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

Fix more edge cases in highlighting.

Kishore Nallanb2c12a9b2cef
diff --git a/include/topster.h b/include/topster.h
index 5550fbd0..19ba746c 100644
--- a/include/topster.h
+++ b/include/topster.h
@@ -16,6 +16,9 @@ struct KV {
uint64_t distinct_key{};
int64_t scores[3]{}; // match score + 2 custom attributes
+ // to be used only in final aggregation
+ uint64_t* query_indices = nullptr;
+
KV(uint8_t field_id, uint16_t queryIndex, uint32_t token_bits, uint64_t key, uint64_t distinct_key,
uint8_t match_score_index, const int64_t *scores):
field_id(field_id), match_score_index(match_score_index),
@@ -27,6 +30,69 @@ struct KV {
}
KV() = default;
+
+ KV(KV& kv) = default;
+
+ KV(KV&& kv) noexcept : field_id(kv.field_id), match_score_index(kv.match_score_index),
+ query_index(kv.query_index), array_index(kv.array_index), token_bits(kv.token_bits),
+ key(kv.key), distinct_key(kv.distinct_key) {
+
+ scores[0] = kv.scores[0];
+ scores[1] = kv.scores[1];
+ scores[2] = kv.scores[2];
+
+ query_indices = kv.query_indices;
+ kv.query_indices = nullptr;
+ }
+
+ KV& operator=(KV&& kv) noexcept {
+ if (this != &kv) {
+ field_id = kv.field_id;
+ match_score_index = kv.match_score_index;
+ query_index = kv.query_index;
+ array_index = kv.array_index;
+ token_bits = kv.token_bits;
+ key = kv.key;
+ distinct_key = kv.distinct_key;
+
+ scores[0] = kv.scores[0];
+ scores[1] = kv.scores[1];
+ scores[2] = kv.scores[2];
+
+ delete[] query_indices;
+ query_indices = kv.query_indices;
+ kv.query_indices = nullptr;
+ }
+
+ return *this;
+ }
+
+ KV& operator=(KV& kv) noexcept {
+ if (this != &kv) {
+ field_id = kv.field_id;
+ match_score_index = kv.match_score_index;
+ query_index = kv.query_index;
+ array_index = kv.array_index;
+ token_bits = kv.token_bits;
+ key = kv.key;
+ distinct_key = kv.distinct_key;
+
+ scores[0] = kv.scores[0];
+ scores[1] = kv.scores[1];
+ scores[2] = kv.scores[2];
+
+ delete[] query_indices;
+ query_indices = kv.query_indices;
+ kv.query_indices = nullptr;
+ }
+
+ return *this;
+ }
+
+ ~KV() {
+ delete [] query_indices;
+ query_indices = nullptr;
+ }
};
/*
diff --git a/src/collection.cpp b/src/collection.cpp
index 44590a26..640e744b 100644
--- a/src/collection.cpp
+++ b/src/collection.cpp
@@ -1405,12 +1405,22 @@ void Collection::aggregate_topster(size_t query_index, Topster& agg_topster, Top
Topster* group_topster = group_topster_entry.second;
for(const auto& map_kv: group_topster->kv_map) {
map_kv.second->query_index += query_index;
+ if(map_kv.second->query_indices != nullptr) {
+ for(size_t i = 0; i < map_kv.second->query_indices[0]; i++) {
+ map_kv.second->query_indices[i+1] += query_index;
+ }
+ }
agg_topster.add(map_kv.second);
}
}
} else {
for(const auto& map_kv: index_topster->kv_map) {
map_kv.second->query_index += query_index;
+ if(map_kv.second->query_indices != nullptr) {
+ for(size_t i = 0; i < map_kv.second->query_indices[0]; i++) {
+ map_kv.second->query_indices[i+1] += query_index;
+ }
+ }
agg_topster.add(map_kv.second);
}
}
@@ -1516,25 +1526,38 @@ void Collection::highlight_result(const field &search_field,
std::vector<art_leaf*> query_suggestion;
std::set<std::string> query_suggestion_tokens;
- for (const art_leaf *token_leaf : searched_queries[field_order_kv->query_index]) {
- // Must search for the token string fresh on that field for the given document since `token_leaf`
- // is from the best matched field and need not be present in other fields of a document.
