diff options
Diffstat (limited to 'src/parse.rs')
-rw-r--r-- | src/parse.rs | 322 |
1 files changed, 259 insertions, 63 deletions
diff --git a/src/parse.rs b/src/parse.rs index 9a60863..1dabd47 100644 --- a/src/parse.rs +++ b/src/parse.rs @@ -1,101 +1,297 @@ -use multipeek::*; +use std::fmt; + use crate::lex::*; -use crate::ast::Expr; +use crate::ast::*; use crate::ast::Binding::*; use crate::ast::Control::*; use crate::ast::Pattern::*; +use Token::*; +use Literal::*; +use Punctuation::*; + +type Input = std::iter::Peekable<std::vec::IntoIter<Token>>; + +#[derive(Clone, Copy)] +struct State { + depth: usize, + step: usize +} + +impl State { + fn indent(&self) -> State { + State { depth: self.depth + self.step, step: self.step } + } + fn dedent(&self) -> State { + State { depth: self.depth - self.step, step: self.step } + } +} /// Convert a basic TokenStream into an AbstractSyntaxTree pub fn astify(input: TokenStream, name: &str) -> Result<Expr> { - use Token::*; - use Literal::*; - use Punctuation::*; + let mut input = input.into_iter().peekable(); + let body = parse_body(&mut input, State { depth: 0, step: 0 })?; + Ok(Expr::Binding(Module{ id: name.to_string(), body })) +} - let mut input = multipeek(input); +/// Parse a series of Exprs, for ex. the body of a function. +fn parse_body(input: &mut Input, state: State) -> Result<Vec<Expr>> { let mut res = Vec::new(); - while let Some(x) = input.peek() { - res.push(parse(&mut input, 0)?); + while input.peek() == Some(&Indent(state.depth)) { + input.next(); + res.push(parse_expr(input, state)?); } - Ok(Expr::Binding(Module{ id: name.to_string(), body: res })) + Ok(res) } -fn parse(input: &mut MultiPeek<std::vec::IntoIter<Token>>, depth: usize) -> Result<Expr> { - use Token::*; - use Literal::*; - use Punctuation::*; - let mut input = input; - match input.peek() { - Some(Word(val)) => match val.as_str() { - "pub" => { - input.next(); - if let Some(Word(val)) = input.peek() { - match val.as_str() { - "const" => parse_const(&mut input, true), - "func" => parse_func(&mut input, true), - "type" => parse_type(&mut input, true), - "mod" => parse_mod(&mut input, true), +/// Expr ::= Let | Var | Const | Func | Type | +/// Mod | Import | Block | Static | +/// For | While | Loop | If | When | Try | Match +fn parse_expr(input: &mut Input, state: State) -> Result<Expr> { + use Keyword::*; + match input.next() { + Some(Key(word)) => match word { + Pub => { + match input.next() { + Some(Key(word)) => match word { + Const => parse_const(input, state, true), + Func => parse_funcdecl(input, state, true), + Type => parse_typedecl(input, state, true), + Mod => parse_mod(input, state, true), _ => return Err("unrecognized keyword following pub".into()), } - } else { - return Err("unrecognized thing following pub".into()); + Some(_) => return Err("unrecognized thing following pub".into()), + None => return Err("end of input".into()), } }, - "let" => parse_let(&mut input), - "var" => parse_var(&mut input), - "const" => parse_const(&mut input, false), - "func" => parse_func(&mut input, false), - "type" => parse_type(&mut input, false), - "mod" => parse_mod(&mut input, false), - "from" | "import" => parse_import(&mut input), - "block" => parse_block(&mut input), - "static" => parse_static(&mut input), - "for" => parse_for(&mut input), - "while" => parse_while(&mut input), - "loop" => parse_loop(&mut input), - "if" => parse_if(&mut input), - "when" => parse_when(&mut input), - "try" => parse_try(&mut input), - "match" => parse_match(&mut input), - _ => parse_line(&mut input), + Let => parse_let(input, state), + Var => parse_var(input, state), + Const => parse_const(input, state, false), + Func => parse_funcdecl(input, state, false), + Type => parse_typedecl(input, state, false), + Mod => parse_mod(input, state, false), + From => parse_import(input, state, true), + Import => parse_import(input, state, false), + Block => parse_block(input, state), + Static => parse_static(input, state), + For => parse_for(input, state), + While => parse_while(input, state), + Loop => parse_loop(input, state), + If => parse_if(input, state), + When => parse_when(input, state), + Try => parse_try(input, state), + Match => parse_match(input, state), + _ => return Err("invalid keyword starting expression".into()), }, - _ => parse_line(&mut input), + _ => todo!