summaryrefslogtreecommitdiff
path: root/src/parse
diff options
context:
space:
mode:
authorLaurenz <laurmaedje@gmail.com>2021-02-17 23:07:28 +0100
committerLaurenz <laurmaedje@gmail.com>2021-02-18 15:06:00 +0100
commit84cdc85ca7494368e7ce2039fcef06ac2d3bd2ed (patch)
tree4820cfbd853fd3e98953f3a23aa398f29761ab53 /src/parse
parente143fd36efaabfcfb42823fe5f7a25dd6de4a960 (diff)
Refresh parser 🌊
Diffstat (limited to 'src/parse')
-rw-r--r--src/parse/collection.rs2
-rw-r--r--src/parse/mod.rs231
-rw-r--r--src/parse/parser.rs132
3 files changed, 171 insertions, 194 deletions
diff --git a/src/parse/collection.rs b/src/parse/collection.rs
index 7ffc4539..ab358f76 100644
--- a/src/parse/collection.rs
+++ b/src/parse/collection.rs
@@ -4,7 +4,7 @@ use super::*;
pub fn args(p: &mut Parser) -> ExprArgs {
let start = p.start();
let items = collection(p, vec![]);
- ExprArgs { span: p.span_from(start), items }
+ ExprArgs { span: p.span(start), items }
}
/// Parse a parenthesized group, which can be either of:
diff --git a/src/parse/mod.rs b/src/parse/mod.rs
index e9cf2a60..1d3b8be7 100644
--- a/src/parse/mod.rs
+++ b/src/parse/mod.rs
@@ -70,8 +70,8 @@ fn node(p: &mut Parser, at_start: &mut bool) -> Option<Node> {
Token::Raw(t) => raw(p, t),
Token::UnicodeEscape(t) => Node::Text(unicode_escape(p, t)),
- // Keywords.
- Token::Let | Token::If | Token::For => {
+ // Hashtag + keyword / identifier.
+ Token::Ident(_) | Token::Let | Token::If | Token::For => {
*at_start = false;
let stmt = token == Token::Let;
let group = if stmt { Group::Stmt } else { Group::Expr };
@@ -100,12 +100,6 @@ fn node(p: &mut Parser, at_start: &mut bool) -> Option<Node> {
return Some(Node::Expr(template(p)));
}
- // Bracket function.
- Token::HashBracket => {
- *at_start = false;
- return Some(Node::Expr(bracket_call(p)?));
- }
-
// Comments.
Token::LineComment(_) | Token::BlockComment(_) => {
p.eat();
@@ -125,7 +119,7 @@ fn node(p: &mut Parser, at_start: &mut bool) -> Option<Node> {
/// Parse a heading.
fn heading(p: &mut Parser) -> Node {
let start = p.start();
- p.assert(&[Token::Eq]);
+ p.assert(Token::Eq);
// Count depth.
let mut level: usize = 0;
@@ -174,108 +168,6 @@ fn unicode_escape(p: &mut Parser, token: TokenUnicodeEscape) -> String {
text
}
-/// Parse a bracketed function call.
-fn bracket_call(p: &mut Parser) -> Option<Expr> {
- p.start_group(Group::Bracket, TokenMode::Code);
-
- // One header is guaranteed, but there may be more (through chaining).
- let mut outer = vec![];
- let mut inner = bracket_subheader(p);
- while p.eat_if(Token::Pipe) {
- if let Some(new) = bracket_subheader(p) {
- outer.extend(inner);
- inner = Some(new);
- }
- }
-
- p.end_group();
-
- let body = match p.peek() {
- Some(Token::LeftBracket) => Some(bracket_body(p)),
- _ => None,
- };
-
- let mut inner = inner?;
- if let Some(body) = body {
- inner.span.expand(body.span());
- inner.args.items.push(ExprArg::Pos(body));
- }
-
- while let Some(mut top) = outer.pop() {
- top.args.items.push(ExprArg::Pos(Expr::Call(inner)));
- inner = top;
- }
-
- Some(Expr::Call(inner))
-}
-
-/// Parse one subheader of a bracketed function call.
