konevsa/typst
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Remove spans from token iterator 🧽

Browse Source
This commit is contained in:
Laurenz 2020-10-02 19:17:47 +02:00
parent 343982c56f
commit 904bc392ab
6 changed files with 319 additions and 282 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/layout/stack.rs
View File

@ -317,7 +317,7 @@ impl StackLayouter {
// Then, we reduce the bounding box for the following layouts. This // Then, we reduce the bounding box for the following layouts. This
// layout uses up space from the origin to the end. Thus, it reduces // layout uses up space from the origin to the end. Thus, it reduces
// the usable space for following layouts at it's origin by its // the usable space for following layouts at its origin by its
// extent along the secondary axis. // extent along the secondary axis.
*bound.get_mut(sys.secondary, GenAlign::Start) += *bound.get_mut(sys.secondary, GenAlign::Start) +=
sys.secondary.factor() * layout.size.secondary(*sys); sys.secondary.factor() * layout.size.secondary(*sys);
@ -345,7 +345,7 @@ impl StackLayouter {
rotation = sys.secondary.axis(); rotation = sys.secondary.axis();
} }
// We reduce the bounding box of this layout at it's end by the // We reduce the bounding box of this layout at its end by the
// accumulated secondary extent of all layouts we have seen so far, // accumulated secondary extent of all layouts we have seen so far,
// which are the layouts after this one since we iterate reversed. // which are the layouts after this one since we iterate reversed.
*bound.get_mut(sys.secondary, GenAlign::End) -= *bound.get_mut(sys.secondary, GenAlign::End) -=
@ -369,7 +369,7 @@ impl StackLayouter {
let align = layout.align; let align = layout.align;
// The space in which this layout is aligned is given by the // The space in which this layout is aligned is given by the
// distances between the borders of it's bounding box. // distances between the borders of its bounding box.
let usable = Size::new(bound.right - bound.left, bound.bottom - bound.top) let usable = Size::new(bound.right - bound.left, bound.bottom - bound.top)
.generalized(sys); .generalized(sys);

427
src/parse/mod.rs
View File

@ -46,55 +46,55 @@ fn tree(p: &mut Parser) -> SynTree {
/// Parse a syntax node. /// Parse a syntax node.
fn node(p: &mut Parser, at_start: bool) -> Option<Spanned<SynNode>> { fn node(p: &mut Parser, at_start: bool) -> Option<Spanned<SynNode>> {
let token = p.eat()?; let start = p.pos();
let span = token.span; let node = match p.eat()? {
Some(match token.v {
// Spaces. // Spaces.
Token::Space(newlines) => { Token::Space(newlines) => {
if newlines < 2 { if newlines < 2 {
SynNode::Space.span_with(span) SynNode::Space
} else { } else {
SynNode::Parbreak.span_with(span) SynNode::Parbreak
} }
} }
Token::Text(text) => SynNode::Text(text.into()).span_with(span), Token::Text(text) => SynNode::Text(text.into()),
// Comments. // Comments.
Token::LineComment(_) | Token::BlockComment(_) => return None, Token::LineComment(_) | Token::BlockComment(_) => return None,
// Markup. // Markup.
Token::Star => SynNode::ToggleBolder.span_with(span), Token::Star => SynNode::ToggleBolder,
Token::Underscore => SynNode::ToggleItalic.span_with(span), Token::Underscore => SynNode::ToggleItalic,
Token::Backslash => SynNode::Linebreak.span_with(span), Token::Backslash => SynNode::Linebreak,
Token::Hashtag => { Token::Hashtag => {
if at_start { if at_start {
heading(p, span.start).map(SynNode::Heading) SynNode::Heading(heading(p, start))
} else { } else {
SynNode::Text(p.get(span).into()).span_with(span) SynNode::Text(p.eaten_from(start).into())
} }
} }
Token::Raw(token) => raw(p, token, span).map(SynNode::Raw), Token::Raw(token) => SynNode::Raw(raw(p, token)),
Token::UnicodeEscape(token) => unicode_escape(p, token, span).map(SynNode::Text), Token::UnicodeEscape(token) => SynNode::Text(unicode_escape(p, token, start)),
// Functions. // Functions.
Token::LeftBracket => { Token::LeftBracket => {
p.jump(span.start); p.jump(start);
bracket_call(p).map(Expr::Call).map(SynNode::Expr) SynNode::Expr(Expr::Call(bracket_call(p)))
} }
// Bad tokens. // Bad tokens.
