badness_parser/
syntax.rs

1//! `SyntaxKind` — the kinds of CST tokens and nodes — and the rowan `Language`
2//! binding for badness's LaTeX surface CST.
3
4use rowan::Language;
5
6/// Kinds of tokens (terminals, from the lexer) and nodes (composites, from the
7/// parser) in the CST.
8///
9/// Token kinds come first, node kinds after; `ROOT` is kept **last** so
10/// [`BadnessLang::kind_from_raw`] can bounds-check the raw discriminant with a
11/// single comparison. Do not add variants after `ROOT`.
12#[allow(non_camel_case_types)]
13#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
14#[repr(u16)]
15pub enum SyntaxKind {
16    // --- Tokens (terminals, produced by the lexer) ---
17    CONTROL_WORD,   // `\foo`  (backslash + ASCII letters)
18    CONTROL_SYMBOL, // `\\`, `\{`, `\%`, `\,` … (backslash + one non-letter)
19    L_BRACE,        // {
20    R_BRACE,        // }
21    L_BRACKET,      // [
22    R_BRACKET,      // ]
23    DOLLAR,         // $
24    AMPERSAND,      // &
25    HASH,           // #
26    CARET,          // ^
27    UNDERSCORE,     // _
28    TILDE,          // ~
29    COMMENT,        // `% …` up to (not including) the line break
30    WHITESPACE,     // spaces / tabs
31    NEWLINE,        // `\n`, `\r\n`, or `\r`
32    WORD,           // a run of ordinary text characters
33    VERB,           // `\verb|…|` / `\verb*|…|` inline verbatim (a single token)
34    VERBATIM_BODY,  // the raw body of a verbatim-like environment (a single token)
35    DOC_MARGIN,     // a `.dtx` documentation line's leading `%` margin (trivia)
36    GUARD,          // a `.dtx` docstrip guard `%<…>` (`%<*t>`/`%</t>`/inline) (trivia)
37    ERROR,          // lexer fallback; the lexer is total, so this is unused today
38
39    // --- Nodes (composites, produced by the Phase 1 parser) ---
40    GROUP,        // { … }
41    OPTIONAL,     // [ … ] optional argument
42    ARGUMENT,     // an argument attached to a command
43    COMMAND,      // a control sequence with its arguments
44    ENVIRONMENT,  // \begin{…} … \end{…}
45    BEGIN,        // \begin{name}
46    END,          // \end{name}
47    NAME_GROUP,   // {name} following \begin / \end
48    INLINE_MATH,  // $ … $   or   \( … \)
49    DISPLAY_MATH, // $$ … $$  or   \[ … \]
50    MATH,         // a math body (the atoms between the delimiters)
51    SCRIPTED,     // a base atom with attached scripts: base (SUBSCRIPT | SUPERSCRIPT)+
52    SUBSCRIPT,    // `_` and its tightly-bound script argument
53    SUPERSCRIPT,  // `^` and its tightly-bound script argument
54    LEFT_RIGHT,   // `\left( … \right)` — a matched delimiter pair wrapping a MATH body
55    PARAGRAPH,    // text delimited by blank lines
56    DOC_COMMENT,  // a bound leading-`%` comment run, grouped before its construct
57    TEXT,         // a run of text and trivia
58    LINE_BREAK,   // `\\`, with a tightly-bound `*` and/or `[len]` (`\\*[2ex]`)
59    ROOT,         // the document root  (keep LAST)
60}
61
62impl SyntaxKind {
63    /// The number of `SyntaxKind` variants. Sound because the enum is
64    /// `#[repr(u16)]` with contiguous discriminants `0..=ROOT` and `ROOT` is kept
65    /// last; used to size kind-indexed tables (e.g. the linter's dispatch table).
66    pub const COUNT: usize = SyntaxKind::ROOT as usize + 1;
67}
68
69impl From<SyntaxKind> for rowan::SyntaxKind {
70    fn from(kind: SyntaxKind) -> Self {
71        Self(kind as u16)
72    }
73}
74
75/// The rowan language marker for badness's CST.
76#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
77pub enum BadnessLang {}
78
79impl Language for BadnessLang {
80    type Kind = SyntaxKind;
81
82    fn kind_from_raw(raw: rowan::SyntaxKind) -> SyntaxKind {
83        assert!(
84            raw.0 <= SyntaxKind::ROOT as u16,
85            "invalid SyntaxKind discriminant: {}",
86            raw.0
87        );
88        // SAFETY: `SyntaxKind` is `#[repr(u16)]` with contiguous discriminants
89        // `0..=ROOT`, and the assert above bounds `raw.0` into that range.
90        unsafe { std::mem::transmute::<u16, SyntaxKind>(raw.0) }
91    }
92
93    fn kind_to_raw(kind: SyntaxKind) -> rowan::SyntaxKind {
94        kind.into()
95    }
96}
97
98pub type SyntaxNode = rowan::SyntaxNode<BadnessLang>;
99pub type SyntaxToken = rowan::SyntaxToken<BadnessLang>;
100pub type SyntaxElement = rowan::SyntaxElement<BadnessLang>;
101
102/// Whitespace and newlines are the only trivia the formatter rewrites; comments
103/// are preserved verbatim and so are *not* collapsible. A shared shape
104/// predicate: the formatter's gap normalization and the semantic layer's expl3
105/// statement segmentation both skip exactly this class.
106pub fn is_collapsible_trivia(kind: SyntaxKind) -> bool {
107    matches!(kind, SyntaxKind::WHITESPACE | SyntaxKind::NEWLINE)
108}
109
110/// Whether a `WORD` token is a single TeX parameter digit (`1`..=`9`) — the
111/// shape that follows `#` in a parameter reference. Reads only the token text.
112pub fn is_param_digit(t: &SyntaxToken) -> bool {
113    matches!(t.text().as_bytes(), [b'1'..=b'9'])
114}
badness_parser/syntax.rs

badness_parser/
syntax.rs
