shrub_parser/
arithmetic.rs

1//! Parser for shell arithmetic expressions.
2
3use crate::{ast, error};
4
5/// Parses a shell arithmetic expression.
6///
7/// # Arguments
8///
9/// * `input` - The arithmetic expression to parse, in string form.
10pub fn parse(input: &str) -> Result<ast::ArithmeticExpr, error::WordParseError> {
11    cacheable_parse(input)
12}
13
14#[cached::macros::cached(max_size = 64, key = "String", convert = r#"{ input.to_owned() }"#)]
15fn cacheable_parse(input: &str) -> Result<ast::ArithmeticExpr, error::WordParseError> {
16    tracing::debug!(target: "arithmetic", "parsing arithmetic expression: '{input}'");
17    arithmetic::full_expression(input)
18        .map_err(|e| error::WordParseError::ArithmeticExpression(e.into()))
19}
20
21peg::parser! {
22    grammar arithmetic() for str {
23        pub(crate) rule full_expression() -> ast::ArithmeticExpr =
24            ![_] { ast::ArithmeticExpr::Literal(0) } /
25            _ e:expression() _ { e }
26
27        pub(crate) rule expression() -> ast::ArithmeticExpr = precedence!{
28            x:(@) _ "," _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Comma, Box::new(x), Box::new(y)) }
29            --
30            x:lvalue() _ "*=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::Multiply, x, Box::new(y)) }
31            x:lvalue() _ "/=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::Divide, x, Box::new(y)) }
32            x:lvalue() _ "%=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::Modulo, x, Box::new(y)) }
33            x:lvalue() _ "+=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::Add, x, Box::new(y)) }
34            x:lvalue() _ "-=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::Subtract, x, Box::new(y)) }
35            x:lvalue() _ "<<=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::ShiftLeft, x, Box::new(y)) }
36            x:lvalue() _ ">>=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::ShiftRight, x, Box::new(y)) }
37            x:lvalue() _ "&=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::BitwiseAnd, x, Box::new(y)) }
38            x:lvalue() _ "|=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::BitwiseOr, x, Box::new(y)) }
39            x:lvalue() _ "^=" _ y:(@) { ast::ArithmeticExpr::BinaryAssignment(ast::BinaryOperator::BitwiseXor, x, Box::new(y)) }
40            x:lvalue() _ "=" _ y:(@) { ast::ArithmeticExpr::Assignment(x, Box::new(y)) }
41            --
42            x:@ _ "?" _ y:expression() _ ":" _ z:(@) { ast::ArithmeticExpr::Conditional(Box::new(x), Box::new(y), Box::new(z)) }
43            --
44            x:(@) _ "||" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::LogicalOr, Box::new(x), Box::new(y)) }
45            --
46            x:(@) _ "&&" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::LogicalAnd, Box::new(x), Box::new(y)) }
47            --
48            x:(@) _ "|" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::BitwiseOr, Box::new(x), Box::new(y)) }
49            --
50            x:(@) _ "^" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::BitwiseXor, Box::new(x), Box::new(y)) }
51            --
52            x:(@) _ "&" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::BitwiseAnd, Box::new(x), Box::new(y)) }
53            --
54            x:(@) _ "==" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Equals, Box::new(x), Box::new(y)) }
55            x:(@) _ "!=" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::NotEquals, Box::new(x), Box::new(y)) }
56            --
57            x:(@) _ "<" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::LessThan, Box::new(x), Box::new(y)) }
58            x:(@) _ ">" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::GreaterThan, Box::new(x), Box::new(y)) }
59            x:(@) _ "<=" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::LessThanOrEqualTo, Box::new(x), Box::new(y)) }
60            x:(@) _ ">=" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::GreaterThanOrEqualTo, Box::new(x), Box::new(y)) }
61            --
62            x:(@) _ "<<" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::ShiftLeft, Box::new(x), Box::new(y)) }
63            x:(@) _ ">>" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::ShiftRight, Box::new(x), Box::new(y)) }
64            --
65            x:(@) _ "+" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Add, Box::new(x), Box::new(y)) }
66            x:(@) _ "-" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Subtract, Box::new(x), Box::new(y)) }
67            --
68            x:(@) _ "*" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Multiply, Box::new(x), Box::new(y)) }
69            x:(@) _ "%" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Modulo, Box::new(x), Box::new(y)) }
70            x:(@) _ "/" _ y:@ { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Divide, Box::new(x), Box::new(y)) }
71            --
72            x:@ _ "**" _ y:(@) { ast::ArithmeticExpr::BinaryOp(ast::BinaryOperator::Power, Box::new(x), Box::new(y)) }
73            --
74            "!" _ x:(@) { ast::ArithmeticExpr::UnaryOp(ast::UnaryOperator::LogicalNot, Box::new(x)) }
75            "~" _ x:(@) { ast::ArithmeticExpr::UnaryOp(ast::UnaryOperator::BitwiseNot, Box::new(x)) }
76            --
77            // NOTE: The negative lookahead leaves `++x` and `--x` to the pre-increment/pre-decrement
78            // operators. As in bash, `++` or `--` before anything but a variable, such as `++0`, is
79            // two unary operators.