_ => { token => {
p.diag_unexpected(token); p.diag_unexpected(token.span_with(start .. p.pos()));
return None; return None;
} }
}) };
Some(node.span_with(start .. p.pos()))
} }
/// Parse a heading. /// Parse a heading.
fn heading(p: &mut Parser, start: Pos) -> Spanned<NodeHeading> { fn heading(p: &mut Parser, start: Pos) -> NodeHeading {
// Parse the section depth. // Parse the section depth.
let count = p.eat_while(|c| c == Token::Hashtag); let count = p.eat_while(|c| c == Token::Hashtag);
let span = (start, p.pos()); let span = Span::new(start, p.pos());
let level = (count.min(5) as u8).span_with(span); let level = (count.min(5) as u8).span_with(span);
if count > 5 { if count > 5 {
p.diag(warning!(span, "section depth larger than 6 has no effect")); p.diag(warning!(span, "section depth larger than 6 has no effect"));
@ -109,26 +109,23 @@ fn heading(p: &mut Parser, start: Pos) -> Spanned<NodeHeading> {
} }
} }
NodeHeading { level, contents }.span_with((start, p.pos())) NodeHeading { level, contents }
} }
/// Parse a raw block. /// Parse a raw block.
fn raw(p: &mut Parser, token: TokenRaw, span: Span) -> Spanned<NodeRaw> { fn raw(p: &mut Parser, token: TokenRaw) -> NodeRaw {
let raw = resolve::resolve_raw(token.text, token.backticks); let raw = resolve::resolve_raw(token.text, token.backticks);
if !token.terminated { if !token.terminated {
p.diag(error!(span.end, "expected backtick(s)")); p.diag(error!(p.pos(), "expected backtick(s)"));
} }
raw.span_with(span) raw
} }
/// Parse a unicode escape sequence. /// Parse a unicode escape sequence.
fn unicode_escape( fn unicode_escape(p: &mut Parser, token: TokenUnicodeEscape, start: Pos) -> String {
p: &mut Parser, let span = Span::new(start, p.pos());
token: TokenUnicodeEscape,
span: Span,
) -> Spanned<String> {
let text = if let Some(c) = resolve::resolve_hex(token.sequence) { let text = if let Some(c) = resolve::resolve_hex(token.sequence) {
c.to_string() c.to_string()
} else { } else {
@ -142,29 +139,28 @@ fn unicode_escape(
p.diag(error!(span.end, "expected closing brace")); p.diag(error!(span.end, "expected closing brace"));
} }
text.span_with(span) text
} }
/// Parse a bracketed function call. /// Parse a bracketed function call.
fn bracket_call(p: &mut Parser) -> Spanned<ExprCall> { fn bracket_call(p: &mut Parser) -> ExprCall {
let before_bracket = p.pos();
p.start_group(Group::Bracket); p.start_group(Group::Bracket);
p.push_mode(TokenMode::Header); p.push_mode(TokenMode::Header);
// One header is guaranteed, but there may be more (through chaining). // One header is guaranteed, but there may be more (through chaining).
let mut outer = vec![]; let mut outer = vec![];
let mut inner = bracket_subheader(p); let mut inner = p.span(|p| bracket_subheader(p));
while p.eat_if(Token::Chain).is_some() { while p.eat_if(Token::Chain) {
outer.push(inner); outer.push(inner);
inner = bracket_subheader(p); inner = p.span(|p| bracket_subheader(p));
} }
p.pop_mode(); p.pop_mode();
p.end_group(); p.end_group();
if p.peek() == Some(Token::LeftBracket) { if p.peek() == Some(Token::LeftBracket) {
let expr = bracket_body(p).map(Lit::Content).map(Expr::Lit); let expr = p.span(|p| Expr::Lit(Lit::Content(bracket_body(p))));
inner.span.expand(expr.span); inner.span.expand(expr.span);
inner.v.args.0.push(LitDictEntry { key: None, expr }); inner.v.args.0.push(LitDictEntry { key: None, expr });
} }
@ -177,26 +173,26 @@ fn bracket_call(p: &mut Parser) -> Spanned<ExprCall> {
inner = top; inner = top;
} }
inner.v.span_with((before_bracket, p.pos())) inner.v
} }
/// Parse one subheader of a bracketed function call. /// Parse one subheader of a bracketed function call.
fn bracket_subheader(p: &mut Parser) -> Spanned<ExprCall> { fn bracket_subheader(p: &mut Parser) -> ExprCall {
p.start_group(Group::Subheader); p.start_group(Group::Subheader);
let before_name = p.pos(); let start = p.pos();
p.skip_white(); p.skip_white();
let name = ident(p).unwrap_or_else(|| { let name = p.span(|p| ident(p)).transpose().unwrap_or_else(|| {
if p.eof() { if p.eof() {
p.diag_expected_at("function name", before_name); p.diag_expected_at("function name", start);
} else { } else {
p.diag_expected("function name"); p.diag_expected("function name");
} }
Ident(String::new()).span_with(before_name) Ident(String::new()).span_with(start)
}); });
p.skip_white(); p.skip_white();
let args = if p.eat_if(Token::Colon).is_some() { let args = if p.eat_if(Token::Colon) {
dict_contents(p).0 dict_contents(p).0
} else { } else {
// Ignore the rest if there's no colon. // Ignore the rest if there's no colon.