- Index* index = indices[field_order_kv->key % num_memory_shards];
- art_leaf *actual_leaf = index->get_token_leaf(search_field.name, &token_leaf->key[0], token_leaf->key_len);
+ size_t qindex = 0;
+
+ do {
+ auto searched_query =
+ (field_order_kv->query_indices == nullptr) ? searched_queries[field_order_kv->query_index] :
+ searched_queries[field_order_kv->query_indices[qindex + 1]];
- //LOG(INFO) << "field: " << search_field.name << ", key: " << token_leaf->key;
+ for (art_leaf* token_leaf : searched_query) {
+ // Must search for the token string fresh on that field for the given document since `token_leaf`
+ // is from the best matched field and need not be present in other fields of a document.
+ Index* index = indices[field_order_kv->key % num_memory_shards];
+ art_leaf* actual_leaf = index->get_token_leaf(search_field.name, &token_leaf->key[0], token_leaf->key_len);
+
+ if(actual_leaf == nullptr) {
+ continue;
+ }
- if(actual_leaf != nullptr) {
- query_suggestion.push_back(actual_leaf);
- std::string token(reinterpret_cast<char*>(actual_leaf->key), actual_leaf->key_len-1);
- query_suggestion_tokens.insert(token);
- std::vector<uint16_t> positions;
uint32_t doc_index = actual_leaf->values->ids.indexOf(field_order_kv->key);
- auto doc_indices = new uint32_t[1];
- doc_indices[0] = doc_index;
- leaf_to_indices.push_back(doc_indices);
+ if(doc_index != actual_leaf->values->ids.getLength()) {
+ query_suggestion.push_back(actual_leaf);
+ std::string token(reinterpret_cast<char*>(actual_leaf->key), actual_leaf->key_len - 1);
+ //LOG(INFO) << "field: " << search_field.name << ", key: " << token;
+ query_suggestion_tokens.insert(token);
+ auto doc_indices = new uint32_t[1];
+ doc_indices[0] = doc_index;
+ leaf_to_indices.push_back(doc_indices);
+ }
}
- }
+
+ qindex++;
+ } while(field_order_kv->query_indices != nullptr && qindex < field_order_kv->query_indices[0]);
+
if(query_suggestion.size() != q_tokens.size()) {
// can happen for compound query matched across 2 fields when some tokens are dropped
@@ -1543,6 +1566,8 @@ void Collection::highlight_result(const field &search_field,
continue;
}
+ query_suggestion_tokens.insert(q_token);
+
Index* index = indices[field_order_kv->key % num_memory_shards];
art_leaf *actual_leaf = index->get_token_leaf(search_field.name,
reinterpret_cast<const unsigned char *>(q_token.c_str()),
@@ -1638,9 +1663,10 @@ void Collection::highlight_result(const field &search_field,
highlight.matched_tokens.emplace_back();
std::vector<std::string>& matched_tokens = highlight.matched_tokens.back();
+ bool found_first_match = false;
while(tokenizer.next(raw_token, raw_token_index, tok_start, tok_end)) {
- if(token_offsets.empty()) {
+ if(!found_first_match) {
if(snippet_start_window.size() == highlight_affix_num_tokens + 1) {
snippet_start_window.pop_front();
}
@@ -1648,7 +1674,10 @@ void Collection::highlight_result(const field &search_field,
snippet_start_window.push_back(tok_start);
}
- if (token_hits.count(raw_token) != 0 ||
+ bool token_already_found = token_hits.count(raw_token) != 0;