(), // what can i do with this?? match line here } } +/// Let ::= 'let' Pattern Annotation? '=' Expr +fn parse_let(input: &mut Input, state: State) -> Result<Expr> { + let id = parse_pattern(input, state)?; + let mut kind = None; + if let Some(Sep(Colon)) = input.peek() { + input.next(); + kind = Some(parse_typedesc(input, state)?); + } + if input.next() != Some(Sep(Equals)) { + return Err("= not following binding".into()) + } + let value = Box::new(parse_expr(input, state)?); + Ok(Expr::Binding(Let { id, kind, value })) +} +/// Var ::= 'var' Pattern Annotation? ('=' Expr)? +fn parse_var(input: &mut Input, state: State) -> Result<Expr> { + let id = parse_pattern(input, state)?; + let mut kind = None; + if let Some(Sep(Colon)) = input.peek() { + input.next(); + kind = Some(parse_typedesc(input, state)?); + } + let mut value = None; + if input.next() != Some(Sep(Equals)) { + value = Some(Box::new(parse_expr(input, state)?)); + } + Ok(Expr::Binding(Var { id, kind, value })) +} // Const ::= 'pub'? 'const' Pattern Annotation? '=' Expr -fn parse_const(input: &mut MultiPeek<std::vec::IntoIter<Token>>, public: bool) -> Result<Expr> { todo!() } +fn parse_const(input: &mut Input, state: State, public: bool) -> Result<Expr> { + let id = parse_pattern(input, state)?; + let mut kind = None; + if let Some(Sep(Colon)) = input.peek() { + input.next(); + kind = Some(parse_typedesc(input, state)?); + } + if input.next() != Some(Sep(Equals)) { + return Err("= not following binding".into()) + } + let value = Box::new(parse_expr(input, state)?); + Ok(Expr::Binding(Const { public, id, kind, value })) +} // Func ::= 'pub'? ('func' | 'proc') Ident Generics? Parameters? (':' TypeDesc) '=' Body -fn parse_func(input: &mut MultiPeek<std::vec::IntoIter<Token>>, public: bool) -> Result<Expr> { todo!() } +fn parse_funcdecl(input: &mut Input, state: State, public: bool) -> Result<Expr> { todo!() } // TypeDecl ::= 'pub'? 'type' Pattern Generics? '=' 'distinct'? 'ref'? TypeDesc -fn parse_type(input: &mut MultiPeek<std::vec::IntoIter<Token>>, public: bool) -> Result<Expr> { todo!() } +fn parse_typedecl(input: &mut Input, state: State, public: bool) -> Result<Expr> { + let pattern = parse_pattern(input, state)?; + todo!() +} // Mod ::= 'pub'? 'mod' Ident ':' Body -fn parse_mod(input: &mut MultiPeek<std::vec::IntoIter<Token>>, public: bool) -> Result<Expr> { todo!() } +fn parse_mod(input: &mut Input, state: State, public: bool) -> Result<Expr> { + match input.next() { + Some(Word(id)) => { + match input.next() { + Some(Sep(Colon)) => { + let body = parse_body(input, state.indent())?; + Ok(Expr::Binding(Module { id, body })) + }, + _ => return Err("unexpected token following mod label".into()), + } + }, + _ => return Err("unexpected thing following mod keyword".into()), + } +} -// Let ::= 'let' Pattern Annotation? '=' Expr -fn parse_let(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } -// Var ::= 'var' Pattern Annotation? ('=' Expr)? -fn parse_var(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } // Import ::= ('from' Ident)? 'import' Ident (',' Ident)* ('as' Ident)? -fn parse_import(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_import(input: &mut Input, state: State, from_scope: bool) -> Result<Expr> { + let mut from = None; + if from_scope { + match input.next() { + Some(Word(id)) => from = Some(id), + _ => return Err("identifier not following from keyword".into()) + } + if input.next() != Some(Key(Keyword::Import)) { + return Err("expected import to follow from".into()) + } + } + todo!() +} // Block ::= 'block' Ident? ':' Body -fn parse_block(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_block(input: &mut Input, state: State) -> Result<Expr> { // todo: body + offset + match input.next() { + Some(Sep(Colon)) => { + let id = None; + let body = parse_body(input, state.indent())?; + Ok(Expr::Control(Block { id, body })) + }, + Some(Word(label)) => { + match input.next() { + Some(Sep(Colon)) => { + let id = Some(label); + let body = parse_body(input, state.indent())?; + Ok(Expr::Control(Block { id, body })) + }, + _ => return Err("unexpected token following block label".into()), + } + }, + _ => return Err("unexpected thing following block keyword".into()), + } +} // Static ::= 'static' ':' Body -fn parse_static(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_static(input: &mut Input, state: State) -> Result<Expr> { + if