Newlines are complicated, y'all 😱

Co-authored-by: laurmaedje@outlook.de <laurmaedje@outlook.de>

4 files changed, 184 insertions, 171 deletions

diff --git a/src/layout/tree.rs b/src/layout/tree.rs
index 714cfe27..16a2930a 100644
--- a/src/layout/tree.rs
+++ b/src/layout/tree.rs
@@ -3,7 +3,7 @@
 use crate::style::LayoutStyle;
 use crate::syntax::decoration::Decoration;
 use crate::syntax::span::{Span, Spanned};
-use crate::syntax::tree::{CallExpr, SyntaxNode, SyntaxTree, CodeBlockExpr};
+use crate::syntax::tree::{CallExpr, SyntaxNode, SyntaxTree, Code};
 use crate::{DynFuture, Feedback, Pass};
 use super::line::{LineContext, LineLayouter};
 use super::text::{layout_text, TextContext};
@@ -63,10 +63,7 @@ impl<'a> TreeLayouter<'a> {
         match &node.v {
             SyntaxNode::Spacing => self.layout_space(),
             SyntaxNode::Linebreak => self.layouter.finish_line(),
-            SyntaxNode::Parbreak => self.layouter.add_secondary_spacing(
-                self.style.text.paragraph_spacing(),
-                SpacingKind::PARAGRAPH,
-            ),
+            SyntaxNode::Parbreak => self.layout_parbreak(),
 
             SyntaxNode::ToggleItalic => {
                 self.style.text.italic = !self.style.text.italic;
@@ -84,7 +81,7 @@ impl<'a> TreeLayouter<'a> {
             }
 
             SyntaxNode::Raw(lines) => self.layout_raw(lines).await,
-            SyntaxNode::CodeBlock(block) => self.layout_code(block).await,
+            SyntaxNode::Code(block) => self.layout_code(block).await,
 
             SyntaxNode::Call(call) => {
                 self.layout_call(Spanned::new(call, node.span)).await;
@@ -99,6 +96,13 @@ impl<'a> TreeLayouter<'a> {
         );
     }
 
+    fn layout_parbreak(&mut self) {
+        self.layouter.add_secondary_spacing(
+            self.style.text.paragraph_spacing(),
+            SpacingKind::PARAGRAPH,
+        );
+    }
+
     async fn layout_text(&mut self, text: &str) {
         self.layouter.add(
             layout_text(
@@ -133,19 +137,16 @@ impl<'a> TreeLayouter<'a> {
         self.style.text.fallback = fallback;
     }
 
-    async fn layout_code(&mut self, block: &CodeBlockExpr) {
-        let fallback = self.style.text.fallback.clone();
-        self.style.text.fallback
-            .list_mut()
-            .insert(0, "monospace".to_string());
-        self.style.text.fallback.flatten();
-
-        for line in &block.raw {
-            self.layout_text(line).await;
-            self.layouter.finish_line();
+    async fn layout_code(&mut self, code: &Code) {
+        if code.block {
+            self.layout_parbreak();
         }
 
-        self.style.text.fallback = fallback;
+        self.layout_raw(&code.lines).await;
+
+        if code.block {
+            self.layout_parbreak()
+        }
     }
 
     async fn layout_call(&mut self, call: Spanned<&CallExpr>) {
diff --git a/src/syntax/parsing.rs b/src/syntax/parsing.rs
index 76509fae..0d12f6e1 100644
--- a/src/syntax/parsing.rs
+++ b/src/syntax/parsing.rs
@@ -7,14 +7,9 @@ use crate::color::RgbaColor;
 use crate::compute::table::SpannedEntry;
 use super::decoration::Decoration;
 use super::span::{Pos, Span, Spanned};
-use super::tokens::{is_newline_char, Token, TokenMode, Tokens, is_identifier};
+use super::tokens::{is_newline_char, Token, TokenMode, Tokens};
 use super::tree::{
-    CallExpr,
-    Expr,
-    SyntaxNode,
-    SyntaxTree,
-    TableExpr,
-    CodeBlockExpr,
+    CallExpr, Expr, SyntaxNode, SyntaxTree, TableExpr, Code,
 };
 use super::Ident;
 
