iowo/crates/lang/src/parser/ast/lossless/parser.rs

470 lines
15 KiB
Rust

use std::borrow::Borrow;
use rowan::{
Checkpoint, GreenNode, GreenNodeBuilder, GreenNodeData, GreenTokenData, Language, NodeOrToken,
};
use crate::parser::{
ast::lossless::{lex::SyntaxKind::*, Lang},
Span,
};
use super::lex::{self, SyntaxKind};
pub mod parser_to_events {
use chumsky::Parser;
use crate::parser::ast::lossless::lex::SyntaxKind::{self, *};
#[derive(Debug, PartialEq, Eq)]
pub(super) enum Event {
StartNode,
StartErr(SyntaxError),
EatToken,
FinishNode,
FinishErr,
}
#[derive(Debug, PartialEq, Eq)]
enum SyntaxError {
Expected(SyntaxKind),
AttrExpectedValue,
/// guessed if there's a newline and attr on next line without comma
/// should then suggest comma after attr
ExpectedCommaBetweenAttrs,
}
pub fn to_events(src: &str) -> Vec<Event> {
let mut tokens = lex::lex(src);
parser().parse(tokens)
}
pub fn parser() -> impl Parser<'static, SyntaxKind, Vec<Event>> {
let whitespace = just(WHITESPACE).or(NEWLINE).repeated().collect::<Vec<_>>();
whitespace
}
}
#[derive(PartialEq, Eq)]
pub struct Parse {
pub green_node: GreenNode,
}
impl std::fmt::Debug for Parse {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
debug_print_green_node(NodeOrToken::Node(self.green_node.borrow()), f, 0)
}
}
fn debug_print_green_node(
node: NodeOrToken<&GreenNodeData, &GreenTokenData>,
f: &mut std::fmt::Formatter<'_>,
lvl: i32,
) -> std::fmt::Result {
for _ in 0..lvl {
f.write_str(" ")?;
}
match node {
NodeOrToken::Node(n) => {
writeln!(f, "{:?} {{", Lang::kind_from_raw(node.kind()));
for c in n.children() {
debug_print_green_node(c, f, lvl + 1)?;
}
for _ in 0..lvl {
f.write_str(" ")?;
}
f.write_str("}\n")
}
NodeOrToken::Token(t) => {
writeln!(f, "{:?} {:?};", Lang::kind_from_raw(t.kind()), t.text())
}
}
}
#[derive(Debug)]
struct Parser<'src> {
tokens: Vec<(SyntaxKind, &'src str)>,
builder: GreenNodeBuilder<'src>,
errors: Vec<SyntaxError>,
}
#[derive(Debug, PartialEq, Eq)]
enum SyntaxError {
Expected(SyntaxKind),
AttrExpectedValue,
/// guessed if there's a newline and attr on next line without comma
/// should then suggest comma after attr
ExpectedCommaBetweenAttrs,
}
pub fn parse(src: &str) -> Parse {
let mut tokens = lex::lex(src);
Parser {
tokens,
builder: GreenNodeBuilder::new(),
errors: Vec::new(),
}
.parse()
}
impl Parser<'_> {
fn parse(mut self) -> Parse {
self.start_node(ROOT);
match self.expr(None) {
expr::ExprRes::Ok => (),
expr::ExprRes::Eof => (),
expr::ExprRes::NoExpr => todo!(),
}
self.builder.finish_node();
Parse {
green_node: self.builder.finish(),
}
}
fn start_node(&mut self, kind: SyntaxKind) {
self.builder.start_node(kind.into());
}
fn finish_node(&mut self) {
self.builder.finish_node();
}
/// Advance one token, adding it to the current branch of the tree builder.
