Unicode escape error moved to tokenizer

6 files changed, 78 insertions, 91 deletions

diff --git a/src/parse/mod.rs b/src/parse/mod.rs
index ce992834..8775e8a1 100644
--- a/src/parse/mod.rs
+++ b/src/parse/mod.rs
@@ -54,7 +54,10 @@ where
     while !p.eof() && f(p) {
         markup_node(p, &mut at_start);
         if let Some(node) = p.last_child() {
-            at_start &= matches!(node.kind(), &NodeKind::Space(_) | &NodeKind::Parbreak | &NodeKind::LineComment | &NodeKind::BlockComment);
+            at_start &= matches!(node.kind(),
+                &NodeKind::Space(_) | &NodeKind::Parbreak |
+                &NodeKind::LineComment | &NodeKind::BlockComment
+            );
         }
     }
 
@@ -88,22 +91,8 @@ fn markup_node(p: &mut Parser, at_start: &mut bool) {
         | NodeKind::Emph
         | NodeKind::Strong
         | NodeKind::Linebreak
-        | NodeKind::Raw(_) => p.eat(),
-
-        NodeKind::UnicodeEscape(u) => {
-            if u.character.is_none() {
-                let src = p.peek_src();
-                p.convert(NodeKind::Error(
-                    ErrorPosition::Full,
-                    "invalid unicode escape sequence".into(),
-                ));
-                p.start();
-                p.end(NodeKind::Text(src.into()));
-                return;
-            }
-
-            p.eat();
-        }
+        | NodeKind::Raw(_)
+        | NodeKind::UnicodeEscape(_) => p.eat(),
 
         NodeKind::Eq if *at_start => heading(p),
         NodeKind::ListBullet if *at_start => list_node(p),
@@ -503,9 +492,8 @@ fn item(p: &mut Parser) -> NodeKind {
 /// Convert a collection into an array, producing errors for anything other than
 /// expressions.
 fn array(p: &mut Parser, items: usize) {
-    p.start_with(items);
     p.filter_children(
-        0,
+        p.child_count() - items,
         |x| match x.kind() {
             NodeKind::Named | NodeKind::ParameterSink => false,
             _ => true,
@@ -522,15 +510,14 @@ fn array(p: &mut Parser, items: usize) {
         },
     );
 
-    p.end(NodeKind::Array)
+    p.convert_with(items, NodeKind::Array);
 }
 
 /// Convert a collection into a dictionary, producing errors for anything other
 /// than named pairs.
 fn dict(p: &mut Parser, items: usize) {
-    p.start_with(items);
     p.filter_children(
-        0,
+        p.child_count() - items,
         |x| {
             x.kind() == &NodeKind::Named
                 || x.kind().is_parenthesis()
@@ -547,7 +534,7 @@ fn dict(p: &mut Parser, items: usize) {
             ),
         },
     );
-    p.end(NodeKind::Dict);
+    p.convert_with(items, NodeKind::Dict);
 }
 
 /// Convert a collection into a list of parameters, producing errors for
@@ -684,8 +671,7 @@ fn let_expr(p: &mut Parser) {
                 return;
             }
 
-            p.start_with(p.child_count() - offset);
-            p.end(NodeKind::Closure)
+            p.convert_with(p.child_count() - offset, NodeKind::Closure);
         }
     }
 
diff --git a/src/parse/parser.rs b/src/parse/parser.rs
index e6fcc1ae..240de43d 100644
--- a/src/parse/parser.rs
+++ b/src/parse/parser.rs
@@ -186,9 +186,27 @@ impl<'s> Parser<'s> {
     }
 
     pub fn convert(&mut self, kind: NodeKind) {
-        self.start();
-        self.eat();
-        self.end(kind);
+        let len = self.tokens.index() - self.next_start;
+
+        self.children.push(
+            GreenNode::with_child(
+                kind,
+                len,
+                GreenData::new(self.next.clone().unwrap(), len),
+            )
+            .into(),
+        );
+        self.fast_forward();
+        self.success = true;
+    }
+
+    pub fn convert_with(&mut self, preserve: usize, kind: NodeKind) {
+        let preserved: Vec<_> =
+            self.children.drain(self.children.len() - preserve ..).collect();
+        let len = preserved.iter().map(|c| c.len()).sum();
+        self.children
+            .push(GreenNode::with_children(kind, len, preserved).into());
+        self.success = true;
     }
 
     /// End the current node and undo its existence, inling all accumulated
diff --git a/src/parse/tokens.rs b/src/parse/tokens.rs
index 7c500ce7..1d2e32ec 100644
--- a/src/parse/tokens.rs
+++ b/src/parse/tokens.rs
@@ -200,7 +200,7 @@ impl<'s> Tokens<'s> {
             TABLE.get(c as usize).copied().unwrap_or_else(|| c.is_whitespace())
         });
 