-fn bracket_subheader(p: &mut Parser) -> Option<ExprCall> {
- p.start_group(Group::Subheader, TokenMode::Code);
- let name = ident(p);
- let args = args(p);
- let span = p.end_group();
- Some(ExprCall {
- span,
- callee: Box::new(Expr::Ident(name?)),
- args,
- })
-}
-
-/// Parse the body of a bracketed function call.
-fn bracket_body(p: &mut Parser) -> Expr {
- p.start_group(Group::Bracket, TokenMode::Markup);
- let tree = Rc::new(tree(p));
- let span = p.end_group();
- Expr::Template(ExprTemplate { span, tree })
-}
-
-/// Parse an expression.
-fn expr(p: &mut Parser) -> Option<Expr> {
- expr_with(p, 0)
-}
-
-/// Parse an expression with operators having at least the minimum precedence.
-fn expr_with(p: &mut Parser, min_prec: usize) -> Option<Expr> {
- let start = p.start();
- let mut lhs = match p.eat_map(UnOp::from_token) {
- Some(op) => {
- let prec = op.precedence();
- let expr = Box::new(expr_with(p, prec)?);
- Expr::Unary(ExprUnary { span: p.span_from(start), op, expr })
- }
- None => primary(p)?,
- };
-
- loop {
- let op = match p.peek().and_then(BinOp::from_token) {
- Some(binop) => binop,
- None => break,
- };
-
- let mut prec = op.precedence();
- if prec < min_prec {
- break;
- }
-
- p.eat();
- match op.associativity() {
- Associativity::Left => prec += 1,
- Associativity::Right => {}
- }
-
- let rhs = match expr_with(p, prec) {
- Some(rhs) => Box::new(rhs),
- None => break,
- };
-
- let span = lhs.span().join(rhs.span());
- lhs = Expr::Binary(ExprBinary { span, lhs: Box::new(lhs), op, rhs });
- }
-
- Some(lhs)
-}
-
/// Parse a primary expression.
fn primary(p: &mut Parser) -> Option<Expr> {
if let Some(expr) = literal(p) {
@@ -289,10 +181,10 @@ fn primary(p: &mut Parser) -> Option<Expr> {
span: p.eat_span(),
string: string.into(),
};
- if p.peek() == Some(Token::LeftParen) {
- Some(paren_call(p, ident))
- } else {
- Some(Expr::Ident(ident))
+
+ match p.peek_direct() {
+ Some(Token::LeftParen) | Some(Token::LeftBracket) => Some(call(p, ident)),
+ _ => Some(Expr::Ident(ident)),
}
}
@@ -304,7 +196,6 @@ fn primary(p: &mut Parser) -> Option<Expr> {
// Structures.
Some(Token::LeftBrace) => block(p, true),
Some(Token::LeftBracket) => Some(template(p)),
- Some(Token::HashBracket) => bracket_call(p),
Some(Token::LeftParen) => Some(parenthesized(p)),
// Nothing.
@@ -327,7 +218,12 @@ fn literal(p: &mut Parser) -> Option<Expr> {
Token::Angle(val, unit) => LitKind::Angle(val, unit),
Token::Percent(p) => LitKind::Percent(p),
Token::Color(color) => LitKind::Color(color),
- Token::Str(token) => LitKind::Str(string(p, token)),
+ Token::Str(token) => LitKind::Str({
+ if !token.terminated {
+ p.expected_at("quote", p.peek_span().end);
+ }
+ resolve::resolve_string(token.string)
+ }),
_ => return None,
};
Some(Expr::Lit(Lit { span: p.eat_span(), kind }))
@@ -360,30 +256,83 @@ fn block(p: &mut Parser, scopes: bool) -> Option<Expr> {
Some(Expr::Block(ExprBlock { span, exprs, scoping: scopes }))
}
-/// Parse a parenthesized function call.
-fn paren_call(p: &mut Parser, name: Ident) -> Expr {
- p.start_group(Group::Paren, TokenMode::Code);
- let args = args(p);
- p.end_group();
+/// Parse an expression.