@@ -88,28 +83,27 @@ impl Parser<'_> {
                     if !terminated {
                         error!(
                             @self.feedback, Span::at(token.span.end),
-                            "expected code block to close",
+                            "expected backticks",
                         );
                     }
-                    let mut valid_ident = false;
-                    let mut lang = lang.map(|s| s.map(|v| {
-                        if is_identifier(v) {
-                            valid_ident = true;
-                        }
-                        Ident(v.to_string())
-                    }));
-
-                    if !valid_ident {
-                        if let Some(l) = lang {
-                            error!(
-                                @self.feedback, l.span,
-                                "expected language to be a valid identifier",
-                            );
+
+                    let lang = lang.and_then(|lang| {
+                        if let Some(ident) = Ident::new(lang.v) {
+                            Some(Spanned::new(ident, lang.span))
+                        } else {
+                            error!(@self.feedback, lang.span, "invalid identifier");
+                            None
                         }
-                        lang = None;
+                    });
+
+                    let mut lines = unescape_code(raw);
+                    let block = lines.len() > 1;
+
+                    if lines.last().map(|s| s.is_empty()).unwrap_or(false) {
+                        lines.pop();
                     }
 
-                    self.with_span(SyntaxNode::CodeBlock(CodeBlockExpr { raw: unescape_code(raw), lang }))
+                    self.with_span(SyntaxNode::Code(Code { lang, lines, block }))
                 }
 
                 Token::Text(text) => {
@@ -624,102 +618,107 @@ fn unescape_string(string: &str) -> String {
 /// Unescape raw markup and split it into into lines.
 fn unescape_raw(raw: &str) -> Vec<String> {
     let mut iter = raw.chars().peekable();
-    let mut line = String::new();
-    let mut lines = Vec::new();
+    let mut text = String::new();
 
     while let Some(c) = iter.next() {
         if c == '\\' {
-            match iter.next() {
-                Some('`') => line.push('`'),
-                Some(c) => { line.push('\\'); line.push(c); }
-                None => line.push('\\'),
-            }
-        } else if is_newline_char(c) {
-            if c == '\r' && iter.peek() == Some(&'\n') {
-                iter.next();
-            }
+            if let Some(c) = iter.next() {
+                if c != '\\' && c != '`' {
+                    text.push('\\');
+                }
 
-            lines.push(std::mem::take(&mut line));
+                text.push(c);
+            } else {
+                text.push('\\');
+            }
         } else {
-            line.push(c);
+            text.push(c);
         }
     }
 
-    lines.push(line);
-    lines
+    split_lines(&text)
 }
 
 /// Unescape raw markup and split it into into lines.
 fn unescape_code(raw: &str) -> Vec<String> {
     let mut iter = raw.chars().peekable();
-    let mut line = String::new();
-    let mut lines = Vec::new();
-    let mut backticks: usize = 0;
-
-    // This assignment is used in line 731, 733;
-    // the compiler does not want to acknowledge that, however.
-    #[allow(unused_assignments)]
-    let mut update_backtick_count = true;
+    let mut text = String::new();
+    let mut backticks = 0u32;
+    let mut update_backtick_count;
 
     while let Some(c) = iter.next() {
         update_backtick_count = true;
-        if is_newline_char(c) {
-            if c == '\r' && iter.peek() == Some(&'\n') {
-                iter.next();
-            }
 
-            lines.push(std::mem::take(&mut line));
-        } else {
-            if c == '\\' && backticks > 0 {
-                let mut tail = String::new();
-                let mut escape_success = false;
-
-                let mut backticks_after_slash: u8 = 0;
-
-                while let Some(&s) = iter.peek() {
-                    match s {
-                        '\\' => {
-                            if backticks_after_slash == 0 {
-                                tail.push(s);
-                            } else {
-                                // Pattern like `\`\` should fail
-                                // escape and just be printed verbantim.
-                                break;
-                            }
+        if c == '\\' && backticks > 0 {
+            let mut tail = String::new();
+            let mut escape_success = false;
+            let mut backticks_after_slash = 0u32;
+
+            while let Some(&s) = iter.peek() {
+                match s {
+                    '\\' => {
+                        if backticks_after_slash == 0 {
+                            tail.push('\\');
+                        } else {
+                            // Pattern like `\`\` should fail
+                            // escape and just be printed verbantim.
+                            break;
                         }
-                        '`'  => {
-                            tail.push(s);
-                            backticks_after_slash += 1;
-                            if backticks_after_slash == 2 {
-                                escape_success = true;
-                                iter.next();
-                                break;
-                            }
+                    }
+                    '`' => {
+                        tail.push(s);
+                        backticks_after_slash += 1;
+                        if backticks_after_slash == 2 {
+                            escape_success = true;
+                            iter.next();
+                            break;
                         }
-                        _    => { break }
                     }
-
-                    iter.next();
+                    _ => break,
                 }
 
