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
4/// Kinds of tokens (terminals, from the lexer) and nodes (composites, from the
5/// parser) in the CST.
6///
7/// Token kinds come first, node kinds after; `ROOT` is kept **last** so
8/// [`rowan::Language::kind_from_raw`] can bounds-check the raw discriminant with
9/// a single comparison. Keep the hidden end marker immediately after `ROOT`.
10#[allow(non_camel_case_types)]
11#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
12#[repr(u16)]
13pub enum SyntaxKind {
14    // --- Tokens (terminals, produced by the lexer) ---
15    CONTROL_WORD,   // `\foo`  (backslash + ASCII letters)
16    CONTROL_SYMBOL, // `\\`, `\{`, `\%`, `\,` … (backslash + one non-letter)
17    L_BRACE,        // {
18    R_BRACE,        // }
19    L_BRACKET,      // [
20    R_BRACKET,      // ]
21    DOLLAR,         // $
22    AMPERSAND,      // &
23    HASH,           // #
24    CARET,          // ^
25    UNDERSCORE,     // _
26    TILDE,          // ~
27    COMMENT,        // `% …` up to (not including) the line break
28    WHITESPACE,     // spaces / tabs
29    NEWLINE,        // `\n`, `\r\n`, or `\r`
30    WORD,           // a run of ordinary text characters
31    VERB,           // `\verb|…|` / `\verb*|…|` inline verbatim (a single token)
32    VERBATIM_BODY,  // the raw body of a verbatim-like environment (a single token)
33    DOC_MARGIN,     // a `.dtx` documentation line's leading `%` margin (trivia)
34    GUARD,          // a `.dtx` docstrip guard `%<…>` (`%<*t>`/`%</t>`/inline) (trivia)
35    ERROR,          // lexer fallback; the lexer is total, so this is unused today
36
37    // Nodes
38    GROUP,       // { … }
39    OPTIONAL,    // [ … ] optional argument
40    ARGUMENT,    // an argument attached to a command
41    COMMAND,     // a control sequence with its arguments
42    ENVIRONMENT, // \begin{…} … \end{…}
43    BEGIN,       // \begin{name}
44    END,         // \end{name}
45    NAME_GROUP,  // {name} following \begin / \end
46    // `\if…\else…\or…\fi`, when the shape gate pairs it. The construct is a run
47    // of `CONDITIONAL_BRANCH` nodes followed by the closing `\fi` as the last
48    // child, mirroring `ENVIRONMENT > BEGIN … END`. The `\if` test's extent is
49    // not statically resolvable (`\ifnum\radius>5` scans ⟨number⟩⟨rel⟩⟨number⟩
50    // by TeX's own scanner), so the opener and its test ride the *first* branch
51    // rather than a head node of their own.
52    CONDITIONAL,
53    // One branch of a `CONDITIONAL`. The first holds the opener, its test, and
54    // the then-body; every later one *starts with* its `\else`/`\or` divider, so
55    // a consumer finds the boundaries positionally and never by command name.
56    CONDITIONAL_BRANCH,
57    INLINE_MATH,  // $ … $   or   \( … \)
58    DISPLAY_MATH, // $$ … $$  or   \[ … \]
59    MATH,         // a math body (the atoms between the delimiters)
60    SCRIPTED,     // a base atom with attached scripts: base (SUBSCRIPT | SUPERSCRIPT)+
61    SUBSCRIPT,    // `_` and its tightly-bound script argument
62    SUPERSCRIPT,  // `^` and its tightly-bound script argument
63    LEFT_RIGHT,   // `\left( … \right)` — a matched delimiter pair wrapping a MATH body
64    PARAGRAPH,    // text delimited by blank lines
65    DOC_COMMENT,  // a bound leading-`%` comment run, grouped before its construct
66    TEXT,         // a run of text and trivia
67    LINE_BREAK,   // `\\`, with a tightly-bound `*` and/or `[len]` (`\\*[2ex]`)
68    // A `;`-terminated statement in a curated `statementBody` environment body
69    // (the TikZ/pgf picture family). Owns everything from the statement's first
70    // non-trivia element through the top-level WORD carrying the terminating
71    // `;`, so layout can derive statement boundaries from structure rather than
72    // authored newlines. A run that never reaches a `;` is left as plain
73    // paragraph content — recognition is retrospective and degrades silently.
74    STATEMENT,
75    ROOT, // the document root  (keep LAST)
76    #[doc(hidden)]
77    __LAST,
78}
79
80impl SyntaxKind {
81    /// The number of `SyntaxKind` variants. Sound because the enum is
82    /// `#[repr(u16)]` with contiguous discriminants `0..=ROOT` and `ROOT` is kept
83    /// last; used to size kind-indexed tables (e.g. the linter's dispatch table).
84    pub const COUNT: usize = SyntaxKind::__LAST as usize;
85}
86
87/// The rowan language marker for badness's CST.
88#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
89pub enum BadnessLang {}
90
91impl_rowan_lang!(BadnessLang, SyntaxKind, "LaTeX");
92
93pub type SyntaxNode = rowan::SyntaxNode<BadnessLang>;
94pub type SyntaxToken = rowan::SyntaxToken<BadnessLang>;
95pub type SyntaxElement = rowan::SyntaxElement<BadnessLang>;
96
97/// Whitespace and newlines are the only trivia the formatter rewrites; comments
98/// are preserved verbatim and so are *not* collapsible. A shared shape
99/// predicate: the formatter's gap normalization and the semantic layer's expl3
100/// statement segmentation both skip exactly this class.
101pub fn is_collapsible_trivia(kind: SyntaxKind) -> bool {
102    matches!(kind, SyntaxKind::WHITESPACE | SyntaxKind::NEWLINE)
103}
104
105/// Trivia ignored by syntax readers: collapsible whitespace plus comments.
106pub fn is_trivia(kind: SyntaxKind) -> bool {
107    is_collapsible_trivia(kind) || kind == SyntaxKind::COMMENT
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
