Add spectrum scan time and scans.scan_start_time

include/mzpeak/metadata/scans.h

@@ -0,0 +1,45 @@
+/*
+
+This file is part of the mzpeak project.  It is subject to the license
+specified in the LICENSE file which can be found in the top-level
+directory of this repository.
+
+*/
+
+#pragma once
+
+#include <memory>
+#include <vector>
+
+namespace MzPeak::Util {
+class Parquet;
+}
+
+namespace MzPeak::Metadata {
+
+/**
+ * The method of precursor-ion selection and activation.
+ */
+class Scans final {
+public:
+  /// Data for all scans.
+  struct Raw {
+    std::vector<float> scan_start_time;
+  };
+
+  /// Default constructor;
+  Scans();
+
+  /// Constructor.
+  Scans(std::unique_ptr<Util::Parquet>, uint64_t);
+
+  /**
+   * Return the raw table data for all scans.
+   */
+  const Raw& raw() const { return raw_; }
+
+private:
+  Raw raw_;
+};
+
+} // namespace MzPeak::Metadata

include/mzpeak/metadata/spectrum.h

@@ -40,8 +40,14 @@ class Spectrum final {
    */
   const std::vector<double>& delta_model() const { return delta_model_; }
 
+  /**
+   * The scan time in minutes.
+   */
+  std::optional<double> scan_time() const { return scan_time_; };
+
 private:
   std::optional<uint8_t> ms_level_;
+  std::optional<double> scan_time_;
   std::vector<double> delta_model_;
 };
 

include/mzpeak/schema/data_kind.h

@@ -8,44 +8,79 @@ directory of this repository.
 
 #pragma once
 
+#include <optional>
 #include <string>
+#include <variant>
 
 namespace MzPeak::Schema {
 
 /**
  * Indicates how data is encoded in a parquet file.
  */
-enum class DataKind {
-  /// Files that contain signal data.
-  DataArray,
-
-  /// Processed data (e.g. centroided).  It is expected that the
-  /// unprocessed version of the data is present in the same MzPeak
-  /// file as a DataArray.
-  Peaks,
-
-  /// Metadata relating to one of the other files.
-  Metadata,
-
-  /// Non-standard file that can't be decoded by this library.
-  /// However, users of this library can access the raw bytes of
-  /// this file.
-  Proprietary,
-
-  /// Non-standard file that can't be decoded by this library.
-  /// However, users of this library can access the raw bytes of
-  /// this file.
-  Other
-};
+class DataKind final {
+public:
+  enum Type {
+    /// Files that contain signal data.
+    DataArray,
 
-/**
- * Convert a DataKind to a string.
- */
-std::string data_kind_to_string(DataKind);
+    /// Processed data (e.g. centroided).  It is expected that the
+    /// unprocessed version of the data is present in the same MzPeak
+    /// file as a DataArray.
+    Peaks,
 
-/**
- * Parse a DataKind from a string view.
- */
-DataKind data_kind_from_string(std::string_view);
+    /// Metadata relating to one of the other files.
+    Metadata,
+
+    /// A scan or acquisition from the original raw file used to
+    /// create a spectrum.
+    Scans,
+
+    /// The method of precursor-ion selection and activation.
+    Precursors,
+
+    /// An ion isolated for dissociation.
+    SelectedIons,
+
+    /// When describing single reaction monitoring (SRM) or multiple
+    /// reaction monitoring (MRM) experiments, each product ion is
+    /// isolated separately with a different isolation window. This
+    /// table is usually empty or absent
+    Products,
+
+    /// Non-standard file that can't be decoded by this library.
+    /// However, users of this library can access the raw bytes of
+    /// this file.
+    Proprietary,
+  };
+
+  // Internal storage type.
+  using value_type = std::variant<Type, std::string>;
+
+  /// Constructor.
+  DataKind(std::string_view);
+
+  /// Constructor.
+  DataKind(Type);
+
+  /**
+   * Return the string representation of the data kind.
+   */
+  std::string to_string() const;
+
+  /**
+   * Return the enumerated type of the data kind if it is known.
+   */
+  std::optional<Type> type() const;
+
+  /**
+   * Return `true` if this data kind is a known holder of metadata.
+   * For example, the `Scans` and `Precursors` type are both
+   * considered to be metadata.
+   */
+  bool is_metadata() const;
+
+private:
+  value_type val_;
+};
 
 } // namespace MzPeak::Schema

include/mzpeak/schema/file.h

@@ -35,11 +35,6 @@ class File final {
   /// Conversion from JSON.
   explicit File(const json::object&);
 
-  /// Return `true` if this file is associated with the given file.
-  /// For example, if this file is a DataArray and the other file is a
-  /// Metadata file with a similar name.
-  bool is_associated_with(const File&) const;
-
   /// Equality operator.
   bool operator==(const File&) const;
 
@@ -57,7 +52,7 @@ class File final {
 
 private:
   std::string file_name_;
-  DataKind data_kind_ = DataKind::Other;
+  DataKind data_kind_ = DataKind("other");
   EntityType entity_type_ = EntityType::Other;
   std::vector<Column> columns_;
 };

include/mzpeak/spectrum.h

@@ -12,6 +12,7 @@ top-level directory of this repository.
 #include <vector>
 
 #include "mzpeak/data/array_index.h"
+#include "mzpeak/metadata/scans.h"
 
 namespace MzPeak {
 
@@ -50,6 +51,22 @@ class Spectrum final {
    */
   uint8_t ms_level() const;
 
+  /**
+   * Scan time in minutes.
+   *
+   * If the scan time was not stored in the spectra metadata then it
+   * will be fetched from the scans array.
+   *
+   * If the scan metadata is also empty then this method will return
+   * 0.0.
+   */
+  double scan_time();
+
+  /**
+   * Return scan details.
+   */
+  const Metadata::Scans& scans();
+
 protected:
   friend class Spectra;
 
@@ -66,6 +83,8 @@ class Spectrum final {
   std::vector<double> mz_;
   std::vector<float> intensity_;
   uint8_t ms_level_;
+  std::optional<double> scan_time_;
+  std::optional<Metadata::Scans> scans_;
 };
 
 } // namespace MzPeak

include/mzpeak/util/manager.h

@@ -34,7 +34,13 @@ class Manager final {
   /**
    * Find a file given its name.
    */
-  std::vector<Schema::File>::const_iterator find_file(std::string_view name) const;
+  std::vector<Schema::File>::const_iterator find_file(std::string_view) const;
+
+  /**
+   * Find a file given its `EntityType` and `DataKind`.
+   */
+  std::vector<Schema::File>::const_iterator find_file(Schema::EntityType,
+                                                      Schema::DataKind::Type) const;
 
   /**
    * Open a Parquet file from the mzPeak archive.