input.next() != Some(Sep(Colon)) { + return Err("colon must follow static invocation".into()); + } + let body = parse_body(input, state.indent())?; + Ok(Expr::Control(Static { body })) +} + // For ::= 'for' Pattern 'in' Expr ':' Body -fn parse_for(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_for(input: &mut Input, state: State) -> Result<Expr> { + let binding = parse_pattern(input, state)?; + if input.next() != Some(Key(Keyword::In)) { + return Err("expected in keyword after for pattern".into()); + } + let range = Box::new(parse_expr(input, state)?); + if input.next() != Some(Sep(Colon)) { + return Err("expected colon after in expression".into()); + } + let body = parse_body(input, state.indent())?; + Ok(Expr::Control(For { binding, range, body })) +} // While ::= 'while' Expr ':' Body -fn parse_while(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_while(input: &mut Input, state: State) -> Result<Expr> { + let cond = Box::new(parse_expr(input, state)?); + if input.next() != Some(Sep(Colon)) { + return Err("expected colon after while keyword".into()); + } + let body = parse_body(input, state.indent())?; + Ok(Expr::Control(While { cond, body })) +} // Loop ::= 'loop' ':' Body -fn parse_loop(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_loop(input: &mut Input, state: State) -> Result<Expr> { + if input.next() != Some(Sep(Colon)) { + return Err("expected colon after loop keyword".into()); + } + let body = parse_body(input, state.indent())?; + Ok(Expr::Control(Loop { body })) +} + // If ::= 'if' Expr ':' Body ('elif' Expr ':' Body)* ('else' ':' Body)? -fn parse_if(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_if(input: &mut Input, state: State) -> Result<Expr> { + let mut branches = Vec::new(); + branches.push(parse_cond_branch(input, state)?); + while input.peek() == Some(&Key(Keyword::Elif)) { + input.next(); + branches.push(parse_cond_branch(input, state)?); + } + let mut else_body = None; + if input.peek() == Some(&Key(Keyword::Else)) { + input.next(); + else_body = Some(parse_body(input, state.indent())?); + } + Ok(Expr::Control(If { branches, else_body })) +} // When ::= 'when' Expr ':' Body ('elif' Expr ':' Body)* ('else' ':' Body)? -fn parse_when(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_when(input: &mut Input, state: State) -> Result<Expr> { + let mut branches = Vec::new(); + branches.push(parse_cond_branch(input, state)?); + while input.peek() == Some(&Key(Keyword::Elif)) { + input.next(); + branches.push(parse_cond_branch(input, state)?); + } + let mut else_body = None; + if input.peek() == Some(&Key(Keyword::Else)) { + input.next(); + else_body = Some(parse_body(input, state.indent())?); + } + let mut body = Vec::new(); + body.push(Expr::Control(If { branches, else_body })); + Ok(Expr::Control(Static { body })) +} // Try ::= 'try' ':' Body ('except' Ident (',' Ident)* ':' Body) ('finally' ':' Body)? -fn parse_try(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_try(input: &mut Input, state: State) -> Result<Expr> { + if input.next() != Some(Sep(Colon)) { + return Err("expected colon after try keyword".into()); + } + let body = parse_body(input, state.indent())?; + let catches = Vec::new(); + while input.peek() == Some(&Key(Keyword::Catch)) { + input.next(); + todo!(); + } + let mut finally = None; + if input.peek() == Some(&Key(Keyword::Finally)) { + input.next(); + if input.next() != Some(Sep(Colon)) { + return Err("expected colon after try keyword".into()); + } + finally = Some(parse_body(input, state.indent())?); + } + Ok(Expr::Control(Try { body, catches, finally })) +} // Match ::= 'match' Expr ('of' Pattern (',' Pattern)* ('where' Expr)? ':' Body)+ -fn parse_match(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_match(input: &mut Input, state: State) -> Result<Expr> { + let item = parse_pattern(input, state)?; + let mut branches = Vec::new(); + while input.peek() == Some(&Key(Keyword::Of)) { + input.next(); + todo!(); + } + Ok(Expr::Control(Match { item, branches })) +} -fn parse_line(input: &mut MultiPeek<std::vec::IntoIter<Token>>) -> Result<Expr> { todo!() } +fn parse_typedesc(input: &mut Input, state: State) -> Result<Type> { todo!() } +fn parse_pattern(input: &mut Input, state: State) -> Result<Pattern> { todo!() } +fn parse_cond_branch(input: &mut Input, state: State) -> Result<CondBranch> { todo!() } // lex, parse, expand, compile? |