schala/schala-lang/language/src/parsing.rs
2019-06-13 02:27:11 -07:00

1810 lines
60 KiB
Rust

//! # Parsing
//! This module is where the recursive-descent parsing methods live.
//!
//!
//! # Schala EBNF Grammar
//! This document is the authoritative grammar of Schala, represented in something approximating
//! Extended Backus-Naur form. Terminal productions are in "double quotes", or UPPERCASE
//! if they represent a class of tokens rather than an specific string, or are otherwise
//! unreprsentable in ASCII.
//!
//! ## Top level structure
//!
//! ```
//! program := (statement delimiter)* EOF
//! delimiter := NEWLINE | ";"
//! statement := expression | declaration
//! block := "{" (statement delimiter)* "}"
//! declaration := type_declaration | func_declaration | binding_declaration | impl_declaration
//! ```
//! ## Declarations
//!
//! ### Types
//! ```
//! type_declaration := "type" type_declaration_body
//! type_declaration_body := "alias" type_alias | "mut"? type_singleton_name "=" type_body
//! type_alias := IDENTIFIER "=" type_name
//! type_body := variant_specifier ("|" variant_specifier)*
//! variant_specifier := IDENTIFIER | IDENTIFIER "{" typed_identifier_list "}" | IDENTIFIER "(" type_name* ")"
//! typed_identifier_list := typed_identifier*
//! typed_identifier := IDENTIFIER type_anno
//! ```
//! ### Functions
//!
//! ```
//! func_declaration := func_signature func_body
//! func_body := ε | nonempty_func_body
//! nonempty_func_body := "{" (statement delimiter)* "}"
//! func_signature := "fn" func_name formal_param_list type_anno+
//! func_name := IDENTIFIER | operator
//! formal_param_list := "(" (formal_param ",")* ")"
//! formal_param := IDENTIFIER type_anno+
//! ```
//!
//! ### Variable bindings
//! ```binding_declaration := "let" "mut"? IDENTIFIER "=" expresion```
//!
//! ### Interfaces
//!
//! ```
//! interface_declaration := "interface" type_singleton_name signature_block
//! impl_declaration := "impl" type_singleton_name decl_block | "impl" type_singleton_name "for" type_name decl_block
//! decl_block := "{" (func_declaration)* "}"
//! signature_block := "{" (func_signature)* "}"
//! ```
//!
//! ### Type Annotations
//!
//! ```
//! type_anno := ":" type_name
//! type_name := type_singleton_name | "(" type_names ")"
//! type_names := ε | type_name (, type_name)*
//! type_singleton_name = IDENTIFIER (type_params)*
//! type_params := "<" type_name (, type_name)* ">"
//! ```
//!
//! ## Expressions
//! ```
//! expression := precedence_expr type_anno+
//! precedence_expr := prefix_expr
//! prefix_expr := prefix_op call_expr
//! prefix_op := "+" | "-" | "!" | "~"
//! call_expr := index_expr ( "(" invocation_list ")" )* | ε
//! invocation_list := invocation_argument ("," invocation_argument)* | ε
//! invocation_argument := expression | IDENTIFIER "=" expression | "_"
//! index_expr := primary ( "[" (expression ("," (expression)* | ε) "]" )*
//! primary := literal | paren_expr | if_expr | for_expr | while_expr | identifier_expr | lambda_expr | anonymous_struct | list_expr
//! expr_or_block := "{" (statement delimiter)* "}" | expr
//! ```
//!
//! ### Primary expressions
//!
//! ```
//! list_expr := "[" (expression, ",")* "]"
//! lambda_expr := "\\" lambda_param_list type_anno+ nonempty_func_body
//! lambda_param_list := formal_param_list | formal_param
//! paren_expr := "(" paren_inner ")"
//! paren_inner := (expression ",")*
//! identifier_expr := named_struct | IDENTIFIER
//! ```
//!
//! ## Literals
//! ```
//! literal := "true" | "false" | number_literal | STR_LITERAL
//! named_struct := IDENTIFIER record_block
//! record_block := "{" (record_entry, ",")* | "}" //TODO support anonymus structs, update syntax
//! record_entry := IDENTIFIER ":" expression
//! anonymous_struct := TODO
//! number_literal := int_literal | float_literal
//! int_literal = ("0x" | "0b") digits
//! float_literal := digits ("." digits)
//! digits := (digit_group underscore*)+
//! digit_group := DIGIT+
//! ```
//!
//! ### Patterns
//! ```
//! pattern := "(" (pattern, ",")* ")" | simple_pattern
//! simple_pattern := pattern_literal | record_pattern | tuple_struct_pattern
//! pattern_literal := "true" | "false" | signed_number_literal | STR_LITERAL | IDENTIFIER
//! signed_number_literal := "-"? number_literal
//! record_pattern := IDENTIFIER "{" (record_pattern_entry, ",")* "}"
//! record_pattern_entry := IDENTIFIER | IDENTIFIER ":" Pattern
//! tuple_struct_pattern := IDENTIFIER "(" (pattern, ",")* ")"
//! ```
//!
//! ### If-expressions
//! ```
//! if_expr := "if" discriminator ("then" condititional | "is" simple_pattern_match | guard_block)
//! discriminator := precedence_expr (operator)+
//! conditional := expr_or_block else_clause
//! simple_pattern_match := pattern "then" conditional
//! else_clause := ε | "else" expr_or_block
//! guard_block := "{" (guard_arm, ",")* "}"
//! guard_arm := guard "->" expr_or_block
//! guard := "is" pattern | (operator)+ precedence_expr
//! ```
//!
//! ### While expressions
//! ```
//! while_expr := "while" while_cond "{" (statement delimiter)* "}"
//! while_cond := ε | expression | expression "is" pattern //TODO maybe is-expresions should be primary
//! ```
//!
//! //TODO this implies there must be at least one enumerator, which the parser doesn"t support right
//! //this second, and maybe should fail later anyway
//! ### For-expressions
//! ```
//! for_expr := "for" (enumerator | "{" enumerators "}") for_expr_body
//! for_expr_body := "return" expression | "{" (statement delimiter)* "}"
//! enumerators := enumerator ("," enumerators)*
//! enumerator := identifier "<-" expression | identifier "=" expression //TODO add guards, etc.
//! ```
//!
use std::rc::Rc;
use std::iter::Peekable;
use std::vec::IntoIter;
use crate::tokenizing::*;
use crate::tokenizing::Kw::*;
use crate::tokenizing::TokenKind::*;
use crate::ast::*;
use crate::builtin::{BinOp, PrefixOp};
/// Represents a parsing error
#[derive(Debug)]
pub struct ParseError {
pub msg: String,
pub token: Token
}
impl ParseError {
fn new_with_token<T, M>(msg: M, token: Token) -> ParseResult<T> where M: Into<String> {
Err(ParseError { msg: msg.into(), token })
}
}
/// Represents either a successful parsing result or a ParseError
pub type ParseResult<T> = Result<T, ParseError>;
#[derive(Debug)]
pub struct ParseRecord {
production_name: String,
next_token: String,
level: u32,
}
/// Main data structure for doing parsing.
pub struct Parser {
token_handler: TokenHandler,
parse_record: Vec<ParseRecord>,
parse_level: u32,
restrictions: ParserRestrictions,
}
struct ParserRestrictions {
no_struct_literal: bool
}
struct TokenHandler {
tokens: Peekable<IntoIter<Token>>,
end_of_file: (usize, usize),
}
impl TokenHandler {
fn new(tokens: Vec<Token>) -> TokenHandler {
let end_of_file = match tokens.last() {
None => (0, 0),
Some(t) => (t.line_num, t.char_num)
};
let tokens = tokens.into_iter().peekable();
TokenHandler { tokens, end_of_file }
}
fn peek_kind(&mut self) -> TokenKind {
self.tokens.peek().map(|ref t| { t.kind.clone() }).unwrap_or(TokenKind::EOF)
}
fn peek(&mut self) -> Token {
self.tokens.peek().map(|t: &Token| { t.clone()}).unwrap_or(Token { kind: TokenKind::EOF, line_num: self.end_of_file.0, char_num: self.end_of_file.1})
}
fn next(&mut self) -> Token {
self.tokens.next().unwrap_or(Token { kind: TokenKind::EOF, line_num: self.end_of_file.0, char_num: self.end_of_file.1})
}
}
impl Parser {
/// Create a new parser initialized with some tokens.
