diff --git a/src/parse/mod.rs b/src/parse/mod.rs index 8775e8a17..22288d01a 100644 --- a/src/parse/mod.rs +++ b/src/parse/mod.rs @@ -520,7 +520,7 @@ fn dict(p: &mut Parser, items: usize) { p.child_count() - items, |x| { x.kind() == &NodeKind::Named - || x.kind().is_parenthesis() + || x.kind().is_paren() || x.kind() == &NodeKind::Comma || x.kind() == &NodeKind::Colon }, @@ -550,7 +550,7 @@ fn params(p: &mut Parser, count: usize, allow_parens: bool) { ), _ => false, } - || (allow_parens && x.kind().is_parenthesis()), + || (allow_parens && x.kind().is_paren()), |_| (ErrorPosition::Full, "expected identifier".into()), ); } diff --git a/src/syntax/expr.rs b/src/syntax/ast.rs similarity index 83% rename from src/syntax/expr.rs rename to src/syntax/ast.rs index 1439cbdb6..bdd0767de 100644 --- a/src/syntax/expr.rs +++ b/src/syntax/ast.rs @@ -1,8 +1,164 @@ -use super::{Ident, Markup, NodeKind, RedNode, RedRef, Span, TypedNode}; +use super::{Ident, NodeKind, RedNode, RedRef, Span, TypedNode}; use crate::geom::{AngularUnit, LengthUnit}; use crate::node; use crate::util::EcoString; +node! { + /// The syntactical root capable of representing a full parsed document. + Markup +} + +impl Markup { + pub fn nodes<'a>(&'a self) -> impl Iterator + 'a { + self.0.children().filter_map(RedRef::cast) + } +} + +/// A single piece of markup. +#[derive(Debug, Clone, PartialEq)] +pub enum MarkupNode { + /// Whitespace containing less than two newlines. + Space, + /// A forced line break: `\`. + Linebreak, + /// A paragraph break: Two or more newlines. + Parbreak, + /// Strong text was enabled / disabled: `*`. + Strong, + /// Emphasized text was enabled / disabled: `_`. + Emph, + /// Plain text. + Text(EcoString), + /// A raw block with optional syntax highlighting: `` `...` ``. + Raw(RawNode), + /// A section heading: `= Introduction`. + Heading(HeadingNode), + /// An item in an unordered list: `- ...`. + List(ListNode), + /// An item in an enumeration (ordered list): `1. ...`. + Enum(EnumNode), + /// An expression. + Expr(Expr), +} + +impl TypedNode for MarkupNode { + fn cast_from(node: RedRef) -> Option { + match node.kind() { + NodeKind::Space(_) => Some(MarkupNode::Space), + NodeKind::Linebreak => Some(MarkupNode::Linebreak), + NodeKind::Parbreak => Some(MarkupNode::Parbreak), + NodeKind::Strong => Some(MarkupNode::Strong), + NodeKind::Emph => Some(MarkupNode::Emph), + NodeKind::Text(s) => Some(MarkupNode::Text(s.clone())), + NodeKind::UnicodeEscape(u) => Some(MarkupNode::Text(u.character.into())), + NodeKind::EnDash => Some(MarkupNode::Text(EcoString::from("\u{2013}"))), + NodeKind::EmDash => Some(MarkupNode::Text(EcoString::from("\u{2014}"))), + NodeKind::NonBreakingSpace => { + Some(MarkupNode::Text(EcoString::from("\u{00A0}"))) + } + NodeKind::Raw(_) => node.cast().map(MarkupNode::Raw), + NodeKind::Heading => node.cast().map(MarkupNode::Heading), + NodeKind::List => node.cast().map(MarkupNode::List), + NodeKind::Enum => node.cast().map(MarkupNode::Enum), + NodeKind::Error(_, _) => None, + _ => node.cast().map(MarkupNode::Expr), + } + } +} + +/// A raw block with optional syntax highlighting: `` `...` ``. +#[derive(Debug, Clone, PartialEq)] +pub struct RawNode { + /// An optional identifier specifying the language to syntax-highlight in. + pub lang: Option, + /// The raw text, determined as the raw string between the backticks trimmed + /// according to the above rules. + pub text: EcoString, + /// Whether the element is block-level, that is, it has 3+ backticks + /// and contains at least one newline. + pub block: bool, +} + +impl TypedNode for RawNode { + fn cast_from(node: RedRef) -> Option { + match node.kind() { + NodeKind::Raw(raw) => { + let span = node.span(); + let start = span.start + raw.backticks as usize; + Some(Self { + block: raw.block, + lang: raw.lang.as_ref().and_then(|x| { + let span = Span::new(span.source, start, start + x.len()); + Ident::new(x, span) + }), + text: raw.text.clone(), + }) + } + _ => None, + } + } +} + +node! { + /// A section heading: `= Introduction`. + Heading => HeadingNode +} + +impl HeadingNode { + /// The contents of the heading. + pub fn body(&self) -> Markup { + self.0 + .cast_first_child() + .expect("heading node is missing markup body") + } + + /// The section depth (numer of equals signs). + pub fn level(&self) -> u8 { + self.0 + .children() + .find_map(|node| match node.kind() { + NodeKind::HeadingLevel(heading) => Some(*heading), + _ => None, + }) + .expect("heading node is missing heading level") + } +} + +node! { + /// An item in an unordered list: `- ...