Struct GrammarData 

pub struct GrammarData {
    pub allow_empty_rule_ids: Vec<i32>,
    pub optimized: bool,
    pub complete_fsm: CompactFsm,
    pub per_rule_fsms: Vec<Option<CompactFsmWithStartEnd>>,
    pub per_rule_fsm_hashes: Vec<Option<u64>>,
    pub per_rule_fsm_new_state_ids: Vec<Vec<(i32, i32)>>,
    /* private fields */
}

The BNF grammar AST. Equivalent to xgrammar’s Grammar::Impl.

The FSM acceleration fields (complete_fsm, per_rule_fsms, per_rule_fsm_hashes, per_rule_fsm_new_state_ids) are populated by the W3 grammar-functor passes (GrammarFSMBuilder / GrammarFSMHasher); they are Default::default() until then.

Fields

allow_empty_rule_ids: Vec<i32>

Rule ids permitted to match the empty string.

optimized: bool

Whether grammar optimization passes have run.

complete_fsm: CompactFsm

The single FSM holding every rule’s states/edges, built by GrammarFSMBuilder. Empty until the FSM builder runs.

per_rule_fsms: Vec<Option<CompactFsmWithStartEnd>>

Per-rule FSM views into complete_fsm. None for a rule whose body could not be expressed as an FSM. Empty until built.

per_rule_fsm_hashes: Vec<Option<u64>>

Per-rule structural hash of the rule’s FSM, or None when the rule could not be hashed. Set by GrammarFSMHasher.

per_rule_fsm_new_state_ids: Vec<Vec<(i32, i32)>>

Per-rule (original_state_id, new_state_id) mapping produced alongside the hash by GrammarFSMHasher.

Implementations

impl GrammarData

pub fn new() -> Self

An empty grammar with no root.

pub fn from_parts( rules: Vec<Rule>, expr_data: Vec<i32>, expr_indptr: Vec<i32>, root_rule_id: i32, ) -> Self

Construct directly from CSR parts — used by the grammar builder (W2) and the deserializer.

pub fn set_rule_name(&mut self, rule_id: i32, name: impl Into<String>)

Mutable access to a rule’s name — used by RootRuleRenamer.

pub fn set_expr_data(&mut self, expr_id: i32, index: usize, value: i32)

Borrowed view of expr expr_id as (type, &mut data[..]). Used by RepetitionNormalizer to rewrite a Repeat payload in place.

pub fn num_rules(&self) -> i32

Number of rules.

pub fn rule(&self, rule_id: i32) -> &Rule

The rule with the given id. Panics if out of bounds (matches the C++ XGRAMMAR_DCHECK contract).

pub fn rule_mut(&mut self, rule_id: i32) -> &mut Rule

Mutable access to a rule.

pub fn rules(&self) -> &[Rule]

All rules.

pub fn root_rule_id(&self) -> i32

The root rule id (-1 if unset).

pub fn set_root_rule_id(&mut self, id: i32)

Set the root rule id.

pub fn root_rule(&self) -> &Rule

The root rule. Panics if the root is unset/out of bounds.

pub fn num_exprs(&self) -> i32

Number of expressions.

pub fn append_expr(&mut self, kind: GrammarExprType, data: &[i32]) -> i32

Append a raw (type, data[]) expression to the CSR store and return its new expr id. Used by LookaheadAssertionAnalyzer, which adds Sequence exprs to an already-built grammar.

pub fn expr(&self, expr_id: i32) -> GrammarExpr<'_>

Borrowed view of expression expr_id. The CSR slot is [type, data_len, data0, …]; this returns (type, &data[..]).

pub fn byte_string(&self, expr_id: i32) -> String

Decode a ByteString expression to its String.

pub fn tag_dispatch(&self, expr_id: i32) -> TagDispatch

Decode a TagDispatch expression. Panics if expr_id is not a tag dispatch (matches the C++ XGRAMMAR_DCHECK).

Trait Implementations

impl Clone for GrammarData

fn clone(&self) -> GrammarData

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GrammarData

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

Formats the value using the given formatter.

impl Default for GrammarData

fn default() -> GrammarData

Returns the “default value” for a type.

impl From<GrammarData> for Grammar

fn from(data: GrammarData) -> Self

Converts to this type from the input type.