big_code_analysis_ast/getter/
rust.rs

1//! `Getter` implementation for Rust.
2#![allow(clippy::wildcard_imports, clippy::enum_glob_use)]
3
4use super::*;
5
6impl Getter for RustCode {
7    fn get_func_space_name<'a, 'tree>(
8        node: &Node<'tree>,
9        code: &'a [u8],
10        _ancestors: Ancestors<'tree, '_>,
11    ) -> Option<&'a str> {
12        // we're in a function or in a class or an impl
13        // for an impl: we've  'impl ... type {...'
14        if let Some(name) = node
15            .child_by_field_name("name")
16            .or_else(|| node.child_by_field_name("type"))
17        {
18            node_text(code, &name)
19        } else {
20            Some("<anonymous>")
21        }
22    }
23
24    fn get_space_kind(node: &Node) -> SpaceKind {
25        use Rust::*;
26
27        match node.kind_id().into() {
28            FunctionItem | ClosureExpression => SpaceKind::Function,
29            TraitItem => SpaceKind::Trait,
30            ImplItem => SpaceKind::Impl,
31            SourceFile => SpaceKind::Unit,
32            _ => SpaceKind::Unknown,
33        }
34    }
35
36    fn get_op_type<'a>(node: &Node<'a>, ancestors: Ancestors<'a, '_>) -> TokenRole {
37        use Rust::*;
38
39        match node.kind_id().into() {
40            // `||` is treated as an operator only if it's part of a binary expression.
41            // This prevents misclassification inside macros where closures without arguments (e.g., `let closure = || { /* ... */ };`)
42            // are not recognized as `ClosureExpression` and their `||` node is identified as `PIPEPIPE` instead of `ClosureParameters`.
43            //
44            // Similarly, exclude `/` when it corresponds to the third slash in `///` (`OuterDocCommentMarker`)
45            PIPEPIPE | SLASH => match ancestors.parent(node) {
46                Some(parent) if matches!(parent.kind_id().into(), BinaryExpression) => {
47                    TokenRole::Operator
48                }
49                _ => TokenRole::Unknown,
50            },
51            // Ensure `!` is counted as an operator unless it belongs to an `InnerDocCommentMarker` `//!`
52            BANG => match ancestors.parent(node) {
53                Some(parent) if !matches!(parent.kind_id().into(), InnerDocCommentMarker) => {
54                    TokenRole::Operator
55                }
56                _ => TokenRole::Unknown,
57            },
58            // COLONCOLON (`::`) is the path-segment separator. C++, Java,
59            // C#, and Kotlin all classify it as an operator; omitting it
60            // here (issue #394) silently dropped every path expression
61            // (`std::collections::HashMap`, `Vec::new`, `T::method`) into
62            // TokenRole::Unknown, deflating n1/N1 for path-heavy code.
63            //
64            // The 14 declaration/visibility keywords (Const, Static, Enum,
65            // Struct, Trait, Impl, Use, Mod, Pub, Type, Union, Where,
66            // Extern, Dyn) were inconsistently absent — the impl already
67            // accepted 17 other keywords (As, Async, Await, Break, …, Fn).
68            // Including them brings declaration-heavy code in line with
69            // statement-heavy code.
70            LPAREN | LBRACE | LBRACK | As | EQGT | PLUS | STAR | Async | Await | Break
71            | Continue | Else | For | If | In | Let | Loop | Match | Return | Unsafe | While
72            | EQ | COMMA | DASHGT | QMARK | LT | GT | AMP | MutableSpecifier | DOTDOT
73            | DOTDOTEQ | DASH | AMPAMP | PIPE | CARET | EQEQ | BANGEQ | LTEQ | GTEQ | LTLT
74            | GTGT | PERCENT | PLUSEQ | DASHEQ | STAREQ | SLASHEQ | PERCENTEQ | AMPEQ | PIPEEQ
75            | CARETEQ | LTLTEQ | GTGTEQ | Move | DOT | PrimitiveType | PrimitiveType2
76            | PrimitiveType3 | PrimitiveType4 | PrimitiveType5 | PrimitiveType6
77            | PrimitiveType7 | PrimitiveType8 | PrimitiveType9 | PrimitiveType10
78            | PrimitiveType11 | PrimitiveType12 | PrimitiveType13 | PrimitiveType14
79            | PrimitiveType15 | PrimitiveType16 | PrimitiveType17 | Fn | SEMI | COLONCOLON
80            | Const | Static | Enum | Struct | Trait | Impl | Use | Mod | Pub | Type | Union
81            | Where | Extern | Dyn => TokenRole::Operator,
82            // FieldIdentifier (e.g. `p.x`) and TypeIdentifier (e.g. `Vec`,
83            // `HashMap`) are operand-class names — C++ and Go classify them
84            // the same way (see arms ~588 and ~862 below). Omitting them
85            // here silently dropped both into TokenRole::Unknown,
86            // deflating n2/N2 and the derived vocabulary/volume/effort
87            // estimates (issue #390).
88            Identifier | TypeIdentifier | FieldIdentifier | StringLiteral | RawStringLiteral
89            | IntegerLiteral | FloatLiteral | BooleanLiteral | Zelf | CharLiteral | UNDERSCORE => {
90                TokenRole::Operand
91            }
92            _ => TokenRole::Unknown,
93        }
94    }
95
96    get_operator!(Rust);
97}
big_code_analysis_ast/getter/rust.rs

big_code_analysis_ast/getter/
rust.rs
