300 lines
10 KiB
Rust
300 lines
10 KiB
Rust
use frontend::ast::{
|
||
BinaryOp, Binding, DeclKind, Expr, ExprKind, Function, Program, StmtKind, UnaryOp,
|
||
};
|
||
use frontend::{DiagnosticKind, Span, TokenKind, lex, parse, parse_expression};
|
||
|
||
fn function(program: &Program, index: usize) -> &Function {
|
||
let DeclKind::Function(function) = &program.declarations[index].kind else {
|
||
panic!("expected function")
|
||
};
|
||
function
|
||
}
|
||
|
||
fn binding(function: &Function, index: usize) -> &Binding {
|
||
let StmtKind::Binding(binding) = &function.body.statements[index].kind else {
|
||
panic!("expected binding")
|
||
};
|
||
binding
|
||
}
|
||
|
||
fn result(function: &Function) -> &Expr {
|
||
let StmtKind::Return(returned) = &function.body.statements.last().unwrap().kind else {
|
||
panic!("expected return")
|
||
};
|
||
returned.value.as_ref().unwrap()
|
||
}
|
||
|
||
fn binary(expression: &Expr, expected: BinaryOp) -> (&Expr, &Expr) {
|
||
let ExprKind::Binary {
|
||
operation,
|
||
left,
|
||
right,
|
||
} = &expression.kind
|
||
else {
|
||
panic!("expected a binary expression: {expression:?}");
|
||
};
|
||
assert_eq!(*operation, expected);
|
||
(left, right)
|
||
}
|
||
|
||
fn number(expression: &Expr, expected: &str) {
|
||
assert_eq!(expression.kind, ExprKind::Number(expected.into()));
|
||
}
|
||
|
||
#[test]
|
||
fn unicode_tokens_refer_to_exact_original_byte_ranges() {
|
||
let source = "// 日本語\r\nlet 商品_２ = \"こんにちは🌸\"";
|
||
let tokens = lex(source).unwrap();
|
||
assert_eq!(tokens.len(), 4);
|
||
assert_eq!(tokens[0].kind, TokenKind::Name);
|
||
assert_eq!(&source[tokens[0].span.start..tokens[0].span.end], "let");
|
||
assert_eq!(&source[tokens[1].span.start..tokens[1].span.end], "商品_２");
|
||
assert_eq!(tokens[2].kind, TokenKind::Equal);
|
||
assert_eq!(tokens[3].kind, TokenKind::String);
|
||
for token in tokens {
|
||
assert!(source.is_char_boundary(token.span.start));
|
||
assert!(source.is_char_boundary(token.span.end));
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn decimal_spelling_and_minus_tokens_are_preserved_without_float_conversion() {
|
||
let source = "-000123456789012345678901234567890.00100 / 7 // comment";
|
||
let tokens = lex(source).unwrap();
|
||
assert_eq!(
|
||
tokens.iter().map(|token| token.kind).collect::<Vec<_>>(),
|
||
vec![
|
||
TokenKind::Minus,
|
||
TokenKind::Number,
|
||
TokenKind::Slash,
|
||
TokenKind::Number,
|
||
]
|
||
);
|
||
assert_eq!(
|
||
&source[tokens[1].span.start..tokens[1].span.end],
|
||
"000123456789012345678901234567890.00100"
|
||
);
|
||
let expression = parse_expression(source).unwrap();
|
||
let (negative, divisor) = binary(&expression, BinaryOp::Divide);
|
||
let ExprKind::Unary {
|
||
operation: UnaryOp::Negate,
|
||
operand,
|
||
} = &negative.kind
|
||
else {
|
||
panic!("expected unary minus");
|
||
};
|
||
number(operand, "000123456789012345678901234567890.00100");
|
||
number(divisor, "7");
|
||
}
|
||
|
||
#[test]
|
||
fn strings_decode_only_the_supported_escapes_and_keep_comments_inside_strings() {
|
||
let expression = parse_expression(r#""商品\n\t\"\\// text""#).unwrap();
|
||
assert_eq!(
|
||
expression.kind,
|
||
ExprKind::String("商品\n\t\"\\// text".into())
|
||
);
|
||
assert_eq!(
|
||
expression.span,
|
||
Span::new(0, r#""商品\n\t\"\\// text""#.len())
|
||
);
|
||
}
|
||
|
||
#[test]
|
||