pub fn new(initial_input: Vec<Token>) -> Parser {
Parser {
token_handler: TokenHandler::new(initial_input),
parse_record: vec![],
parse_level: 0,
restrictions: ParserRestrictions { no_struct_literal: false }
}
}
/// Parse all loaded tokens up to this point.
pub fn parse(&mut self) -> ParseResult<AST> {
self.program()
}
/*
pub fn parse_with_new_tokens(&mut self, new_tokens: Vec<Token>) -> ParseResult<AST> {
}
*/
pub fn format_parse_trace(self) -> Vec<String> {
self.parse_record.into_iter().map(|r| {
let mut indent = String::new();
for _ in 0..r.level {
indent.push(' ');
}
format!("{}Production `{}`, token: {}", indent, r.production_name, r.next_token)
}).collect()
}
}
macro_rules! print_token_pattern {
($tokenpattern:pat) => { stringify!($tokenpattern) }
}
macro_rules! expect {
($self:expr, $token_kind:pat) => { expect!($self, $token_kind if true) };
($self:expr, $expected_kind:pat if $cond:expr) => {
{
let tok = $self.token_handler.peek();
match tok.get_kind() {
$expected_kind if $cond => $self.token_handler.next(),
actual_kind => {
let msg = format!("Expected {}, got {:?}", print_token_pattern!($expected_kind), actual_kind);
return ParseError::new_with_token(msg, tok);
}
}
}
}
}
macro_rules! delimited {
($self:expr, $start:pat, $parse_fn:ident, $( $delim:pat )|+, $end:pat, nonstrict) => {
delimited!($self, $start, $parse_fn, $( $delim )|*, $end, false)
};
($self:expr, $start:pat, $parse_fn:ident, $( $delim:pat )|+, $end:pat) => {
delimited!($self, $start, $parse_fn, $( $delim )|*, $end, true)
};
($self:expr, $start:pat, $parse_fn:ident, $( $delim:pat )|+, $end:pat, $strictness:expr) => {
{
expect!($self, $start);
let mut acc = vec![];
loop {
let peek = $self.token_handler.peek();
match peek.get_kind() {
$end | EOF => break,
_ => (),
}
if !$strictness {
match peek.get_kind() {
$( $delim )|* => { $self.token_handler.next(); continue },
_ => ()
}
}
acc.push($self.$parse_fn()?);
match $self.token_handler.peek().get_kind() {
$( $delim )|* => { $self.token_handler.next(); continue },
_ if $strictness => break,
_ => continue,
};
}
expect!($self, $end);
acc
}
};
}
impl Parser {
/// `program := (statement delimiter)* EOF`
/// `delimiter := NEWLINE | ';'`
#[recursive_descent_method]
fn program(&mut self) -> ParseResult<AST> {
let mut statements = Vec::new();
loop {
match self.token_handler.peek().get_kind() {
EOF => break,
Newline | Semicolon => {
self.token_handler.next();
continue;
},
_ => statements.push(
Meta::new(self.statement()?)
),
}
}
Ok(AST(statements))
}
/// `statement := expression | declaration`
#[recursive_descent_method]
fn statement(&mut self) -> ParseResult<Statement> {
//TODO handle error recovery here
match self.token_handler.peek().get_kind() {
Keyword(Type) => self.type_declaration().map(|decl| { Statement::Declaration(decl) }),
Keyword(Func)=> self.func_declaration().map(|func| { Statement::Declaration(func) }),
Keyword(Let) => self.binding_declaration().map(|decl| Statement::Declaration(decl)),
Keyword(Interface) => self.interface_declaration().map(|decl| Statement::Declaration(decl)),
Keyword(Impl) => self.impl_declaration().map(|decl| Statement::Declaration(decl)),
_ => self.expression().map(|expr| { Statement::ExpressionStatement(expr.into()) } ),
}
}
#[recursive_descent_method]
fn type_declaration(&mut self) -> ParseResult<Declaration> {
expect!(self, Keyword(Type));
self.type_declaration_body()
}
#[recursive_descent_method]
fn type_declaration_body(&mut self) -> ParseResult<Declaration> {
if let Keyword(Alias) = self.token_handler.peek_kind() {
self.type_alias()
} else {
let mutable = if let Keyword(Mut) = self.token_handler.peek_kind() {
self.token_handler.next();
true
} else {
false
};
let name = self.type_singleton_name()?;
expect!(self, Operator(ref c) if **c == "=");
let body = self.type_body()?;
Ok(Declaration::TypeDecl { name, body, mutable})
}
}
#[recursive_descent_method]
fn type_alias(&mut self) -> ParseResult<Declaration> {
expect!(self, Keyword(Alias));
let alias = self.identifier()?;
expect!(self, Operator(ref c) if **c == "=");
let original = self.identifier()?;
Ok(Declaration::TypeAlias(alias, original))
}
#[recursive_descent_method]
fn type_body(&mut self) -> ParseResult<TypeBody> {
let mut variants = Vec::new();
variants.push(self.variant_specifier()?);
loop {
if let Pipe = self.token_handler.peek_kind() {
self.token_handler.next();
variants.push(self.variant_specifier()?);
} else {
break;
}
}
Ok(TypeBody(variants))
}
#[recursive_descent_method]
fn variant_specifier(&mut self) -> ParseResult<Variant> {
use self::Variant::*;
let name = self.identifier()?;
match self.token_handler.peek_kind() {
LParen => {
let tuple_members = delimited!(self, LParen, type_name, Comma, RParen);
Ok(TupleStruct(name, tuple_members))
},
LCurlyBrace => {
let typed_identifier_list = delimited!(self, LCurlyBrace, typed_identifier, Comma, RCurlyBrace);
Ok(Record {name, members: typed_identifier_list })
},
_ => Ok(UnitStruct(name))
}
}
#[recursive_descent_method]
fn typed_identifier(&mut self) -> ParseResult<(Rc<String>, TypeIdentifier)> {
let identifier = self.identifier()?;
expect!(self, Colon);
let type_name = self.type_name()?;
Ok((identifier, type_name))
}
#[recursive_descent_method]
fn func_declaration(&mut self) -> ParseResult<Declaration> {
let signature = self.func_signature()?;
if let LCurlyBrace = self.token_handler.peek_kind() {
let statements = self.nonempty_func_body()?;
Ok(Declaration::FuncDecl(signature, statements))
} else {
Ok(Declaration::FuncSig(signature))
}
}
#[recursive_descent_method]
fn func_signature(&mut self) -> ParseResult<Signature> {
expect!(self, Keyword(Func));
let (name, operator) = match self.token_handler.peek_kind() {
Operator(s) => {
let name = s.clone();
self.token_handler.next();
(name, true)
},
_ => (self.identifier()?, false)
};
let params = self.formal_param_list()?;
let type_anno = match self.token_handler.peek_kind() {
Colon => Some(self.type_anno()?),
_ => None,
};
Ok(Signature { name, operator, params, type_anno })
}
#[recursive_descent_method]
fn nonempty_func_body(&mut self) -> ParseResult<Vec<Meta<Statement>>> {
let statements = delimited!(self, LCurlyBrace, statement, Newline | Semicolon, RCurlyBrace, nonstrict);
Ok(statements.into_iter().map(|s| Meta::new(s)).collect())
}
#[recursive_descent_method]
fn formal_param_list(&mut self) -> ParseResult<Vec<FormalParam>> {
Ok(delimited!(self, LParen, formal_param, Comma, RParen))
}
#[recursive_descent_method]
fn formal_param(&mut self) -> ParseResult<FormalParam> {
let name = self.identifier()?;
let ty = match self.token_handler.peek_kind() {
Colon => Some(self.type_anno()?),
_ => None
};
Ok((name, ty))
}
#[recursive_descent_method]
fn binding_declaration(&mut self) -> ParseResult<Declaration> {
expect!(self, Keyword(Kw::Let));
let constant = match self.token_handler.peek_kind() {
Keyword(Kw::Mut) => {
self.token_handler.next();
false
}
_ => true
};
let name = self.identifier()?;
let type_anno = if let Colon = self.token_handler.peek_kind() {
Some(self.type_anno()?)