+
+ // ensures that the `snippet_start_offset` is always from a matched token, and not from query suggestion
+ if ((found_first_match && token_already_found) ||
(match_offset_index < match.offsets.size() &&
match.offsets[match_offset_index].offset == raw_token_index)) {
@@ -1661,9 +1690,15 @@ void Collection::highlight_result(const field &search_field,
} while(match_offset_index < match.offsets.size() &&
match.offsets[match_offset_index - 1].offset == match.offsets[match_offset_index].offset);
- if(token_offsets.size() == 1) {
+ if(!found_first_match) {
snippet_start_offset = snippet_start_window.front();
}
+
+ found_first_match = true;
+
+ } else if(query_suggestion_tokens.find(raw_token) != query_suggestion_tokens.end()) {
+ token_offsets.emplace(tok_start, tok_end);
+ token_hits.insert(raw_token);
}
if(raw_token_index == last_valid_offset + highlight_affix_num_tokens) {
@@ -1693,6 +1728,11 @@ void Collection::highlight_result(const field &search_field,
auto offset_it = token_offsets.begin();
std::stringstream highlighted_text;
+ // tokens from query might occur before actual snippet start offset: we skip that
+ while(offset_it != token_offsets.end() && offset_it->first < snippet_start_offset) {
+ offset_it++;
+ }
+
for(size_t i = snippet_start_offset; i <= snippet_end_offset; i++) {
if(offset_it != token_offsets.end()) {
if (i == offset_it->first) {
diff --git a/src/index.cpp b/src/index.cpp
index a36a7adc..ee8debdf 100644
--- a/src/index.cpp
+++ b/src/index.cpp
@@ -995,6 +995,11 @@ void Index::search_candidates(const uint8_t & field_id,
field_num_results += filtered_results_size;
+ /*if(filtered_results_size != 0) {
+ LOG(INFO) << size_t(field_id) << " - " << log_query.str() << ", filtered_results_size: " << filtered_results_size
+ << ", popcount: " << (__builtin_popcount(token_bits) - 1);
+ }*/
+
delete[] filtered_result_ids;
delete[] result_ids;
} else {
@@ -1695,13 +1700,18 @@ void Index::search(const std::vector<query_tokens_t>& field_query_tokens,
auto& kvs = seq_id_kvs.second; // each `kv` can be from a different field
std::sort(kvs.begin(), kvs.end(), Topster::is_greater);
+ kvs[0]->query_indices = new uint64_t[kvs.size() + 1];
+ kvs[0]->query_indices[0] = kvs.size();
- // LOG(INFO) << "DOC ID: " << seq_id << ", score: " << kvs[0]->scores[kvs[0]->match_score_index];
+ //LOG(INFO) << "DOC ID: " << seq_id << ", score: " << kvs[0]->scores[kvs[0]->match_score_index];
// to calculate existing aggregate scores across best matching fields
spp::sparse_hash_map<uint8_t, KV*> existing_field_kvs;
- for(const auto kv: kvs) {
- existing_field_kvs.emplace(kv->field_id, kv);
+ for(size_t kv_i = 0; kv_i < kvs.size(); kv_i++) {
+ existing_field_kvs.emplace(kvs[kv_i]->field_id, kvs[kv_i]);
+ kvs[0]->query_indices[kv_i+1] = kvs[kv_i]->query_index;
+ /*LOG(INFO) << "kv_i: " << kv_i << ", kvs[kv_i]->query_index: " << kvs[kv_i]->query_index << ", "
+ << "searched_query: " << searched_queries[kvs[kv_i]->query_index][0];*/
}
uint32_t token_bits = (uint32_t(1) << 31); // top most bit set to guarantee atleast 1 bit set