+fn expr(p: &mut Parser) -> Option<Expr> {
+ expr_with(p, 0)
+}
+
+/// Parse an expression with operators having at least the minimum precedence.
+fn expr_with(p: &mut Parser, min_prec: usize) -> Option<Expr> {
+ let start = p.start();
+ let mut lhs = match p.eat_map(UnOp::from_token) {
+ Some(op) => {
+ let prec = op.precedence();
+ let expr = Box::new(expr_with(p, prec)?);
+ Expr::Unary(ExprUnary { span: p.span(start), op, expr })
+ }
+ None => primary(p)?,
+ };
+
+ loop {
+ let op = match p.peek().and_then(BinOp::from_token) {
+ Some(binop) => binop,
+ None => break,
+ };
+
+ let mut prec = op.precedence();
+ if prec < min_prec {
+ break;
+ }
+
+ p.eat();
+ match op.associativity() {
+ Associativity::Left => prec += 1,
+ Associativity::Right => {}
+ }
+
+ let rhs = match expr_with(p, prec) {
+ Some(rhs) => Box::new(rhs),
+ None => break,
+ };
+
+ let span = lhs.span().join(rhs.span());
+ lhs = Expr::Binary(ExprBinary { span, lhs: Box::new(lhs), op, rhs });
+ }
+
+ Some(lhs)
+}
+
+/// Parse a function call.
+fn call(p: &mut Parser, name: Ident) -> Expr {
+ let mut args = match p.peek_direct() {
+ Some(Token::LeftParen) => {
+ p.start_group(Group::Paren, TokenMode::Code);
+ let args = args(p);
+ p.end_group();
+ args
+ }
+ _ => ExprArgs {
+ span: Span::at(name.span.end),
+ items: vec![],
+ },
+ };
+
+ if p.peek_direct() == Some(Token::LeftBracket) {
+ let body = template(p);
+ args.items.push(ExprArg::Pos(body));
+ }
+
Expr::Call(ExprCall {
- span: p.span_from(name.span.start),
+ span: p.span(name.span.start),
callee: Box::new(Expr::Ident(name)),
args,
})
}
-/// Parse a string.
-fn string(p: &mut Parser, token: TokenStr) -> String {
- if !token.terminated {
- p.expected_at("quote", p.peek_span().end);
- }
- resolve::resolve_string(token.string)
-}
-
/// Parse a let expression.
fn expr_let(p: &mut Parser) -> Option<Expr> {
let start = p.start();
- p.assert(&[Token::Let]);
+ p.assert(Token::Let);
let mut expr_let = None;
if let Some(binding) = ident(p) {
@@ -393,7 +342,7 @@ fn expr_let(p: &mut Parser) -> Option<Expr> {
}
expr_let = Some(Expr::Let(ExprLet {
- span: p.span_from(start),
+ span: p.span(start),
binding,
init: init.map(Box::new),
}))
@@ -405,18 +354,24 @@ fn expr_let(p: &mut Parser) -> Option<Expr> {
/// Parse an if expresion.
fn expr_if(p: &mut Parser) -> Option<Expr> {
let start = p.start();
- p.assert(&[Token::If]);
+ p.assert(Token::If);
let mut expr_if = None;
if let Some(condition) = expr(p) {
if let Some(if_body) = body(p) {
let mut else_body = None;
- if p.eat_if(Token::Else) {
+
+ // We are in code mode but still want to react to `#else` if the
+ // outer mode is markup.