`. + List => ListNode +} + +impl ListNode { + /// The contents of the list item. + pub fn body(&self) -> Markup { + self.0.cast_first_child().expect("list node is missing body") + } +} + +node! { + /// An item in an enumeration (ordered list): `1. ...`. + Enum => EnumNode +} + +impl EnumNode { + /// The contents of the list item. + pub fn body(&self) -> Markup { + self.0.cast_first_child().expect("enumeration node is missing body") + } + + /// The number, if any. + pub fn number(&self) -> Option { + self.0 + .children() + .find_map(|node| match node.kind() { + NodeKind::EnumNumbering(num) => Some(num.clone()), + _ => None, + }) + .expect("enumeration node is missing number") + } +} + /// An expression. #[derive(Debug, Clone, PartialEq)] pub enum Expr { diff --git a/src/syntax/markup.rs b/src/syntax/markup.rs deleted file mode 100644 index f43a618a0..000000000 --- a/src/syntax/markup.rs +++ /dev/null @@ -1,159 +0,0 @@ -use super::{Expr, Ident, NodeKind, RedNode, RedRef, Span, TypedNode}; -use crate::node; -use crate::util::EcoString; - -node! { - /// The syntactical root capable of representing a full parsed document. - Markup -} - -impl Markup { - pub fn nodes<'a>(&'a self) -> impl Iterator + 'a { - self.0.children().filter_map(RedRef::cast) - } -} - -/// A single piece of markup. -#[derive(Debug, Clone, PartialEq)] -pub enum MarkupNode { - /// Whitespace containing less than two newlines. - Space, - /// A forced line break: `\`. - Linebreak, - /// A paragraph break: Two or more newlines. - Parbreak, - /// Strong text was enabled / disabled: `*`. - Strong, - /// Emphasized text was enabled / disabled: `_`. - Emph, - /// Plain text. - Text(EcoString), - /// A raw block with optional syntax highlighting: `` `...` ``. - Raw(RawNode), - /// A section heading: `= Introduction`. - Heading(HeadingNode), - /// An item in an unordered list: `- ...`. - List(ListNode), - /// An item in an enumeration (ordered list): `1. ...`. - Enum(EnumNode), - /// An expression. - Expr(Expr), -} - -impl TypedNode for MarkupNode { - fn cast_from(node: RedRef) -> Option { - match node.kind() { - NodeKind::Space(_) => Some(MarkupNode::Space), - NodeKind::Linebreak => Some(MarkupNode::Linebreak), - NodeKind::Parbreak => Some(MarkupNode::Parbreak), - NodeKind::Strong => Some(MarkupNode::Strong), - NodeKind::Emph => Some(MarkupNode::Emph), - NodeKind::Text(s) => Some(MarkupNode::Text(s.clone())), - NodeKind::UnicodeEscape(u) => Some(MarkupNode::Text(u.character.into())), - NodeKind::EnDash => Some(MarkupNode::Text(EcoString::from("\u{2013}"))), - NodeKind::EmDash => Some(MarkupNode::Text(EcoString::from("\u{2014}"))), - NodeKind::NonBreakingSpace => { - Some(MarkupNode::Text(EcoString::from("\u{00A0}"))) - } - NodeKind::Raw(_) => node.cast().map(MarkupNode::Raw), - NodeKind::Heading => node.cast().map(MarkupNode::Heading), - NodeKind::List => node.cast().map(MarkupNode::List), - NodeKind::Enum => node.cast().map(MarkupNode::Enum), - NodeKind::Error(_, _) => None, - _ => node.cast().map(MarkupNode::Expr), - } - } -} - -/// A raw block with optional syntax highlighting: `` `...` ``. -#[derive(Debug, Clone, PartialEq)] -pub struct RawNode { - /// An optional identifier specifying the language to syntax-highlight in. - pub lang: Option, - /// The raw text, determined as the raw string between the backticks trimmed - /// according to the above rules. - pub text: EcoString, - /// Whether the element is block-level, that is, it has 3+ backticks - /// and contains at least one newline. - pub block: bool, -} - -impl TypedNode for RawNode { - fn cast_from(node: RedRef) -> Option { - match node.kind() { - NodeKind::Raw(raw) => { - let span = node.span(); - let start = span.start + raw.backticks as usize; - Some(Self { - block: raw.block, - lang: raw.lang.as_ref().and_then(|x| { - let span = Span::new(span.source, start, start + x.len()); - Ident::new(x, span) - }), - text: raw.text.clone(), - }) - } - _ => None, - } - } -} - -node! { - /// A section heading: `= Introduction`. - Heading => HeadingNode -} - -impl HeadingNode { - /// The contents of the heading. - pub fn body(&self) -> Markup { - self.0 - .cast_first_child() - .expect("heading node is missing markup body") - } - - /// The section depth (numer of equals signs). - pub fn level(&self) -> u8 { - self.0 - .children() - .find_map(|node| match node.kind() { - NodeKind::HeadingLevel(heading) => Some(*heading), - _ => None, - }) - .expect("heading node is missing heading level") - } -} - -node! { - /// An item in an unordered list: `- ...