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

Lowercase unicode while tokenization.

Kishore Nallan339300dcd00b
diff --git a/src/tokenizer.cpp b/src/tokenizer.cpp
index 1d868e01..745e47e8 100644
--- a/src/tokenizer.cpp
+++ b/src/tokenizer.cpp
@@ -96,7 +96,7 @@ bool Tokenizer::next(std::string &token, size_t& token_index, size_t& start_inde
LOG(ERROR) << "Unicode error during parsing: " << errcode;
}
} else {
- token = unicode_text.tempSubString(prev_position, length).toUTF8String(word);
+ token = unicode_text.toLower().tempSubString(prev_position, length).toUTF8String(word);
}
if(!token.empty()) {
diff --git a/test/collection_specific_test.cpp b/test/collection_specific_test.cpp
index 09cd6c4d..6ea594c4 100644
--- a/test/collection_specific_test.cpp
+++ b/test/collection_specific_test.cpp
@@ -1541,15 +1541,28 @@ TEST_F(CollectionSpecificTest, UpdateOfTwoDocsWithSameIdWithinSameBatch) {
spp::sparse_hash_set<std::string>(), 10, "", 30, 4, "title", 20, {}, {}, {}, 0,
"<mark>", "</mark>", {}, 1000, true).get();
- LOG(INFO) << results;
+ collectionManager.drop_collection("coll1");
+}
- ASSERT_EQ(1, results["hits"].size());
- ASSERT_EQ("0", results["hits"][0]["document"]["id"].get<std::string>());
- ASSERT_TRUE(results["hits"][0]["document"].contains("last_chance"));
- ASSERT_EQ(false, results["hits"][0]["document"]["last_chance"].get<bool>());
+TEST_F(CollectionSpecificTest, CyrillicText) {
+ // when the first document containing a token already cannot fit compact posting list
- ASSERT_EQ(1, coll1->_get_index()->_get_numerical_index().at("points")->size());
- ASSERT_EQ(1, coll1->_get_index()->_get_numerical_index().at("last_chance")->size());
+ std::vector<field> fields = {field("title", field_types::STRING, false, false, true, "sr"),};
+
+ Collection* coll1 = collectionManager.create_collection("coll1", 1, fields).get();
+
+ nlohmann::json doc;
+ doc["title"] = "Test Тест";
+ ASSERT_TRUE(coll1->add(doc.dump()).ok());
+
+ doc["title"] = "TEST ТЕСТ";
+ ASSERT_TRUE(coll1->add(doc.dump()).ok());
+
+ auto results = coll1->search("тест", {"title"}, "", {}, {}, {0}, 10, 1, FREQUENCY, {false}).get();
+
+ ASSERT_EQ(2, results["hits"].size());
+ ASSERT_EQ("1", results["hits"][0]["document"]["id"].get<std::string>());
+ ASSERT_EQ("0", results["hits"][1]["document"]["id"].get<std::string>());
collectionManager.drop_collection("coll1");
}
@@ -1779,9 +1792,10 @@ TEST_F(CollectionSpecificTest, VerbatimMatchShouldConsiderTokensMatchedAcrossAll
ASSERT_EQ("3", results["hits"][0]["document"]["id"].get<std::string>());
ASSERT_EQ("2", results["hits"][1]["document"]["id"].get<std::string>());
+ ASSERT_EQ(2, results["hits"].size());
+
collectionManager.drop_collection("coll1");
}
-
TEST_F(CollectionSpecificTest, CustomNumTyposConfiguration) {
// dropped tokens on a single field cannot be deemed as verbatim match