} else {
None
};
expect!(self, Operator(ref o) if **o == "=");
let expr = self.expression()?.into();
Ok(Declaration::Binding { name, constant, type_anno, expr })
}
#[recursive_descent_method]
fn interface_declaration(&mut self) -> ParseResult<Declaration> {
expect!(self, Keyword(Interface));
let name = self.identifier()?;
let signatures = self.signature_block()?;
Ok(Declaration::Interface { name, signatures })
}
#[recursive_descent_method]
fn signature_block(&mut self) -> ParseResult<Vec<Signature>> {
Ok(delimited!(self, LCurlyBrace, func_signature, Newline | Semicolon, RCurlyBrace, nonstrict))
}
#[recursive_descent_method]
fn impl_declaration(&mut self) -> ParseResult<Declaration> {
expect!(self, Keyword(Impl));
let first = self.type_singleton_name()?;
let second = if let Keyword(For) = self.token_handler.peek_kind() {
self.token_handler.next();
Some(self.type_name()?)
} else {
None
};
let block = self.decl_block()?;
Ok(match (first, second) {
(interface_name, Some(type_name)) =>
Declaration::Impl { type_name, interface_name: Some(interface_name), block },
(type_singleton_name, None) =>
Declaration::Impl { type_name: TypeIdentifier::Singleton(type_singleton_name), interface_name: None, block }
})
}
#[recursive_descent_method]
fn decl_block(&mut self) -> ParseResult<Vec<Declaration>> {
Ok(delimited!(self, LCurlyBrace, func_declaration, Newline | Semicolon, RCurlyBrace, nonstrict))
}
#[recursive_descent_method]
fn expression(&mut self) -> ParseResult<Expression> {
let mut expr_body = self.precedence_expr(BinOp::min_precedence())?;
let type_anno = match self.token_handler.peek_kind() {
Colon => Some(self.type_anno()?),
_ => None
};
if let Some(_) = expr_body.1 {
return ParseError::new_with_token("Bad parse state encountered", self.token_handler.peek());
}
expr_body.1 = type_anno;
Ok(expr_body)
}
#[recursive_descent_method]
fn type_anno(&mut self) -> ParseResult<TypeIdentifier> {
expect!(self, Colon);
self.type_name()
}
#[recursive_descent_method]
fn type_name(&mut self) -> ParseResult<TypeIdentifier> {
use self::TypeIdentifier::*;
Ok(match self.token_handler.peek_kind() {
LParen => Tuple(delimited!(self, LParen, type_name, Comma, RParen)),
_ => Singleton(self.type_singleton_name()?),
})
}
#[recursive_descent_method]
fn type_singleton_name(&mut self) -> ParseResult<TypeSingletonName> {
Ok(TypeSingletonName {
name: self.identifier()?,
params: match self.token_handler.peek_kind() {
LAngleBracket => delimited!(self, LAngleBracket, type_name, Comma, RAngleBracket),
_ => vec![],
}
})
}
// this implements Pratt parsing, see http://journal.stuffwithstuff.com/2011/03/19/pratt-parsers-expression-parsing-made-easy/
fn precedence_expr(&mut self, precedence: i32) -> ParseResult<Expression> {
let record = ParseRecord {
production_name: "precedence_expr".to_string(),
next_token: format!("{}", self.token_handler.peek().to_string_with_metadata()),
level: self.parse_level,
};
self.parse_level += 1;
self.parse_record.push(record);
let mut lhs = self.prefix_expr()?;
loop {
let new_precedence = match BinOp::get_precedence_from_token(&self.token_handler.peek_kind()) {
Some(p) => p,
None => break,
};
if precedence >= new_precedence {
break;
}
let next_tok = self.token_handler.next();
let operation = match BinOp::from_sigil_token(&next_tok.kind) {
Some(sigil) => sigil,
None => unreachable!()
};
let rhs = self.precedence_expr(new_precedence)?;
lhs = Expression(ExpressionKind::BinExp(operation, bx!(lhs.into()), bx!(rhs.into())), None);
}
self.parse_level -= 1;
Ok(lhs)
}
#[recursive_descent_method]
fn prefix_expr(&mut self) -> ParseResult<Expression> {
match self.token_handler.peek_kind() {
Operator(ref op) if PrefixOp::is_prefix(&*op) => {
let sigil = match self.token_handler.next().kind {
Operator(op) => op,
_ => unreachable!(),
};
let expr = self.primary()?;
Ok(Expression(
ExpressionKind::PrefixExp(PrefixOp::from_sigil(sigil.as_str()), bx!(expr.into())),
None
))
},
_ => self.call_expr()
}
}
#[recursive_descent_method]
fn call_expr(&mut self) -> ParseResult<Expression> {
let mut expr = self.index_expr()?;
while let LParen = self.token_handler.peek_kind() {
let arguments = delimited!(self, LParen, invocation_argument, Comma, RParen);
let arguments = arguments.into_iter().map(|s| Meta::new(s)).collect();
expr = Expression(ExpressionKind::Call { f: bx!(expr.into()), arguments }, None); //TODO none is incorrect
}
Ok(expr)
}
#[recursive_descent_method]
fn invocation_argument(&mut self) -> ParseResult<InvocationArgument> {
Ok(match self.token_handler.peek_kind() {
Underscore => {
self.token_handler.next();
InvocationArgument::Ignored
},
Identifier(s) => {
self.token_handler.next();
match self.token_handler.peek_kind() {
Operator(ref op) if **op == "=" => {
self.token_handler.next();
let expr = self.expression()?;
InvocationArgument::Keyword { name: s.clone(), expr }
},
_ => {
let expr = self.expression()?;
InvocationArgument::Positional(expr)
}
}
},
_ => InvocationArgument::Positional(self.expression()?)
})
}
#[recursive_descent_method]
fn index_expr(&mut self) -> ParseResult<Expression> {
let primary = self.primary()?;
Ok(if let LSquareBracket = self.token_handler.peek_kind() {
let indexers = delimited!(self, LSquareBracket, expression, Comma, RSquareBracket)
.into_iter().map(|ex| ex.into()).collect();
Expression(ExpressionKind::Index {
indexee: bx!(Expression(primary.0, None).into()),
indexers,
}, None)
} else {
primary
})
}
#[recursive_descent_method]
fn primary(&mut self) -> ParseResult<Expression> {
match self.token_handler.peek_kind() {
LCurlyBrace => self.curly_brace_expr(),
Backslash => self.lambda_expr(),
LParen => self.paren_expr(),
LSquareBracket => self.list_expr(),
Keyword(Kw::If) => self.if_expr(),
Keyword(Kw::For) => self.for_expr(),
Keyword(Kw::While) => self.while_expr(),
Identifier(_) => self.identifier_expr(),
_ => self.literal(),
}
}
#[recursive_descent_method]
fn list_expr(&mut self) -> ParseResult<Expression> {
let exprs = delimited!(self, LSquareBracket, expression, Comma, RSquareBracket)
.into_iter().map(|ex| ex.into()).collect();
Ok(Expression(ExpressionKind::ListLiteral(exprs), None))
}
#[recursive_descent_method]
fn curly_brace_expr(&mut self) -> ParseResult<Expression> {
ParseError::new_with_token("Not implemented", self.token_handler.peek())
}
#[recursive_descent_method]
fn lambda_expr(&mut self) -> ParseResult<Expression> {
expect!(self, Backslash);
let params = self.lambda_param_list()?;
let type_anno = match self.token_handler.peek_kind() {
Colon => Some(self.type_anno()?),
_ => None,
};
let body = self.nonempty_func_body()?;
Ok(Expression(ExpressionKind::Lambda { params, type_anno, body }, None)) //TODO need to handle types somehow
}
#[recursive_descent_method]
fn lambda_param_list(&mut self) -> ParseResult<Vec<FormalParam>> {
if let LParen = self.token_handler.peek_kind() {
self.formal_param_list()
} else {
let single_param = self.formal_param()?;
Ok(vec![single_param])
}
}
#[recursive_descent_method]
fn paren_expr(&mut self) -> ParseResult<Expression> {
use self::ExpressionKind::*;
let old_struct_value = self.restrictions.no_struct_literal;
self.restrictions.no_struct_literal = false;
let output = {
let mut inner = delimited!(self, LParen, expression, Comma, RParen);
match inner.len() {
0 => Ok(Expression(TupleLiteral(vec![]), None)),
1 => Ok(inner.pop().unwrap()),
_ => {
let inner: Vec<Meta<Expression>> = inner.into_iter().map(|ex| ex.into()).collect();
Ok(Expression(TupleLiteral(inner), None))
}
}
};
self.restrictions.no_struct_literal = old_struct_value;
output
}
#[recursive_descent_method]
fn identifier_expr(&mut self) -> ParseResult<Expression> {
use self::ExpressionKind::*;
let identifier = self.identifier()?;
Ok(match self.token_handler.peek_kind() {
LCurlyBrace if !self.restrictions.no_struct_literal => {
let fields = self.record_block()?;
Expression(NamedStruct { name: identifier, fields }, None)
},
_ => Expression(Value(identifier), None)
})
}
#[recursive_descent_method]
fn record_block(&mut self) -> ParseResult<Vec<(Rc<String>, Meta<Expression>)>> {
Ok(
delimited!(self, LCurlyBrace, record_entry, Comma, RCurlyBrace)
.into_iter().map(|(s, ex)| (s, ex.into())).collect()
)
}
#[recursive_descent_method]
fn record_entry(&mut self) -> ParseResult<(Rc<String>, Expression)> {
let field_name = self.identifier()?;
expect!(self, Colon);
let value = self.expression()?;
Ok((field_name, value))
}
#[recursive_descent_method]
fn if_expr(&mut self) -> ParseResult<Expression> {
expect!(self, Keyword(Kw::If));
let discriminator = Box::new({
self.restrictions.no_struct_literal = true;
let x = self.discriminator();
self.restrictions.no_struct_literal = false;
x?