`. - List => ListNode -} - -impl ListNode { - /// The contents of the list item. - pub fn body(&self) -> Markup { - self.0.cast_first_child().expect("list node is missing body") - } -} - -node! { - /// An item in an enumeration (ordered list): `1. ...`. - Enum => EnumNode -} - -impl EnumNode { - /// The contents of the list item. - pub fn body(&self) -> Markup { - self.0.cast_first_child().expect("enumeration node is missing body") - } - - /// The number, if any. - pub fn number(&self) -> Option { - self.0 - .children() - .find_map(|node| match node.kind() { - NodeKind::EnumNumbering(num) => Some(num.clone()), - _ => None, - }) - .expect("enumeration node is missing number") - } -} diff --git a/src/syntax/mod.rs b/src/syntax/mod.rs index 9d4beb6cb..9fd2b21d2 100644 --- a/src/syntax/mod.rs +++ b/src/syntax/mod.rs @@ -1,450 +1,38 @@ //! Syntax types. -mod expr; +mod ast; mod ident; -mod markup; mod pretty; mod span; -mod token; use std::fmt; use std::fmt::{Debug, Display, Formatter}; use std::mem; use std::rc::Rc; -pub use expr::*; +pub use ast::*; pub use ident::*; -pub use markup::*; pub use pretty::*; pub use span::*; -pub use token::*; use crate::geom::{AngularUnit, LengthUnit}; use crate::source::SourceId; use crate::util::EcoString; -#[derive(Debug, Clone, PartialEq)] -pub enum NodeKind { - /// A left square bracket: `[`. - LeftBracket, - /// A right square bracket: `]`. - RightBracket, - /// A left curly brace: `{`. - LeftBrace, - /// A right curly brace: `}`. - RightBrace, - /// A left round parenthesis: `(`. - LeftParen, - /// A right round parenthesis: `)`. - RightParen, - /// An asterisk: `*`. - Star, - /// A comma: `,`. - Comma, - /// A semicolon: `;`. - Semicolon, - /// A colon: `:`. - Colon, - /// A plus: `+`. - Plus, - /// A hyphen: `-`. - Minus, - /// A slash: `/`. - Slash, - /// A single equals sign: `=`. - Eq, - /// Two equals signs: `==`. - EqEq, - /// An exclamation mark followed by an equals sign: `!=`. - ExclEq, - /// A less-than sign: `<`. - Lt, - /// A less-than sign followed by an equals sign: `<=`. - LtEq, - /// A greater-than sign: `>`. - Gt, - /// A greater-than sign followed by an equals sign: `>=`. - GtEq, - /// A plus followed by an equals sign: `+=`. - PlusEq, - /// A hyphen followed by an equals sign: `-=`. - HyphEq, - /// An asterisk followed by an equals sign: `*=`. - StarEq, - /// A slash followed by an equals sign: `/=`. - SlashEq, - /// Two dots: `..`. - Dots, - /// An equals sign followed by a greater-than sign: `=>`. - Arrow, - /// The `not` operator. - Not, - /// The `and` operator. - And, - /// The `or` operator. - Or, - /// The `with` operator. - With, - /// The `with` expression: `with (1)`. - WithExpr, - /// The none literal: `none`. - None, - /// The auto literal: `auto`. - Auto, - /// The `let` keyword. - Let, - /// The `if` keyword. - If, - /// The `else` keyword. - Else, - /// The `for` keyword. - For, - /// The `in` keyword. - In, - /// The `while` keyword. - While, - /// The `break` keyword. - Break, - /// The `continue` keyword. - Continue, - /// The `return` keyword. - Return, - /// The `import` keyword. - Import, - /// The `include` keyword. - Include, - /// The `from` keyword. - From, - /// One or more whitespace characters. - Space(usize), - /// A consecutive non-markup string. - Text(EcoString), - /// A slash and the letter "u" followed by a hexadecimal unicode entity - /// enclosed in curly braces: `\u{1F5FA}`. - UnicodeEscape(UnicodeEscapeToken), - /// An arbitrary number of backticks followed by inner contents, terminated - /// with the same number of backticks: `` `...` ``. - Raw(Rc), - /// Dollar signs surrounding inner contents. - Math(Rc), - /// A numbering: `23.`. - /// - /// Can also exist without the number: `.`. - EnumNumbering(Option), - /// An identifier: `center`. - Ident(EcoString), - /// A boolean: `true`, `false`. - Bool(bool), - /// An integer: `120`. - Int(i64), - /// A floating-point number: `1.2`, `10e-4`. - Float(f64), - /// A length: `12pt`, `3cm`. - Length(f64, LengthUnit), - /// An angle: `90deg`. - Angle(f64, AngularUnit), - /// A percentage: `50%`. - /// - /// _Note_: `50%` is stored as `50.0` here, as in the corresponding - /// [literal](super::Lit::Percent). - Percentage(f64), - /// A fraction unit: `3fr`. - Fraction(f64), - /// A quoted string: `"..."