-                if !escape_success {
-                    line.push(c);
-                    backticks = backticks_after_slash as usize;
-                    update_backtick_count = false;
-                } else {
-                    backticks = 0;
-                }
+                iter.next();
+            }
 
-                line.push_str(&tail);
+            if !escape_success {
+                text.push(c);
+                backticks = backticks_after_slash;
+                update_backtick_count = false;
             } else {
-                line.push(c);
+                backticks = 0;
             }
+
+            text.push_str(&tail);
+        } else {
+            text.push(c);
         }
 
-        if update_backtick_count && c == '`' {
-            backticks += 1;
-        } else if update_backtick_count {
-            backticks = 0;
+        if update_backtick_count {
+            if c == '`' {
+                backticks += 1;
+            } else {
+                backticks = 0;
+            }
+        }
+    }
+
+    split_lines(&text)
+}
+
+fn split_lines(text: &str) -> Vec<String> {
+    let mut iter = text.chars().peekable();
+    let mut line = String::new();
+    let mut lines = Vec::new();
+
+    while let Some(c) = iter.next() {
+        if is_newline_char(c) {
+            if c == '\r' && iter.peek() == Some(&'\n') {
+                iter.next();
+            }
+
+            lines.push(std::mem::take(&mut line));
+        } else {
+            line.push(c);
         }
     }
 
@@ -753,13 +752,23 @@ mod tests {
         };
     }
 
-
-    fn Lang(text: &str) -> Option<Spanned<Ident>> { Some(Spanned::zero(Ident(text.to_string()))) }
-
     macro_rules! C {
-        ($lang:expr, $($line:expr),* $(,)?) => {
-            SyntaxNode::CodeBlock(CodeBlockExpr { raw: vec![$($line.to_string()) ,*], lang: $lang })
-        };
+        (None, $($line:expr),* $(,)?) => {{
+            let lines = vec![$($line.to_string()) ,*];
+            SyntaxNode::Code(Code {
+                lang: None,
+                block: lines.len() > 1,
+                lines,
+            })
+        }};
+        (Some($lang:expr), $($line:expr),* $(,)?) => {{
+            let lines = vec![$($line.to_string()) ,*];
+            SyntaxNode::Code(Code {
+                lang: Some(Into::<Spanned<&str>>::into($lang).map(|s| Ident(s.to_string()))),
+                block: lines.len() > 1,
+                lines,
+            })
+        }};
     }
 
     macro_rules! F {
@@ -896,6 +905,7 @@ mod tests {
         }
 
         test("raw\\`",     vec!["raw`"]);
+        test("raw\\\\`",   vec!["raw\\`"]);
         test("raw\ntext",  vec!["raw", "text"]);
         test("a\r\nb",     vec!["a", "b"]);
         test("a\n\nb",     vec!["a", "", "b"]);
@@ -942,16 +952,16 @@ mod tests {
         t!("`hi\\`du`"   => R!["hi`du"]);
 
         t!("```java System.out.print```" => C![
-            Lang("java"), "System.out.print"
-            ]);
+            Some("java"), "System.out.print"
+        ]);
         t!("``` console.log(\n\"alert\"\n)" => C![
             None, "console.log(", "\"alert\"", ")"
-            ]);
+        ]);
         t!("```typst \r\n Typst uses `\\`` to indicate code blocks" => C![
-            Lang("typst"), " Typst uses ``` to indicate code blocks"
-            ]);
-        e!("``` hi\nyou"      => s(1,3, 1,3, "expected code block to close"));
-        e!("```🌍 hi\nyou```" => s(0,3, 0,4, "expected language to be a valid identifier"));
+            Some("typst"), " Typst uses ``` to indicate code blocks"
+        ]);
+        e!("``` hi\nyou"      => s(1,3, 1,3, "expected backticks"));
+        e!("```🌍 hi\nyou```" => s(0,3, 0,4, "invalid identifier"));
         t!("💜\n\n 🌍"       => T("💜"), P, T("🌍"));
 
