gazelle/
grammar.rs

1//! Grammar types - both public AST and internal representation types.
2
3/// An interned symbol ID for O(1) lookups.
4/// Layout:
5/// - IDs 0..num_terminals: terminals (EOF is always terminal 0)
6/// - IDs num_terminals.. onwards: non-terminals
7#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
8pub struct SymbolId(pub(crate) u32);
9
10impl SymbolId {
11    /// The EOF symbol ID (always 0).
12    pub const EOF: SymbolId = SymbolId(0);
13
14    /// Create a SymbolId from a raw u32.
15    #[doc(hidden)]
16    pub const fn new(id: u32) -> Self {
17        SymbolId(id)
18    }
19}
20
21// ============================================================================
22// Public AST types for grammar definitions (require alloc)
23// ============================================================================
24
25use alloc::string::String;
26use alloc::vec::Vec;
27
28/// A grammar definition, typically produced by [`parse_grammar`](crate::parse_grammar)
29/// or built programmatically with fields.
30#[derive(Debug, Clone, PartialEq, Eq)]
31pub struct Grammar {
32    /// Name of the start symbol.
33    pub start: String,
34    /// Expected number of reduce/reduce conflicts.
35    pub expect_rr: usize,
36    /// Expected number of shift/reduce conflicts.
37    pub expect_sr: usize,
38    /// Terminal definitions.
39    pub terminals: Vec<TerminalDef>,
40    /// Grammar rules (productions).
41    pub rules: Vec<Rule>,
42}
43
44/// How a terminal's shift/reduce conflicts are resolved.
45#[derive(Debug, Clone, Copy, PartialEq, Eq)]
46pub enum TerminalKind {
47    /// Normal terminal — conflicts are reported as errors.
48    Plain,
49    /// `prec` — resolved at runtime by comparing `Precedence` levels.
50    Prec,
51    /// `shift` — conflicts are resolved statically in favor of shift.
52    Shift,
53    /// `reduce` — conflicts are resolved statically in favor of reduce.
54    Reduce,
55    /// `conflict` — resolved at runtime by the lexer passing `Resolve::Shift` or `Resolve::Reduce`.
56    Conflict,
57}
58
59/// A terminal definition in the grammar.
60#[derive(Debug, Clone, PartialEq, Eq)]
61pub struct TerminalDef {
62    /// Terminal name (e.g., "NUM", "PLUS").
63    pub name: String,
64    /// Whether this terminal carries a typed payload.
65    pub has_type: bool,
66    /// How shift/reduce conflicts on this terminal are resolved.
67    pub kind: TerminalKind,
68    /// Optional regex pattern for automatic lexer generation.
69    pub pattern: Option<String>,
70}
71
72/// A rule (production) in the grammar.
73#[derive(Debug, Clone, PartialEq, Eq)]
74pub struct Rule {
75    /// Non-terminal name (left-hand side).
76    pub name: String,
77    /// Alternatives (right-hand sides).
78    pub alts: Vec<Alt>,
79}
80
81/// An alternative (right-hand side) of a rule.
82#[derive(Debug, Clone, PartialEq, Eq)]
83pub struct Alt {
84    /// Terms in this alternative.
85    pub terms: Vec<Term>,
86    /// Action name (e.g., `=> binop`).
87    pub name: String,
88}
89
90/// A term in a grammar rule.
91#[derive(Debug, Clone, PartialEq, Eq, Hash, PartialOrd, Ord)]
92pub enum Term {
93    /// Plain symbol reference.
94    Symbol(String),
95    /// `?` - optional (zero or one).
96    Optional(String),
97    /// `*` - zero or more.
98    ZeroOrMore(String),
99    /// `+` - one or more.
100    OneOrMore(String),
101    /// `%` - one or more separated by the given symbol.
102    SeparatedBy { symbol: String, sep: String },
103    /// `_` - empty production marker.
104    Empty,
105}
gazelle/grammar.rs

gazelle/
grammar.rs