@ -207,171 +203,26 @@ fn bracket_subheader(p: &mut Parser) -> Spanned<ExprCall> {
LitDict::new() LitDict::new()
}; };
ExprCall { name, args }.span_with(p.end_group()) p.end_group();
ExprCall { name, args }
} }
/// Parse the body of a bracketed function call. /// Parse the body of a bracketed function call.
fn bracket_body(p: &mut Parser) -> Spanned<SynTree> { fn bracket_body(p: &mut Parser) -> SynTree {
p.start_group(Group::Bracket); p.start_group(Group::Bracket);
p.push_mode(TokenMode::Body); p.push_mode(TokenMode::Body);
let tree = tree(p); let tree = tree(p);
p.pop_mode(); p.pop_mode();
tree.span_with(p.end_group()) p.end_group();
} tree
/// Parse an expression: `term (+ term)*`.
fn expr(p: &mut Parser) -> Option<Spanned<Expr>> {
binops(p, "summand", term, |token| match token {
Token::Plus => Some(BinOp::Add),
Token::Hyphen => Some(BinOp::Sub),
_ => None,
})
}
/// Parse a term: `factor (* factor)*`.
fn term(p: &mut Parser) -> Option<Spanned<Expr>> {
binops(p, "factor", factor, |token| match token {
Token::Star => Some(BinOp::Mul),
Token::Slash => Some(BinOp::Div),
_ => None,
})
}
/// Parse binary operations of the from `a (<op> b)*`.
fn binops(
p: &mut Parser,
operand_name: &str,
operand: fn(&mut Parser) -> Option<Spanned<Expr>>,
op: fn(Token) -> Option<BinOp>,
) -> Option<Spanned<Expr>> {
let mut lhs = operand(p)?;
loop {
p.skip_white();
if let Some(op) = p.eat_map(op) {
p.skip_white();
if let Some(rhs) = operand(p) {
let span = lhs.span.join(rhs.span);
let expr = Expr::Binary(ExprBinary {
lhs: lhs.map(Box::new),
op,
rhs: rhs.map(Box::new),
});
lhs = expr.span_with(span);
p.skip_white();
} else {
let span = lhs.span.join(op.span);
p.diag(error!(span, "missing right {}", operand_name));
break;
}
} else {
break;
}
}
Some(lhs)
}
/// Parse a factor of the form `-?value`.
fn factor(p: &mut Parser) -> Option<Spanned<Expr>> {
if let Some(op) = p.eat_map(|token| match token {
Token::Hyphen => Some(UnOp::Neg),
_ => None,
}) {
p.skip_white();
if let Some(expr) = factor(p) {
let span = op.span.join(expr.span);
let expr = Expr::Unary(ExprUnary { op, expr: expr.map(Box::new) });
Some(expr.span_with(span))
} else {
p.diag(error!(op.span, "missing factor"));
None
}
} else {
value(p)
}
}
/// Parse a value.
fn value(p: &mut Parser) -> Option<Spanned<Expr>> {
let Spanned { v: token, span } = p.eat()?;
Some(match token {
// Bracketed function call.
Token::LeftBracket => {
p.jump(span.start);
let call = bracket_call(p);
let span = call.span;
let node = call.map(Expr::Call).map(SynNode::Expr);
Expr::Lit(Lit::Content(vec![node])).span_with(span)
}
// Content expression.
Token::LeftBrace => {
p.jump(span.start);
content(p).map(Lit::Content).map(Expr::Lit)
}
// Dictionary or just a parenthesized expression.
Token::LeftParen => {
p.jump(span.start);
parenthesized(p)
}
// Function or just ident.
Token::Ident(id) => {
let ident = Ident(id.into()).span_with(span);
p.skip_white();
if p.peek() == Some(Token::LeftParen) {
paren_call(p, ident).map(Expr::Call)
} else {
ident.map(Lit::Ident).map(Expr::Lit)
}
}
// Atomic values.
Token::Bool(b) => Expr::Lit(Lit::Bool(b)).span_with(span),
Token::Number(f) => Expr::Lit(Lit::Float(f)).span_with(span),
Token::Length(l) => Expr::Lit(Lit::Length(l)).span_with(span),
Token::Hex(hex) => color(p, hex, span).map(Lit::Color).map(Expr::Lit),
Token::Str(token) => string(p, token, span).map(Lit::Str).map(Expr::Lit),
// No value.
_ => {
p.jump(span.start);
return None;
}
})
}
// Parse a content expression: `{...}`.
fn content(p: &mut Parser) -> Spanned<SynTree> {
p.start_group(Group::Brace);
p.push_mode(TokenMode::Body);
let tree = tree(p);
p.pop_mode();
tree.span_with(p.end_group())
}
/// Parse a parenthesized expression: `(a + b)`, `(1, key="value").
fn parenthesized(p: &mut Parser) -> Spanned<Expr> {
p.start_group(Group::Paren);
let (dict, coercable) = dict_contents(p);
let expr = if coercable {
dict.0.into_iter().next().expect("dict is coercable").expr.v
} else {
Expr::Lit(Lit::Dict(dict))
};
expr.span_with(p.end_group())
} }
/// Parse a parenthesized function call. /// Parse a parenthesized function call.
fn paren_call(p: &mut Parser, name: Spanned<Ident>) -> Spanned<ExprCall> { fn paren_call(p: &mut Parser, name: Spanned<Ident>) -> ExprCall {
p.start_group(Group::Paren); p.start_group(Group::Paren);
let args = dict_contents(p).0; let args = dict_contents(p).0;
let span = name.span.join(p.end_group()); p.end_group();
ExprCall { name, args }.span_with(span) ExprCall { name, args }
} }
/// Parse the contents of a dictionary. /// Parse the contents of a dictionary.