fn malformed_strings_numbers_and_unknown_characters_report_lexical_errors() {
|
||
for source in [
|
||
"\"unfinished",
|
||
"\"bad\\q\"",
|
||
"\"bad\nline\"",
|
||
"\"bad\rline\"",
|
||
"\"bad\0\"",
|
||
"\"trailing\\",
|
||
"1e3",
|
||
"12name",
|
||
"1_000",
|
||
"'text'",
|
||
";",
|
||
"🌸",
|
||
"\0",
|
||
] {
|
||
let error = lex(source).unwrap_err();
|
||
assert_eq!(error.kind, DiagnosticKind::Lexical, "{source:?}");
|
||
assert!(error.span.start <= error.span.end);
|
||
assert!(error.span.end <= source.len());
|
||
assert!(source.is_char_boundary(error.span.start));
|
||
assert!(source.is_char_boundary(error.span.end));
|
||
}
|
||
let source = "\"日\\q\"";
|
||
let error = lex(source).unwrap_err();
|
||
assert_eq!(&source[error.span.start..error.span.end], "\\q");
|
||
assert_eq!(error.location(source).column, 3);
|
||
}
|
||
|
||
#[test]
|
||
fn arithmetic_precedence_and_left_associativity_form_the_expected_tree() {
|
||
let expression = parse_expression("1 - 2 - 3 * 4 / 5 + 6").unwrap();
|
||
let (subtract, last) = binary(&expression, BinaryOp::Add);
|
||
number(last, "6");
|
||
let (first_subtract, divide) = binary(subtract, BinaryOp::Subtract);
|
||
let (first, second) = binary(first_subtract, BinaryOp::Subtract);
|
||
number(first, "1");
|
||
number(second, "2");
|
||
let (multiply, fifth) = binary(divide, BinaryOp::Divide);
|
||
number(fifth, "5");
|
||
let (third, fourth) = binary(multiply, BinaryOp::Multiply);
|
||
number(third, "3");
|
||
number(fourth, "4");
|
||
let grouped = parse_expression("(1 + 2) * 3").unwrap();
|
||
let (sum, _) = binary(&grouped, BinaryOp::Multiply);
|
||
binary(sum, BinaryOp::Add);
|
||
assert_eq!(sum.span, Span::new(0, 7));
|
||
let nested = parse_expression("--1").unwrap();
|
||
let ExprKind::Unary { operand, .. } = nested.kind else {
|
||
panic!("expected outer negate");
|
||
};
|
||
assert!(matches!(
|
||
operand.kind,
|
||
ExprKind::Unary {
|
||
operation: UnaryOp::Negate,
|
||
..
|
||
}
|
||
));
|
||
}
|
||
|
||
#[test]
|
||
fn member_and_call_chains_bind_more_tightly_than_unary_minus() {
|
||
let expression = parse_expression("-factory(1).build(.title)(\"日\", 2)").unwrap();
|
||
let ExprKind::Unary {
|
||
operation: UnaryOp::Negate,
|
||
operand,
|
||
} = expression.kind
|
||
else {
|
||
panic!("expected negate");
|
||
};
|
||
let ExprKind::Call { callee, arguments } = operand.kind else {
|
||
panic!("expected last call");
|
||
};
|
||
assert_eq!(arguments.len(), 2);
|
||
assert_eq!(arguments[0].kind, ExprKind::String("日".into()));
|
||
let ExprKind::Call { callee, arguments } = callee.kind else {
|
||
panic!("expected build call");
|
||
};
|
||
let ExprKind::Symbol(name) = &arguments[0].kind else {
|
||
panic!("expected a symbol");
|
||
};
|
||
assert_eq!(name.text, "title");
|
||
let ExprKind::Member { object, name } = callee.kind else {
|
||
panic!("expected member");
|
||
};
|
||
assert_eq!(name.text, "build");
|
||
let ExprKind::Call { callee, arguments } = object.kind else {
|
||
panic!("expected factory call");
|
||
};
|
||
assert_eq!(callee.kind, ExprKind::Name("factory".into()));
|
||
number(&arguments[0], "1");
|
||
}
|
||
|
||
#[test]
|
||
fn parses_typed_functions_bindings_and_multiple_top_level_definitions() {
|
||
let example = parse(include_str!("../examples/basic.endor")).unwrap();
|
||
assert_eq!(example.declarations.len(), 2);
|
||
let source = r#"
|
||
// Basic draft syntax, including a Unicode name.