});
let body = Box::new(match self.token_handler.peek_kind() {
Keyword(Kw::Then) => self.conditional()?,
Keyword(Kw::Is) => self.simple_pattern_match()? ,
_ => self.guard_block()?
});
Ok(Expression(ExpressionKind::IfExpression { discriminator, body }, None))
}
#[recursive_descent_method]
fn discriminator(&mut self) -> ParseResult<Discriminator> {
let lhs = self.prefix_expr()?;
let ref next = self.token_handler.peek_kind();
Ok(if let Some(op) = BinOp::from_sigil_token(next) {
Discriminator::BinOp(lhs, op)
} else {
Discriminator::Simple(lhs)
})
}
#[recursive_descent_method]
fn conditional(&mut self) -> ParseResult<IfExpressionBody> {
expect!(self, Keyword(Kw::Then));
let then_clause = self.expr_or_block()?;
let else_clause = self.else_clause()?;
Ok(IfExpressionBody::SimpleConditional(then_clause, else_clause))
}
#[recursive_descent_method]
fn simple_pattern_match(&mut self) -> ParseResult<IfExpressionBody> {
expect!(self, Keyword(Kw::Is));
let pat = self.pattern()?;
expect!(self, Keyword(Kw::Then));
let then_clause = self.expr_or_block()?;
let else_clause = self.else_clause()?;
Ok(IfExpressionBody::SimplePatternMatch(pat, then_clause, else_clause))
}
#[recursive_descent_method]
fn else_clause(&mut self) -> ParseResult<Option<Block>> {
Ok(if let Keyword(Kw::Else) = self.token_handler.peek_kind() {
self.token_handler.next();
Some(self.expr_or_block()?)
} else {
None
})
}
#[recursive_descent_method]
fn guard_block(&mut self) -> ParseResult<IfExpressionBody> {
//TODO - delimited! isn't sophisticated enough to do thisa
//let guards = delimited!(self, LCurlyBrace, guard_arm, Comma, RCurlyBrace);
expect!(self, LCurlyBrace);
let mut guards = vec![];
loop {
match self.token_handler.peek_kind() {
RCurlyBrace | EOF => break,
Semicolon | Newline => { self.token_handler.next(); continue},
_ => {
let guard_arm = self.guard_arm()?;
guards.push(guard_arm);
loop {
match self.token_handler.peek_kind() {
Semicolon | Newline => { self.token_handler.next(); continue; },
_ => break,
}
}
if let RCurlyBrace = self.token_handler.peek_kind() {
break;
}
expect!(self, Comma);
}
}
}
expect!(self, RCurlyBrace);
Ok(IfExpressionBody::GuardList(guards))
}
#[recursive_descent_method]
fn guard_arm(&mut self) -> ParseResult<GuardArm> {
let guard = self.guard()?;
expect!(self, Operator(ref c) if **c == "->");
let body = self.expr_or_block()?;
Ok(GuardArm { guard, body })
}
#[recursive_descent_method]
fn guard(&mut self) -> ParseResult<Guard> {
Ok(match self.token_handler.peek_kind() {
Keyword(Kw::Is) => {
self.token_handler.next();
let pat = self.pattern()?;
Guard::Pat(pat)
},
ref tok if BinOp::from_sigil_token(tok).is_some() => {
let op = BinOp::from_sigil_token(&self.token_handler.next().kind).unwrap();
let precedence = op.get_precedence();
let Expression(expr, _) = self.precedence_expr(precedence)?;
Guard::HalfExpr(HalfExpr { op: Some(op), expr })
},
_ => {
//TODO - I think there's a better way to do this involving the precedence of ->
let Expression(expr, _) = self.prefix_expr()?;
Guard::HalfExpr(HalfExpr { op: None, expr })
}
})
}
#[recursive_descent_method]
fn pattern(&mut self) -> ParseResult<Pattern> {
if let LParen = self.token_handler.peek_kind() {
let tuple_pattern_variants = delimited!(self, LParen, pattern, Comma, RParen);
Ok(Pattern::TuplePattern(tuple_pattern_variants))
} else {
self.simple_pattern()
}
}
#[recursive_descent_method]
fn simple_pattern(&mut self) -> ParseResult<Pattern> {
Ok({
let tok = self.token_handler.peek();
match tok.get_kind() {
Identifier(_) => {
let id = self.identifier()?;
match self.token_handler.peek_kind() {
LCurlyBrace => {
let members = delimited!(self, LCurlyBrace, record_pattern_entry, Comma, RCurlyBrace);
Pattern::Record(id, members)
},
LParen => {
let members = delimited!(self, LParen, pattern, Comma, RParen);
Pattern::TupleStruct(id, members)
},
_ => Pattern::Literal(PatternLiteral::VarPattern(id))
}
},
Keyword(Kw::True) => {
self.token_handler.next();
Pattern::Literal(PatternLiteral::BoolPattern(true))
},
Keyword(Kw::False) => {
self.token_handler.next();
Pattern::Literal(PatternLiteral::BoolPattern(false))
},
StrLiteral(s) => {
self.token_handler.next();
Pattern::Literal(PatternLiteral::StringPattern(s))
},
DigitGroup(_) | HexLiteral(_) | BinNumberSigil | Period => self.signed_number_literal()?,
Operator(ref op) if **op == "-" => self.signed_number_literal()?,
Underscore => {
self.token_handler.next();
Pattern::Ignored
},
other => return ParseError::new_with_token(format!("{:?} is not a valid Pattern", other), tok)
}
})
}
#[recursive_descent_method]
fn signed_number_literal(&mut self) -> ParseResult<Pattern> {
let neg = match self.token_handler.peek_kind() {
Operator(ref op) if **op == "-" => {
self.token_handler.next();
true
},
_ => false
};
let Expression(expr_type, _) = self.number_literal()?;
Ok(Pattern::Literal(PatternLiteral::NumPattern { neg, num: expr_type }))
}
#[recursive_descent_method]
fn record_pattern_entry(&mut self) -> ParseResult<(Rc<String>, Pattern)> {
let name = self.identifier()?;
Ok(match self.token_handler.peek_kind() {
Colon => {
expect!(self, Colon);
let pat = self.pattern()?;
(name, pat)
},
_ => (name.clone(), Pattern::Literal(PatternLiteral::StringPattern(name.clone())))
})
}
#[recursive_descent_method]
fn block(&mut self) -> ParseResult<Block> {
let block = delimited!(self, LCurlyBrace, statement, Newline | Semicolon, RCurlyBrace, nonstrict);
Ok(block.into_iter().map(|s| { Meta::new(s) }).collect())
}
#[recursive_descent_method]
fn expr_or_block(&mut self) -> ParseResult<Block> {
match self.token_handler.peek_kind() {
LCurlyBrace => self.block(),
_ => {
let expr = self.expression()?;
Ok(vec![Meta::new(Statement::ExpressionStatement(expr.into()))])
}
}
}
#[recursive_descent_method]
fn while_expr(&mut self) -> ParseResult<Expression> {
use self::ExpressionKind::*;
expect!(self, Keyword(Kw::While));
let condition = {
self.restrictions.no_struct_literal = true;
let x = self.while_cond();
self.restrictions.no_struct_literal = false;
x?.map(|expr| bx!(expr.into()))
};
let body = self.block()?;
Ok(Expression(WhileExpression {condition, body}, None))
}
#[recursive_descent_method]
fn while_cond(&mut self) -> ParseResult<Option<Expression>> {
Ok(match self.token_handler.peek_kind() {
LCurlyBrace => None,
_ => Some(self.expression()?),
})
}
#[recursive_descent_method]
fn for_expr(&mut self) -> ParseResult<Expression> {
expect!(self, Keyword(Kw::For));
let enumerators = if let LCurlyBrace = self.token_handler.peek_kind() {
delimited!(self, LCurlyBrace, enumerator, Comma | Newline, RCurlyBrace)
} else {
let single_enum = {
self.restrictions.no_struct_literal = true;
let s = self.enumerator();
self.restrictions.no_struct_literal = false;
s?