fn bump(&mut self) {
let (kind, text) = self.tokens.pop().unwrap();
self.builder.token(kind.into(), text);
}
fn syntax_err(&mut self, err: SyntaxError) {
let (_, text) = self.tokens.pop().unwrap();
self.builder.token(PARSE_ERR.into(), text);
self.errors.push(err);
}
fn syntax_err_by_checkpoint(&mut self, checkpoint: Checkpoint, err: SyntaxError) {
self.builder.start_node_at(checkpoint, PARSE_ERR.into());
self.finish_node();
self.errors.push(err);
}
fn expected(&mut self, expected: SyntaxKind) {
self.syntax_err(SyntaxError::Expected(expected))
}
/// Peek at the first unprocessed token
fn current(&self) -> Option<SyntaxKind> {
self.tokens.last().map(|(kind, _)| *kind)
}
fn next(&self) -> Option<SyntaxKind> {
self.tokens
.get(self.tokens.len() - 2)
.map(|(kind, _)| *kind)
}
fn skip_ws(&mut self) {
while self.current() == Some(WHITESPACE) || self.current() == Some(NEWLINE) {
self.bump()
}
}
fn skip_ws_without_newlines(&mut self) {
while self.current() == Some(WHITESPACE) {
self.bump()
}
}
}
mod expr {
use rowan::Checkpoint;
use super::{attrset::AttrsetRes, instr::NodeRes, Parser};
use crate::parser::{ast::lossless::lex::SyntaxKind::*, Span};
impl Parser<'_> {
pub(super) fn expr(&mut self, start: Option<Checkpoint>) -> ExprRes {
self.skip_ws();
let start = start.unwrap_or_else(|| self.builder.checkpoint());
match self.current() {
Some(IDENT) => {
let expr_res = match self.instr() {
NodeRes::Ok => ExprRes::Ok,
NodeRes::Eof => ExprRes::Eof,
};
self.builder.start_node_at(start, EXPR.into());
self.finish_node();
expr_res
}
Some(_) => self.atom(Some(start)),
None => ExprRes::Eof,
}
}
pub(super) fn atom(&mut self, start: Option<Checkpoint>) -> ExprRes {
self.skip_ws();
let start = start.unwrap_or_else(|| self.builder.checkpoint());
match self.current() {
Some(INT_NUM | FLOAT_NUM | STRING) => {
self.bump();
self.builder.start_node_at(start, EXPR.into());
self.finish_node();
ExprRes::Ok
}
Some(L_CURLY) => match self.attrset(start) {
AttrsetRes::Ok => ExprRes::Ok,
AttrsetRes::Eof => ExprRes::Eof,
},
Some(L_PAREN) => {
self.builder.start_node_at(start, PARENTHESIZED_EXPR.into());
self.bump();
self.expr(None);
self.skip_ws();
match self.current() {
Some(R_PAREN) => ExprRes::Ok,
Some(_) => todo!(),
None => ExprRes::Eof,
}
}
Some(_) => ExprRes::NoExpr,
None => ExprRes::Eof,
}
}
}
pub enum ExprRes {
Ok,
Eof,
/// isnt an expression
NoExpr,
}
}
mod attrset {
use chumsky::container::Container;
use rowan::Checkpoint;
use super::{expr::ExprRes, instr::NodeRes, Parser};
use crate::parser::{
ast::lossless::{lex::SyntaxKind::*, parser::SyntaxError},
Span,
};
impl Parser<'_> {
pub(super) fn attrset(&mut self, checkpoint: Checkpoint) -> AttrsetRes {
assert_eq!(self.current(), Some(L_CURLY));
self.bump();
self.skip_ws();
match self.current() {
Some(R_CURLY) => {
self.builder.start_node_at(checkpoint, ATTR_SET.into());
self.bump();
self.finish_node();
AttrsetRes::Ok
}
Some(_) => {
self.builder.start_node_at(checkpoint, ATTR_SET.into());
let res = match self.attrs() {
AttrRes::Eof => AttrsetRes::Eof,
AttrRes::RCurly | AttrRes::Ok => {
println!("curr: {:?}", self.current());
AttrsetRes::Ok
}
};
self.finish_node();
res
}
None => AttrsetRes::Eof,
}
// self.start_node(ATTR);
}
fn attrs(&mut self) -> AttrRes {
