refactor: better logging and checks for initial ndjson file checks in…

… preprocessing (#490)
GenSpectrum · Jun 18, 2024 · d10a2f7 · d10a2f7
1 parent ba97f0d
commit d10a2f7
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 2 deletions.
diff --git a/src/silo/preprocessing/metadata_info.cpp b/src/silo/preprocessing/metadata_info.cpp
@@ -89,11 +89,13 @@ bool MetadataInfo::isNdjsonFileEmpty(const std::filesystem::path& ndjson_file) {
    duckdb::DuckDB duck_db(nullptr);
    duckdb::Connection connection(duck_db);
 
-   auto result = connection.Query(fmt::format(
+   const std::string query = fmt::format(
       "SELECT COUNT(*) "
       "FROM (SELECT * FROM read_json_auto(\"{}\") LIMIT 1);",
       ndjson_file.string()
-   ));
+   );
+   SPDLOG_DEBUG("ndjson emptiness-check: {}", query);
+   auto result = connection.Query(query);
 
    auto row_count_value = result->GetValue<int64_t>(0, 0);
    const int64_t row_count = duckdb::BigIntValue::Get(row_count_value);

diff --git a/src/silo/preprocessing/preprocessor.cpp b/src/silo/preprocessing/preprocessor.cpp
@@ -128,12 +128,21 @@ void Preprocessor::buildTablesFromNdjsonInput(const std::filesystem::path& file_
       boost::join(MetadataInfo::getMetadataSQLTypes(database_config), ",")
    ));
 
+   SPDLOG_DEBUG("build - checking whether the file '{}' exists: ", file_name.string());
    if (!std::filesystem::exists(file_name)) {
       throw silo::preprocessing::PreprocessingException(
          fmt::format("The specified input file {} does not exist.", file_name.string())
       );
    }
 
+   SPDLOG_DEBUG("build - checking whether the file '{}' is not a directory: ", file_name.string());
+   if (std::filesystem::is_directory(file_name)) {
+      throw silo::preprocessing::PreprocessingException(
+         fmt::format("The specified input file {} is a directory.", file_name.string())
+      );
+   }
+
+   SPDLOG_DEBUG("build - checking whether the file '{}' is empty: ", file_name.string());
    if (MetadataInfo::isNdjsonFileEmpty(file_name)) {
       SPDLOG_WARN(
          "The specified input file {} is empty. Ignoring its content.", file_name.string()