`. - Str(StrToken), - /// Two slashes followed by inner contents, terminated with a newline: - /// `//\n`. - LineComment, - /// A slash and a star followed by inner contents, terminated with a star - /// and a slash: `/**/`. - /// - /// The comment can contain nested block comments. - BlockComment, - /// Tokens that appear in the wrong place. - Error(ErrorPosition, EcoString), - /// Unknown character sequences. - Unknown(EcoString), - /// Template markup. - Markup, - /// A forced line break: `\`. - Linebreak, - /// A paragraph break: Two or more newlines. - Parbreak, - /// Strong text was enabled / disabled: `*`. - Strong, - /// Emphasized text was enabled / disabled: `_`. - Emph, - /// A non-breaking space: `~`. - NonBreakingSpace, - /// An en-dash: `--`. - EnDash, - /// An em-dash: `---`. - EmDash, - /// A section heading: `= Introduction`. - Heading, - /// A heading's level: `=`, `==`, `===`, etc. - HeadingLevel(u8), - /// An item in an unordered list: `- ...`. - List, - /// The bullet character of an item in an unordered list: `-`. - ListBullet, - /// An item in an enumeration (ordered list): `1. ...`. - Enum, - /// An array expression: `(1, "hi", 12cm)`. - Array, - /// A dictionary expression: `(thickness: 3pt, pattern: dashed)`. - Dict, - /// A named argument: `thickness: 3pt`. - Named, - /// A template expression: `[*Hi* there!]`. - Template, - /// A grouped expression: `(1 + 2)`. - Group, - /// A block expression: `{ let x = 1; x + 2 }`. - Block, - /// A unary operation: `-x`. - Unary, - /// A binary operation: `a + b`. - Binary, - /// An invocation of a function: `f(x, y)`. - Call, - /// A function call's argument list: `(x, y)`. - CallArgs, - /// A closure expression: `(x, y) => z`. - Closure, - /// A closure's parameters: `(x, y)`. - ClosureParams, - /// A parameter sink: `..x`. - ParameterSink, - /// A for loop expression: `for x in y { ... }`. - ForExpr, - /// A while loop expression: `while x { ... }`. - WhileExpr, - /// An if expression: `if x { ... }`. - IfExpr, - /// A let expression: `let x = 1`. - LetExpr, - /// A for loop's destructuring pattern: `x` or `x, y`. - ForPattern, - /// The import expression: `import x from "foo.typ"`. - ImportExpr, - /// Items to import: `a, b, c`. - ImportItems, - /// The include expression: `include "foo.typ"`. - IncludeExpr, -} - -#[derive(Debug, Copy, Clone, PartialEq, Eq)] -pub enum ErrorPosition { - /// At the start of the node. - Start, - /// Over the full width of the node. - Full, - /// At the end of the node. - End, -} - -impl Display for NodeKind { - fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result { - f.pad(self.as_str()) - } -} - -impl NodeKind { - pub fn is_parenthesis(&self) -> bool { - match self { - Self::LeftParen => true, - Self::RightParen => true, - _ => false, - } - } - - pub fn is_bracket(&self) -> bool { - match self { - Self::LeftBracket => true, - Self::RightBracket => true, - _ => false, - } - } - - pub fn is_brace(&self) -> bool { - match self { - Self::LeftBrace => true, - Self::RightBrace => true, - _ => false, - } - } - - pub fn is_error(&self) -> bool { - matches!(self, NodeKind::Error(_, _)) - } - - pub fn as_str(&self) -> &'static str { - match self { - Self::LeftBracket => "opening bracket", - Self::RightBracket => "closing bracket", - Self::LeftBrace => "opening brace", - Self::RightBrace => "closing brace", - Self::LeftParen => "opening paren", - Self::RightParen => "closing paren", - Self::Star => "star", - Self::Comma => "comma", - Self::Semicolon => "semicolon", - Self::Colon => "colon", - Self::Plus => "plus", - Self::Minus => "minus", - Self::Slash => "slash", - Self::Eq => "assignment operator", - Self::EqEq => "equality operator", - Self::ExclEq => "inequality operator", - Self::Lt => "less-than operator", - Self::LtEq => "less-than or equal operator", - Self::Gt => "greater-than operator", - Self::GtEq => "greater-than or equal operator", - Self::PlusEq => "add-assign operator", - Self::HyphEq => "subtract-assign operator", - Self::StarEq => "multiply-assign operator", - Self::SlashEq => "divide-assign operator", - Self::Dots => "dots", - Self::Arrow => "arrow", - Self::Not => "operator `not`", - Self::And => "operator `and`", - Self::Or => "operator `or`", - Self::With => "operator `with`", - Self::WithExpr => "`with` expression", - Self::None => "`none`", - Self::Auto => "`auto`", - Self::Let => "keyword `let`", - Self::If => "keyword `if`", - Self::Else => "keyword `else`", - Self::For => "keyword `for`", - Self::In => "keyword `in`", - Self::While => "keyword `while`", - Self::Break => "keyword `break`", - Self::Continue => "keyword `continue`", - Self::Return => "keyword `return`", - Self::Import => "keyword `import`", - Self::Include => "keyword `include`", - Self::From => "keyword `from`", - Self::Space(_) => "space", - Self::Math(_) => "math formula", - Self::EnumNumbering(_) => "numbering", - Self::Str(_) => "string", - Self::LineComment => "line comment", - Self::BlockComment => "block comment", - Self::Markup => "markup", - Self::Linebreak => "forced linebreak", - Self::Parbreak => "paragraph break", - Self::Strong => "strong", - Self::Emph => "emphasis", - Self::Text(_) => "text", - Self::NonBreakingSpace => "non-breaking space", - Self::EnDash => "en dash", - Self::EmDash => "em dash", - Self::UnicodeEscape(_) => "unicode escape sequence", - Self::Raw(_) => "raw block", - Self::Heading => "heading", - Self::HeadingLevel(_) => "heading level", - Self::List => "list", - Self::ListBullet => "list bullet", - Self::Enum => "enum", - Self::Ident(_) => "identifier", - Self::Bool(_) => "boolean", - Self::Int(_) => "integer", - Self::Float(_) => "float", - Self::Length(_, _) => "length", - Self::Angle(_, _) => "angle", - Self::Percentage(_) => "percentage", - Self::Fraction(_) => "`fr` value", - Self::Array => "array", - Self::Dict => "dictionary", - Self::Named => "named argument", - Self::Template => "template", - Self::Group => "group", - Self::Block => "block", - Self::Unary => "unary expression", - Self::Binary => "binary expression", - Self::Call => "call", - Self::CallArgs => "call arguments", - Self::Closure => "closure", - Self::ClosureParams => "closure parameters", - Self::ParameterSink => "parameter sink", - Self::ForExpr => "for-loop expression", - Self::WhileExpr => "while-loop expression", - Self::IfExpr => "if expression", - Self::LetExpr => "let expression", - Self::ForPattern => "for-loop destructuring pattern", - Self::ImportExpr => "import expression", - Self::ImportItems => "import items", - Self::IncludeExpr => "include expression", - Self::Unknown(src) => match src.as_str() { - "*/" => "end of block comment", - _ => "invalid token", - }, - Self::Error(_, _) => "parse error", - } - } -} - -/// A syntactical node. -#[derive(Clone, PartialEq)] -pub struct GreenNode { - /// Node metadata. - data: GreenData, - /// This node's children, losslessly make up this node. - children: Vec, -} - -/// Data shared between [`GreenNode`]s and [`GreenToken`]s. -#[derive(Clone, PartialEq)] -pub struct GreenData { - /// What kind of node this is (each kind would have its own struct in a - /// strongly typed AST). - kind: NodeKind, - /// The byte length of the node in the source. - len: usize, - /// Whether this node or any of its children are erroneous. - erroneous: bool, -} - -impl GreenData { - pub fn new(kind: NodeKind, len: usize) -> Self { - Self { len, erroneous: kind.is_error(), kind } - } - - pub fn kind(&self) -> &NodeKind { - &self.kind - } - - pub fn len(&self) -> usize { - self.len - } - - pub fn erroneous(&self) -> bool { - self.erroneous - } -} - -impl From for Green { - fn from(token: GreenData) -> Self { - Self::Token(token) - } -} - /// Children of a [`GreenNode`]. #[derive(Clone, PartialEq)] pub enum Green { - /// A terminal owned token. - Token(GreenData), /// A non-terminal node in an Rc. Node(Rc), + /// A terminal owned token. + Token(GreenData), } impl Green { fn data(&self) -> &GreenData { match self { - Green::Token(t) => &t, Green::Node(n) => &n.data, + Green::Token(t) => &t, } } @@ -462,12 +50,41 @@ impl Green { pub fn children(&self) -> &[Green] { match self { - Green::Token(_) => &[], Green::Node(n) => &n.children(), + Green::Token(_) => &[], } } } +impl Default for Green { + fn default() -> Self { + Self::Token(GreenData::new(NodeKind::None, 0)) + } +} + +impl Debug for Green { + fn fmt(&self, f: &mut Formatter) -> fmt::Result { + write!(f, "{:?