Struct DictionaryEncoding 

pub struct DictionaryEncoding { /* private fields */ }

Stores repeated strings efficiently by referencing them with integer codes.

Each unique string appears once in the dictionary. Values are stored as LE u32 indices pointing into that dictionary, refcounted as bytes::Bytes so heap-owned and mmap-backed columns share the same type (revised D7).

Implementations

impl DictionaryEncoding

pub fn new(dictionary: Arc<[Arc<str>]>, codes: Vec<u32>) -> Self

Creates a new dictionary encoding from a dictionary and codes (legacy Vec<u32> input).

pub fn from_bytes_storage( dictionary: Arc<[Arc<str>]>, codes_bytes: Bytes, code_count: usize, ) -> Self

Constructs a dictionary encoding from pre-encoded bytes (Phase 3c entry point).

codes_bytes must be code_count * 4 bytes of LE u32 values.

pub fn with_nulls(self, null_bitmap: Vec<u64>) -> Self

Adds a null bitmap to this encoding (legacy Vec<u64> input).

pub fn with_null_bytes(self, null_bitmap: Bytes) -> Self

Adds a pre-encoded null bitmap (Phase 3c entry point).

pub fn len(&self) -> usize

Returns the number of values.

pub fn is_empty(&self) -> bool

Returns whether the encoding is empty.

pub fn dictionary_size(&self) -> usize

Returns the number of unique strings in the dictionary.

pub fn dictionary(&self) -> &Arc<[Arc<str>]>

Returns the dictionary.

pub fn codes_bytes(&self) -> Bytes

Returns the encoded codes as raw LE u32 bytes (always materialised).

Phase 3b: codes storage is bytes::Bytes. Use Self::code_at for indexed access; this returns the raw byte storage for serializers that write the storage out directly.

pub fn as_codes_slice(&self) -> Option<&[u32]>

Returns a direct &[u32] slice when the codes live in RAM.

pub fn as_null_words_slice(&self) -> Option<&[u64]>

Returns a direct &[u64] view of the null bitmap when it lives in RAM. None when there is no null bitmap or the bitmap is mmap-backed.

pub fn code_count(&self) -> usize

Number of u32 codes stored.

pub fn code_at(&self, idx: usize) -> Option<u32>

Returns the code at idx, or None if out of range.

pub fn codes(&self) -> Vec<u32>

Returns the codes as a materialized Vec<u32> (allocates).

Prefer Self::code_at or Self::code_count for reads. This exists for callers that need a contiguous slice and accept the allocation (e.g., legacy serialization paths).

pub fn is_null(&self, index: usize) -> bool

Returns whether the value at index is null.

pub fn get(&self, index: usize) -> Option<&str>

Returns the string value at the given index.

Returns None if the value is null.

pub fn get_code(&self, index: usize) -> Option<u32>

Returns the code at the given index.

pub fn iter(&self) -> impl Iterator<Item = Option<&str>>

Iterates over all values, yielding Option<&str>.

pub fn compression_ratio(&self) -> f64

Returns the compression ratio (original size / compressed size).

pub fn encode(&self, value: &str) -> Option<u32>

Encodes a lookup value into a code, if it exists in the dictionary.

pub fn filter_by_code(&self, predicate: impl Fn(u32) -> bool) -> Vec<usize>

Returns the row offsets where the code matches predicate and the row is not null.

Branches once on the storage variants so the per-row body is a tight loop over native slices in the common in-memory case.

Trait Implementations

impl Clone for DictionaryEncoding

fn clone(&self) -> DictionaryEncoding

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for DictionaryEncoding

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.