diff --git a/test/collection_locale_test.cpp b/test/collection_locale_test.cpp
index c6f5b63a..75fe60b1 100644
--- a/test/collection_locale_test.cpp
+++ b/test/collection_locale_test.cpp
@@ -263,7 +263,7 @@ TEST_F(CollectionLocaleTest, SearchAgainstThaiTextExactMatch) {
ASSERT_EQ("ติดกับดัก<mark>ราย</mark><mark>ได้</mark>ปานกลาง",
results["hits"][0]["highlights"][0]["snippet"].get<std::string>());
- ASSERT_EQ("ข้อมูลรายคนหรือ<mark>ราย</mark>บริษัทในการเชื่อมโยงส่วน<mark>ได้</mark>ส่วนเสีย",
+ ASSERT_EQ("ข้อมูล<mark>ราย</mark>คนหรือ<mark>ราย</mark>บริษัทในการเชื่อมโยงส่วน<mark>ได้</mark>ส่วนเสีย",
results["hits"][1]["highlights"][0]["snippet"].get<std::string>());
}
diff --git a/test/collection_manager_test.cpp b/test/collection_manager_test.cpp
index 8b3f507a..ac87bfc4 100644
--- a/test/collection_manager_test.cpp
+++ b/test/collection_manager_test.cpp
@@ -232,7 +232,6 @@ TEST_F(CollectionManagerTest, RestoreRecordsOnRestart) {
std::vector<std::string> facets;
nlohmann::json results = collection1->search("thomas", search_fields, "", facets, sort_fields, {0}, 10, 1, FREQUENCY, {false}).get();
- LOG(INFO) << results;
ASSERT_EQ(4, results["hits"].size());
std::unordered_map<std::string, field> schema = collection1->get_schema();
diff --git a/test/collection_specific_test.cpp b/test/collection_specific_test.cpp
index c1855479..bf74f410 100644
--- a/test/collection_specific_test.cpp
+++ b/test/collection_specific_test.cpp
@@ -648,3 +648,75 @@ TEST_F(CollectionSpecificTest, HandleBadCharactersInStringGracefully) {
collectionManager.drop_collection("coll1");
}
+
+TEST_F(CollectionSpecificTest, HighlightSecondaryFieldWithPrefixMatch) {
+ std::vector<field> fields = {field("title", field_types::STRING, false),
+ field("description", field_types::STRING, false),
+ field("points", field_types::INT32, false),};
+
+ Collection* coll1 = collectionManager.create_collection("coll1", 1, fields, "points").get();
+
+ nlohmann::json doc1;
+ doc1["id"] = "0";
+ doc1["title"] = "Functions and Equations";
+ doc1["description"] = "Use a function to solve an equation.";
+ doc1["points"] = 100;
+
+ nlohmann::json doc2;
+ doc2["id"] = "1";
+ doc2["title"] = "Function of effort";
+ doc2["description"] = "Learn all about it.";
+ doc2["points"] = 100;
+
+ ASSERT_TRUE(coll1->add(doc1.dump()).ok());
+ ASSERT_TRUE(coll1->add(doc2.dump()).ok());
+
+ auto results = coll1->search("function", {"title", "description"}, "", {}, {}, {0}, 10,
+ 1, FREQUENCY, {true, true},
+ 10, spp::sparse_hash_set<std::string>(),
+ spp::sparse_hash_set<std::string>(), 10, "", 30, 4, "", 40, {}, {}, {}, 0,
+ "<mark>", "</mark>", {1, 1}).get();
+
+ ASSERT_EQ("0", results["hits"][0]["document"]["id"].get<std::string>());
+
+ ASSERT_EQ(2, results["hits"][0]["highlights"].size());
+
+ ASSERT_EQ("<mark>Functions</mark> and Equations",
+ results["hits"][0]["highlights"][0]["snippet"].get<std::string>());
+
+ ASSERT_EQ("Use a <mark>function</mark> to solve an equation.",
+ results["hits"][0]["highlights"][1]["snippet"].get<std::string>());
+