}: {}", self.kind(), self.len())?; + if let Self::Node(n) = self { + if !n.children.is_empty() { + f.write_str(" ")?; + f.debug_list().entries(&n.children).finish()?; + } + } + + Ok(()) + } +} + +/// A syntactical node. +#[derive(Clone, PartialEq)] +pub struct GreenNode { + /// Node metadata. + data: GreenData, + /// This node's children, losslessly make up this node. + children: Vec, +} + impl GreenNode { pub fn new(kind: NodeKind, len: usize) -> Self { Self { @@ -503,23 +120,39 @@ impl From> for Green { } } -impl Default for Green { - fn default() -> Self { - Self::Token(GreenData::new(NodeKind::None, 0)) +/// Data shared between [`GreenNode`]s and [`GreenToken`]s. +#[derive(Clone, PartialEq)] +pub struct GreenData { + /// What kind of node this is (each kind would have its own struct in a + /// strongly typed AST). + kind: NodeKind, + /// The byte length of the node in the source. + len: usize, + /// Whether this node or any of its children are erroneous. + erroneous: bool, +} + +impl GreenData { + pub fn new(kind: NodeKind, len: usize) -> Self { + Self { len, erroneous: kind.is_error(), kind } + } + + pub fn kind(&self) -> &NodeKind { + &self.kind + } + + pub fn len(&self) -> usize { + self.len + } + + pub fn erroneous(&self) -> bool { + self.erroneous } } -impl Debug for Green { - fn fmt(&self, f: &mut Formatter) -> fmt::Result { - write!(f, "{:?}: {}", self.kind(), self.len())?; - if let Self::Node(n) = self { - if !n.children.is_empty() { - f.write_str(" ")?; - f.debug_list().entries(&n.children).finish()?; - } - } - - Ok(()) +impl From for Green { + fn from(token: GreenData) -> Self { + Self::Token(token) } } @@ -678,6 +311,408 @@ pub trait TypedNode: Sized { fn cast_from(value: RedRef) -> Option; } +#[derive(Debug, Clone, PartialEq)] +pub enum NodeKind { + /// A left square bracket: `[`. + LeftBracket, + /// A right square bracket: `]`. + RightBracket, + /// A left curly brace: `{`. + LeftBrace, + /// A right curly brace: `}`. + RightBrace, + /// A left round parenthesis: `(`. + LeftParen, + /// A right round parenthesis: `)`. + RightParen, + /// An asterisk: `*`. + Star, + /// A comma: `,`. + Comma, + /// A semicolon: `;`. + Semicolon, + /// A colon: `:`. + Colon, + /// A plus: `+`. + Plus, + /// A hyphen: `-`. + Minus, + /// A slash: `/`. + Slash, + /// A single equals sign: `=`. + Eq, + /// Two equals signs: `==`. + EqEq, + /// An exclamation mark followed by an equals sign: `!=`. + ExclEq, + /// A less-than sign: `<`. + Lt, + /// A less-than sign followed by an equals sign: `<=`. + LtEq, + /// A greater-than sign: `>`. + Gt, + /// A greater-than sign followed by an equals sign: `>=`. + GtEq, + /// A plus followed by an equals sign: `+=`. + PlusEq, + /// A hyphen followed by an equals sign: `-=`. + HyphEq, + /// An asterisk followed by an equals sign: `*=`. + StarEq, + /// A slash followed by an equals sign: `/=`. + SlashEq, + /// The `not` operator. + Not, + /// The `and` operator. + And, + /// The `or` operator. + Or, + /// The `with` operator. + With, + /// Two dots: `..`. + Dots, + /// An equals sign followed by a greater-than sign: `=>`. + Arrow, + /// The none literal: `none`. + None, + /// The auto literal: `auto`. + Auto, + /// The `let` keyword. + Let, + /// The `if` keyword. + If, + /// The `else` keyword. + Else, + /// The `for` keyword. + For, + /// The `in` keyword. + In, + /// The `while` keyword. + While, + /// The `break` keyword. + Break, + /// The `continue` keyword. + Continue, + /// The `return` keyword. + Return, + /// The `import` keyword. + Import, + /// The `include` keyword. + Include, + /// The `from` keyword. + From, + /// Template markup. + Markup, + /// One or more whitespace characters. + Space(usize), + /// A forced line break: `\`. + Linebreak, + /// A paragraph break: Two or more newlines. + Parbreak, + /// A consecutive non-markup string. + Text(EcoString), + /// A non-breaking space: `~`. + NonBreakingSpace, + /// An en-dash: `--`. + EnDash, + /// An em-dash: `---`. + EmDash, + /// A slash and the letter "u" followed by a hexadecimal unicode entity + /// enclosed in curly braces: `\u{1F5FA}`. + UnicodeEscape(UnicodeEscapeToken), + /// Strong text was enabled / disabled: `*`. + Strong, + /// Emphasized text was enabled / disabled: `_`. + Emph, + /// A section heading: `= Introduction`. + Heading, + /// A heading's level: `=`, `==`, `===`, etc. + HeadingLevel(u8), + /// An item in an enumeration (ordered list): `1. ...`. + Enum, + /// A numbering: `23.`. + /// + /// Can also exist without the number: `.`. + EnumNumbering(Option), + /// An item in an unordered list: `- ...`. + List, + /// The bullet character of an item in an unordered list: `-`. + ListBullet, + /// An arbitrary number of backticks followed by inner contents, terminated + /// with the same number of backticks: `` `...