From 855a4928e27a02cb0dd9f8ce2e0f6b9b8e5319ed Mon Sep 17 00:00:00 2001
From: "Jonas Kellerer (TNG)" <jonas.kellerer@tngtech.com>
Date: Tue, 5 Mar 2024 13:00:40 +0100
Subject: [PATCH] fix: nucleotide symbol equals with dot From
 https://github.com/GenSpectrum/cov-spectrum-website/issues/938

---
 src/silo/common/nucleotide_symbols.cpp        |  1 +
 src/silo/common/nucleotide_symbols.test.cpp   |  2 +-
 .../test/amino_acid_symbol_equals.test.cpp    |  2 +-
 .../test/nucleotide_symbol_equals.test.cpp    | 72 +++++++++++++++++++
 4 files changed, 75 insertions(+), 2 deletions(-)
 create mode 100644 src/silo/test/nucleotide_symbol_equals.test.cpp

diff --git a/src/silo/common/nucleotide_symbols.cpp b/src/silo/common/nucleotide_symbols.cpp
index d31a17312..92758889d 100644
--- a/src/silo/common/nucleotide_symbols.cpp
+++ b/src/silo/common/nucleotide_symbols.cpp
@@ -46,6 +46,7 @@ char Nucleotide::symbolToChar(Nucleotide::Symbol symbol) {
 std::optional<Nucleotide::Symbol> Nucleotide::charToSymbol(char character) {
    switch (character) {
       case '.':
+         return std::nullopt;
       case '-':
          return Symbol::GAP;
       case 'A':
diff --git a/src/silo/common/nucleotide_symbols.test.cpp b/src/silo/common/nucleotide_symbols.test.cpp
index bc11e0512..7983d7856 100644
--- a/src/silo/common/nucleotide_symbols.test.cpp
+++ b/src/silo/common/nucleotide_symbols.test.cpp
@@ -7,7 +7,7 @@ TEST(NucleotideSymbol, enumShouldHaveSameLengthAsArrayOfSymbols) {
 }
 
 TEST(NucleotideSymbol, conversionFromCharacter) {
-   EXPECT_EQ(silo::Nucleotide::charToSymbol('.'), silo::Nucleotide::Symbol::GAP);
+   EXPECT_EQ(silo::Nucleotide::charToSymbol('.'), std::nullopt);
    EXPECT_EQ(silo::Nucleotide::charToSymbol('-'), silo::Nucleotide::Symbol::GAP);
    EXPECT_EQ(silo::Nucleotide::charToSymbol('A'), silo::Nucleotide::Symbol::A);
    EXPECT_EQ(silo::Nucleotide::charToSymbol('N'), silo::Nucleotide::Symbol::N);
diff --git a/src/silo/test/amino_acid_symbol_equals.test.cpp b/src/silo/test/amino_acid_symbol_equals.test.cpp
index 8b0040ec4..9c4f9157e 100644
--- a/src/silo/test/amino_acid_symbol_equals.test.cpp
+++ b/src/silo/test/amino_acid_symbol_equals.test.cpp
@@ -23,7 +23,7 @@ const nlohmann::json DATA_WITH_M = createDataWithAminoAcidSequence("M*");
 const nlohmann::json DATA_WITH_B = createDataWithAminoAcidSequence("B*");
 
 const auto DATABASE_CONFIG =
-   DatabaseConfig{"segmenet1", {"dummy name", {{"primaryKey", ValueType::STRING}}, "primaryKey"}};
+   DatabaseConfig{"segment1", {"dummy name", {{"primaryKey", ValueType::STRING}}, "primaryKey"}};
 
 const auto REFERENCE_GENOMES = ReferenceGenomes{
    {{"segment1", "A"}},
diff --git a/src/silo/test/nucleotide_symbol_equals.test.cpp b/src/silo/test/nucleotide_symbol_equals.test.cpp
new file mode 100644
index 000000000..204b87517
--- /dev/null
+++ b/src/silo/test/nucleotide_symbol_equals.test.cpp
@@ -0,0 +1,72 @@
+#include <nlohmann/json.hpp>
+
+#include "silo/test/query_fixture.test.h"
+
+using silo::ReferenceGenomes;
+using silo::config::DatabaseConfig;
+using silo::config::ValueType;
+using silo::test::QueryTestData;
+using silo::test::QueryTestScenario;
+
+#include <boost/uuid/uuid_generators.hpp>
+#include <boost/uuid/uuid_io.hpp>
+
+using boost::uuids::random_generator;
+
+nlohmann::json createDataWithNucleotideSequence(const std::string& nucleotideSequence) {
+   random_generator generator;
+   const auto request_id = generator();
+
+   return {
+      {"metadata", {{"primaryKey", "id_" + to_string(request_id)}}},
+      {"alignedNucleotideSequences", {{"segment1", nucleotideSequence}}},
+      {"unalignedNucleotideSequences", {{"segment1", nullptr}}},
+      {"alignedAminoAcidSequences", {{"gene1", nullptr}}}
+   };
+}
+const nlohmann::json DATA_SAME_AS_REFERENCE = createDataWithNucleotideSequence("ATGCN");
+const nlohmann::json DATA_WITH_ALL_N = createDataWithNucleotideSequence("NNNNN");
+const nlohmann::json DATA_WITH_ALL_MUTATED = createDataWithNucleotideSequence("CATTT");
+
+const auto DATABASE_CONFIG =
+   DatabaseConfig{"segment1", {"dummy name", {{"primaryKey", ValueType::STRING}}, "primaryKey"}};
+
+const auto REFERENCE_GENOMES = ReferenceGenomes{
+   {{"segment1", "ATGCN"}},
+   {{"gene1", "M*"}},
+};
+
+const QueryTestData TEST_DATA{
+   {DATA_SAME_AS_REFERENCE, DATA_SAME_AS_REFERENCE, DATA_WITH_ALL_N, DATA_WITH_ALL_MUTATED},
+   DATABASE_CONFIG,
+   REFERENCE_GENOMES
+};
+
+nlohmann::json createNucleotideSymbolEqualsQuery(const std::string& symbol, int position) {
+   return {
+      {"action", {{"type", "Aggregated"}}},
+      {"filterExpression",
+       {{"type", "NucleotideEquals"},
+        {"position", position},
+        {"symbol", symbol},
+        {"sequenceName", "segment1"}}}
+   };
+}
+
+const QueryTestScenario NUCLEOTIDE_EQUALS_WITH_SYMBOL = {
+   "nucleotideEqualsWithSymbol",
+   createNucleotideSymbolEqualsQuery("C", 1),
+   {{{"count", 1}}}
+};
+
+const QueryTestScenario NUCLEOTIDE_EQUALS_WITH_DOT = {
+   "nucleotideEqualsWithDot",
+   createNucleotideSymbolEqualsQuery(".", 1),
+   {{{"count", 2}}}
+};
+
+QUERY_TEST(
+   NucleotideSymbolEquals,
+   TEST_DATA,
+   ::testing::Values(NUCLEOTIDE_EQUALS_WITH_SYMBOL, NUCLEOTIDE_EQUALS_WITH_DOT)
+);