Struct SharedDictionary 

pub struct SharedDictionary { /* private fields */ }

Shared dictionary state that can be cloned cheaply across tokenizers.

Holds the raw dictionary bytes (Arc<DictData>) and a pre-parsed DoubleArrayTrie (Arc<DoubleArrayTrie>). Creating a new Lexicon from this struct is a 16-byte Arc::clone + lightweight offset parsing, not a 150 MB trie copy.

Implementations

impl SharedDictionary

pub fn new(data: Vec<u8>) -> Result<Self, String>

Build a SharedDictionary from owned (already inhibited) dictionary bytes.

Callers must apply inhibit_connection_in_data before passing data.

pub fn from_mmap(mmap: Mmap) -> Result<Self, String>

Build a SharedDictionary from a memory-mapped file.

Use this when no connection inhibitions are needed — the OS manages the pages and they can be shared / evicted without heap pressure.

pub fn create_lexicon(&self) -> (Lexicon, usize)

Create a Lexicon that shares the pre-built trie.

pub fn create_grammar(&self) -> Result<Grammar, String>

Re-parse the grammar section from the shared data.

pub fn data(&self) -> &Arc<DictData>

Get a reference to the shared data Arc.

pub fn has_synonyms(&self) -> bool

Whether this dictionary version includes synonym group IDs.