@ -405,7 +256,7 @@ fn dict_contents(p: &mut Parser) -> (LitDict, bool) {
break; break;
} }
if p.eat_if(Token::Comma).is_none() { if !p.eat_if(Token::Comma) {
p.diag_expected_at("comma", behind); p.diag_expected_at("comma", behind);
} }
@ -418,7 +269,7 @@ fn dict_contents(p: &mut Parser) -> (LitDict, bool) {
/// Parse a single entry in a dictionary. /// Parse a single entry in a dictionary.
fn dict_entry(p: &mut Parser) -> Option<LitDictEntry> { fn dict_entry(p: &mut Parser) -> Option<LitDictEntry> {
if let Some(ident) = ident(p) { if let Some(ident) = p.span(|p| ident(p)).transpose() {
p.skip_white(); p.skip_white();
match p.peek() { match p.peek() {
// Key-value pair. // Key-value pair.
@ -438,7 +289,11 @@ fn dict_entry(p: &mut Parser) -> Option<LitDictEntry> {
// Function call. // Function call.
Some(Token::LeftParen) => Some(LitDictEntry { Some(Token::LeftParen) => Some(LitDictEntry {
key: None, key: None,
expr: paren_call(p, ident).map(Expr::Call), expr: {
let start = ident.span.start;
let call = paren_call(p, ident);
Expr::Call(call).span_with(start .. p.pos())
},
}), }),
// Just an identifier. // Just an identifier.
@ -454,8 +309,160 @@ fn dict_entry(p: &mut Parser) -> Option<LitDictEntry> {
} }
} }
/// Parse an expression: `term (+ term)*`.
fn expr(p: &mut Parser) -> Option<Spanned<Expr>> {
binops(p, "summand", term, |token| match token {
Token::Plus => Some(BinOp::Add),
Token::Hyphen => Some(BinOp::Sub),
_ => None,
})
}
/// Parse a term: `factor (* factor)*`.
fn term(p: &mut Parser) -> Option<Spanned<Expr>> {
binops(p, "factor", factor, |token| match token {
Token::Star => Some(BinOp::Mul),
Token::Slash => Some(BinOp::Div),
_ => None,
})
}
/// Parse binary operations of the from `a (<op> b)*`.
fn binops(
p: &mut Parser,
operand_name: &str,
operand: fn(&mut Parser) -> Option<Spanned<Expr>>,
op: fn(Token) -> Option<BinOp>,
) -> Option<Spanned<Expr>> {
let mut lhs = operand(p)?;
loop {
p.skip_white();
if let Some(op) = p.span(|p| p.eat_map(op)).transpose() {
p.skip_white();
if let Some(rhs) = operand(p) {
let span = lhs.span.join(rhs.span);
let expr = Expr::Binary(ExprBinary {
lhs: lhs.map(Box::new),
op,
rhs: rhs.map(Box::new),
});
lhs = expr.span_with(span);
p.skip_white();
} else {
let span = lhs.span.join(op.span);
p.diag(error!(span, "missing right {}", operand_name));
break;
}
} else {
break;
}
}
Some(lhs)
}
/// Parse a factor of the form `-?value`.
fn factor(p: &mut Parser) -> Option<Spanned<Expr>> {
let op = |token| match token {
Token::Hyphen => Some(UnOp::Neg),
_ => None,
};
p.span(|p| {
if let Some(op) = p.span(|p| p.eat_map(op)).transpose() {
p.skip_white();
if let Some(expr) = factor(p) {
Some(Expr::Unary(ExprUnary { op, expr: expr.map(Box::new) }))
} else {
p.diag(error!(op.span, "missing factor"));
None
}
} else {
value(p)
}
})
.transpose()
}
/// Parse a value.
fn value(p: &mut Parser) -> Option<Expr> {
let start = p.pos();
Some(match p.eat()? {
// Bracketed function call.
Token::LeftBracket => {
p.jump(start);
let node = p.span(|p| SynNode::Expr(Expr::Call(bracket_call(p))));
Expr::Lit(Lit::Content(vec![node]))
}
// Content expression.
Token::LeftBrace => {
p.jump(start);
Expr::Lit(Lit::Content(content(p)))
}
// Dictionary or just a parenthesized expression.
Token::LeftParen => {
p.jump(start);
parenthesized(p)
}
// Function or just ident.
Token::Ident(id) => {
let ident = Ident(id.into());
let after = p.pos();
p.skip_white();
if p.peek() == Some(Token::LeftParen) {
let name = ident.span_with(start .. after);
Expr::Call(paren_call(p, name))
} else {
Expr::Lit(Lit::Ident(ident))
}
}
// Atomic values.