+ if match p.outer_mode() {
+ TokenMode::Markup => p.eat_if(Token::Invalid("#else")),
+ TokenMode::Code => p.eat_if(Token::Else),
+ } {
else_body = body(p);
}
expr_if = Some(Expr::If(ExprIf {
- span: p.span_from(start),
+ span: p.span(start),
condition: Box::new(condition),
if_body: Box::new(if_body),
else_body: else_body.map(Box::new),
@@ -430,7 +385,7 @@ fn expr_if(p: &mut Parser) -> Option<Expr> {
/// Parse a for expression.
fn expr_for(p: &mut Parser) -> Option<Expr> {
let start = p.start();
- p.assert(&[Token::For]);
+ p.assert(Token::For);
let mut expr_for = None;
if let Some(pattern) = for_pattern(p) {
@@ -438,7 +393,7 @@ fn expr_for(p: &mut Parser) -> Option<Expr> {
if let Some(iter) = expr(p) {
if let Some(body) = body(p) {
expr_for = Some(Expr::For(ExprFor {
- span: p.span_from(start),
+ span: p.span(start),
pattern,
iter: Box::new(iter),
body: Box::new(body),
diff --git a/src/parse/parser.rs b/src/parse/parser.rs
index 7c660182..5d390bc1 100644
--- a/src/parse/parser.rs
+++ b/src/parse/parser.rs
@@ -10,6 +10,8 @@ pub struct Parser<'s> {
pub diags: DiagSet,
/// An iterator over the source tokens.
tokens: Tokens<'s>,
+ /// The stack of open groups.
+ groups: Vec<GroupEntry>,
/// The next token.
next: Option<Token<'s>>,
/// The peeked token.
@@ -19,21 +21,20 @@ pub struct Parser<'s> {
next_start: Pos,
/// The end position of the last (non-whitespace if in code mode) token.
last_end: Pos,
- /// The stack of open groups.
- groups: Vec<GroupEntry>,
}
/// A logical group of tokens, e.g. `[...]`.
+#[derive(Debug, Copy, Clone)]
struct GroupEntry {
/// The start position of the group. Used by `Parser::end_group` to return
/// The group's full span.
- start: Pos,
+ pub start: Pos,
/// The kind of group this is. This decides which tokens will end the group.
- /// For example, a [`GroupKind::Paren`] will be ended by
+ /// For example, a [`Group::Paren`] will be ended by
/// [`Token::RightParen`].
- kind: Group,
+ pub kind: Group,
/// The mode the parser was in _before_ the group started.
- prev_mode: TokenMode,
+ pub outer_mode: TokenMode,
}
/// A group, confined by optional start and end delimiters.
@@ -60,13 +61,13 @@ impl<'s> Parser<'s> {
let mut tokens = Tokens::new(src, TokenMode::Markup);
let next = tokens.next();
Self {
- tokens,
+ diags: DiagSet::new(),
next,
+ tokens,
+ last_end: Pos::ZERO,
peeked: next,
next_start: Pos::ZERO,
- last_end: Pos::ZERO,
groups: vec![],
- diags: DiagSet::new(),
}
}
@@ -118,16 +119,16 @@ impl<'s> Parser<'s> {
self.groups.push(GroupEntry {
start: self.next_start,
kind,
- prev_mode: self.tokens.mode(),
+ outer_mode: self.tokens.mode(),
});
self.tokens.set_mode(mode);
self.repeek();
match kind {
- Group::Paren => self.assert(&[Token::LeftParen]),
- Group::Bracket => self.assert(&[Token::HashBracket, Token::LeftBracket]),
- Group::Brace => self.assert(&[Token::LeftBrace]),
+ Group::Paren => self.assert(Token::LeftParen),
+ Group::Bracket => self.assert(Token::LeftBracket),
+ Group::Brace => self.assert(Token::LeftBrace),
Group::Subheader => {}
Group::Stmt => {}
Group::Expr => {}
@@ -141,7 +142,7 @@ impl<'s> Parser<'s> {
pub fn end_group(&mut self) -> Span {
let prev_mode = self.tokens.mode();
let group = self.groups.pop().expect("no started group");
- self.tokens.set_mode(group.prev_mode);
+ self.tokens.set_mode(group.outer_mode);
self.repeek();
let mut rescan = self.tokens.mode() != prev_mode;
@@ -173,6 +174,62 @@ impl<'s> Parser<'s> {
Span::new(group.start, self.last_end)
}
+ /// The tokenization mode outside of the current group.
+ ///
+ /// For example, this would be [`Markup`] if we are in a [`Code`] group that
+ /// is embedded in a [`Markup`] group.