let mut res = AttrRes::Ok;
while res == AttrRes::Ok {
println!("it: {:?}", self.tokens.last());
match self.attr() {
AttrRes::Ok => {
self.skip_ws_without_newlines();
println!(
"a: {:?}, {:?}",
self.tokens.last(),
self.tokens.get(self.tokens.len() - 2)
);
println!("errs: {:?}", self.errors);
res = AttrRes::Ok;
let checkpoint_previous_end = self.builder.checkpoint();
res = match self.current() {
Some(COMMA) => {
self.bump();
AttrRes::Ok
}
Some(R_CURLY) => {
self.bump();
res = AttrRes::Ok;
break;
}
Some(NEWLINE) => {
self.skip_ws();
println!(
"b: {:?}, {:?}",
self.tokens.last(),
self.tokens.get(self.tokens.len() - 2)
);
match self.current() {
Some(COMMA) => {
self.bump();
AttrRes::Ok
}
Some(R_CURLY) => {
self.bump();
res = AttrRes::Ok;
break;
}
Some(IDENT) => {
println!("wtf");
self.syntax_err_by_checkpoint(
checkpoint_previous_end,
SyntaxError::ExpectedCommaBetweenAttrs,
);
// self.syntax_err(SyntaxError::ExpectedCommaBetweenAttrs);
AttrRes::Ok
}
Some(_) => {
self.bump();
AttrRes::Ok
}
None => {
res = AttrRes::Eof;
break;
}
}
}
Some(_) => {
self.bump();
println!(
"c: {:?}, {:?}",
self.tokens.last(),
self.tokens.get(self.tokens.len() - 2)
);
AttrRes::Ok
}
None => {
res = AttrRes::Eof;
break;
}
}
}
AttrRes::Eof => {
res = AttrRes::Eof;
break;
}
AttrRes::RCurly => {
res = AttrRes::RCurly;
break;
}
}
}
println!("toks_left: {:?}", self.tokens);
res
}
fn attr(&mut self) -> AttrRes {
self.skip_ws();
self.start_node(ATTR);
self.start_node(ATTR_NAME);
match self.current() {
Some(IDENT) => self.bump(),
Some(R_CURLY) => return AttrRes::Ok,
Some(_) => self.expected(IDENT),
None => return AttrRes::Eof,
}
self.finish_node();
self.skip_ws();
match self.current() {
Some(COLON) => self.bump(),
Some(R_CURLY) => {
self.expected(COLON);
return AttrRes::RCurly;
}
Some(_) => self.expected(COLON),
None => return AttrRes::Eof,
}
self.skip_ws();
self.start_node(ATTR_VALUE);
match self.expr(None) {
ExprRes::Ok => self.bump(),
ExprRes::Eof => return AttrRes::Eof,
ExprRes::NoExpr => match self.current() {
Some(COMMA) => self.syntax_err(SyntaxError::AttrExpectedValue),
Some(R_CURLY) => {
self.syntax_err(SyntaxError::AttrExpectedValue);
return AttrRes::RCurly;
}
Some(_) => self.expected(EXPR),
None => unreachable!(),
},
}
self.finish_node();
self.finish_node();
AttrRes::Ok
}
}
#[derive(PartialEq, Eq)]
pub enum AttrsetRes {
Ok,
Eof,
}
#[derive(PartialEq, Eq)]
enum AttrRes {
Ok,
Eof,
RCurly,
}
}
mod instr {
use super::Parser;
use crate::parser::{
ast::lossless::{lex::SyntaxKind::*, parser::expr::ExprRes},
Span,
};
impl Parser<'_> {
pub(super) fn instr(&mut self) -> NodeRes {
assert_eq!(self.current(), Some(IDENT));
self.skip_ws();
self.start_node(INSTR);
self.instr_name();
// used to count positionals
let mut i = 0;
let params_checkpoint = self.builder.checkpoint();
loop {
match self.expr(None) {
ExprRes::Ok => {
i += 1;
continue;
}
ExprRes::NoExpr | ExprRes::Eof => break,
}
}
if i >= 1 {
self.builder
.start_node_at(params_checkpoint, INSTR_PARAMS.into());
self.finish_node();
}
self.finish_node();
NodeRes::Ok
}
fn instr_name(&mut self) {
self.start_node(INSTR_NAME);
while self.current() == Some(IDENT) {
self.bump();
self.skip_ws_without_newlines();
}
self.finish_node();
}
}
pub(super) enum NodeRes {
Ok,
Eof,
}
}