+ collectionManager.drop_collection("coll1");
+}
+
+TEST_F(CollectionSpecificTest, HighlightWithDropTokens) {
+ std::vector<field> fields = {field("description", field_types::STRING, false),
+ field("points", field_types::INT32, false),};
+
+ Collection* coll1 = collectionManager.create_collection("coll1", 1, fields, "points").get();
+
+ nlohmann::json doc1;
+ doc1["id"] = "0";
+ doc1["description"] = "HPE Aruba AP-575 802.11ax Wireless Access Point - TAA Compliant - 2.40 GHz, "
+ "5 GHz - MIMO Technology - 1 x Network (RJ-45) - Gigabit Ethernet - Bluetooth 5";
+ doc1["points"] = 100;
+
+ ASSERT_TRUE(coll1->add(doc1.dump()).ok());
+
+ auto results = coll1->search("HPE Aruba AP-575 Technology Gigabit Bluetooth 5", {"description"}, "", {}, {}, {0}, 10,
+ 1, FREQUENCY, {true},
+ 10, spp::sparse_hash_set<std::string>(),
+ spp::sparse_hash_set<std::string>(), 10, "", 30, 4, "description", 40, {}, {}, {}, 0,
+ "<mark>", "</mark>").get();
+
+ ASSERT_EQ(1, results["hits"][0]["highlights"].size());
+ ASSERT_EQ("0", results["hits"][0]["document"]["id"].get<std::string>());
+
+ ASSERT_EQ("<mark>HPE</mark> <mark>Aruba</mark> <mark>AP-575</mark> 802.11ax Wireless Access Point - "
+ "TAA Compliant - 2.40 GHz, <mark>5</mark> GHz - MIMO <mark>Technology</mark> - 1 x Network (RJ-45) - "
+ "<mark>Gigabit</mark> Ethernet - <mark>Bluetooth</mark> <mark>5</mark>",
+ results["hits"][0]["highlights"][0]["snippet"].get<std::string>());
+
+ collectionManager.drop_collection("coll1");
+}
diff --git a/test/collection_test.cpp b/test/collection_test.cpp
index 0e9221a2..8f573e6e 100644
--- a/test/collection_test.cpp
+++ b/test/collection_test.cpp
@@ -746,23 +746,23 @@ TEST_F(CollectionTest, ArrayStringFieldHighlight) {
ASSERT_STREQ(id.c_str(), result_id.c_str());
}
- ASSERT_EQ(4, results["hits"][0]["highlights"][0].size());
- ASSERT_STREQ(results["hits"][0]["highlights"][0]["field"].get<std::string>().c_str(), "tags");
- ASSERT_EQ(2, results["hits"][0]["highlights"][0]["snippets"].size());
- ASSERT_STREQ("<mark>truth</mark>", results["hits"][0]["highlights"][0]["snippets"][0].get<std::string>().c_str());
- ASSERT_STREQ("plain <mark>truth</mark>", results["hits"][0]["highlights"][0]["snippets"][1].get<std::string>().c_str());
- ASSERT_EQ(2, results["hits"][0]["highlights"][0]["matched_tokens"].size());
- ASSERT_STREQ("truth", results["hits"][0]["highlights"][0]["matched_tokens"][0][0].get<std::string>().c_str());
- ASSERT_STREQ("truth", results["hits"][0]["highlights"][0]["matched_tokens"][1][0].get<std::string>().c_str());
- ASSERT_EQ(2, results["hits"][0]["highlights"][0]["indices"].size());
- ASSERT_EQ(1, results["hits"][0]["highlights"][0]["indices"][0]);
- ASSERT_EQ(2, results["hits"][0]["highlights"][0]["indices"][1]);
+ ASSERT_EQ(3, results["hits"][0]["highlights"][0].size());
+ ASSERT_STREQ("title", results["hits"][0]["highlights"][0]["field"].get<std::string>().c_str());
+ ASSERT_STREQ("Plain <mark>Truth</mark>", results["hits"][0]["highlights"][0]["snippet"].get<std::string>().c_str());
+ ASSERT_EQ(1, results["hits"][0]["highlights"][0]["matched_tokens"].size());