-        NodeKind::Text(resolve_string(self.s.eaten_from(start)))
+        NodeKind::Text(self.s.eaten_from(start).into())
     }
 
     fn whitespace(&mut self) -> NodeKind {
@@ -243,10 +243,16 @@ impl<'s> Tokens<'s> {
                     let sequence: EcoString = self.s.eat_while(|c| c.is_ascii_alphanumeric()).into();
 
                     if self.s.eat_if('}') {
-                        NodeKind::UnicodeEscape(Rc::new(UnicodeEscapeToken {
-                            character: resolve_hex(&sequence),
-                            sequence,
-                        }))
+                        if let Some(character) = resolve_hex(&sequence) {
+                            NodeKind::UnicodeEscape(UnicodeEscapeToken {
+                                character,
+                            })
+                        } else {
+                            NodeKind::Error(
+                                ErrorPosition::Full,
+                                "invalid unicode escape sequence".into(),
+                            )
+                        }
                     } else {
                         NodeKind::Error(
                             ErrorPosition::End,
@@ -560,35 +566,21 @@ mod tests {
     use Option::None;
     use TokenMode::{Code, Markup};
 
-    fn UnicodeEscape(sequence: &str, terminated: bool) -> NodeKind {
-        if terminated {
-            NodeKind::UnicodeEscape(Rc::new(UnicodeEscapeToken {
-                character: resolve_hex(sequence),
-                sequence: sequence.into(),
-            }))
-        } else {
-            NodeKind::Error(ErrorPosition::End, "expected closing brace".into())
-        }
+    fn UnicodeEscape(character: char) -> NodeKind {
+        NodeKind::UnicodeEscape(UnicodeEscapeToken { character })
     }
 
-    fn Raw(
-        text: &str,
-        lang: Option<&str>,
-        backticks_left: u8,
-        err_msg: Option<&str>,
-        block: bool,
-    ) -> NodeKind {
-        match err_msg {
-            None => NodeKind::Raw(Rc::new(RawToken {
-                text: text.into(),
-                lang: lang.map(Into::into),
-                backticks: backticks_left,
-                block,
-            })),
-            Some(msg) => {
-                NodeKind::Error(ErrorPosition::End, format!("expected {}", msg).into())
-            }
-        }
+    fn Error(pos: ErrorPosition, message: &str) -> NodeKind {
+        NodeKind::Error(pos, message.into())
+    }
+
+    fn Raw(text: &str, lang: Option<&str>, backticks_left: u8, block: bool) -> NodeKind {
+        NodeKind::Raw(Rc::new(RawToken {
+            text: text.into(),
+            lang: lang.map(Into::into),
+            backticks: backticks_left,
+            block,
+        }))
     }
 
     fn Math(formula: &str, display: bool, err_msg: Option<&str>) -> NodeKind {
@@ -795,16 +787,16 @@ mod tests {
         t!(Markup[" /"]: r#"\""# => Text(r"\"), Text("\""));
 
         // Test basic unicode escapes.
-        t!(Markup: r"\u{}"     => UnicodeEscape("", true));
-        t!(Markup: r"\u{2603}" => UnicodeEscape("2603", true));
-        t!(Markup: r"\u{P}"    => UnicodeEscape("P", true));
+        t!(Markup: r"\u{}"     => Error(ErrorPosition::Full, "invalid unicode escape sequence"));
+        t!(Markup: r"\u{2603}" => UnicodeEscape('☃'));
+        t!(Markup: r"\u{P}"    => Error(ErrorPosition::Full, "invalid unicode escape sequence"));
 
         // Test unclosed unicode escapes.
-        t!(Markup[" /"]: r"\u{"     => UnicodeEscape("", false));
-        t!(Markup[" /"]: r"\u{1"    => UnicodeEscape("1", false));
-        t!(Markup[" /"]: r"\u{26A4" => UnicodeEscape("26A4", false));
-        t!(Markup[" /"]: r"\u{1Q3P" => UnicodeEscape("1Q3P", false));
-        t!(Markup: r"\u{1🏕}"       => UnicodeEscape("1", false), Text("🏕"), RightBrace);
+        t!(Markup[" /"]: r"\u{"     => Error(ErrorPosition::End, "expected closing brace"));
+        t!(Markup[" /"]: r"\u{1"    => Error(ErrorPosition::End, "expected closing brace"));
+        t!(Markup[" /"]: r"\u{26A4" => Error(ErrorPosition::End, "expected closing brace"));
+        t!(Markup[" /"]: r"\u{1Q3P" => Error(ErrorPosition::End, "expected closing brace"));
+        t!(Markup: r"\u{1🏕}"       => Error(ErrorPosition::End, "expected closing brace"), Text("🏕"), RightBrace);
     }
 