` ``. + Raw(Rc), + /// Dollar signs surrounding inner contents. + Math(Rc), + /// An identifier: `center`. + Ident(EcoString), + /// A boolean: `true`, `false`. + Bool(bool), + /// An integer: `120`. + Int(i64), + /// A floating-point number: `1.2`, `10e-4`. + Float(f64), + /// A length: `12pt`, `3cm`. + Length(f64, LengthUnit), + /// An angle: `90deg`. + Angle(f64, AngularUnit), + /// A percentage: `50%`. + /// + /// _Note_: `50%` is stored as `50.0` here, as in the corresponding + /// [literal](super::Lit::Percent). + Percentage(f64), + /// A fraction unit: `3fr`. + Fraction(f64), + /// A quoted string: `"..."`. + Str(StrToken), + /// An array expression: `(1, "hi", 12cm)`. + Array, + /// A dictionary expression: `(thickness: 3pt, pattern: dashed)`. + Dict, + /// A named argument: `thickness: 3pt`. + Named, + /// A grouped expression: `(1 + 2)`. + Group, + /// A unary operation: `-x`. + Unary, + /// A binary operation: `a + b`. + Binary, + /// An invocation of a function: `f(x, y)`. + Call, + /// A function call's argument list: `(x, y)`. + CallArgs, + /// A closure expression: `(x, y) => z`. + Closure, + /// A closure's parameters: `(x, y)`. + ClosureParams, + /// A parameter sink: `..x`. + ParameterSink, + /// A template expression: `[*Hi* there!]`. + Template, + /// A block expression: `{ let x = 1; x + 2 }`. + Block, + /// A for loop expression: `for x in y { ... }`. + ForExpr, + /// A while loop expression: `while x { ... }`. + WhileExpr, + /// An if expression: `if x { ... }`. + IfExpr, + /// A let expression: `let x = 1`. + LetExpr, + /// The `with` expression: `with (1)`. + WithExpr, + /// A for loop's destructuring pattern: `x` or `x, y`. + ForPattern, + /// The import expression: `import x from "foo.typ"`. + ImportExpr, + /// Items to import: `a, b, c`. + ImportItems, + /// The include expression: `include "foo.typ"`. + IncludeExpr, + /// Two slashes followed by inner contents, terminated with a newline: + /// `//\n`. + LineComment, + /// A slash and a star followed by inner contents, terminated with a star + /// and a slash: `/**/`. + /// + /// The comment can contain nested block comments. + BlockComment, + /// Tokens that appear in the wrong place. + Error(ErrorPosition, EcoString), + /// Unknown character sequences. + Unknown(EcoString), +} + +#[derive(Debug, Copy, Clone, PartialEq, Eq)] +pub enum ErrorPosition { + /// At the start of the node. + Start, + /// Over the full width of the node. + Full, + /// At the end of the node. + End, +} + +/// A quoted string token: `"..."`. +#[derive(Debug, Clone, PartialEq)] +#[repr(transparent)] +pub struct StrToken { + /// The string inside the quotes. + pub string: EcoString, +} + +/// A raw block token: `` `...` ``. +#[derive(Debug, Clone, PartialEq)] +pub struct RawToken { + /// The raw text in the block. + pub text: EcoString, + /// The programming language of the raw text. + pub lang: Option, + /// The number of opening backticks. + pub backticks: u8, + /// Whether to display this as a block. + pub block: bool, +} + +/// A math formula token: `$2pi + x$` or `$[f'(x) = x^2]$`. +#[derive(Debug, Clone, PartialEq)] +pub struct MathToken { + /// The formula between the dollars. + pub formula: EcoString, + /// Whether the formula is display-level, that is, it is surrounded by + /// `$[..]`. + pub display: bool, +} + +/// A unicode escape sequence token: `\u{1F5FA}`. +#[derive(Debug, Clone, PartialEq)] +#[repr(transparent)] +pub struct UnicodeEscapeToken { + /// The resulting unicode character. + pub character: char, +} + +impl Display for NodeKind { + fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result { + f.pad(self.as_str()) + } +} + +impl NodeKind { + pub fn is_paren(&self) -> bool { + match self { + Self::LeftParen => true, + Self::RightParen => true, + _ => false, + } + } + + pub fn is_bracket(&self) -> bool { + match self { + Self::LeftBracket => true, + Self::RightBracket => true, + _ => false, + } + } + + pub fn is_brace(&self) -> bool { + match self { + Self::LeftBrace => true, + Self::RightBrace => true, + _ => false, + } + } + + pub fn is_error(&self) -> bool { + matches!