Struct FastFieldWriter 

pub struct FastFieldWriter {
    pub column_type: FastFieldColumnType,
    pub multi: bool,
    /* private fields */
}

Collects values during indexing and serializes a single fast-field column.

Supports both single-valued and multi-valued columns. For multi-valued columns, values are stored in a flat array with an offset column that maps doc_id → value range.

Fields

column_type: FastFieldColumnType

multi: bool

Whether this is a multi-valued column.

Implementations

impl FastFieldWriter

pub fn new_numeric(column_type: FastFieldColumnType) -> Self

Create a writer for a single-valued numeric column (u64/i64/f64).

pub fn new_numeric_multi(column_type: FastFieldColumnType) -> Self

Create a writer for a multi-valued numeric column.

pub fn new_text() -> Self

Create a writer for a single-valued text ordinal column.

pub fn new_text_multi() -> Self

Create a writer for a multi-valued text ordinal column.

pub fn add_u64(&mut self, doc_id: u32, value: u64)

Record a numeric value for doc_id. Fills gaps with 0. For single-value mode only.

pub fn add_i64(&mut self, doc_id: u32, value: i64)

Record an i64 value (zigzag-encoded).

pub fn add_f64(&mut self, doc_id: u32, value: f64)

Record an f64 value (sortable-encoded).

pub fn add_text(&mut self, doc_id: u32, value: &str)

Record a text value (dictionary-encoded at build time).

pub fn pad_to(&mut self, num_docs: u32)

Ensure the column covers num_docs entries.

Absent entries are filled with FAST_FIELD_MISSING for single-value columns, or with empty offset ranges for multi-value columns.

pub fn num_docs(&self) -> u32

Number of documents in this column.

pub fn serialize( &mut self, writer: &mut dyn Write, data_offset: u64, ) -> Result<(FastFieldTocEntry, u64)>

Serialize column data using blocked format with auto-selecting codec.

Writes a single block: [num_blocks(4)] BlockIndexEntry [block_data] [block_dict?] Returns (toc_entry, total_bytes_written).