};
vec![single_enum]
};
let body = Box::new(self.for_expr_body()?);
Ok(Expression(ExpressionKind::ForExpression { enumerators, body }, None))
}
#[recursive_descent_method]
fn enumerator(&mut self) -> ParseResult<Enumerator> {
let id = self.identifier()?;
expect!(self, Operator(ref c) if **c == "<-");
let generator = self.expression()?.into();
Ok(Enumerator { id, generator })
}
#[recursive_descent_method]
fn for_expr_body(&mut self) -> ParseResult<ForBody> {
use self::ForBody::*;
let tok = self.token_handler.peek();
Ok(match tok.get_kind() {
LCurlyBrace => {
let statements = delimited!(self, LCurlyBrace, statement, Newline | Semicolon, RCurlyBrace, nonstrict);
StatementBlock(statements.into_iter().map(|s| Meta::new(s)).collect())
},
Keyword(Kw::Return) => {
self.token_handler.next();
MonadicReturn(self.expression()?.into())
},
_ => return ParseError::new_with_token("for expressions must end in a block or 'return'", tok),
})
}
#[recursive_descent_method]
fn identifier(&mut self) -> ParseResult<Rc<String>> {
let tok = self.token_handler.next();
match tok.get_kind() {
Identifier(s) => Ok(s),
p => ParseError::new_with_token(format!("Expected an identifier, got {:?}", p), tok),
}
}
#[recursive_descent_method]
fn literal(&mut self) -> ParseResult<Expression> {
use self::ExpressionKind::*;
let tok = self.token_handler.peek();
match tok.get_kind() {
DigitGroup(_) | HexLiteral(_) | BinNumberSigil | Period => self.number_literal(),
Keyword(Kw::True) => {
self.token_handler.next();
Ok(Expression(BoolLiteral(true), None))
},
Keyword(Kw::False) => {
self.token_handler.next();
Ok(Expression(BoolLiteral(false), None))
},
StrLiteral(s) => {
self.token_handler.next();
Ok(Expression(StringLiteral(s.clone()), None))
}
e => ParseError::new_with_token(format!("Expected a literal expression, got {:?}", e), tok),
}
}
#[recursive_descent_method]
fn number_literal(&mut self) -> ParseResult<Expression> {
match self.token_handler.peek_kind() {
HexLiteral(_) | BinNumberSigil => self.int_literal(),
_ => self.float_literal(),
}
}
#[recursive_descent_method]
fn int_literal(&mut self) -> ParseResult<Expression> {
use self::ExpressionKind::*;
let tok = self.token_handler.next();
match tok.get_kind() {
BinNumberSigil => {
let digits = self.digits()?;
let n = parse_binary(digits, tok)?;
Ok(Expression(NatLiteral(n), None))
},
HexLiteral(text) => {
let digits: String = text.chars().filter(|c| c.is_digit(16)).collect();
let n = parse_hex(digits, tok)?;
Ok(Expression(NatLiteral(n), None))
},
_ => return ParseError::new_with_token("Expected '0x' or '0b'", tok),
}
}
#[recursive_descent_method]
fn float_literal(&mut self) -> ParseResult<Expression> {
use self::ExpressionKind::*;
let tok = self.token_handler.peek();
let mut digits = self.digits()?;
if let Period = self.token_handler.peek_kind() {
self.token_handler.next();
digits.push_str(".");
digits.push_str(&self.digits()?);
match digits.parse::<f64>() {
Ok(f) => Ok(Expression(FloatLiteral(f), None)),
Err(e) => ParseError::new_with_token(format!("Float failed to parse with error: {}", e), tok),
}
} else {
match digits.parse::<u64>() {
Ok(d) => Ok(Expression(NatLiteral(d), None)),
Err(e) => ParseError::new_with_token(format!("Integer failed to parse with error: {}", e), tok),
}
}
}
#[recursive_descent_method]
fn digits(&mut self) -> ParseResult<String> {
let mut ds = String::new();
loop {
match self.token_handler.peek_kind() {
Underscore => { self.token_handler.next(); continue; },
DigitGroup(ref s) => { self.token_handler.next(); ds.push_str(s)},
_ => break,
}
}
Ok(ds)
}
}
fn parse_binary(digits: String, tok: Token) -> ParseResult<u64> {
let mut result: u64 = 0;
let mut multiplier = 1;
for d in digits.chars().rev() {
match d {
'1' => result += multiplier,
'0' => (),
_ => return ParseError::new_with_token("Encountered a character not '1' or '0 while parsing a binary literal", tok),
}
multiplier = match multiplier.checked_mul(2) {
Some(m) => m,
None => return ParseError::new_with_token("This binary expression will overflow", tok)
}
}
Ok(result)
}
fn parse_hex(digits: String, tok: Token) -> ParseResult<u64> {
let mut result: u64 = 0;
let mut multiplier: u64 = 1;
for d in digits.chars().rev() {
match d.to_digit(16) {
Some(n) => result += n as u64 * multiplier,
None => return ParseError::new_with_token("Encountered a non-hex digit in a hex literal", tok),
}
multiplier = match multiplier.checked_mul(16) {
Some(m) => m,
None => return ParseError::new_with_token("This hex expression will overflow", tok)
}
}
Ok(result)
}
#[cfg(test)]
mod parse_tests {
use ::std::rc::Rc;
use super::tokenize;
use super::ParseResult;
use crate::builtin::{PrefixOp, BinOp};
use crate::ast::{AST, Meta, Expression, Statement, IfExpressionBody, Discriminator, Pattern, PatternLiteral, TypeBody, Enumerator, ForBody, InvocationArgument};
use super::Statement::*;
use super::Declaration::*;
use super::Signature;
use super::TypeIdentifier::*;
use super::TypeSingletonName;
use super::ExpressionKind::*;
use super::Variant::*;
use super::ForBody::*;
fn parse(input: &str) -> ParseResult<AST> {
let tokens: Vec<crate::tokenizing::Token> = tokenize(input);
let mut parser = super::Parser::new(tokens);
parser.parse()
}
macro_rules! parse_test {
($string:expr, $correct:expr) => { assert_eq!(parse($string).unwrap(), $correct) };
}
macro_rules! parse_test_wrap_ast {
($string:expr, $correct:expr) => { parse_test!($string, AST(vec![$correct])) }
}
macro_rules! parse_error {
($string:expr) => { assert!(parse($string).is_err()) }
}
macro_rules! val {
($var:expr) => { Value(Rc::new($var.to_string())) }
}
macro_rules! ty {
($name:expr) => { Singleton(tys!($name)) }
}
macro_rules! tys {
($name:expr) => { TypeSingletonName { name: Rc::new($name.to_string()), params: vec![] } };
}
macro_rules! ex {
($expr_type:expr) => { Expression($expr_type, None) };
(m $expr_type:expr) => { Meta::new(Expression($expr_type, None)) };