+ ASSERT_STREQ("Truth", results["hits"][0]["highlights"][0]["matched_tokens"][0].get<std::string>().c_str());
- ASSERT_EQ(3, results["hits"][0]["highlights"][1].size());
- ASSERT_STREQ("title", results["hits"][0]["highlights"][1]["field"].get<std::string>().c_str());
- ASSERT_STREQ("Plain <mark>Truth</mark>", results["hits"][0]["highlights"][1]["snippet"].get<std::string>().c_str());
- ASSERT_EQ(1, results["hits"][0]["highlights"][1]["matched_tokens"].size());
- ASSERT_STREQ("Truth", results["hits"][0]["highlights"][1]["matched_tokens"][0].get<std::string>().c_str());
+ ASSERT_EQ(4, results["hits"][0]["highlights"][1].size());
+ ASSERT_STREQ(results["hits"][0]["highlights"][1]["field"].get<std::string>().c_str(), "tags");
+ ASSERT_EQ(2, results["hits"][0]["highlights"][1]["snippets"].size());
+ ASSERT_STREQ("<mark>truth</mark>", results["hits"][0]["highlights"][1]["snippets"][0].get<std::string>().c_str());
+ ASSERT_STREQ("plain <mark>truth</mark>", results["hits"][0]["highlights"][1]["snippets"][1].get<std::string>().c_str());
+ ASSERT_EQ(2, results["hits"][0]["highlights"][1]["matched_tokens"].size());
+ ASSERT_STREQ("truth", results["hits"][0]["highlights"][1]["matched_tokens"][0][0].get<std::string>().c_str());
+ ASSERT_STREQ("truth", results["hits"][0]["highlights"][1]["matched_tokens"][1][0].get<std::string>().c_str());
+ ASSERT_EQ(2, results["hits"][0]["highlights"][1]["indices"].size());
+ ASSERT_EQ(1, results["hits"][0]["highlights"][1]["indices"][0]);
+ ASSERT_EQ(2, results["hits"][0]["highlights"][1]["indices"][1]);
ASSERT_EQ(3, results["hits"][1]["highlights"][0].size());
ASSERT_STREQ("title", results["hits"][1]["highlights"][0]["field"].get<std::string>().c_str());
@@ -2456,15 +2456,15 @@ TEST_F(CollectionTest, SearchHighlightFieldFully) {
spp::sparse_hash_set<std::string>(), 10, "", 5, 5, "title, tags").get();
ASSERT_EQ(2, res["hits"][0]["highlights"].size());
- ASSERT_EQ("<mark>LAZY</mark>", res["hits"][0]["highlights"][0]["values"][0].get<std::string>());
ASSERT_EQ("The quick brown fox jumped over the <mark>lazy</mark> dog and ran straight to the forest to sleep.",
- res["hits"][0]["highlights"][1]["value"].get<std::string>());
+ res["hits"][0]["highlights"][0]["value"].get<std::string>());
+ ASSERT_EQ(1, res["hits"][0]["highlights"][0]["matched_tokens"].size());
+ ASSERT_STREQ("lazy", res["hits"][0]["highlights"][0]["matched_tokens"][0].get<std::string>().c_str());
- ASSERT_EQ(1, res["hits"][0]["highlights"][1]["matched_tokens"].size());
- ASSERT_STREQ("lazy", res["hits"][0]["highlights"][1]["matched_tokens"][0].get<std::string>().c_str());
-
- ASSERT_EQ(1, res["hits"][0]["highlights"][0]["values"][0].size());
- ASSERT_STREQ("<mark>LAZY</mark>", res["hits"][0]["highlights"][0]["values"][0].get<std::string>().c_str());
+ ASSERT_EQ(1, res["hits"][0]["highlights"][1]["values"].size());
+ ASSERT_EQ("<mark>LAZY</mark>", res["hits"][0]["highlights"][1]["values"][0].get<std::string>());
+ ASSERT_EQ(1, res["hits"][0]["highlights"][1]["snippets"].size());
