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

components/core/src/clp_s/ArchiveReader.hpp

@@ -10,6 +10,7 @@
 #include <utility>
 #include <vector>
 
+#include <nlohmann/json_fwd.hpp>
 #include <ystdlib/error_handling/Result.hpp>
 
 #include <clp_s/ArchiveReaderAdaptor.hpp>
@@ -160,8 +161,8 @@ class ArchiveReader {
     void close();
 
     /**
-     * @return The schema ids in the archive. It also defines the order that tables should be read
-     * in to avoid seeking backwards.
+     * @return The schema ids in the archive. It also defines the order that tables should be
+     * read in to avoid seeking backwards.
      */
     [[nodiscard]] std::vector<int32_t> const& get_schema_ids() const { return m_schema_ids; }
 
@@ -182,6 +183,16 @@ 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:
     /**
      * Reads archive metadata and prepares the archive reader for subsequent archive reads.
@@ -239,13 +250,14 @@ class ArchiveReader {
     );
 
     /**
-     * Reads a table with given ID from the packed stream reader. If read_stream is called multiple
-     * times in a row for the same stream_id a cached buffer is returned. This function allows the
-     * caller to ask for the same buffer to be reused to read multiple different tables: this can
-     * save memory allocations, but can only be used when tables are read one at a time.
+     * Reads a table with given ID from the packed stream reader. If read_stream is called
+     * multiple times in a row for the same stream_id a cached buffer is returned. This function
+     * allows the caller to ask for the same buffer to be reused to read multiple different
+     * tables: this can save memory allocations, but can only be used when tables are read one
+     * at a time.
      * @param stream_id
-     * @param reuse_buffer when true the same buffer is reused across invocations, overwriting data
-     * returned previous calls to read_stream
+     * @param reuse_buffer when true the same buffer is reused across invocations, overwriting
+     * data returned previous calls to read_stream
      * @return a buffer containing the decompressed stream identified by stream_id
      */
     std::shared_ptr<char[]> read_stream(size_t stream_id, bool reuse_buffer);

components/core/src/clp_s/ArchiveReaderAdaptor.cpp

@@ -168,6 +168,9 @@ 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;
 }
@@ -362,4 +365,13 @@ 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

@@ -101,6 +101,13 @@ 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.
@@ -188,6 +195,7 @@ 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());
     }
 }