(m $expr_type:expr, $type_anno:expr) => { Meta::new(Expression($expr_type, Some($type_anno))) };
(s $expr_text:expr) => {
{
let tokens: Vec<crate::tokenizing::Token> = tokenize($expr_text);
let mut parser = super::Parser::new(tokens);
parser.expression().unwrap()
}
};
}
macro_rules! inv {
($expr_type:expr) => { Meta::new(InvocationArgument::Positional($expr_type)) }
}
macro_rules! binexp {
($op:expr, $lhs:expr, $rhs:expr) => { BinExp(BinOp::from_sigil($op), bx!(Expression($lhs, None).into()), bx!(Expression($rhs, None).into())) }
}
macro_rules! prefexp {
($op:expr, $lhs:expr) => { PrefixExp(PrefixOp::from_sigil($op), bx!(Expression($lhs, None).into())) }
}
macro_rules! exst {
($expr_type:expr) => { Meta::new(Statement::ExpressionStatement(Expression($expr_type, None).into())) };
($expr_type:expr, $type_anno:expr) => { Meta::new(Statement::ExpressionStatement(Expression($expr_type, Some($type_anno)).into())) };
($op:expr, $lhs:expr, $rhs:expr) => { Meta::new(Statement::ExpressionStatement(ex!(binexp!($op, $lhs, $rhs)))) };
(s $statement_text:expr) => {
{
let tokens: Vec<crate::tokenizing::Token> = tokenize($statement_text);
let mut parser = super::Parser::new(tokens);
Meta::new(parser.statement().unwrap())
}
}
}
#[test]
fn parsing_number_literals_and_binexps() {
parse_test_wrap_ast! { ".2", exst!(FloatLiteral(0.2)) };
parse_test_wrap_ast! { "8.1", exst!(FloatLiteral(8.1)) };
parse_test_wrap_ast! { "0b010", exst!(NatLiteral(2)) };
parse_test_wrap_ast! { "0b0_1_0_", exst!(NatLiteral(2)) }
parse_test_wrap_ast! {"0xff", exst!(NatLiteral(255)) };
parse_test_wrap_ast! {"0xf_f_", exst!(NatLiteral(255)) };
parse_test_wrap_ast! {"0xf_f_+1", exst!(binexp!("+", NatLiteral(255), NatLiteral(1))) };
parse_test! {"3; 4; 4.3", AST(
vec![exst!(NatLiteral(3)), exst!(NatLiteral(4)),
exst!(FloatLiteral(4.3))])
};
parse_test!("1 + 2 * 3", AST(vec!
[
exst!(binexp!("+", NatLiteral(1), binexp!("*", NatLiteral(2), NatLiteral(3))))
]));
parse_test!("1 * 2 + 3", AST(vec!
[
exst!(binexp!("+", binexp!("*", NatLiteral(1), NatLiteral(2)), NatLiteral(3)))
]));
parse_test!("1 && 2", AST(vec![exst!(binexp!("&&", NatLiteral(1), NatLiteral(2)))]));
parse_test!("1 + 2 * 3 + 4", AST(vec![exst!(
binexp!("+",
binexp!("+", NatLiteral(1), binexp!("*", NatLiteral(2), NatLiteral(3))),
NatLiteral(4)))]));
parse_test!("(1 + 2) * 3", AST(vec!
[exst!(binexp!("*", binexp!("+", NatLiteral(1), NatLiteral(2)), NatLiteral(3)))]));
parse_test!(".1 + .2", AST(vec![exst!(binexp!("+", FloatLiteral(0.1), FloatLiteral(0.2)))]));
parse_test!("1 / 2", AST(vec![exst!(binexp!("/", NatLiteral(1), NatLiteral(2)))]));
}
#[test]
fn parsing_tuples() {
parse_test!("()", AST(vec![exst!(TupleLiteral(vec![]))]));
parse_test!("(\"hella\", 34)", AST(vec![exst!(
TupleLiteral(
vec![ex!(s r#""hella""#).into(), ex!(s "34").into()]
)
)]));
parse_test!("((1+2), \"slough\")", AST(vec![exst!(TupleLiteral(vec![
ex!(binexp!("+", NatLiteral(1), NatLiteral(2))).into(),
ex!(StringLiteral(rc!(slough))).into(),
]))]))
}
#[test]
fn parsing_identifiers() {
parse_test!("a", AST(vec![exst!(val!("a"))]));
parse_test!("some_value", AST(vec![exst!(val!("some_value"))]));
parse_test!("a + b", AST(vec![exst!(binexp!("+", val!("a"), val!("b")))]));
//parse_test!("a[b]", AST(vec![Expression(
//parse_test!("a[]", <- TODO THIS NEEDS TO FAIL
//parse_test("a()[b]()[d]")
//TODO fix this parsing stuff
/*
parse_test! { "perspicacity()[a]", AST(vec![
exst!(Index {
indexee: bx!(ex!(Call { f: bx!(ex!(val!("perspicacity"))), arguments: vec![] })),
indexers: vec![ex!(val!("a"))]
})
])
}
*/
parse_test!("a[b,c]", AST(vec![exst!(Index { indexee: bx!(ex!(m val!("a"))), indexers: vec![ex!(m val!("b")), ex!(m val!("c"))]} )]));
parse_test!("None", AST(vec![exst!(val!("None"))]));
parse_test!("Pandas { a: x + y }", AST(vec![
exst!(NamedStruct { name: rc!(Pandas), fields: vec![(rc!(a), ex!(m binexp!("+", val!("x"), val!("y"))))]})
]));
parse_test! { "Pandas { a: n, b: q, }",
AST(vec![
exst!(NamedStruct { name: rc!(Pandas), fields:
vec![(rc!(a), ex!(m val!("n"))), (rc!(b), ex!(m val!("q")))]
}
)
])
};
}
#[test]
fn parsing_complicated_operators() {
parse_test!("a <- b", AST(vec![exst!(binexp!("<-", val!("a"), val!("b")))]));
parse_test!("a || b", AST(vec![exst!(binexp!("||", val!("a"), val!("b")))]));
parse_test!("a<>b", AST(vec![exst!(binexp!("<>", val!("a"), val!("b")))]));
parse_test!("a.b.c.d", AST(vec![exst!(binexp!(".",
binexp!(".",
binexp!(".", val!("a"), val!("b")),
val!("c")),
val!("d")))]));
parse_test!("-3", AST(vec![exst!(prefexp!("-", NatLiteral(3)))]));
parse_test!("-0.2", AST(vec![exst!(prefexp!("-", FloatLiteral(0.2)))]));
parse_test!("!3", AST(vec![exst!(prefexp!("!", NatLiteral(3)))]));
parse_test!("a <- -b", AST(vec![exst!(binexp!("<-", val!("a"), prefexp!("-", val!("b"))))]));
parse_test!("a <--b", AST(vec![exst!(binexp!("<--", val!("a"), val!("b")))]));
}
#[test]
fn parsing_functions() {
parse_test!("fn oi()", AST(vec![Meta::new(Declaration(FuncSig(Signature { name: rc!(oi), operator: false, params: vec![], type_anno: None })))]));
parse_test!("oi()", AST(vec![exst!(Call { f: bx!(ex!(m val!("oi"))), arguments: vec![] })]));
parse_test!("oi(a, 2 + 2)", AST(vec![exst!(Call
{ f: bx!(ex!(m val!("oi"))),
arguments: vec![inv!(ex!(val!("a"))).into(), inv!(ex!(binexp!("+", NatLiteral(2), NatLiteral(2)))).into()]
})]));
parse_error!("a(b,,c)");
parse_test!("fn a(b, c: Int): Int", AST(vec![Meta::new(Declaration(
FuncSig(Signature { name: rc!(a), operator: false, params: vec![
(rc!(b), None), (rc!(c), Some(ty!("Int")))
], type_anno: Some(ty!("Int")) })))]));
parse_test!("fn a(x) { x() }", AST(vec![Meta::new(Declaration(
FuncDecl(Signature { name: rc!(a), operator: false, params: vec![(rc!(x),None)], type_anno: None },
vec![exst!(Call { f: bx!(ex!(m val!("x"))), arguments: vec![] })])))]));
parse_test!("fn a(x) {\n x() }", AST(vec![Meta::new(Declaration(
FuncDecl(Signature { name: rc!(a), operator: false, params: vec![(rc!(x),None)], type_anno: None },
