Skip to content

Commit

Permalink
Include nested reference collections. (#1683)
Browse files Browse the repository at this point in the history 
* Fix `Collection::prune_ref_doc` segfault.

* Allow including nested collections if a reference to it is present.
  • Loading branch information
@happy-san
happy-san committed Apr 23, 2024
1 parent 495f91c commit f1e6403
Show file tree
Hide file tree
Showing 2 changed files with 114 additions and 6 deletions.
19 changes: 14 additions & 5 deletions src/collection.cpp
View file
Expand Up @@ -5187,9 +5187,14 @@ Option<bool> Collection::prune_ref_doc(nlohmann::json& doc,
}

// Include nested join references.
if (!ref_include_exclude.nested_join_includes.empty() && !references.coll_to_references->empty()) {
if (!ref_include_exclude.nested_join_includes.empty()) {
// Passing empty references in case the nested include collection is not joined, but it still can be included
// if we have a reference to it.
std::map<std::string, reference_filter_result_t> refs;
auto nested_include_exclude_op = include_references(nest_ref_doc ? doc[key] : doc, ref_doc_seq_id,
ref_collection.get(), references.coll_to_references[0],
ref_collection.get(),
references.coll_to_references == nullptr ? refs :
references.coll_to_references[0],
ref_include_exclude.nested_join_includes);
if (!nested_include_exclude_op.ok()) {
return nested_include_exclude_op;
Expand Down Expand Up @@ -5247,10 +5252,14 @@ Option<bool> Collection::prune_ref_doc(nlohmann::json& doc,
}

// Include nested join references.
if (!ref_include_exclude.nested_join_includes.empty() &&
references.coll_to_references != nullptr && !references.coll_to_references->empty()) {
if (!ref_include_exclude.nested_join_includes.empty()) {
// Passing empty references in case the nested include collection is not joined, but it still can be included
// if we have a reference to it.
std::map<std::string, reference_filter_result_t> refs;
auto nested_include_exclude_op = include_references(nest_ref_doc ? doc[key].at(i) : doc, ref_doc_seq_id,
ref_collection.get(), references.coll_to_references[i],
ref_collection.get(),
references.coll_to_references == nullptr ? refs :
references.coll_to_references[i],
ref_include_exclude.nested_join_includes);
if (!nested_include_exclude_op.ok()) {
return nested_include_exclude_op;
Expand Down
101 changes: 100 additions & 1 deletion test/collection_join_test.cpp
View file
Expand Up @@ -2258,7 +2258,6 @@ TEST_F(CollectionJoinTest, FilterByNestedReferences) {
ASSERT_TRUE(search_op.ok());

res_obj = nlohmann::json::parse(json_res);
LOG(INFO) << res_obj.dump();
// coll_d_2 -> coll_c_2 -> coll_b_0, coll_b_1
//
// coll_d_1 -> coll_c_1 -> coll_b_1
Expand All @@ -2279,6 +2278,106 @@ TEST_F(CollectionJoinTest, FilterByNestedReferences) {
ASSERT_EQ(1, res_obj["hits"][1]["document"]["Coll_C"][0]["Coll_B"].size());
ASSERT_EQ("coll_b_1", res_obj["hits"][1]["document"]["Coll_C"][0]["Coll_B"][0]["title"]);

auto doc = R"({
"title": "coll_b_3",
"ref_coll_a": "0"
})"_json;
auto doc_add_op = collectionManager.get_collection("Coll_B")->add(doc.dump());
if (!doc_add_op.ok()) {
LOG(INFO) << doc_add_op.error();
}
ASSERT_TRUE(doc_add_op.ok());

doc = R"({
"title": "coll_c_4",
"ref_coll_b": ["3"]
})"_json;
doc_add_op = collectionManager.get_collection("Coll_C")->add(doc.dump());
if (!doc_add_op.ok()) {
LOG(INFO) << doc_add_op.error();
}
ASSERT_TRUE(doc_add_op.ok());

