Sync metadata with the draft spec

bin/mzp-inspect.cpp

@@ -14,6 +14,7 @@ top-level directory of this repository.
 
 #include "mzpeak.h"
 #include "mzpeak/schema/group.h"
+#include "mzpeak/util/manager.h" // IWYU pragma: keep
 
 /******************************************************************************/
 namespace po = boost::program_options;
@@ -22,14 +23,14 @@ namespace po = boost::program_options;
 std::unique_ptr<MzPeak::Util::Parquet> open_parquet_file(MzPeak::Index& index,
                                                          const std::string& file)
 {
-  auto it = std::ranges::find(index.files(), file, &MzPeak::Schema::File::file_name);
+  auto it = index.find(file);
 
   if (it == index.files().end()) {
     std::println(stderr, "file \"{}\" is not in the mzPeak file index", file);
     return nullptr;
   }
 
-  return index.parquet(*it);
+  return index.manager()->parquet(*it);
 }
 
 /******************************************************************************/
@@ -39,7 +40,7 @@ int print_array_index(MzPeak::Index& index, const std::string& file)
   if (parquet == nullptr) return 1;
 
   auto fmd = parquet->file_metadata();
-  auto et = parquet->index_file().entity_type;
+  auto et = parquet->index_file().entity_type();
   auto key = MzPeak::Schema::entity_type_to_string(et) + "_array_index";
   auto json = parquet->kv_string(fmd, key);
 

include/mzpeak/index.h

@@ -8,14 +8,24 @@ directory of this repository.
 
 #pragma once
 
-#include "mzpeak/io/archive.h"
-#include "mzpeak/schema/file.h"
-#include "mzpeak/spectra.h"
-#include "mzpeak/util/parquet.h"
+#include <memory>
+#include <vector>
 
 namespace MzPeak {
 
-// Internal implementation.
+namespace IO {
+class Archive;
+}
+
+namespace Schema {
+class File;
+}
+
+namespace Util {
+class Manager;
+}
+
+class Spectra;
 
 /**
  * Read-only access to the index inside a MzPeak archive.
@@ -25,27 +35,28 @@ class Index {
   /// Constructor.
   Index(std::unique_ptr<MzPeak::IO::Archive>);
 
-  /// Destructor.
-  ~Index();
-
   /**
    * Return a list of files found in the index.
    */
   const std::vector<Schema::File>& files() const;
 
+  /**
+   * Find a file in the mzPeak archive with the given name.
+   */
+  std::vector<Schema::File>::const_iterator find(const std::string_view&) const;
+
   /**
    * Access the spectra in the file.
    */
   Spectra spectra() const;
 
   /**
-   * Open a Parquet file directly.
+   * Access the low-level MzPeak Manager object.
    */
-  std::unique_ptr<Util::Parquet> parquet(const Schema::File&) const;
+  std::shared_ptr<Util::Manager> manager() const;
 
 protected:
-  struct Impl;
-  std::unique_ptr<Impl> impl_;
+  std::shared_ptr<Util::Manager> manager_;
 };
 
 } // namespace MzPeak

include/mzpeak/metadata/spectrum.h

@@ -13,27 +13,18 @@ top-level directory of this repository.
 #include <vector>
 
 namespace MzPeak::Util {
-class Slice;
-}
-
-namespace MzPeak::Schema {
-class Group;
+class Parquet;
 }
 
 namespace MzPeak::Metadata {
 
-class Table;
-
 /**
  * Metadata from the spectrum table.
  */
 class Spectrum final {
 public:
   /// Constructor.
-  Spectrum(std::shared_ptr<Table>, uint64_t);
-
-  /// Destructor.
-  ~Spectrum() = default;
+  Spectrum(std::unique_ptr<Util::Parquet>, uint64_t);
 
   /**
    * Return the spectrum level (MS:1000511).
@@ -50,10 +41,6 @@ class Spectrum final {
   const std::vector<double>& delta_model() const { return delta_model_; }
 
 private:
-  std::shared_ptr<Table> table_;
-  std::shared_ptr<Schema::Group> group_;
-  uint64_t index_;
-
   std::optional<uint8_t> ms_level_;
   std::vector<double> delta_model_;
 };

include/mzpeak/schema/file.h

@@ -17,13 +17,20 @@ directory of this repository.
 namespace MzPeak::Schema {
 namespace json = boost::json;
 
-struct File {
+/**
+ * A description of a file in the mzPeak archive.
+ */
+class File final {
+public:
+  struct Column {
+    std::string name;
+    std::string path;
+    std::optional<std::string> accession;
+    std::optional<std::string> unit;
+  };
 
   /// Constructor from a file name.
-  explicit File(const std::string& name)
-      : file_name(name)
-  {
-  }
+  explicit File(const std::string& name);
 
   /// Conversion from JSON.
   explicit File(const json::object&);
@@ -34,16 +41,25 @@ struct File {
   bool is_associated_with(const File&) const;
 
   /// Equality operator.
-  bool operator==(const File&) const = default;
+  bool operator==(const File&) const;
 
   /// The name of this file.
-  std::string file_name;
+  const std::string& file_name() const { return file_name_; }
 
   /// This file's data kind.
-  DataKind data_kind = DataKind::Other;
+  DataKind data_kind() const { return data_kind_; }
 
