refactor: some more logging during preprocessing

include/silo/preprocessing/preprocessing_config.h

@@ -4,6 +4,8 @@
 #include <filesystem>
 #include <string>
 
+#include <fmt/core.h>
+
 namespace silo::preprocessing {
 
 struct InputDirectory {
@@ -68,4 +70,13 @@ std::filesystem::path createPath(
 
 }  // namespace silo::preprocessing
 
+template <>
+struct [[maybe_unused]] fmt::formatter<silo::preprocessing::PreprocessingConfig>
+    : fmt::formatter<std::string> {
+   [[maybe_unused]] static auto format(
+      const silo::preprocessing::PreprocessingConfig& preprocessing_config,
+      format_context& ctx
+   ) -> decltype(ctx.out());
+};
+
 #endif  // SILO_PREPROCESSING_CONFIG_H

src/silo/prepare_dataset.cpp

@@ -168,11 +168,8 @@ void partitionSequenceFile(
    std::unordered_map<std::string, std::string>& key_to_chunk,
    std::string_view reference_sequence
 ) {
-   SPDLOG_INFO("partitioning sequences file to {}", output_folder.string());
-
    auto chunk_to_seq_writer =
       getSequenceWritersForChunks(output_folder, chunk_names, reference_sequence);
-   SPDLOG_DEBUG("Created file streams in folder {}", output_folder.string());
 
    writeSequenceChunks(sequence_in, key_to_chunk, chunk_to_seq_writer);
 }
@@ -214,6 +211,12 @@ void silo::partitionData(
 
       create_directory(nuc_folder);
 
+      SPDLOG_INFO(
+         "partitioning nucleotide sequences from {} to {}",
+         sequence_filename.string(),
+         nuc_folder.string()
+      );
+
       partitionSequenceFile(
          sequence_input, nuc_folder, chunk_names, key_to_chunk, reference_genome
       );
@@ -229,6 +232,12 @@ void silo::partitionData(
 
       create_directory(aa_folder);
 
+      SPDLOG_INFO(
+         "partitioning amino acid sequences from {} to {}",
+         sequence_filename.string(),
+         aa_folder.string()
+      );
+
       partitionSequenceFile(sequence_input, aa_folder, chunk_names, key_to_chunk, reference_genome);
    }
 

src/silo/preprocessing/preprocessing_config.cpp

@@ -64,4 +64,23 @@ PreprocessingConfig::PreprocessingConfig(
    sorted_partition_folder = createOutputPath(output_directory, sorted_partition_folder_.folder);
    serialization_folder = createOutputPath(output_directory, serialization_folder_.folder);
 }
-}  // namespace silo::preprocessing
+}  // namespace silo::preprocessing
+
+[[maybe_unused]] auto fmt::formatter<silo::preprocessing::PreprocessingConfig>::format(
+   const silo::preprocessing::PreprocessingConfig& preprocessing_config,
+   fmt::format_context& ctx
+) -> decltype(ctx.out()) {
+   return format_to(
+      ctx.out(),
+      "PreprocessingConfig[input directory: {}, pango_lineage_definition_file: {}, "
+      "metadata_file: {}, partition_folder: {}, sorted_partition_folder: {}, "
+      "serialization_folder: {}, reference_genome_file: {}]",
+      preprocessing_config.input_directory.string(),
+      preprocessing_config.pango_lineage_definition_file.string(),
+      preprocessing_config.metadata_file.string(),
+      preprocessing_config.partition_folder.string(),
+      preprocessing_config.sorted_partition_folder.string(),
+      preprocessing_config.serialization_folder.string(),
+      preprocessing_config.reference_genome_file.string()
+   );
+}

src/silo/preprocessing/preprocessing_config_reader.cpp

@@ -73,6 +73,8 @@ struct convert<PreprocessingConfig> {
          reference_genome_filename
       );
 
+      SPDLOG_TRACE("Resulting preprocessing config: {}", config);
+
       return true;
    }
 };