     #[test]
@@ -894,22 +886,22 @@ mod tests {
     #[test]
     fn test_tokenize_raw_blocks() {
         // Test basic raw block.
-        t!(Markup: "``"     => Raw("", None, 1, None, false));
-        t!(Markup: "`raw`"  => Raw("raw", None, 1, None, false));
-        t!(Markup[""]: "`]" => Raw("]", None, 1, Some("1 backtick"), false));
+        t!(Markup: "``"     => Raw("", None, 1, false));
+        t!(Markup: "`raw`"  => Raw("raw", None, 1, false));
+        t!(Markup[""]: "`]" => Error(ErrorPosition::End, "expected 1 backtick"));
 
         // Test special symbols in raw block.
-        t!(Markup: "`[brackets]`" => Raw("[brackets]", None, 1, None, false));
-        t!(Markup[""]: r"`\`` "   => Raw(r"\", None, 1, None, false), Raw(" ", None, 1, Some("1 backtick"), false));
+        t!(Markup: "`[brackets]`" => Raw("[brackets]", None, 1, false));
+        t!(Markup[""]: r"`\`` "   => Raw(r"\", None, 1, false), Error(ErrorPosition::End, "expected 1 backtick"));
 
         // Test separated closing backticks.
-        t!(Markup: "```not `y`e`t```" => Raw("`y`e`t", Some("not"), 3, None, false));
+        t!(Markup: "```not `y`e`t```" => Raw("`y`e`t", Some("not"), 3, false));
 
         // Test more backticks.
-        t!(Markup: "``nope``"             => Raw("", None, 1, None, false), Text("nope"), Raw("", None, 1, None, false));
-        t!(Markup: "````🚀````"           => Raw("", Some("🚀"), 4, None, false));
-        t!(Markup[""]: "`````👩‍🚀````noend" => Raw("````noend", Some("👩‍🚀"), 5, Some("5 backticks"), false));
-        t!(Markup[""]: "````raw``````"    => Raw("", Some("raw"), 4, None, false), Raw("", None, 1, None, false));
+        t!(Markup: "``nope``"             => Raw("", None, 1, false), Text("nope"), Raw("", None, 1, false));
+        t!(Markup: "````🚀````"           => Raw("", Some("🚀"), 4, false));
+        t!(Markup[""]: "`````👩‍🚀````noend" => Error(ErrorPosition::End, "expected 5 backticks"));
+        t!(Markup[""]: "````raw``````"    => Raw("", Some("raw"), 4, false), Raw("", None, 1, false));
     }
 
     #[test]
diff --git a/src/syntax/markup.rs b/src/syntax/markup.rs
index 49b2a519..f43a618a 100644
--- a/src/syntax/markup.rs
+++ b/src/syntax/markup.rs
@@ -1,7 +1,6 @@
 use super::{Expr, Ident, NodeKind, RedNode, RedRef, Span, TypedNode};
 use crate::node;
 use crate::util::EcoString;
-use std::fmt::Write;
 
 node! {
     /// The syntactical root capable of representing a full parsed document.
@@ -50,14 +49,7 @@ impl TypedNode for MarkupNode {
             NodeKind::Strong => Some(MarkupNode::Strong),
             NodeKind::Emph => Some(MarkupNode::Emph),
             NodeKind::Text(s) => Some(MarkupNode::Text(s.clone())),
-            NodeKind::UnicodeEscape(u) => Some(MarkupNode::Text(match u.character {
-                Some(c) => c.into(),
-                None => {
-                    let mut eco = EcoString::with_capacity(u.sequence.len() + 4);
-                    write!(&mut eco, "\\u{{{}}}", u.sequence).unwrap();
-                    eco
-                }
-            })),
+            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 => {
diff --git a/src/syntax/mod.rs b/src/syntax/mod.rs
index afa0ab86..9d4beb6c 100644
--- a/src/syntax/mod.rs
+++ b/src/syntax/mod.rs
@@ -121,7 +121,7 @@ pub enum NodeKind {
     Text(EcoString),
     /// A slash and the letter "u" followed by a hexadecimal unicode entity
     /// enclosed in curly braces: `\u{1F5FA}`.
-    UnicodeEscape(Rc<UnicodeEscapeToken>),
+    UnicodeEscape(UnicodeEscapeToken),
     /// An arbitrary number of backticks followed by inner contents, terminated
     /// with the same number of backticks: `` `...` ``.
     Raw(Rc<RawToken>),
diff --git a/src/syntax/token.rs b/src/syntax/token.rs
index 5a621495..4f43bb4f 100644
--- a/src/syntax/token.rs
+++ b/src/syntax/token.rs
@@ -33,9 +33,8 @@ pub struct MathToken {
 
 /// A unicode escape sequence token: `\u{1F5FA}`.
 #[derive(Debug, Clone, PartialEq)]
+#[repr(transparent)]
 pub struct UnicodeEscapeToken {
-    /// The escape sequence between the braces.
-    pub sequence: EcoString,
     /// The resulting unicode character.
-    pub character: Option<char>,
+    pub character: char,
 }