Initial implementation of hictk metadata command

docs/generate_cli_reference.sh

@@ -24,6 +24,7 @@ subcommands=(
   fix-mcool
   load
   merge
+  metadata
   rename-chromosomes
   validate
   zoomify

src/hictk/CMakeLists.txt

@@ -21,6 +21,7 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_fix_mcool.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_load.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_merge.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_metadata.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_rename_chromosomes.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_validate.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cli/cli_zoomify.cpp
@@ -33,6 +34,7 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/fix_mcool/fix_mcool.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/load/load.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/merge/merge.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/metadata/metadata.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/rename_chromosomes/rename_chromosomes.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/validate/validate.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/zoomify/zoomify.cpp)

src/hictk/cli/cli.cpp

@@ -42,6 +42,8 @@
       _subcommand = subcommand::load;
     } else if (_cli.get_subcommand("merge")->parsed()) {
       _subcommand = subcommand::merge;
+    } else if (_cli.get_subcommand("metadata")->parsed()) {
+      _subcommand = subcommand::metadata;
     } else if (_cli.get_subcommand("rename-chromosomes")->parsed()) {
       _subcommand = subcommand::rename_chromosomes;
     } else if (_cli.get_subcommand("validate")->parsed()) {
@@ -92,6 +94,8 @@
       return "load";
     case merge:
       return "merge";
+    case metadata:
+      return "metadata";
     case rename_chromosomes:
       return "rename-chromosomes";
     case validate:
@@ -116,6 +120,7 @@
   make_fix_mcool_subcommand();
   make_load_subcommand();
   make_merge_subcommand();
+  make_metadata_subcommand();
   make_rename_chromosomes_subcommand();
   make_validate_subcommand();
   make_zoomify_subcommand();
@@ -141,6 +146,8 @@
     case merge:
       validate_merge_subcommand();
       break;
+    case metadata:
+      break;
     case rename_chromosomes:
       validate_rename_chromosomes_subcommand();
       break;
@@ -174,6 +181,9 @@
     case merge:
       transform_args_merge_subcommand();
       break;
+    case metadata:
+      transform_args_metadata_subcommand();
+      break;
     case rename_chromosomes:
       transform_args_rename_chromosomes_subcommand();
       break;

src/hictk/cli/cli_metadata.cpp

@@ -0,0 +1,66 @@
+// Copyright (C) 2024 Roberto Rossini <roberros@uio.no>
+//
+// SPDX-License-Identifier: MIT
+
+#include <fmt/format.h>
+#include <fmt/ranges.h>
+#include <spdlog/spdlog.h>
+
+#include <CLI/CLI.hpp>
+#include <cassert>
+#include <cstddef>
+#include <cstdint>
+#include <filesystem>
+#include <limits>
+#include <stdexcept>
+#include <string>
+#include <variant>
+#include <vector>
+
+#include "hictk/tools/cli.hpp"
+#include "hictk/tools/config.hpp"
+
+namespace hictk::tools {
+
+void Cli::make_metadata_subcommand() {
+  auto& sc = *_cli.add_subcommand("metadata", "Print file metadata to stdout.")
+                  ->fallthrough()
+                  ->preparse_callback([this]([[maybe_unused]] std::size_t i) {
+                    assert(_config.index() == 0);
+                    _config = MetadataConfig{};
+                  });
+
+  _config = MetadataConfig{};
+  auto& c = std::get<MetadataConfig>(_config);
+
+  // clang-format off
+  sc.add_option(
+      "uri",
+      c.uri,
+      "Path to a .hic or .[ms]cool file (Cooler URI syntax supported).")
+      ->check(IsValidCoolerFile | IsValidMultiresCoolerFile | IsValidSingleCellCoolerFile | IsValidHiCFile)
+      ->required();
+
+  sc.add_option(
+      "-f,--output-format",
+      c.output_format,
+      "Format used to return file metadata.\n"
+      "Should be one of: json, toml, tsv, or yaml.")
+      ->check(CLI::IsMember({"json", "tsv", "toml", "yaml"}))
+      ->capture_default_str();
+  // clang-format on
+
+  _config = std::monostate{};
+}
+
+void Cli::transform_args_metadata_subcommand() {
+  auto& c = std::get<MetadataConfig>(_config);
+
+  c.input_format = infer_input_format(c.uri);
+
+  // in spdlog, high numbers correspond to low log levels
+  assert(c.verbosity > 0 && c.verbosity < 5);
+  c.verbosity = static_cast<std::uint8_t>(spdlog::level::critical) - c.verbosity;
+}
+
+}  // namespace hictk::tools

src/hictk/include/hictk/tools/cli.hpp

@@ -205,6 +205,7 @@ class Cli {
     fix_mcool,
     load,
     merge,
+    metadata,
     rename_chromosomes,
     validate,
     zoomify,
@@ -235,6 +236,7 @@ class Cli {
   void make_fix_mcool_subcommand();
   void make_load_subcommand();
   void make_merge_subcommand();
+  void make_metadata_subcommand();
   void make_rename_chromosomes_subcommand();
   void make_validate_subcommand();
   void make_zoomify_subcommand();
@@ -259,6 +261,7 @@ class Cli {
   void transform_args_fix_mcool_subcommand();
   void transform_args_load_subcommand();
   void transform_args_merge_subcommand();
+  void transform_args_metadata_subcommand();
   void transform_args_rename_chromosomes_subcommand();
   void transform_args_zoomify_subcommand();
   void transform_args();

src/hictk/include/hictk/tools/config.hpp

@@ -182,6 +182,14 @@ struct MergeConfig {
   std::uint8_t verbosity{4};
 };
 
+struct MetadataConfig {
+  std::filesystem::path uri{};
+  std::string input_format{};
+  std::string output_format{"json"};
+
+  std::uint8_t verbosity{2};
+};
+
 struct RenameChromosomesConfig {
   std::string uri{};
   std::filesystem::path path_to_name_mappings{};
@@ -227,6 +235,7 @@ using Config = std::variant<std::monostate,
                             FixMcoolConfig,
                             LoadConfig,
                             MergeConfig,
+                            MetadataConfig,
                             RenameChromosomesConfig,
                             ValidateConfig,
                             ZoomifyConfig>;

src/hictk/include/hictk/tools/tools.hpp

@@ -16,6 +16,7 @@ namespace hictk::tools {
 [[nodiscard]] int fix_mcool_subcmd(const FixMcoolConfig& c);
 [[nodiscard]] int load_subcmd(const LoadConfig& c);
 [[nodiscard]] int merge_subcmd(const MergeConfig& c);
+[[nodiscard]] int metadata_subcmd(const MetadataConfig& c);
 [[nodiscard]] int rename_chromosomes_subcmd(const RenameChromosomesConfig& c);
 [[nodiscard]] int validate_subcmd(const ValidateConfig& c);
 [[nodiscard]] int zoomify_subcmd(const ZoomifyConfig& c);

src/hictk/main.cpp

@@ -112,6 +112,8 @@
         return load_subcmd(std::get<LoadConfig>(config));
       case sc::merge:
         return merge_subcmd(std::get<MergeConfig>(config));
+      case sc::metadata:
+        return metadata_subcmd(std::get<MetadataConfig>(config));
       case sc::rename_chromosomes:
         return rename_chromosomes_subcmd(std::get<RenameChromosomesConfig>(config));
       case sc::validate: