fix: insertion search e2e and insertion column tests, dont allow non-…

…empty value for insertion search
GenSpectrum · Jul 24, 2023 · 19af04d · 19af04d
1 parent ae900da
commit 19af04d
Show file tree

Hide file tree

Showing 6 changed files with 33 additions and 21 deletions.
diff --git a/...test/queries/insertionContains_empty.json → ...validQueries/insertionContains_empty.json b/...test/queries/insertionContains_empty.json → ...validQueries/insertionContains_empty.json
@@ -11,9 +11,8 @@
       "value": ""
     }
   },
-  "expectedQueryResult": [
-    {
-      "count": 17
-    }
-  ]
+  "expectedError": {
+    "error": "Bad request",
+    "message": "The field 'value' in an InsertionContains expression must not be an empty string"
+  }
 }
diff --git a/endToEndTests/test/invalidQueries/insertionContains_invalidPattern.json b/endToEndTests/test/invalidQueries/insertionContains_invalidPattern.json
@@ -7,6 +7,7 @@
     "filterExpression": {
       "type": "InsertionContains",
       "column": "insertions",
+      "position": 25701,
       "value": "CC+++"
     }
   },

diff --git a/endToEndTests/test/queries/insertionContains_not_exact1.json b/endToEndTests/test/queries/insertionContains_not_exact1.json
@@ -7,7 +7,7 @@
     "filterExpression": {
       "type": "InsertionContains",
       "column": "insertions",
-      "position": 25701,
+      "position": 22339,
       "value": ".*GCT.*GGT.*"
     }
   },

diff --git a/src/silo/query_engine/filter_expressions/insertion_contains.cpp b/src/silo/query_engine/filter_expressions/insertion_contains.cpp
@@ -88,12 +88,16 @@ void from_json(const nlohmann::json& json, std::unique_ptr<InsertionContains>& f
       json.contains("value"), "The field 'value' is required in an InsertionContains expression"
    )
    CHECK_SILO_QUERY(
-      json["value"].is_string() || json["value"].is_null(),
-      "The field 'value' in an InsertionContains expression needs to be a string or null"
+      json["value"].is_string() && !json["value"].is_null(),
+      "The field 'value' in an InsertionContains expression needs to be a string"
    )
    const std::string& column_name = json["column"];
-   uint32_t position = json["position"];
-   const std::string& value = json["value"].is_null() ? "" : json["value"].get<std::string>();
+   const uint32_t position = json["position"];
+   const std::string& value = json["value"].get<std::string>();
+   CHECK_SILO_QUERY(
+      !value.empty(),
+      "The field 'value' in an InsertionContains expression must not be an empty string"
+   )
    CHECK_SILO_QUERY(
       validateRegexPattern(value),
       "The field 'value' in the InsertionContains expression does not contain a valid regex "

diff --git a/src/silo/storage/column/insertion_column.test.cpp b/src/silo/storage/column/insertion_column.test.cpp
@@ -32,26 +32,29 @@ TEST(InsertionColumn, shouldReturnTheCorrectSearchedValues) {
    InsertionColumnPartition under_test(lookup);
 
    under_test.insert("25701:ACCA");
-   under_test.insert("2301:CCG");
-   under_test.insert("2301:CCG");
-   under_test.insert("19832:TTACAT,25701:ACCA");
+   under_test.insert("25701:CCG");
+   under_test.insert("25701:CCG");
+   under_test.insert("25701:TTACAT,25701:ACCA");
    under_test.insert("25701:ACCA");
-   under_test.insert("19832:TTACAT,25701:ACCA,29903:AGCTGTTCAG");
+   under_test.insert("25701:TTACAT,25701:ACCA,25701:AGCTGTTCAG");
 
    under_test.buildInsertionIndex();
 
-   const auto result1 = under_test.search(".*CC.*");
+   const auto result1 = under_test.search(25701, ".*CC.*");
    ASSERT_EQ(*result1, roaring::Roaring({0, 1, 2, 3, 4, 5}));
 
-   const auto result2 = under_test.search(".*TTA.*CAT.*");
+   const auto result2 = under_test.search(25701, ".*TTA.*CAT.*");
    ASSERT_EQ(*result2, roaring::Roaring({3, 5}));
 
-   const auto result3 = under_test.search(".*AGC.*TGT.*TCA.*G.*");
+   const auto result3 = under_test.search(25701, ".*AGC.*TGT.*TCA.*G.*");
    ASSERT_EQ(*result3, roaring::Roaring({5}));
 
-   const auto result4 = under_test.search(".*AGC.*TG.*T.*T.*C.*AG.*");
+   const auto result4 = under_test.search(25701, ".*AGC.*TG.*T.*T.*C.*AG.*");
    ASSERT_EQ(*result4, roaring::Roaring({5}));
 
-   const auto result5 = under_test.search(".*TTT.*AAA.*");
+   const auto result5 = under_test.search(25701, ".*TTT.*AAA.*");
+   ASSERT_EQ(*result5, roaring::Roaring());
+
+   const auto result6 = under_test.search(100, ".*TTT.*AAA.*");
    ASSERT_EQ(*result5, roaring::Roaring());
 }
diff --git a/src/silo/storage/column/insertion_index.cpp b/src/silo/storage/column/insertion_index.cpp
@@ -215,13 +215,18 @@ std::unique_ptr<roaring::Roaring> InsertionIndex::search(
    const auto three_mers = extractThreeMers(search_pattern);
    const std::regex regex_search_pattern(search_pattern);
 
+   const auto insertion_pos_it = insertion_positions.find(position);
+   if (insertion_pos_it == insertion_positions.end()) {
+      return std::make_unique<roaring::Roaring>();
+   }
+
    if (!three_mers.empty()) {
-      return insertion_positions.at(position).searchWithThreeMerIndex(
+      return insertion_pos_it->second.searchWithThreeMerIndex(
          three_mers, regex_search_pattern
       );
    }
    auto result = std::make_unique<roaring::Roaring>();
-   for (const auto& insertion : insertion_positions.at(position).insertions) {
+   for (const auto& insertion : insertion_pos_it->second.insertions) {
       if (std::regex_search(insertion.value, regex_search_pattern)) {
          *result |= insertion.sequence_ids;
       }