vec![exst!(Call { f: bx!(ex!(m val!("x"))), arguments: vec![] })])))]));
let multiline = r#"
fn a(x) {
x()
}
"#;
parse_test!(multiline, AST(vec![Meta::new(Declaration(
FuncDecl(Signature { name: rc!(a), operator: false, params: vec![(rc!(x),None)], type_anno: None },
vec![exst!(Call { f: bx!(ex!(m val!("x"))), arguments: vec![] })])))]));
let multiline2 = r#"
fn a(x) {
x()
}
"#;
parse_test!(multiline2, AST(vec![Meta::new(Declaration(
FuncDecl(Signature { name: rc!(a), operator: false, params: vec![(rc!(x),None)], type_anno: None },
vec![exst!(s "x()")])))]));
}
#[test]
fn parsing_bools() {
parse_test!("false", AST(vec![exst!(BoolLiteral(false))]));
parse_test!("true", AST(vec![exst!(BoolLiteral(true))]));
}
#[test]
fn parsing_strings() {
parse_test!(r#""hello""#, AST(vec![exst!(StringLiteral(rc!(hello)))]));
}
#[test]
fn parsing_types() {
parse_test!("type Yolo = Yolo", AST(vec![Meta::new(Declaration(TypeDecl { name: tys!("Yolo"), body: TypeBody(vec![UnitStruct(rc!(Yolo))]), mutable: false} ))]));
parse_test!("type mut Yolo = Yolo", AST(vec![Meta::new(Declaration(TypeDecl { name: tys!("Yolo"), body: TypeBody(vec![UnitStruct(rc!(Yolo))]), mutable: true} ))]));
parse_test!("type alias Sex = Drugs", AST(vec![Meta::new(Declaration(TypeAlias(rc!(Sex), rc!(Drugs))))]));
parse_test!("type Sanchez = Miguel | Alejandro(Int, Option<a>) | Esperanza { a: Int, b: String }",
AST(vec![Meta::new(Declaration(TypeDecl{
name: tys!("Sanchez"),
body: TypeBody(vec![
UnitStruct(rc!(Miguel)),
TupleStruct(rc!(Alejandro), vec![
Singleton(TypeSingletonName { name: rc!(Int), params: vec![] }),
Singleton(TypeSingletonName { name: rc!(Option), params: vec![Singleton(TypeSingletonName { name: rc!(a), params: vec![] })] }),
]),
Record{
name: rc!(Esperanza),
members: vec![
(rc!(a), Singleton(TypeSingletonName { name: rc!(Int), params: vec![] })),
(rc!(b), Singleton(TypeSingletonName { name: rc!(String), params: vec![] })),
]
}
]),
mutable: false
}))]));
parse_test!("type Jorge<a> = Diego | Kike(a)", AST(vec![
Meta::new(Declaration(TypeDecl{
name: TypeSingletonName { name: rc!(Jorge), params: vec![Singleton(TypeSingletonName { name: rc!(a), params: vec![] })] },
body: TypeBody(vec![UnitStruct(rc!(Diego)), TupleStruct(rc!(Kike), vec![Singleton(TypeSingletonName { name: rc!(a), params: vec![] })])]),
mutable: false
}
))]));
}
#[test]
fn parsing_bindings() {
parse_test!("let mut a = 10", AST(vec![Meta::new(Declaration(Binding { name: rc!(a), constant: false, type_anno: None, expr: ex!(m NatLiteral(10)) } ))]));
parse_test!("let a = 2 + 2", AST(vec![Meta::new(Declaration(Binding { name: rc!(a), constant: true, type_anno: None, expr: ex!(m binexp!("+", NatLiteral(2), NatLiteral(2))) }) )]));
parse_test!("let a: Nat = 2 + 2", AST(vec![Meta::new(Declaration(
Binding { name: rc!(a), constant: true, type_anno: Some(Singleton(TypeSingletonName { name: rc!(Nat), params: vec![] })),
expr: Meta::new(ex!(binexp!("+", NatLiteral(2), NatLiteral(2)))) }
))]));
}
#[test]
fn parsing_block_expressions() {
parse_test! {
"if a() then { b(); c() }", AST(vec![exst!(
IfExpression {
discriminator: bx! {
Discriminator::Simple(ex!(Call { f: bx!(ex!(m val!("a"))), arguments: vec![]}))
},
body: bx! {
IfExpressionBody::SimpleConditional(
vec![exst!(Call { f: bx!(ex!(m val!("b"))), arguments: vec![]}), exst!(Call { f: bx!(ex!(m val!("c"))), arguments: vec![] })],
None
)
}
}
)])
};
parse_test! {
"if a() then { b(); c() } else { q }", AST(vec![exst!(
IfExpression {
discriminator: bx! {
Discriminator::Simple(ex!(Call { f: bx!(ex!(m val!("a"))), arguments: vec![]}))
},
body: bx! {
IfExpressionBody::SimpleConditional(
vec![exst!(Call { f: bx!(ex!(m val!("b"))), arguments: vec![]}), exst!(Call { f: bx!(ex!(m val!("c"))), arguments: vec![] })],
Some(
vec![exst!(val!("q"))],
)
)
}
}
)])
};
/*
parse_test!("if a() then { b(); c() }", AST(vec![exst!(
IfExpression(bx!(ex!(Call { f: bx!(ex!(val!("a"))), arguments: vec![]})),
vec![exst!(Call { f: bx!(ex!(val!("b"))), arguments: vec![]}), exst!(Call { f: bx!(ex!(val!("c"))), arguments: vec![] })],
None)
)]));
parse_test!(r#"
if true then {
const a = 10
b
} else {
c
}"#,
AST(vec![exst!(IfExpression(bx!(ex!(BoolLiteral(true))),
vec![Declaration(Binding { name: rc!(a), constant: true, expr: ex!(NatLiteral(10)) }),
exst!(val!(rc!(b)))],
Some(vec![exst!(val!(rc!(c)))])))])
);
parse_test!("if a { b } else { c }", AST(vec![exst!(
IfExpression(bx!(ex!(val!("a"))),
vec![exst!(val!("b"))],
Some(vec![exst!(val!("c"))])))]));
parse_test!("if (A {a: 1}) { b } else { c }", AST(vec![exst!(
IfExpression(bx!(ex!(NamedStruct { name: rc!(A), fields: vec![(rc!(a), ex!(NatLiteral(1)))]})),
vec![exst!(val!("b"))],
Some(vec![exst!(val!("c"))])))]));
parse_error!("if A {a: 1} { b } else { c }");
*/
}
#[test]
fn parsing_interfaces() {
parse_test!("interface Unglueable { fn unglue(a: Glue); fn mar(): Glue }", AST(vec![
Meta::new(Declaration(Interface {
name: rc!(Unglueable),
signatures: vec![
Signature { name: rc!(unglue), operator: false, params: vec![(rc!(a), Some(Singleton(TypeSingletonName { name: rc!(Glue), params: vec![] })))], type_anno: None },
Signature { name: rc!(mar), operator: false, params: vec![], type_anno: Some(Singleton(TypeSingletonName { name: rc!(Glue), params: vec![] })) },
]
}))
]));
}
#[test]
fn parsing_impls() {
parse_test!("impl Heh { fn yolo(); fn swagg(); }", AST(vec![
Meta::new(
Declaration(Impl {
type_name: ty!("Heh"),
interface_name: None,
block: vec![
FuncSig(Signature { name: rc!(yolo), operator: false, params: vec![], type_anno: None }),
FuncSig(Signature { name: rc!(swagg), operator: false, params: vec![], type_anno: None })
] }))]));
parse_test!("impl Mondai for Lollerino { fn yolo(); fn swagg(); }", AST(vec![
Meta::new(Declaration(Impl {
type_name: ty!("Lollerino"),
interface_name: Some(TypeSingletonName { name: rc!(Mondai), params: vec![] }),
block: vec![