Token::Bool(b) => Expr::Lit(Lit::Bool(b)),
Token::Number(f) => Expr::Lit(Lit::Float(f)),
Token::Length(l) => Expr::Lit(Lit::Length(l)),
Token::Hex(hex) => Expr::Lit(Lit::Color(color(p, hex, start))),
Token::Str(token) => Expr::Lit(Lit::Str(string(p, token))),
// No value.
_ => {
p.jump(start);
return None;
}
})
}
// Parse a content expression: `{...}`.
fn content(p: &mut Parser) -> SynTree {
p.start_group(Group::Brace);
p.push_mode(TokenMode::Body);
let tree = tree(p);
p.pop_mode();
p.end_group();
tree
}
/// Parse a parenthesized expression: `(a + b)`, `(1, key="value").
fn parenthesized(p: &mut Parser) -> Expr {
p.start_group(Group::Paren);
let (dict, coercable) = dict_contents(p);
let expr = if coercable {
dict.0.into_iter().next().expect("dict is coercable").expr.v
} else {
Expr::Lit(Lit::Dict(dict))
};
p.end_group();
expr
}
/// Parse an identifier. /// Parse an identifier.
fn ident(p: &mut Parser) -> Option<Spanned<Ident>> { fn ident(p: &mut Parser) -> Option<Ident> {
p.eat_map(|token| match token { p.eat_map(|token| match token {
Token::Ident(id) => Some(Ident(id.into())), Token::Ident(id) => Some(Ident(id.into())),
_ => None, _ => None,
@ -463,23 +470,21 @@ fn ident(p: &mut Parser) -> Option<Spanned<Ident>> {
} }
/// Parse a color. /// Parse a color.
fn color(p: &mut Parser, hex: &str, span: Span) -> Spanned<RgbaColor> { fn color(p: &mut Parser, hex: &str, start: Pos) -> RgbaColor {
RgbaColor::from_str(hex) RgbaColor::from_str(hex).unwrap_or_else(|_| {
.unwrap_or_else(|_| { // Heal color by assuming black.
// Heal color by assuming black. p.diag(error!(start .. p.pos(), "invalid color"));
p.diag(error!(span, "invalid color")); RgbaColor::new_healed(0, 0, 0, 255)
RgbaColor::new_healed(0, 0, 0, 255) })
})
.span_with(span)
} }
/// Parse a string. /// Parse a string.
fn string(p: &mut Parser, token: TokenStr, span: Span) -> Spanned<String> { fn string(p: &mut Parser, token: TokenStr) -> String {
if !token.terminated { if !token.terminated {
p.diag_expected_at("quote", span.end); p.diag_expected_at("quote", p.pos());
} }
resolve::resolve_string(token.string).span_with(span) resolve::resolve_string(token.string)
} }
#[cfg(test)] #[cfg(test)]

110
src/parse/parser.rs
View File

@ -2,15 +2,16 @@ use std::fmt::{self, Debug, Formatter};
use super::{Scanner, TokenMode, Tokens}; use super::{Scanner, TokenMode, Tokens};
use crate::diagnostic::Diagnostic; use crate::diagnostic::Diagnostic;
use crate::syntax::{Decoration, Pos, Span, Spanned, Token}; use crate::syntax::{Decoration, Pos, Span, SpanWith, Spanned, Token};
use crate::Feedback; use crate::Feedback;
/// A convenient token-based parser. /// A convenient token-based parser.
pub struct Parser<'s> { pub struct Parser<'s> {
tokens: Tokens<'s>, tokens: Tokens<'s>,
peeked: Option<Spanned<Token<'s>>>, peeked: Option<Token<'s>>,
modes: Vec<TokenMode>, modes: Vec<TokenMode>,
groups: Vec<(Pos, Group)>, groups: Vec<Group>,
pos: Pos,
f: Feedback, f: Feedback,
} }
@ -22,6 +23,7 @@ impl<'s> Parser<'s> {
peeked: None, peeked: None,
modes: vec![], modes: vec![],
groups: vec![], groups: vec![],
pos: Pos::ZERO,
f: Feedback::new(), f: Feedback::new(),
} }
} }
@ -39,12 +41,14 @@ impl<'s> Parser<'s> {
/// Eat the next token and add a diagnostic that it was not the expected /// Eat the next token and add a diagnostic that it was not the expected
/// `thing`. /// `thing`.
pub fn diag_expected(&mut self, thing: &str) { pub fn diag_expected(&mut self, thing: &str) {
let before = self.pos();
if let Some(found) = self.eat() { if let Some(found) = self.eat() {
let after = self.pos();
self.diag(error!( self.diag(error!(
found.span, before .. after,
"expected {}, found {}", "expected {}, found {}",
thing, thing,
found.v.name(), found.name(),
)); ));
} else { } else {
self.diag_expected_at(thing, self.pos()); self.diag_expected_at(thing, self.pos());
@ -89,25 +93,24 @@ impl<'s> Parser<'s> {
/// # Panics /// # Panics
/// This panics if the next token does not start the given group. /// This panics if the next token does not start the given group.
pub fn start_group(&mut self, group: Group) { pub fn start_group(&mut self, group: Group) {
let start = self.pos();
match group { match group {
Group::Paren => self.eat_assert(Token::LeftParen), Group::Paren => self.eat_assert(Token::LeftParen),
Group::Bracket => self.eat_assert(Token::LeftBracket), Group::Bracket => self.eat_assert(Token::LeftBracket),
Group::Brace => self.eat_assert(Token::LeftBrace), Group::Brace => self.eat_assert(Token::LeftBrace),
Group::Subheader => {} Group::Subheader => {}
} }
self.groups.push((start, group)); self.groups.push(group);
} }
/// Ends the parsing of a group and returns the span of the whole group. /// Ends the parsing of a group and returns the span of the whole group.
