Method write_segment

nodedb-columnar/src/writer/segment_writer.rs:65–161

Encode a drained memtable into a segment byte buffer. `schema` is the column schema, `columns` are the drained column data, `row_count` is the total number of rows. When `kek` is `Some`, the assembled plaintext segment is wrapped in an AES-256-GCM encrypted `SEGC` envelope before being returned. When `None`, the raw `NDBS` segment bytes are returned.

(
        &self,
        schema: &ColumnarSchema,
        columns: &[ColumnData],
        row_count: usize,
        kek: Option<&nodedb_wal::crypto::WalEncryptionKey>,
    )

    /// AES-256-GCM encrypted `SEGC` envelope before being returned. When
    /// `None`, the raw `NDBS` segment bytes are returned.
    pub fn write_segment(
        &self,
        schema: &ColumnarSchema,
        columns: &[ColumnData],
        row_count: usize,
        kek: Option<&nodedb_wal::crypto::WalEncryptionKey>,
    ) -> Result<Vec<u8>, ColumnarError> {
        if row_count == 0 {
            return Err(ColumnarError::EmptyMemtable);
        }
        if columns.len() != schema.columns.len() {
            return Err(ColumnarError::SchemaMismatch {
                expected: schema.columns.len(),
                got: columns.len(),
            });
        }

        let mut buf = Vec::new();

        // 1. Write header.
        buf.extend_from_slice(&SegmentHeader::current().to_bytes());

        // 2. Encode each column's blocks.
        let _metas_guard = self
            .governor
            .as_ref()
            .map(|g| {
                g.reserve(
                    EngineId::Columnar,
                    columns.len() * std::mem::size_of::<ColumnMeta>(),
                )
            })
            .transpose()?;
        let mut column_metas = Vec::with_capacity(columns.len());

        for (i, (col_def, col_data)) in schema.columns.iter().zip(columns.iter()).enumerate() {
            let col_start = buf.len() as u64;

            // Select codec for this column type.
            let codec = select_codec_for_profile(&col_def.column_type, self.profile_tag);

            // Encode blocks.
            let block_stats = encode_column_blocks(
                &mut buf,
                col_data,
                &col_def.column_type,
                codec,
                row_count,
                self.governor.as_ref(),
            )?;

            let col_end = buf.len() as u64;

            // For DictEncoded columns, the codec stored in meta is DeltaFastLanesLz4 (IDs),
            // and the dictionary strings are stored in the meta for reader reconstruction.
            let (effective_codec, dictionary) = match col_data {
                ColumnData::DictEncoded { dictionary, .. } => (
                    ResolvedColumnCodec::DeltaFastLanesLz4,