|
||
function discountedPrice(price: Number, discount: Number): Number {
|
||
let amount = price - discount
|
||
return amount
|
||
}
|
||
function 表示(): String { return "こんにちは" }
|
||
"#;
|
||
let program = parse(source).unwrap();
|
||
assert_eq!(program.declarations.len(), 2);
|
||
let first = function(&program, 0);
|
||
assert_eq!(first.name.text, "discountedPrice");
|
||
assert_eq!(first.parameters.len(), 2);
|
||
assert_eq!(first.parameters[0].type_name.text, "Number");
|
||
assert_eq!(first.return_type.text, "Number");
|
||
assert_eq!(binding(first, 0).name.text, "amount");
|
||
binary(&binding(first, 0).value, BinaryOp::Subtract);
|
||
assert_eq!(result(first).kind, ExprKind::Name("amount".into()));
|
||
assert_eq!(
|
||
&source[first.body.statements.last().unwrap().span.start
|
||
..first.body.statements.last().unwrap().span.end],
|
||
"return amount"
|
||
);
|
||
assert_eq!(
|
||
result(function(&program, 1)).kind,
|
||
ExprKind::String("こんにちは".into())
|
||
);
|
||
assert_eq!(
|
||
&source[function(&program, 1).span.start..function(&program, 1).span.end],
|
||
"function 表示(): String { return \"こんにちは\" }"
|
||
);
|
||
// The output owns its text and can outlive the input buffer.
|
||
let owned = {
|
||
let source = String::from("function f(): String { return \"ok\" }");
|
||
parse(&source).unwrap()
|
||
};
|
||
assert_eq!(function(&owned, 0).name.text, "f");
|
||
}
|
||
|
||
#[test]
|
||
fn syntax_words_are_contextual_and_block_layout_does_not_depend_on_newlines() {
|
||
let program =
|
||
parse("function class(function: Type): Type { let agent = function return agent }")
|
||
.unwrap();
|
||
assert_eq!(function(&program, 0).name.text, "class");
|
||
assert_eq!(function(&program, 0).parameters[0].name.text, "function");
|
||
assert_eq!(binding(function(&program, 0), 0).name.text, "agent");
|
||
let multiline =
|
||
parse("function f\r\n(\r\nx: Number\r\n): Number\r\n{\r\nreturn x\r\n}").unwrap();
|
||
assert_eq!(
|
||
result(function(&multiline, 0)).kind,
|
||
ExprKind::Name("x".into())
|
||
);
|
||
}
|
||
|
||
#[test]
|
||
fn parser_requires_complete_input_and_rejects_unimplemented_or_malformed_syntax() {
|
||
for source in [
|
||
"",
|
||
"// empty",
|
||
"let x = 1",
|
||
"use atelier:shop.data",
|
||
"agent Worker {}",
|
||
"function f() { return 1 }",
|
||
"function f(x): Number { return x }",
|
||
"function f(x: Number,): Number { return x }",
|
||
"function f(): Number { return 1 let x = 2 }",
|
||
"function f(): Number { return 1 } extra",
|
||
"function f(): Number { function g(): Number { return 1 } return 2 }",
|
||
] {
|
||
assert!(parse(source).is_err(), "{source}");
|
||
}
|
||
for source in [
|
||
"", "x y", "1 +", "+1", ".5", "1.", "f(1,)", "f(,1)", "f(1 2)", "(1", "obj.", "[1]",
|
||
"a == b", "1.2.3",
|
||
] {
|
||
assert!(parse_expression(source).is_err(), "{source}");
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn syntax_errors_map_token_indexes_to_utf8_source_and_eof_locations() {
|
||
let source = "function f(): String {\r\n let 商品 = \"日\"\r\n return )\r\n}";
|
||
let error = parse(source).unwrap_err();
|
||
assert_eq!(error.kind, DiagnosticKind::Syntax);
|
||
assert_eq!(&source[error.span.start..error.span.end], ")");
|
||
assert_eq!(error.location(source).line, 3);
|
||
assert_eq!(error.location(source).column, 10);
|
||
assert!(!error.expected.is_empty());
|
||
let source = "function f(): Number { return 1\r\n ";
|
||
let error = parse(source).unwrap_err();
|
||
assert_eq!(error.span, Span::new(source.len(), source.len()));
|
||
assert_eq!(error.location(source).line, 2);
|
||
assert_eq!(error.location(source).column, 3);
|
||
let source = "\r\n\r\n🌸";
|
||
let error = lex(source).unwrap_err();
|
||
assert_eq!(error.location(source).line, 3);
|
||
assert_eq!(error.location(source).column, 1);
|
||
}
|