   /// This file's entity type.
-  EntityType entity_type = EntityType::Other;
+  EntityType entity_type() const { return entity_type_; }
+
+  /// Column definitions for this file.
+  const std::vector<Column>& columns() const { return columns_; }
+
+private:
+  std::string file_name_;
+  DataKind data_kind_ = DataKind::Other;
+  EntityType entity_type_ = EntityType::Other;
+  std::vector<Column> columns_;
 };
 
 } // namespace MzPeak::Schema

include/mzpeak/schema/group.h

@@ -14,10 +14,12 @@ top-level directory of this repository.
 #include <string>
 
 #include "mzpeak/schema/cv.h"
+#include "mzpeak/schema/file.h"
 #include "mzpeak/util/types.h"
 
 // Forward declarations.
 namespace parquet::schema {
+class Node;
 class GroupNode;
 } // namespace parquet::schema
 
@@ -87,16 +89,10 @@ class Group final {
      */
     index_type absolute_index() const;
 
-    /**
-     * The name of this field using underscores to replace spaces and
-     * other special characters.
-     */
-    const std::string& name() const;
-
     /**
      * The name of this field as recognized by parquet.
      */
-    const std::string& schema_name() const;
+    const std::string& name() const;
 
     /**
      * The structural type this field represents.
@@ -129,7 +125,6 @@ class Group final {
     index_type rel_index_;
     index_type abs_index_;
     std::string schema_name_;
-    std::string clean_name_;
     std::optional<CVType> cv_type_;
     std::optional<CVUnit> cv_unit_;
     std::optional<Util::Type> type_;
@@ -140,8 +135,13 @@ class Group final {
   /// Fields are stored in a map for quick look-up using their name.
   using field_map_t = std::map<std::string, std::shared_ptr<Field>>;
 
+  /// Constructor for the root group to hold all of the top-level
+  /// columns that are not in a separate struct/group.
+  explicit Group(const parquet::schema::GroupNode&, const Schema::File&);
+
   /// Constructor from a parquet schema descriptor.
   explicit Group(const parquet::schema::GroupNode&,
+                 const Schema::File&,
                  index_type index,
                  index_type offset);
 
@@ -176,8 +176,27 @@ class Group final {
    */
   const field_map_t& fields() const;
 
+  /**
+   * Return true if this is the root group.
+   *
+   * There is only one root group and its fields represent the
+   * top-level columns that are not themselves members of a group or
+   * struct.
+   */
+  bool is_root() const;
+
+  /**
+   * Return a schema path to the given field.  Mostly useful for error
+   * messages.
+   */
+  std::string path(const Field&) const;
+
 private:
+  void
+  make_fields(const parquet::schema::GroupNode&, const Schema::File&, index_type);
+
   std::string name_;
+  bool is_root_;
   index_type index_;
   field_map_t fields_;
 };

include/mzpeak/spectra.h

@@ -16,8 +16,8 @@ namespace MzPeak::Data {
 class Signals;
 } // namespace MzPeak::Data
 
-namespace MzPeak::Metadata {
-class Table;
+namespace MzPeak::Util {
+class Manager;
 }
 
 namespace MzPeak {
@@ -28,12 +28,12 @@ namespace MzPeak {
 class Spectra final : public Util::EnumerableProxy<Spectrum> {
 public:
   /// Low-level constructor from a Parquet file.
-  explicit Spectra(std::unique_ptr<Data::Signals>, std::unique_ptr<Metadata::Table>);
+  explicit Spectra(std::unique_ptr<Data::Signals>, std::shared_ptr<Util::Manager>);
 
 private:
   // Internal data access.
   std::shared_ptr<Data::Signals> data_;
-  std::shared_ptr<Metadata::Table> meta_;
+  std::shared_ptr<Util::Manager> manager_;
 
   // Function to fetch a specific spectrum.
   Spectrum fetch(uint64_t);

include/mzpeak/spectrum.h

@@ -9,27 +9,28 @@ top-level directory of this repository.
 #pragma once
 
 #include <memory>
+#include <vector>
 
 #include "mzpeak/data/array_index.h"
-#include "mzpeak/data/encoding.h"
-#include "mzpeak/data/signals.h"
-#include "mzpeak/metadata/spectrum.h"
-#include "mzpeak/metadata/table.h"
-#include "mzpeak/util/slice.h"
 
 namespace MzPeak {
 
-// Forward declaration.
+namespace Data {
+class Signals;
+} // namespace Data
+
+namespace Util {
+class Slice;
+class Manager;
+} // namespace Util
+
 class Spectra;
 
 /**
  * Access to a single spectrum in an MzPeak file.
  */
 class Spectrum final {
 public:
-  /// The type of decoder used.
-  using decoder_type = Data::Encoding::Decoder<double>;
-
   /// Destructor.
   ~Spectrum() = default;
 
@@ -54,18 +55,17 @@ class Spectrum final {
 
   /// Internal constructor.
   Spectrum(uint64_t index,
+           std::shared_ptr<Util::Manager>,
            std::shared_ptr<Data::Signals>,
            const std::vector<Data::ArrayIndex::Dimension>&,
-           std::unique_ptr<Util::Slice>,
-           std::shared_ptr<Metadata::Table>);
+           std::unique_ptr<Util::Slice>);
 
 private:
   uint64_t index_;
-  std::shared_ptr<Metadata::Table> md_table_;
-  Metadata::Spectrum md_spec_;
-  decoder_type decoder_;
+  std::shared_ptr<Util::Manager> manager_;
   std::vector<double> mz_;
   std::vector<float> intensity_;
+  uint8_t ms_level_;
 };
 
 } // namespace MzPeak