+ ///
+ /// [`Markup`]: TokenMode::Markup
+ /// [`Code`]: TokenMode::Code
+ pub fn outer_mode(&mut self) -> TokenMode {
+ self.groups.last().map_or(TokenMode::Markup, |group| group.outer_mode)
+ }
+
+ /// Whether the end of the source string or group is reached.
+ pub fn eof(&self) -> bool {
+ self.peek().is_none()
+ }
+
+ /// Peek at the next token without consuming it.
+ pub fn peek(&self) -> Option<Token<'s>> {
+ self.peeked
+ }
+
+ /// Peek at the next token if it follows immediately after the last one
+ /// without any whitespace in between.
+ pub fn peek_direct(&self) -> Option<Token<'s>> {
+ if self.next_start == self.last_end {
+ self.peeked
+ } else {
+ None
+ }
+ }
+
+ /// Peek at the span of the next token.
+ ///
+ /// Has length zero if `peek()` returns `None`.
+ pub fn peek_span(&self) -> Span {
+ Span::new(
+ self.next_start,
+ if self.eof() { self.next_start } else { self.tokens.pos() },
+ )
+ }
+
+ /// Peek at the source of the next token.
+ pub fn peek_src(&self) -> &'s str {
+ self.get(self.peek_span())
+ }
+
+ /// Checks whether the next token fulfills a condition.
+ ///
+ /// Returns `false` if there is no next token.
+ pub fn check<F>(&self, f: F) -> bool
+ where
+ F: FnOnce(Token<'s>) -> bool,
+ {
+ self.peek().map_or(false, f)
+ }
+
/// Consume the next token.
pub fn eat(&mut self) -> Option<Token<'s>> {
let token = self.peek()?;
@@ -210,8 +267,8 @@ impl<'s> Parser<'s> {
Span::new(start, self.last_end)
}
- /// Consume the next token if it is the given one and produce an error if
- /// not.
+ /// Consume the next token if it is the given one and produce a diagnostic
+ /// if not.
pub fn expect(&mut self, t: Token) -> bool {
let eaten = self.eat_if(t);
if !eaten {
@@ -221,9 +278,9 @@ impl<'s> Parser<'s> {
}
/// Consume the next token, debug-asserting that it is one of the given ones.
- pub fn assert(&mut self, ts: &[Token]) {
+ pub fn assert(&mut self, t: Token) {
let next = self.eat();
- debug_assert!(next.map_or(false, |n| ts.contains(&n)));
+ debug_assert_eq!(next, Some(t));
}
/// Skip whitespace and comment tokens.
@@ -238,41 +295,6 @@ impl<'s> Parser<'s> {
}
}
- /// Peek at the next token without consuming it.
- pub fn peek(&self) -> Option<Token<'s>> {
- self.peeked
- }
-
- /// Peek at the span of the next token.
- ///
- /// Has length zero if `peek()` returns `None`.
- pub fn peek_span(&self) -> Span {
- Span::new(
- self.next_start,
- if self.eof() { self.next_start } else { self.tokens.pos() },
- )
- }
-
- /// Peek at the source of the next token.
- pub fn peek_src(&self) -> &'s str {
- self.get(self.peek_span())
- }
-
- /// Checks whether the next token fulfills a condition.
- ///
- /// Returns `false` if there is no next token.
- pub fn check<F>(&self, f: F) -> bool
- where
- F: FnOnce(Token<'s>) -> bool,
- {
- self.peek().map_or(false, f)
- }
-
- /// Whether the end of the source string or group is reached.
- pub fn eof(&self) -> bool {
- self.peek().is_none()
- }
-
/// The position at which the next token starts.
pub fn start(&self) -> Pos {
self.next_start
@@ -285,8 +307,8 @@ impl<'s> Parser<'s> {
self.last_end
}
- /// The span from
- pub fn span_from(&self, start: Pos) -> Span {
+ /// The span from `start` to the end of the last token.
+ pub fn span(&self, start: Pos) -> Span {
Span::new(start, self.last_end)
}