doc = R"({
"title": "coll_d_3",
"ref_coll_c": ["4"]
})"_json;
doc_add_op = collectionManager.get_collection("Coll_D")->add(doc.dump());
if (!doc_add_op.ok()) {
LOG(INFO) << doc_add_op.error();
}
ASSERT_TRUE(doc_add_op.ok());

req_params = {
{"collection", "Coll_D"},
{"q", "coll_d_3"},
{"query_by", "title"},
{"filter_by", "$Coll_C(id:*)"},
// We will be able to include Coll_A document since we join on Coll_C that has reference to Coll_B that in
// turn has a reference to Coll_A.
{"include_fields", "title, $Coll_C(title), $Coll_B(title, $Coll_A(title))"}
};
search_op = collectionManager.do_search(req_params, embedded_params, json_res, now_ts);
ASSERT_TRUE(search_op.ok());

res_obj = nlohmann::json::parse(json_res);
// coll_d_3 -> coll_c_4 -> coll_b_3 -> coll_a_0
ASSERT_EQ(1, res_obj["found"].get<size_t>());
ASSERT_EQ(1, res_obj["hits"].size());
ASSERT_EQ(3, res_obj["hits"][0]["document"].size());
ASSERT_EQ("coll_d_3", res_obj["hits"][0]["document"]["title"]);

ASSERT_EQ(1, res_obj["hits"][0]["document"]["Coll_C"].size());
ASSERT_EQ("coll_c_4", res_obj["hits"][0]["document"]["Coll_C"]["title"]);

ASSERT_EQ(1, res_obj["hits"][0]["document"]["Coll_B"].size());
ASSERT_EQ("coll_b_3", res_obj["hits"][0]["document"]["Coll_B"][0]["title"]);
ASSERT_EQ(1, res_obj["hits"][0]["document"]["Coll_B"][0]["Coll_A"].size());
ASSERT_EQ("coll_a_0", res_obj["hits"][0]["document"]["Coll_B"][0]["Coll_A"]["title"]);

schema_json =
R"({
"name": "Coll_E",
"fields": [
{"name": "title", "type": "string"},
{"name": "ref_coll_b", "type": "string", "reference": "Coll_B.id"}
]
})"_json;
collection_create_op = collectionManager.create_collection(schema_json);
doc = R"({
"title": "coll_e_0",
"ref_coll_b": "3"
})"_json;
doc_add_op = collectionManager.get_collection("Coll_E")->add(doc.dump());
if (!doc_add_op.ok()) {
LOG(INFO) << doc_add_op.error();
}
ASSERT_TRUE(doc_add_op.ok());

req_params = {
{"collection", "Coll_D"},
{"q", "coll_d_3"},
{"query_by", "title"},
{"filter_by", "$Coll_C(id:*)"},
// We won't be able to include Coll_E document since we neither join on it nor we have any reference to it.
{"include_fields", "title, $Coll_C(title), $Coll_B(title, $Coll_E(title))"}
};
search_op = collectionManager.do_search(req_params, embedded_params, json_res, now_ts);
ASSERT_TRUE(search_op.ok());

res_obj = nlohmann::json::parse(json_res);
ASSERT_EQ(1, res_obj["found"].get<size_t>());
ASSERT_EQ(1, res_obj["hits"].size());
ASSERT_EQ(3, res_obj["hits"][0]["document"].size());
ASSERT_EQ("coll_d_3", res_obj["hits"][0]["document"]["title"]);

ASSERT_EQ(1, res_obj["hits"][0]["document"]["Coll_C"].size());
ASSERT_EQ("coll_c_4", res_obj["hits"][0]["document"]["Coll_C"]["title"]);

ASSERT_EQ(1, res_obj["hits"][0]["document"]["Coll_B"].size());
ASSERT_EQ("coll_b_3", res_obj["hits"][0]["document"]["Coll_B"][0]["title"]);
ASSERT_EQ(0, res_obj["hits"][0]["document"]["Coll_B"][0].count("Coll_E"));

schema_json =
R"({
"name": "products",
Expand Down

0 comments on commit f1e6403

Please sign in to comment.