/// ///
/// # Panics /// # Panics
/// This panics if no group was started. /// This panics if no group was started.
pub fn end_group(&mut self) -> Span { pub fn end_group(&mut self) {
// Check that we are indeed at the end of the group. // Check that we are indeed at the end of the group.
debug_assert_eq!(self.peek(), None, "unfinished group"); debug_assert_eq!(self.peek(), None, "unfinished group");
let (start, group) = self.groups.pop().expect("unstarted group"); let group = self.groups.pop().expect("unstarted group");
let end = match group { let end = match group {
Group::Paren => Some(Token::RightParen), Group::Paren => Some(Token::RightParen),
Group::Bracket => Some(Token::RightBracket), Group::Bracket => Some(Token::RightBracket),
@ -119,14 +122,12 @@ impl<'s> Parser<'s> {
// This `peek()` can't be used directly because it hides the end of // This `peek()` can't be used directly because it hides the end of
// group token. To circumvent this, we drop down to `self.peeked`. // group token. To circumvent this, we drop down to `self.peeked`.
self.peek(); self.peek();
if self.peeked.map(|s| s.v) == Some(token) { if self.peeked == Some(token) {
self.peeked = None; self.bump();
} else { } else {
self.diag(error!(self.pos(), "expected {}", token.name())); self.diag(error!(self.pos(), "expected {}", token.name()));
} }
} }
Span::new(start, self.pos())
} }
/// Skip whitespace tokens. /// Skip whitespace tokens.
@ -136,34 +137,43 @@ impl<'s> Parser<'s> {
}); });
} }
/// Execute `f` and return the result alongside the span of everything `f`
/// ate.
pub fn span<T>(&mut self, f: impl FnOnce(&mut Self) -> T) -> Spanned<T> {
let start = self.pos;
f(self).span_with(start .. self.pos)
}
/// Consume the next token. /// Consume the next token.
pub fn eat(&mut self) -> Option<Spanned<Token<'s>>> { pub fn eat(&mut self) -> Option<Token<'s>> {
self.peek()?; self.peek()?;
self.peeked.take() self.bump()
} }
/// Consume the next token if it is the given one. /// Consume the next token if it is the given one.
pub fn eat_if(&mut self, t: Token) -> Option<Spanned<Token<'s>>> { pub fn eat_if(&mut self, t: Token) -> bool {
if self.peek()? == t { self.peeked.take() } else { None } if self.peek() == Some(t) {
self.bump();
true
} else {
false
}
} }
/// Consume the next token if the closure maps it a to `Some`-variant. /// Consume the next token if the closure maps it a to `Some`-variant.
pub fn eat_map<T>( pub fn eat_map<T>(&mut self, f: impl FnOnce(Token<'s>) -> Option<T>) -> Option<T> {
&mut self,
mut f: impl FnMut(Token<'s>) -> Option<T>,
) -> Option<Spanned<T>> {
let token = self.peek()?; let token = self.peek()?;
if let Some(t) = f(token) { let out = f(token);
self.peeked.take().map(|spanned| spanned.map(|_| t)) if out.is_some() {
} else { self.bump();
None
} }
out
} }
/// Consume the next token, debug-asserting that it is the given one. /// Consume the next token, debug-asserting that it is the given one.
pub fn eat_assert(&mut self, t: Token) { pub fn eat_assert(&mut self, t: Token) {
let next = self.eat(); let next = self.eat();
debug_assert_eq!(next.map(|s| s.v), Some(t)); debug_assert_eq!(next, Some(t));
} }
/// Consume tokens while the condition is true. /// Consume tokens while the condition is true.
@ -182,7 +192,7 @@ impl<'s> Parser<'s> {
if f(t) { if f(t) {
break; break;
} }
self.peeked = None; self.bump();
count += 1; count += 1;
} }
count count
@ -191,11 +201,11 @@ impl<'s> Parser<'s> {
/// Peek at the next token without consuming it. /// Peek at the next token without consuming it.
pub fn peek(&mut self) -> Option<Token<'s>> { pub fn peek(&mut self) -> Option<Token<'s>> {
let token = match self.peeked { let token = match self.peeked {
Some(token) => token.v, Some(token) => token,
None => { None => {
let token = self.tokens.next()?; let token = self.tokens.next()?;
self.peeked = Some(token); self.peeked = Some(token);
token.v token
} }
}; };
@ -207,7 +217,7 @@ impl<'s> Parser<'s> {
_ => return Some(token), _ => return Some(token),
}; };
if self.groups.iter().rev().any(|&(_, g)| g == group) { if self.groups.contains(&group) {
None None
} else { } else {
Some(token) Some(token)
@ -217,7 +227,7 @@ impl<'s> Parser<'s> {
/// Checks whether the next token fulfills a condition. /// Checks whether the next token fulfills a condition.
