From 4a7967812e9db30efd3612113749bcc5b955b7b6 Mon Sep 17 00:00:00 2001
From: Noah Hellman <noah@hllmn.net>
Date: Tue, 17 Jan 2023 22:36:10 +0100
Subject: [PATCH] lex: fix buggy peek

---
 src/inline.rs | 23 ++++++++++++++++-------
 src/lex.rs    | 24 ++++++++++++++++++------
 2 files changed, 34 insertions(+), 13 deletions(-)

diff --git a/src/inline.rs b/src/inline.rs
index a80d611..34e6c0a 100644
--- a/src/inline.rs
+++ b/src/inline.rs
@@ -67,7 +67,7 @@ pub struct Event {
     pub span: Span,
 }
 
-pub struct Parser<I> {
+pub struct Parser<I: Iterator + Clone> {
     /// Lexer, hosting upcoming source.
     lexer: lex::Lexer<I>,
     /// Span of current event.
@@ -177,9 +177,10 @@ impl<I: Iterator<Item = char> + Clone> Parser<I> {
                             Some(lex::Kind::Open(Delimiter::BraceEqual))
                         )
                     {
-                        let mut ahead = self.lexer.inner().clone();
+                        let mut ahead = self.lexer.chars();
                         let mut end = false;
                         let len = (&mut ahead)
+                            .skip(2) // {=
                             .take_while(|c| {
                                 if *c == '{' {
                                     return false;
@@ -191,8 +192,16 @@ impl<I: Iterator<Item = char> + Clone> Parser<I> {
                             })
                             .count();
                         if len > 0 && end {
+                            let tok = self.eat();
+                            debug_assert_eq!(
+                                tok,
+                                Some(lex::Token {
+                                    kind: lex::Kind::Open(Delimiter::BraceEqual),
+                                    len: 2,
+                                })
+                            );
                             self.lexer = lex::Lexer::new(ahead);
-                            let span_format = Span::by_len(self.span.end() + "{=".len(), len);
+                            let span_format = Span::by_len(self.span.end(), len);
                             kind = RawFormat;
                             self.events[opener_event].kind = EventKind::Enter(kind);
                             self.events[opener_event].span = span_format;
@@ -238,7 +247,7 @@ impl<I: Iterator<Item = char> + Clone> Parser<I> {
                 .back()
                 .map_or(false, |e| e.kind == EventKind::Str)
         {
-            let mut ahead = self.lexer.inner().clone();
+            let mut ahead = self.lexer.chars();
             let (mut attr_len, mut has_attr) = attr::valid(std::iter::once('{').chain(&mut ahead));
             attr_len -= 1; // rm {
             if attr_len > 0 {
@@ -296,7 +305,7 @@ impl<I: Iterator<Item = char> + Clone> Parser<I> {
 
     fn parse_autolink(&mut self, first: &lex::Token) -> Option<Event> {
         if first.kind == lex::Kind::Sym(Symbol::Lt) {
-            let mut ahead = self.lexer.inner().clone();
+            let mut ahead = self.lexer.chars();
             let mut end = false;
             let mut is_url = false;
             let len = (&mut ahead)
@@ -372,7 +381,7 @@ impl<I: Iterator<Item = char> + Clone> Parser<I> {
                             }
                         }
 
-                        let mut ahead = self.lexer.inner().clone();
+                        let mut ahead = self.lexer.chars();
                         let (mut attr_len, mut has_attr) = attr::valid(&mut ahead);
                         if attr_len > 0 {
                             let span_closer = self.span;
@@ -429,7 +438,7 @@ impl<I: Iterator<Item = char> + Clone> Parser<I> {
     }
 
     fn post_span(&mut self, ty: SpanType, opener_event: usize) -> Option<Event> {
-        let mut ahead = self.lexer.inner().clone();
+        let mut ahead = self.lexer.chars();
         match ahead.next() {
             Some(opener @ ('[' | '(')) => {
                 let img = ty == SpanType::Image;
diff --git a/src/lex.rs b/src/lex.rs
index 833bb32..1390d88 100644
--- a/src/lex.rs
+++ b/src/lex.rs
@@ -82,8 +82,9 @@ impl Sequence {
 }
 
 #[derive(Clone)]
-pub(crate) struct Lexer<I> {
+pub(crate) struct Lexer<I: Iterator + Clone> {
     chars: I,
+    chars_non_peeked: I,
     /// Next character should be escaped.
     escape: bool,
     /// Token to be peeked or next'ed.
@@ -95,26 +96,30 @@ pub(crate) struct Lexer<I> {
 impl<I: Iterator<Item = char> + Clone> Lexer<I> {
     pub fn new(chars: I) -> Lexer<I> {
         Lexer {
-            chars,
+            chars: chars.clone(),
+            chars_non_peeked: chars,
             escape: false,
             next: None,
             len: 0,
         }
     }
 
+    /// NOTE: Peeked [`Kind::Text`] tokens are only one char long, they may be longer when
+    /// consumed.
     pub fn peek(&mut self) -> Option<&Token> {
         if self.next.is_none() {
-            self.next = self.next_token();
+            self.next = self.token();
         }
         self.next.as_ref()
     }
 
-    pub fn inner(&self) -> &I {
-        &self.chars
+    pub fn chars(&self) -> I {
+        self.chars_non_peeked.clone()
     }
 
     fn next_token(&mut self) -> Option<Token> {
         let mut current = self.token();
+        self.chars_non_peeked = self.chars.clone();
 
         // concatenate text tokens
         if let Some(Token { kind: Text, len }) = &mut current {
@@ -145,6 +150,7 @@ impl<I: Iterator<Item = char> + Clone> Lexer<I> {
     }
 
     fn token(&mut self) -> Option<Token> {
+        self.chars_non_peeked = self.chars.clone();
         self.len = 0;
 
         let first = self.eat_char()?;
@@ -271,7 +277,13 @@ impl<I: Iterator<Item = char> + Clone> Iterator for Lexer<I> {
     type Item = Token;
 
     fn next(&mut self) -> Option<Self::Item> {
-        self.next.take().or_else(|| self.next_token())
+        self.next
+            .take()
+            .map(|x| {
+                self.chars_non_peeked = self.chars.clone();
+                x
+            })
+            .or_else(|| self.next_token())
     }
 }