+ ASSERT_EQ("<mark>LAZY</mark>", res["hits"][0]["highlights"][1]["snippets"][0].get<std::string>());
// excluded fields should not be returned in highlights section
spp::sparse_hash_set<std::string> excluded_fields = {"tags"};
@@ -3146,13 +3146,17 @@ TEST_F(CollectionTest, MultiFieldRelevance5) {
ASSERT_STREQ("1", results["hits"][1]["document"]["id"].get<std::string>().c_str());
ASSERT_STREQ("2", results["hits"][2]["document"]["id"].get<std::string>().c_str());
- ASSERT_EQ(1, results["hits"][0]["highlights"].size());
- ASSERT_EQ("country", results["hits"][0]["highlights"][0]["field"].get<std::string>());
- ASSERT_EQ("<mark>Canada</mark>", results["hits"][0]["highlights"][0]["snippet"].get<std::string>());
+ ASSERT_EQ(2, results["hits"][0]["highlights"].size());
+ ASSERT_EQ("field_a", results["hits"][0]["highlights"][0]["field"].get<std::string>());
+ ASSERT_EQ("<mark>Canadia</mark>", results["hits"][0]["highlights"][0]["snippet"].get<std::string>());
+ ASSERT_EQ("country", results["hits"][0]["highlights"][1]["field"].get<std::string>());
+ ASSERT_EQ("<mark>Canada</mark>", results["hits"][0]["highlights"][1]["snippet"].get<std::string>());
- ASSERT_EQ(1, results["hits"][1]["highlights"].size());
- ASSERT_EQ("company_name", results["hits"][1]["highlights"][0]["field"].get<std::string>());
- ASSERT_EQ("<mark>Canaida</mark> Corp", results["hits"][1]["highlights"][0]["snippet"].get<std::string>());
+ ASSERT_EQ(2, results["hits"][1]["highlights"].size());
+ ASSERT_EQ("field_a", results["hits"][1]["highlights"][0]["field"].get<std::string>());
+ ASSERT_EQ("<mark>Canadoo</mark>", results["hits"][1]["highlights"][0]["snippet"].get<std::string>());
+ ASSERT_EQ("company_name", results["hits"][1]["highlights"][1]["field"].get<std::string>());
+ ASSERT_EQ("<mark>Canaida</mark> Corp", results["hits"][1]["highlights"][1]["snippet"].get<std::string>());
ASSERT_EQ(1, results["hits"][2]["highlights"].size());
ASSERT_EQ("field_a", results["hits"][2]["highlights"][0]["field"].get<std::string>());
@@ -3317,7 +3321,7 @@ TEST_F(CollectionTest, MultiFieldHighlighting) {
ASSERT_STREQ("0", results["hits"][0]["document"]["id"].get<std::string>().c_str());
- ASSERT_EQ(2, results["hits"][0]["highlights"].size());
+ ASSERT_EQ(3, results["hits"][0]["highlights"].size());
ASSERT_EQ("name", results["hits"][0]["highlights"][0]["field"].get<std::string>());
ASSERT_EQ("Best Wireless Vehicle <mark>Charger</mark>",
results["hits"][0]["highlights"][0]["snippet"].get<std::string>());
@@ -3326,6 +3330,9 @@ TEST_F(CollectionTest, MultiFieldHighlighting) {
ASSERT_EQ("Easily replenish your cell phone with this wireless <mark>charger.</mark>",
results["hits"][0]["highlights"][1]["snippet"].get<std::string>());
+ ASSERT_EQ("categories", results["hits"][0]["highlights"][2]["field"].get<std::string>());
+ ASSERT_EQ("Car <mark>Chargers</mark>", results["hits"][0]["highlights"][2]["snippets"][0].get<std::string>());
+
results = coll1->search("John With Denver",
{"description"}, "", {}, {}, {0}, 10, 1, FREQUENCY,
{true}, 1, spp::sparse_hash_set<std::string>(),