feat(clp-s): Add get_metadata_for_log_event function to clp_s::ArchiveReader (resolves #2012).

components/core/src/clp_s/ArchiveReader.hpp

@@ -7,6 +7,8 @@
 #include <string_view>
 #include <utility>
 
+#include <nlohmann/json_fwd.hpp>
+
 #include "ArchiveReaderAdaptor.hpp"
 #include "DictionaryReader.hpp"
 #include "InputConfig.hpp"
@@ -159,6 +161,16 @@ class ArchiveReader {
         return get_header().has_deprecated_timestamp_format();
     }
 
+    /**
+     * @param log_event_idx
+     * @return The file-level metadata associated with the record at `log_event_idx`.
+     * @throws ArchiveReaderAdaptor::OperationFailed when `log_event_idx` cannot be mapped to any
+     * metadata.
+     */
+    [[nodiscard]] auto get_metadata_for_log_event(int64_t log_event_idx) -> nlohmann::json const& {
+        return m_archive_reader_adaptor->get_metadata_for_log_event(log_event_idx);
+    }
+
 private:
     /**
      * Initializes a schema reader passed by reference to become a reader for a given schema.

components/core/src/clp_s/ArchiveReaderAdaptor.cpp

@@ -150,6 +150,9 @@ auto ArchiveReaderAdaptor::try_read_range_index(ZstdDecompressor& decompressor,
                 end_index,
                 std::move(range_index_entry.at(RangeIndexWriter::cMetadataFieldsName))
         );
+        if (start_index != end_index) {
+            m_non_empty_range_metadata_map.emplace(end_index, m_range_index.back().fields);
+        }
     }
     return ErrorCodeSuccess;
 }
@@ -332,4 +335,13 @@ void ArchiveReaderAdaptor::checkin_reader_for_section(std::string_view section)
 
     m_current_reader_holder.reset();
 }
+
+auto ArchiveReaderAdaptor::get_metadata_for_log_event(int64_t log_event_idx)
+        -> nlohmann::json const& {
+    auto const it{m_non_empty_range_metadata_map.upper_bound(log_event_idx)};
+    if (m_non_empty_range_metadata_map.end() == it || log_event_idx < 0) {
+        throw OperationFailed(ErrorCodeBadParam, __FILENAME__, __LINE__);
+    }
+    return it->second;
+}
 }  // namespace clp_s

components/core/src/clp_s/ArchiveReaderAdaptor.hpp

@@ -2,6 +2,7 @@
 #define CLP_S_ARCHIVEREADERADAPTOR_HPP
 
 #include <cstddef>
+#include <functional>
 #include <memory>
 #include <optional>
 #include <string>
@@ -89,6 +90,13 @@ class ArchiveReaderAdaptor {
 
     std::vector<RangeIndexEntry> const& get_range_index() const { return m_range_index; }
 
+    /**
+     * @param log_event_idx
+     * @return The file-level metadata associated with the record at `log_event_idx`.
+     * @throws OperationFailed when `log_event_idx` cannot be mapped to any metadata.
+     */
+    [[nodiscard]] auto get_metadata_for_log_event(int64_t log_event_idx) -> nlohmann::json const&;
+
 private:
     /**
      * Tries to read an ArchiveFileInfo packet from the archive metadata.
@@ -176,6 +184,7 @@ class ArchiveReaderAdaptor {
     std::shared_ptr<TimestampDictionaryReader> m_timestamp_dictionary;
     std::shared_ptr<clp::ReaderInterface> m_reader;
     std::vector<RangeIndexEntry> m_range_index;
+    std::map<int64_t, nlohmann::json> m_non_empty_range_metadata_map;
 };
 }  // namespace clp_s
 #endif  // CLP_S_ARCHIVEREADERADAPTOR_HPP

components/core/tests/test-clp_s-range_index.cpp

@@ -20,6 +20,7 @@
 #include "../src/clp/type_utils.hpp"
 #include "../src/clp_s/archive_constants.hpp"
 #include "../src/clp_s/ArchiveReader.hpp"
+#include "../src/clp_s/ArchiveReaderAdaptor.hpp"
 #include "../src/clp_s/ArchiveWriter.hpp"
 #include "../src/clp_s/InputConfig.hpp"
 #include "../src/clp_s/RangeIndexWriter.hpp"
@@ -33,6 +34,7 @@ constexpr std::string_view cTestRangeIndexInputFile{"test_no_floats_sorted.jsonl
 constexpr std::string_view cTestRangeIndexIRInputFile{"test_no_floats_sorted.ir"};
 constexpr std::string_view cTestRangeIndexIRMetadataKey{"test_key"};
 constexpr std::string_view cTestRangeIndexIRMetadataValue{"test_value"};
+constexpr size_t cNumRecordsInInputFile{4};
 
 namespace {
 auto get_test_input_path_relative_to_tests_dir() -> std::filesystem::path;
@@ -138,6 +140,17 @@ void read_and_check_archive_metadata(bool from_ir) {
         REQUIRE_NOTHROW(archive_reader.open(archive_path, clp_s::NetworkAuthOption{}));
         auto const& range_index = archive_reader.get_range_index();
         check_archive_range_index(range_index, from_ir);
+        for (size_t i{}; i < cNumRecordsInInputFile; ++i) {
+            REQUIRE(archive_reader.get_metadata_for_log_event(i) == range_index.at(0).fields);
+        }
+        REQUIRE_THROWS_AS(
+                archive_reader.get_metadata_for_log_event(-1),
+                clp_s::ArchiveReaderAdaptor::OperationFailed
+        );
+        REQUIRE_THROWS_AS(
+                archive_reader.get_metadata_for_log_event(cNumRecordsInInputFile),
+                clp_s::ArchiveReaderAdaptor::OperationFailed
+        );
         REQUIRE_NOTHROW(archive_reader.close());
     }
 }