_doc_id_op.get()});
+ }
+ } else if (document[field_name].is_array()) {
document[reference_helper_field] = nlohmann::json::array();
document[fields::reference_helper_fields] += reference_helper_field;
@@ -219,22 +267,7 @@ Option<bool> Collection::add_reference_helper_fields(nlohmann::json& document, c
"` of type `" + ref_field.type + "`.");
}
- bool is_object_reference_field = flat_fields.count(field_name) != 0;
- std::string object_key;
- bool is_object_array;
- if (is_object_reference_field) {
- object_reference_helper_fields.insert(reference_helper_field);
-
- std::vector<std::string> tokens;
- StringUtils::split(field_name, tokens, ".");
- if (schema.count(tokens[0]) == 0) {
- return Option<bool>(400, "Could not find `" + tokens[0] + "` object/object[] field in the schema.");
- }
- object_key = tokens[0];
- is_object_array = schema.at(object_key).is_array();
- }
-
- if (is_object_reference_field && is_object_array) {
+ if (is_object_array) {
if (!document[field_name].is_array()) {
return Option<bool>(400, "Expected `" + field_name + "` to be an array.");
}
diff --git a/test/collection_join_test.cpp b/test/collection_join_test.cpp
index 9f07c4ad..960799b5 100644
--- a/test/collection_join_test.cpp
+++ b/test/collection_join_test.cpp
@@ -3801,6 +3801,89 @@ TEST_F(CollectionJoinTest, FilterByObjectReferenceField) {
ASSERT_EQ(1, res_obj["hits"][1]["document"]["coll2"][0].count("coll_id"));
ASSERT_EQ("b", res_obj["hits"][1]["document"]["coll2"][0]["coll_id"]);
+ schema_json =
+ R"({
+ "name": "coll3",
+ "fields": [
+ {"name": "coll_id", "type": "string"},
+ {"name": "object.reference_array", "type": "string[]", "reference": "Products.id", "optional": true},
+ {"name": "object", "type": "object"}
+ ],
+ "enable_nested_fields": true
+ })"_json;
+ documents = {
+ R"({
+ "coll_id": "a",
+ "object": {}
+ })"_json,
+ R"({
+ "coll_id": "b",
+ "object": {
+ "reference_array": ["0", "1"]
+ }
+ })"_json
+ };
+
+ collection_create_op = collectionManager.create_collection(schema_json);
+ ASSERT_TRUE(collection_create_op.ok());
+ for (auto const &json: documents) {
+ auto add_op = collection_create_op.get()->add(json.dump());
+ if (!add_op.ok()) {
+ LOG(INFO) << add_op.error();
+ }
+ ASSERT_TRUE(add_op.ok());
+ }
+
+ req_params = {
+ {"collection", "coll3"},
+ {"q", "*"},
+ {"include_fields", "$Products(product_id)"}
+ };
+ search_op_bool = collectionManager.do_search(req_params, embedded_params, json_res, now_ts);
+ ASSERT_TRUE(search_op_bool.ok());
+
+ res_obj = nlohmann::json::parse(json_res);
+ ASSERT_EQ(2, res_obj["found"].get<size_t>());
+ ASSERT_EQ(2, res_obj["hits"].size());
+ ASSERT_EQ(3, res_obj["hits"][0]["document"].size());
+ ASSERT_EQ("b", res_obj["hits"][0]["document"]["coll_id"]);
+ ASSERT_EQ(2, res_obj["hits"][0]["document"]["object"].size());
+ ASSERT_EQ("0", res_obj["hits"][0]["document"]["object"]["reference_array"][0]);
+ ASSERT_EQ("1", res_obj["hits"][0]["document"]["object"]["reference_array"][1]);
+ ASSERT_EQ(1, res_obj["hits"][0]["document"]["object"].count("Products"));
+ ASSERT_EQ(2, res_obj["hits"][0]["document"]["object"]["Products"].size());
+ ASSERT_EQ(1, res_obj["hits"][0]["document"]["object"]["Products"][0].count("product_id"));
+ ASSERT_EQ("product_a", res_obj["hits"][0]["document"]["object"]["Products"][0]["product_id"]);
+ ASSERT_EQ(1, res_obj["hits"][0]["document"]["object"]["Products"][1].count("product_id"));
+ ASSERT_EQ("product_b", res_obj["hits"][0]["document"]["object"]["Products"][1]["product_id"]);
+ ASSERT_EQ(3, res_obj["hits"][1]["document"].size());
+ ASSERT_EQ("a", res_obj["hits"][1]["document"]["coll_id"]);
+ ASSERT_EQ(0, res_obj["hits"][1]["document"]["object"].size());
+
+ req_params = {
+ {"collection", "Products"},
+ {"q", "*"},
+ {"filter_by", "$coll3(id: *)"},
+ {"include_fields", "$coll3(coll_id)"}
+ };
+ search_op_bool = collectionManager.do_search(req_params, embedded_params, json_res, now_ts);
+ ASSERT_TRUE(search_op_bool.ok());
+
+ res_obj = nlohmann::json::parse(json_res);
+ ASSERT_EQ(2, res_obj["found"].get<size_t>());
+ ASSERT_EQ(2, res_obj["hits"].size());
+ ASSERT_EQ(5, res_obj["hits"][0]["document"].size());
+ ASSERT_EQ("product_b", res_obj["hits"][0]["document"]["product_id"]);
+ ASSERT_EQ(1, res_obj["hits"][0]["document"].count("coll3"));
+ ASSERT_EQ(1, res_obj["hits"][0]["document"]["coll3"].size());
+ ASSERT_EQ(1, res_obj["hits"][0]["document"]["coll3"][0].count("coll_id"));
+ ASSERT_EQ("b", res_obj["hits"][0]["document"]["coll3"][0]["coll_id"]);
+ ASSERT_EQ("product_a", res_obj["hits"][1]["document"]["product_id"]);
+ ASSERT_EQ(1, res_obj["hits"][1]["document"].count("coll3"));
+ ASSERT_EQ(1, res_obj["hits"][1]["document"]["coll3"].size());
+ ASSERT_EQ(1, res_obj["hits"][1]["document"]["coll3"][0].count("coll_id"));
+ ASSERT_EQ("b", res_obj["hits"][1]["document"]["coll3"][0]["coll_id"]);
+
schema_json =
R"({
"name": "Portions",