/// ///
/// Returns `false` if there is no next token. /// Returns `false` if there is no next token.
pub fn check(&mut self, f: impl FnMut(Token<'s>) -> bool) -> bool { pub fn check(&mut self, f: impl FnOnce(Token<'s>) -> bool) -> bool {
self.peek().map(f).unwrap_or(false) self.peek().map(f).unwrap_or(false)
} }
@ -229,30 +239,52 @@ impl<'s> Parser<'s> {
/// The position in the string at which the last token ends and next token /// The position in the string at which the last token ends and next token
/// will start. /// will start.
pub fn pos(&self) -> Pos { pub fn pos(&self) -> Pos {
self.peeked.map(|s| s.span.start).unwrap_or_else(|| self.tokens.pos()) self.pos
} }
/// Jump to a position in the source string. /// Jump to a position in the source string.
pub fn jump(&mut self, pos: Pos) { pub fn jump(&mut self, pos: Pos) {
self.tokens.jump(pos); self.tokens.jump(pos);
self.peeked = None; self.bump();
} }
/// Returns the part of the source string that is spanned by the given span. /// Slice a part out of the source string.
pub fn get(&self, span: Span) -> &'s str { pub fn get(&self, span: impl Into<Span>) -> &'s str {
self.scanner().get(span.start.to_usize() .. span.end.to_usize()) self.tokens.scanner().get(span.into().to_range())
}
/// The full source string up to the current index.
pub fn eaten(&self) -> &'s str {
self.tokens.scanner().get(.. self.pos.to_usize())
}
/// The source string from `start` to the current index.
pub fn eaten_from(&self, start: Pos) -> &'s str {
self.tokens.scanner().get(start.to_usize() .. self.pos.to_usize())
}
/// The remaining source string after the current index.
pub fn rest(&self) -> &'s str {
self.tokens.scanner().get(self.pos.to_usize() ..)
} }
/// The underlying scanner. /// The underlying scanner.
pub fn scanner(&self) -> Scanner<'s> { pub fn scanner(&self) -> Scanner<'s> {
self.tokens.scanner() let mut scanner = self.tokens.scanner().clone();
scanner.jump(self.pos.to_usize());
scanner
}
/// Set the position to the tokenizer's position and take the peeked token.
fn bump(&mut self) -> Option<Token<'s>> {
self.pos = self.tokens.pos();
self.peeked.take()
} }
} }
impl Debug for Parser<'_> { impl Debug for Parser<'_> {
fn fmt(&self, f: &mut Formatter) -> fmt::Result { fn fmt(&self, f: &mut Formatter) -> fmt::Result {
let s = self.scanner(); write!(f, "Parser({}|{})", self.eaten(), self.rest())
write!(f, "Parser({}|{})", s.eaten(), s.rest())
} }
} }

2
src/parse/scanner.rs
View File

@ -100,7 +100,7 @@ impl<'s> Scanner<'s> {
/// Checks whether the next char fulfills a condition. /// Checks whether the next char fulfills a condition.
/// ///
/// Returns `false` if there is no next char. /// Returns `false` if there is no next char.
pub fn check(&self, f: impl FnMut(char) -> bool) -> bool { pub fn check(&self, f: impl FnOnce(char) -> bool) -> bool {
self.peek().map(f).unwrap_or(false) self.peek().map(f).unwrap_or(false)
} }

36
src/parse/tokens.rs
View File

@ -5,7 +5,7 @@ use std::fmt::{self, Debug, Formatter};
use super::{is_newline, Scanner}; use super::{is_newline, Scanner};
use crate::length::Length; use crate::length::Length;
use crate::syntax::token::*; use crate::syntax::token::*;
use crate::syntax::{is_ident, Pos, Span, SpanWith, Spanned}; use crate::syntax::{is_ident, Pos};
use TokenMode::*; use TokenMode::*;
@ -53,18 +53,18 @@ impl<'s> Tokens<'s> {
} }
/// The underlying scanner. /// The underlying scanner.
pub fn scanner(&self) -> Scanner<'s> { pub fn scanner(&self) -> &Scanner<'s> {
self.s.clone() &self.s
} }
} }
impl<'s> Iterator for Tokens<'s> { impl<'s> Iterator for Tokens<'s> {
type Item = Spanned<Token<'s>>; type Item = Token<'s>;
/// Parse the next token in the source code. /// Parse the next token in the source code.
fn next(&mut self) -> Option<Self::Item> { fn next(&mut self) -> Option<Self::Item> {
let start = self.s.index(); let start = self.s.index();
let token = match self.s.eat()? { Some(match self.s.eat()? {
// Whitespace with fast path for just a single space. // Whitespace with fast path for just a single space.