80            "+" !("+" _ lvalue()) _ x:(@) { ast::ArithmeticExpr::UnaryOp(ast::UnaryOperator::UnaryPlus, Box::new(x)) }
81            "-" !("-" _ lvalue()) _ x:(@) { ast::ArithmeticExpr::UnaryOp(ast::UnaryOperator::UnaryMinus, Box::new(x)) }
82            --
83            "++" _ x:lvalue() { ast::ArithmeticExpr::UnaryAssignment(ast::UnaryAssignmentOperator::PrefixIncrement, x) }
84            "--" _ x:lvalue() { ast::ArithmeticExpr::UnaryAssignment(ast::UnaryAssignmentOperator::PrefixDecrement, x) }
85            --
86            x:lvalue() _ "++" { ast::ArithmeticExpr::UnaryAssignment(ast::UnaryAssignmentOperator::PostfixIncrement, x) }
87            x:lvalue() _ "--" { ast::ArithmeticExpr::UnaryAssignment(ast::UnaryAssignmentOperator::PostfixDecrement, x) }
88            --
89            n:literal_number() { ast::ArithmeticExpr::Literal(n) }
90            l:lvalue() { ast::ArithmeticExpr::Reference(l) }
91            "(" _ expr:expression() _ ")" { expr }
92        }
93
94        rule lvalue() -> ast::ArithmeticTarget =
95            name:variable_name() "[" index:expression() "]" {
96                ast::ArithmeticTarget::ArrayElement(name.to_owned(), Box::new(index))
97            } /
98            name:variable_name() {
99                ast::ArithmeticTarget::Variable(name.to_owned())
100            }
101
102        rule variable_name() -> &'input str =
103            $(['a'..='z' | 'A'..='Z' | '_'](['a'..='z' | 'A'..='Z' | '_' | '0'..='9']*))
104
105        rule _() -> () = quiet!{[' ' | '\t' | '\n' | '\r']*} {}
106
107        rule literal_number() -> i64 =
108            // Literal with explicit radix (format: <base>#<literal>)
109            radix:decimal_literal() "#" s:$(['0'..='9' | 'a'..='z' | 'A'..='Z' | '@' | '_']+) {?
110                parse_shell_literal_number(s, radix.cast_unsigned())
111            } /
112            // Hex literal
113            "0" ['x' | 'X'] s:$(['0'..='9' | 'a'..='f' | 'A'..='F']*) {?
114                i64::from_str_radix(s, 16).or(Err("i64"))
115            } /
116            // Octal literal
117            s:$("0" ['0'..='8']*) {?
118                i64::from_str_radix(s, 8).or(Err("i64"))
119            } /
120            // Decimal literal
121            decimal_literal()
122
123        rule decimal_literal() -> i64 =
124            s:$(['1'..='9'] ['0'..='9']*) {?
125                // Parse as u64 first, then cast to i64. This handles values like
126                // 9223372036854775808 (i64::MAX + 1) which is needed for INT64_MIN
127                // when preceded by unary minus: -(9223372036854775808) wraps to i64::MIN.
128                s.parse::<u64>().map(|v| v.cast_signed()).or(Err("i64"))
129            }
130    }
131}
132
133fn parse_shell_literal_number(s: &str, radix: u64) -> Result<i64, &'static str> {
134    if !(2..=64).contains(&radix) {
135        return Err("invalid base");
136    }
137
138    // For bases <= 36: case-insensitive (a-z and A-Z both map to 10-35)
139    // For bases > 36 (bash extension):
140    //   0-9 = 0-9, a-z = 10-35, A-Z = 36-61, @ = 62, _ = 63
141    let mut result: i64 = 0;
142
143    for ch in s.chars() {
144        let digit_val = if radix <= 36 {
145            match ch {
146                '0'..='9' => (ch as u64) - ('0' as u64),
147                'a'..='z' => (ch as u64) - ('a' as u64) + 10,
148                'A'..='Z' => (ch as u64) - ('A' as u64) + 10,
149                _ => return Err("invalid digit"),
150            }
151        } else {
152            match ch {
153                '0'..='9' => (ch as u64) - ('0' as u64),
154                'a'..='z' => (ch as u64) - ('a' as u64) + 10,
155                'A'..='Z' => (ch as u64) - ('A' as u64) + 36,
156                '@' => 62,
157                '_' => 63,
158                _ => return Err("invalid digit"),
159            }
160        };
161
162        if digit_val >= radix {
163            return Err("value too great for base");
164        }
165
166        result = result
167            .wrapping_mul(radix.cast_signed())
168            .wrapping_add(digit_val.cast_signed());
169    }
170
171    Ok(result)
172}
shrub_parser/arithmetic.rs

shrub_parser/
arithmetic.rs
