Method BuildIndex

include/storage/tar.hpp:200–269

  private:
    void BuildIndex()
    {
        struct archive *a = archive_read_new();
        archive_read_support_format_tar(a);
        archive_read_support_format_gnutar(a);

        int ret = archive_read_open_filename(a, path.string().c_str(), 10240);
        if (ret != ARCHIVE_OK)
        {
            const char *err = archive_error_string(a);
            std::string errmsg = err ? err : "unknown error";
            archive_read_free(a);
            throw util::RuntimeError(
                path.string(), ErrorCode::FileOpenError, SOURCE_REF, errmsg.c_str());
        }

        struct archive_entry *ae;

        // Collect header positions for all entries in order.
        // archive_read_header_position returns the position of the first header
        // block for an entry. For pax entries with extended attributes (e.g. entries
        // > 8GB), this points to the pax extension header, not the actual file header,
        // so the naive "header_position + 512" calculation is wrong for such entries.
        //
        // Instead, we compute data offsets from consecutive header positions:
        // each entry's padded data ends exactly where the next entry's header begins.
        //   data_offset[i] = header_position[i+1] - ceil_to_512(size[i])
        struct RawEntry
        {
            std::string name;
            std::size_t size;
            std::int64_t header_pos;
            bool is_regular;
        };
        std::vector<RawEntry> all_entries;

        while (archive_read_next_header(a, &ae) == ARCHIVE_OK)
        {
            std::int64_t pos = archive_read_header_position(a);
            bool is_reg = (archive_entry_filetype(ae) == AE_IFREG);
            std::string name;
            std::size_t size = 0;
            if (is_reg)
            {
                name = archive_entry_pathname(ae);
                size = static_cast<std::size_t>(archive_entry_size(ae));
            }
            all_entries.push_back({std::move(name), size, pos, is_reg});
            archive_read_data_skip(a);
        }

        // After ARCHIVE_EOF, header_position points to the end-of-archive marker
        auto end_of_archive_pos = archive_read_header_position(a);

        archive_read_free(a);

        for (std::size_t i = 0; i < all_entries.size(); ++i)
        {