' ' if !self.s.check(|c| c.is_whitespace()) => Token::Space(0), ' ' if !self.s.check(|c| c.is_whitespace()) => Token::Space(0),
c if c.is_whitespace() => { c if c.is_whitespace() => {
@ -109,10 +109,7 @@ impl<'s> Iterator for Tokens<'s> {
// Expressions or just plain text. // Expressions or just plain text.
_ => self.read_text_or_expr(start), _ => self.read_text_or_expr(start),
}; })
let end = self.s.index();
Some(token.span_with(Span::new(start, end)))
} }
} }
@ -298,7 +295,7 @@ fn parse_percent(text: &str) -> Option<f64> {
mod tests { mod tests {
use super::*; use super::*;
use crate::length::Length; use crate::length::Length;
use crate::parse::tests::{check, s}; use crate::parse::tests::check;
use Token::{ use Token::{
BlockComment as BC, Bool, Chain, Hex, Hyphen as Min, Ident as Id, BlockComment as BC, Bool, Chain, Hex, Hyphen as Min, Ident as Id,
@ -317,13 +314,11 @@ mod tests {
Token::UnicodeEscape(TokenUnicodeEscape { sequence, terminated }) Token::UnicodeEscape(TokenUnicodeEscape { sequence, terminated })
} }
macro_rules! t { ($($tts:tt)*) => {test!(@spans=false, $($tts)*)} } macro_rules! t {
macro_rules! ts { ($($tts:tt)*) => {test!(@spans=true, $($tts)*)} } ($mode:expr, $src:expr => $($token:expr),*) => {
macro_rules! test { let exp = vec![$($token),*];
(@spans=$spans:expr, $mode:expr, $src:expr => $($token:expr),*) => {
let exp = vec![$(Into::<Spanned<Token>>::into($token)),*];
let found = Tokens::new($src, $mode).collect::<Vec<_>>(); let found = Tokens::new($src, $mode).collect::<Vec<_>>();
check($src, exp, found, $spans); check($src, exp, found, false);
} }
} }
@ -479,13 +474,4 @@ mod tests {
t!(Header, "(5 - 1) / 2.1" => LP, Num(5.0), S(0), Min, S(0), Num(1.0), RP, t!(Header, "(5 - 1) / 2.1" => LP, Num(5.0), S(0), Min, S(0), Num(1.0), RP,
S(0), Slash, S(0), Num(2.1)); S(0), Slash, S(0), Num(2.1));
} }
#[test]
fn tokenize_with_spans() {
ts!(Body, "hello" => s(0, 5, T("hello")));
ts!(Body, "ab\r\nc" => s(0, 2, T("ab")), s(2, 4, S(1)), s(4, 5, T("c")));
ts!(Body, "// ab\r\n\nf" => s(0, 5, LC(" ab")), s(5, 8, S(2)), s(8, 9, T("f")));
ts!(Body, "/*b*/_" => s(0, 5, BC("b")), s(5, 6, Underscore));
ts!(Header, "a=10" => s(0, 1, Id("a")), s(1, 2, Equals), s(2, 4, Num(10.0)));
}
} }

20
src/syntax/span.rs
View File

@ -1,6 +1,7 @@
//! Mapping of values to the locations they originate from in source code. //! Mapping of values to the locations they originate from in source code.
use std::fmt::{self, Debug, Display, Formatter}; use std::fmt::{self, Debug, Display, Formatter};
use std::ops::Range;
#[cfg(test)] #[cfg(test)]
use std::cell::Cell; use std::cell::Cell;
@ -81,6 +82,14 @@ impl<T> Spanned<T> {
} }
} }
impl<T> Spanned<Option<T>> {
/// Swap the spanned and option.
pub fn transpose(self) -> Option<Spanned<T>> {
let Spanned { v, span } = self;
v.map(|v| v.span_with(span))
}
}
impl<T> Offset for Spanned<T> { impl<T> Offset for Spanned<T> {
fn offset(self, by: Pos) -> Self { fn offset(self, by: Pos) -> Self {
self.map_span(|span| span.offset(by)) self.map_span(|span| span.offset(by))
@ -135,6 +144,11 @@ impl Span {
*self = self.join(other) *self = self.join(other)
} }
/// Convert to a `Range<usize>` for indexing.
pub fn to_range(self) -> Range<usize> {
self.start.to_usize() .. self.end.to_usize()
}
/// When set to `false` comparisons with `PartialEq` ignore spans. /// When set to `false` comparisons with `PartialEq` ignore spans.
#[cfg(test)] #[cfg(test)]
pub(crate) fn set_cmp(cmp: bool) { pub(crate) fn set_cmp(cmp: bool) {
@ -173,12 +187,12 @@ where
} }
} }
impl<T> From<(T, T)> for Span impl<T> From<Range<T>> for Span
where where
T: Into<Pos>, T: Into<Pos>,
{ {
fn from((start, end): (T, T)) -> Self { fn from(range: Range<T>) -> Self {
Self::new(start, end) Self::new(range.start, range.end)
} }
} }