Struct Table 

pub struct Table<P = MemPager>
where
    P: Pager<Blob = EntryHandle> + Send + Sync,
{ /* private fields */ }

Handle for data operations on a table.

Implementations

impl<P> Table<P>

where P: Pager<Blob = EntryHandle> + Send + Sync,

pub fn create_from_columns( display_name: &str, canonical_name: &str, columns: &[PlanColumnSpec], metadata: Arc<MetadataManager<P>>, catalog: Arc<TableCatalog>, store: Arc<ColumnStore<P>>, ) -> LlkvResult<CreateTableResult<P>>

Create a new table from column specifications.

Coordinates metadata persistence, catalog registration, and storage initialization.

pub fn create_from_schema( display_name: &str, canonical_name: &str, schema: &Schema, metadata: Arc<MetadataManager<P>>, catalog: Arc<TableCatalog>, store: Arc<ColumnStore<P>>, ) -> LlkvResult<CreateTableResult<P>>

Create a new table from an Arrow schema (for CREATE TABLE AS SELECT).

pub fn register_sort_index(&self, field_id: FieldId) -> LlkvResult<()>

Register a persisted sort index for the specified user column.

pub fn unregister_sort_index(&self, field_id: FieldId) -> LlkvResult<()>

Remove a persisted sort index for the specified user column if it exists.

pub fn list_registered_indexes( &self, field_id: FieldId, ) -> LlkvResult<Vec<IndexKind>>

List the persisted index kinds registered for the given user column.

pub fn append(&self, batch: &RecordBatch) -> LlkvResult<()>

Append a RecordBatch to the table.

The batch must include:

• A row_id column (type UInt64) with unique row identifiers
• field_id metadata for each user column, mapping to this table’s field IDs

MVCC Columns

If the batch includes created_by or deleted_by columns, they are automatically assigned the correct LogicalFieldId for this table’s MVCC metadata.

Field ID Mapping

Each column’s field_id metadata is converted to a LogicalFieldId by combining it with this table’s ID. This ensures columns from different tables don’t collide in the underlying storage.

Errors

Returns an error if:

• The batch is missing the row_id column
• Any user column is missing field_id metadata
• Field IDs are invalid or malformed
• The underlying storage operation fails

pub fn scan_stream<'a, I, T, F>( &self, projections: I, filter_expr: &Expr<'a, FieldId>, options: ScanStreamOptions<P>, on_batch: F, ) -> LlkvResult<()>

where I: IntoIterator<Item = T>, T: Into<ScanProjection>, F: FnMut(RecordBatch),

Stream one or more projected columns as a sequence of RecordBatches.

• Avoids concat and large materializations.
• Uses the same filter machinery as the old scan to produce row_ids.
• Splits row_ids into fixed-size windows and gathers rows per window to form a small RecordBatch that is sent to on_batch.

pub fn scan_stream_with_exprs<'a, F>( &self, projections: &[ScanProjection], filter_expr: &Expr<'a, FieldId>, options: ScanStreamOptions<P>, on_batch: F, ) -> LlkvResult<()>

where F: FnMut(RecordBatch),

Stream projections using fully resolved expression inputs.

Callers that already parsed expressions into ScanProjection values can use this entry point to skip the iterator conversion performed by Self::scan_stream. The execution semantics and callbacks are identical.

pub fn filter_row_ids<'a>( &self, filter_expr: &Expr<'a, FieldId>, ) -> LlkvResult<Vec<RowId>>

pub fn catalog(&self) -> SysCatalog<'_, P>

pub fn get_table_meta(&self) -> Option<TableMeta>

pub fn get_cols_meta(&self, col_ids: &[FieldId]) -> Vec<Option<ColMeta>>

pub fn schema(&self) -> LlkvResult<Arc<Schema>>

Build and return an Arrow Schema that describes this table.

The returned schema includes the row_id field first, followed by user fields. Each user field has its field_id stored in the field metadata (under the “field_id” key) and the name is taken from the catalog when available or falls back to col_<id>.

pub fn schema_recordbatch(&self) -> LlkvResult<RecordBatch>

Return the table schema formatted as an Arrow RecordBatch suitable for pretty printing. The batch has three columns: name (Utf8), field_id (UInt32) and data_type (Utf8).

pub fn stream_columns( &self, logical_fields: impl Into<Arc<[LogicalFieldId]>>, row_ids: Vec<RowId>, policy: GatherNullPolicy, ) -> LlkvResult<ColumnStream<'_, P>>

Create a streaming view over the provided row IDs for the specified logical fields.

pub fn store(&self) -> &ColumnStore<P>

pub fn table_id(&self) -> TableId

pub fn total_rows_for_col(&self, col_id: FieldId) -> Result<u64>

Return the total number of rows for a given user column id in this table.

This delegates to the ColumnStore descriptor for the logical field that corresponds to (table_id, col_id) and returns the persisted total_row_count.

pub fn total_rows(&self) -> Result<u64>

Return the total number of rows for this table.

Prefer reading the dedicated row-id shadow column if present; otherwise fall back to inspecting any persisted user column descriptor.