FuncSig(Signature { name: rc!(yolo), operator: false, params: vec![], type_anno: None}),
FuncSig(Signature { name: rc!(swagg), operator: false, params: vec![], type_anno: None })
] }))]));
parse_test!("impl Hella<T> for (Alpha, Omega) { }", AST(vec![
Meta::new(Declaration(Impl {
type_name: Tuple(vec![ty!("Alpha"), ty!("Omega")]),
interface_name: Some(TypeSingletonName { name: rc!(Hella), params: vec![ty!("T")] }),
block: vec![]
}))
]));
parse_test!("impl Option<WTFMate> { fn oi() }", AST(vec![
Meta::new(
Declaration(Impl {
type_name: Singleton(TypeSingletonName { name: rc!(Option), params: vec![ty!("WTFMate")]}),
interface_name: None,
block: vec![
FuncSig(Signature { name: rc!(oi), operator: false, params: vec![], type_anno: None }),
]
}))]));
}
#[test]
fn parsing_type_annotations() {
parse_test!("let a = b : Int", AST(vec![
Meta::new(
Declaration(Binding { name: rc!(a), constant: true, type_anno: None, expr:
ex!(m val!("b"), ty!("Int")) }))]));
parse_test!("a : Int", AST(vec![
exst!(val!("a"), ty!("Int"))
]));
parse_test!("a : Option<Int>", AST(vec![
exst!(val!("a"), Singleton(TypeSingletonName { name: rc!(Option), params: vec![ty!("Int")] }))
]));
parse_test!("a : KoreanBBQSpecifier<Kimchi, Option<Bulgogi> >", AST(vec![
exst!(val!("a"), Singleton(TypeSingletonName { name: rc!(KoreanBBQSpecifier), params: vec![
ty!("Kimchi"), Singleton(TypeSingletonName { name: rc!(Option), params: vec![ty!("Bulgogi")] })
] }))
]));
parse_test!("a : (Int, Yolo<a>)", AST(vec![
exst!(val!("a"), Tuple(
vec![ty!("Int"), Singleton(TypeSingletonName {
name: rc!(Yolo), params: vec![ty!("a")]
})]))]));
}
#[test]
fn parsing_lambdas() {
parse_test_wrap_ast! { r#"\(x) { x + 1}"#, exst!(
Lambda { params: vec![(rc!(x), None)], type_anno: None, body: vec![exst!(s "x + 1")] }
)
}
parse_test!(r#"\ (x: Int, y) { a;b;c;}"#, AST(vec![
exst!(Lambda {
params: vec![(rc!(x), Some(ty!("Int"))), (rc!(y), None)],
type_anno: None,
body: vec![exst!(s "a"), exst!(s "b"), exst!(s "c")]
})
]));
parse_test!(r#"\(x){y}(1)"#, AST(vec![
exst!(Call { f: bx!(ex!(m
Lambda {
params: vec![(rc!(x), None)],
type_anno: None,
body: vec![exst!(s "y")] }
)),
arguments: vec![inv!(ex!(NatLiteral(1))).into()] })]));
parse_test_wrap_ast! {
r#"\(x: Int): String { "q" }"#,
exst!(Lambda {
params: vec![(rc!(x), Some(ty!("Int")))],
type_anno: Some(ty!("String")),
body: vec![exst!(s r#""q""#)]
})
}
}
#[test]
fn single_param_lambda() {
parse_test_wrap_ast! {
r"\x { x + 10 }",
exst!(Lambda {
params: vec![(rc!(x), None)],
type_anno: None,
body: vec![exst!(s r"x + 10")]
})
}
parse_test_wrap_ast! {
r"\x: Nat { x + 10 }",
exst!(Lambda {
params: vec![(rc!(x), Some(ty!("Nat")))],
type_anno: None,
body: vec![exst!(s r"x + 10")]
})
}
}
#[test]
fn more_advanced_lambdas() {
parse_test! {
r#"fn wahoo() { let a = 10; \(x) { x + a } };
wahoo()(3) "#, AST(vec![
exst!(s r"fn wahoo() { let a = 10; \(x) { x + a } }"),
exst! {
Call {
f: bx!(ex!(m Call { f: bx!(ex!(m val!("wahoo"))), arguments: vec![] })),
arguments: vec![inv!(ex!(NatLiteral(3))).into()],
}
}
])
}
}
#[test]
fn list_literals() {
parse_test! {
"[1,2]", AST(vec![
exst!(ListLiteral(vec![ex!(m NatLiteral(1)), ex!(m NatLiteral(2))]))])
};
}
#[test]
fn while_expr() {
parse_test! {
"while { }", AST(vec![
exst!(WhileExpression { condition: None, body: vec![] })])
}
parse_test! {
"while a == b { }", AST(vec![
exst!(WhileExpression { condition: Some(bx![ex![m binexp!("==", val!("a"), val!("b"))]]), body: vec![] })])
}
}
#[test]
fn for_expr() {
parse_test! {
"for { a <- maybeValue } return 1", AST(vec![
exst!(ForExpression {
enumerators: vec![Enumerator { id: rc!(a), generator: ex!(m val!("maybeValue")) }],
body: bx!(MonadicReturn(Meta::new(ex!(s "1"))))
})])
}
parse_test! {
"for n <- someRange { f(n); }", AST(vec![
exst!(ForExpression { enumerators: vec![Enumerator { id: rc!(n), generator: ex!(m val!("someRange"))}],
body: bx!(ForBody::StatementBlock(vec![exst!(s "f(n)")]))
})])
}
}
#[test]
fn patterns() {
parse_test_wrap_ast! {
"if x is Some(a) then { 4 } else { 9 }", exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(Pattern::TupleStruct(rc!(Some), vec![Pattern::Literal(PatternLiteral::VarPattern(rc!(a)))]), vec![exst!(s "4")], Some(vec![exst!(s "9")]))) }
)
}
parse_test_wrap_ast! {
"if x is Some(a) then 4 else 9", exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(Pattern::TupleStruct(rc!(Some), vec![Pattern::Literal(PatternLiteral::VarPattern(rc!(a)))]), vec![exst!(s "4")], Some(vec![exst!(s "9")]))) }
)
}
parse_test_wrap_ast! {
"if x is Something { a, b: x } then { 4 } else { 9 }", exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(
Pattern::Record(rc!(Something), vec![
(rc!(a),Pattern::Literal(PatternLiteral::StringPattern(rc!(a)))),
(rc!(b),Pattern::Literal(PatternLiteral::VarPattern(rc!(x))))
]),
vec![exst!(s "4")], Some(vec![exst!(s "9")])))
}
)
}
}
#[test]
fn pattern_literals() {
parse_test_wrap_ast! {
"if x is -1 then 1 else 2",
exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(
Pattern::Literal(PatternLiteral::NumPattern { neg: true, num: NatLiteral(1) }),
vec![exst!(NatLiteral(1))],
Some(vec![exst!(NatLiteral(2))]),
))
}
)
}
parse_test_wrap_ast! {
"if x is 1 then 1 else 2",
exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(
Pattern::Literal(PatternLiteral::NumPattern { neg: false, num: NatLiteral(1) }),
vec![exst!(s "1")],
Some(vec![exst!(s "2")]),
))
}
)
}
parse_test! {
"if x is true then 1 else 2", AST(vec![
exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(
Pattern::Literal(PatternLiteral::BoolPattern(true)),
vec![exst!(NatLiteral(1))],
Some(vec![exst!(NatLiteral(2))]),
))
}
)
])
}
parse_test_wrap_ast! {
"if x is \"gnosticism\" then 1 else 2",
exst!(
IfExpression {
discriminator: bx!(Discriminator::Simple(ex!(s "x"))),
body: bx!(IfExpressionBody::SimplePatternMatch(
Pattern::Literal(PatternLiteral::StringPattern(rc!(gnosticism))),
vec![exst!(s "1")],
Some(vec![exst!(s "2")]),
))
}
)
}
}
}