(self, NodeKind::Error(_, _)) + } + + pub fn as_str(&self) -> &'static str { + match self { + Self::LeftBracket => "opening bracket", + Self::RightBracket => "closing bracket", + Self::LeftBrace => "opening brace", + Self::RightBrace => "closing brace", + Self::LeftParen => "opening paren", + Self::RightParen => "closing paren", + Self::Star => "star", + Self::Comma => "comma", + Self::Semicolon => "semicolon", + Self::Colon => "colon", + Self::Plus => "plus", + Self::Minus => "minus", + Self::Slash => "slash", + Self::Eq => "assignment operator", + Self::EqEq => "equality operator", + Self::ExclEq => "inequality operator", + Self::Lt => "less-than operator", + Self::LtEq => "less-than or equal operator", + Self::Gt => "greater-than operator", + Self::GtEq => "greater-than or equal operator", + Self::PlusEq => "add-assign operator", + Self::HyphEq => "subtract-assign operator", + Self::StarEq => "multiply-assign operator", + Self::SlashEq => "divide-assign operator", + Self::Not => "operator `not`", + Self::And => "operator `and`", + Self::Or => "operator `or`", + Self::With => "operator `with`", + Self::Dots => "dots", + Self::Arrow => "arrow", + Self::None => "`none`", + Self::Auto => "`auto`", + Self::Let => "keyword `let`", + Self::If => "keyword `if`", + Self::Else => "keyword `else`", + Self::For => "keyword `for`", + Self::In => "keyword `in`", + Self::While => "keyword `while`", + Self::Break => "keyword `break`", + Self::Continue => "keyword `continue`", + Self::Return => "keyword `return`", + Self::Import => "keyword `import`", + Self::Include => "keyword `include`", + Self::From => "keyword `from`", + Self::Markup => "markup", + Self::Space(_) => "space", + Self::Linebreak => "forced linebreak", + Self::Parbreak => "paragraph break", + Self::Text(_) => "text", + Self::NonBreakingSpace => "non-breaking space", + Self::EnDash => "en dash", + Self::EmDash => "em dash", + Self::UnicodeEscape(_) => "unicode escape sequence", + Self::Strong => "strong", + Self::Emph => "emphasis", + Self::Heading => "heading", + Self::HeadingLevel(_) => "heading level", + Self::Enum => "enumeration item", + Self::EnumNumbering(_) => "enumeration item numbering", + Self::List => "list item", + Self::ListBullet => "list bullet", + Self::Raw(_) => "raw block", + Self::Math(_) => "math formula", + Self::Ident(_) => "identifier", + Self::Bool(_) => "boolean", + Self::Int(_) => "integer", + Self::Float(_) => "float", + Self::Length(_, _) => "length", + Self::Angle(_, _) => "angle", + Self::Percentage(_) => "percentage", + Self::Fraction(_) => "`fr` value", + Self::Str(_) => "string", + Self::Array => "array", + Self::Dict => "dictionary", + Self::Named => "named argument", + Self::Group => "group", + Self::Unary => "unary expression", + Self::Binary => "binary expression", + Self::Call => "call", + Self::CallArgs => "call arguments", + Self::Closure => "closure", + Self::ClosureParams => "closure parameters", + Self::ParameterSink => "parameter sink", + Self::Template => "template", + Self::Block => "block", + Self::ForExpr => "for-loop expression", + Self::WhileExpr => "while-loop expression", + Self::IfExpr => "`if` expression", + Self::LetExpr => "`let` expression", + Self::WithExpr => "`with` expression", + Self::ForPattern => "for-loop destructuring pattern", + Self::ImportExpr => "`import` expression", + Self::ImportItems => "import items", + Self::IncludeExpr => "`include` expression", + Self::LineComment => "line comment", + Self::BlockComment => "block comment", + Self::Error(_, _) => "parse error", + Self::Unknown(src) => match src.as_str() { + "*/" => "end of block comment", + _ => "invalid token", + }, + } + } +} + #[macro_export] macro_rules! node { ($(#[$attr:meta])* $name:ident) => { diff --git a/src/syntax/token.rs b/src/syntax/token.rs deleted file mode 100644 index 4f43bb4f5..000000000 --- a/src/syntax/token.rs +++ /dev/null @@ -1,40 +0,0 @@ -use crate::util::EcoString; - -/// A quoted string token: `"..."`. -#[derive(Debug, Clone, PartialEq)] -#[repr(transparent)] -pub struct StrToken { - /// The string inside the quotes. - pub string: EcoString, -} - -/// A raw block token: `` `...` ``. -#[derive(Debug, Clone, PartialEq)] -pub struct RawToken { - /// The raw text in the block. - pub text: EcoString, - /// The programming language of the raw text. - pub lang: Option, - /// The number of opening backticks. - pub backticks: u8, - /// Whether to display this as a block. - pub block: bool, -} - -/// A math formula token: `$2pi + x$` or `$[f'(x) = x^2]$`. -#[derive(Debug, Clone, PartialEq)] -pub struct MathToken { - /// The formula between the dollars. - pub formula: EcoString, - /// Whether the formula is display-level, that is, it is surrounded by - /// `$[..]`. - pub display: bool, -} - -/// A unicode escape sequence token: `\u{1F5FA}`. -#[derive(Debug, Clone, PartialEq)] -#[repr(transparent)] -pub struct UnicodeEscapeToken { - /// The resulting unicode character. - pub character: char, -}