change to use enum for file-type and add file_type to JsonParserOption

Author: AVMatthews

components/core/src/clp_s/CommandLineArguments.cpp

@@ -148,6 +148,7 @@ CommandLineArguments::parse_arguments(int argc, char const** argv) {
             po::options_description compression_options("Compression options");
             std::string metadata_db_config_file_path;
             std::string input_path_list_file_path;
+            std::string file_type;
             // clang-format off
             compression_options.add_options()(
                     "compression-level",
@@ -204,9 +205,8 @@ CommandLineArguments::parse_arguments(int argc, char const** argv) {
                     "Do not record log order at ingestion time."
             )(
                     "file-type",
-                    po::value<std::string>(&m_file_type)->value_name("FILE_TYPE")->
-                        default_value(m_file_type),
-                    "The type of file that is to be compressed to archive (e.g Json or IR)"
+                    po::value<std::string>(&file_type)->value_name("FILE_TYPE"),
+                    "The type of file that is to be compressed to archive (e.g json or kv-ir)"
             );
             // clang-format on
 
@@ -260,6 +260,27 @@ CommandLineArguments::parse_arguments(int argc, char const** argv) {
                 throw std::invalid_argument("No input paths specified.");
             }
 
+            constexpr std::string_view cJsonFileType{"json"};
+            constexpr std::string_view cKeyValueIrFileType{"kv-ir"};
+
+            if (parsed_command_line_options.count("file-type") > 0) {
+                if (cJsonFileType == file_type) {
+                    m_file_type = FileType::Json;
+                } else if (cKeyValueIrFileType == file_type) {
+                    m_file_type = FileType::KeyValueIr;
+                    if (m_structurize_arrays) {
+                        SPDLOG_ERROR(
+                                "Invalid combination of arguments; --file-type {} and "
+                                "--structurize-arrays can't be used together",
+                                cKeyValueIrFileType
+                        );
+                        return ParsingResult::Failure;
+                    }
+                } else {
+                    throw std::invalid_argument("Unknown FILE_TYPE: " + file_type);
+                }
+            }
+
             // Parse and validate global metadata DB config
             if (false == metadata_db_config_file_path.empty()) {
                 clp::GlobalMetadataDBConfig metadata_db_config;
@@ -814,5 +835,4 @@ void CommandLineArguments::print_search_usage() const {
                  " [OUTPUT_HANDLER [OUTPUT_HANDLER_OPTIONS]]"
               << std::endl;
 }
-
 }  // namespace clp_s

components/core/src/clp_s/CommandLineArguments.hpp

@@ -36,6 +36,11 @@ class CommandLineArguments {
         Stdout,
     };
 
+    enum class FileType : uint8_t {
+        Json = 0,
+        KeyValueIr
+    };
+
     // Constructors
     explicit CommandLineArguments(std::string const& program_name) : m_program_name(program_name) {}
 
@@ -116,7 +121,7 @@ class CommandLineArguments {
 
     bool get_record_log_order() const { return false == m_disable_log_order; }
 
-    [[nodiscard]] auto get_file_type() const -> std::string { return m_file_type; }
+    [[nodiscard]] auto get_file_type() const -> FileType { return m_file_type; }
 
 private:
     // Methods
@@ -186,7 +191,7 @@ class CommandLineArguments {
     size_t m_target_ordered_chunk_size{};
     size_t m_minimum_table_size{1ULL * 1024 * 1024};  // 1 MB
     bool m_disable_log_order{false};
-    std::string m_file_type{"Json"};
+    FileType m_file_type{FileType::Json};
 
     // Metadata db variables
     std::optional<clp::GlobalMetadataDBConfig> m_metadata_db_config;

components/core/src/clp_s/JsonParser.hpp

@@ -12,6 +12,7 @@
 
 #include "../clp/GlobalMySQLMetadataDB.hpp"
 #include "ArchiveWriter.hpp"
+#include "CommandLineArguments.hpp"
 #include "DictionaryWriter.hpp"
 #include "FileReader.hpp"
 #include "FileWriter.hpp"
@@ -29,6 +30,7 @@ using namespace simdjson;
 namespace clp_s {
 struct JsonParserOption {
     std::vector<std::string> file_paths;
+    CommandLineArguments::FileType input_file_type{CommandLineArguments::FileType::Json};
     std::string timestamp_key;
     std::string archives_dir;
     size_t target_encoded_size{};

components/core/src/clp_s/clp-s.cpp

@@ -72,16 +72,6 @@ bool search_archive(
 );
 
 bool compress(CommandLineArguments const& command_line_arguments) {
-    auto file_type = command_line_arguments.get_file_type();
-    if ("IR" != file_type && "Json" != file_type) {
-        SPDLOG_ERROR("File Type specified is Invalid");
-        return false;
-    }
-    if ("IR" == file_type && command_line_arguments.get_structurize_arrays()) {
-        SPDLOG_ERROR("ERROR: structurized arrays are not supported for IR files");
-        return false;
-    }
-
     auto archives_dir = std::filesystem::path(command_line_arguments.get_archives_dir());
 
     // Create output directory in case it doesn't exist
@@ -98,6 +88,7 @@ bool compress(CommandLineArguments const& command_line_arguments) {
 
     clp_s::JsonParserOption option{};
     option.file_paths = command_line_arguments.get_file_paths();
+    option.input_file_type = command_line_arguments.get_file_type();
     option.archives_dir = archives_dir.string();
     option.target_encoded_size = command_line_arguments.get_target_encoded_size();
     option.max_document_size = command_line_arguments.get_max_document_size();
@@ -123,7 +114,7 @@ bool compress(CommandLineArguments const& command_line_arguments) {
     }
 
     clp_s::JsonParser parser(option);
-    if ("IR" == file_type) {
+    if (CommandLineArguments::FileType::KeyValueIr == option.input_file_type) {
         // Functionality Coming in later PR
         //  -->Call new parsing function in Json Parser to parse IRv2 to archive
         //  -->Check for error from parsing function