         ts!("hi"   => s(0,0, 0,2, T("hi")));
diff --git a/src/syntax/tokens.rs b/src/syntax/tokens.rs
index dbba175d..7ecb05fe 100644
--- a/src/syntax/tokens.rs
+++ b/src/syntax/tokens.rs
@@ -252,7 +252,7 @@ impl<'s> Iterator for Tokens<'s> {
 
             // Style toggles.
             '_' if self.mode == Body => Underscore,
-            '`' if self.mode == Body => self.read_raw_and_code(),
+            '`' if self.mode == Body => self.read_raw_or_code(),
 
             // An escaped thing.
             '\\' if self.mode == Body => self.read_escaped(),
@@ -341,66 +341,67 @@ impl<'s> Tokens<'s> {
         Str { string, terminated }
     }
 
-    fn read_raw_and_code(&mut self) -> Token<'s> {
+    fn read_raw_or_code(&mut self) -> Token<'s> {
         let (raw, terminated) = self.read_until_unescaped('`');
-        if raw.len() == 0 && terminated && self.peek() == Some('`') {
-            // Third tick found; this is a code block
+        if raw.is_empty() && terminated && self.peek() == Some('`') {
+            // Third tick found; this is a code block.
             self.eat();
-            let mut backticks = 0;
-            let mut terminated = true;
-            // Reads the lang tag (until newline or whitespace)
-            let lang_start = self.pos();
-            let (lang_opt, _) = self.read_string_until(
+
+            // Reads the lang tag (until newline or whitespace).
+            let start = self.pos();
+            let lang = self.read_string_until(
                 |c| c == '`' || c.is_whitespace() || is_newline_char(c),
-                 false, 0, 0);
-            let lang_end = self.pos();
+                false, 0, 0,
+            ).0;
+            let end = self.pos();
+            let lang = if !lang.is_empty() {
+                Some(Spanned::new(lang, Span::new(start, end)))
+            } else {
+                None
+            };
+
+            // Skip to start of raw contents.
+            while let Some(c) = self.peek() {
+                if is_newline_char(c) {
+                    self.eat();
+                    if c == '\r' && self.peek() == Some('\n') {
+                        self.eat();
+                    }
+
+                    break;
+                } else if c.is_whitespace() {
+                    self.eat();
+                } else {
+                    break;
+                }
+            }
 
-            #[derive(Debug, PartialEq)]
-            enum WhitespaceIngestion { All, ExceptNewline, Never }
-            let mut ingest_whitespace = WhitespaceIngestion::Never;
-            let mut start = self.index();
+            let start = self.index();
+            let mut backticks = 0u32;
 
             while backticks < 3 {
                 match self.eat() {
                     Some('`') => backticks += 1,
+                    // Escaping of triple backticks.
                     Some('\\') if backticks == 1 && self.peek() == Some('`') => {
                         backticks = 0;
                     }
-                    Some(c) => {
-                        // Remove whitespace between language and content or
-                        // first line break, deal with CRLF and CR line endings.
-                        if ingest_whitespace != WhitespaceIngestion::All
-                        && c == '\n' {
-                            start += 1;
-                            ingest_whitespace = WhitespaceIngestion::All;
-                        } else if ingest_whitespace != WhitespaceIngestion::All
-                        && c == '\r' {
-                            start += 1;
-                            ingest_whitespace = WhitespaceIngestion::ExceptNewline;
-                        } else if ingest_whitespace == WhitespaceIngestion::Never
-                        && c.is_whitespace() {
-                            start += 1;
-                        } else {
-                            ingest_whitespace = WhitespaceIngestion::All;
-                        }
-                    }
-                    None => {
-                        terminated = false;
-                        break;
-                    }
+                    Some(_) => {}
+                    None => break,
                 }
             }
-            let end = self.index() - (if terminated { 3 } else { 0 });
 
-            return Code {
-                lang: if lang_opt.len() == 0 { None } else {
-                    Some(Spanned::new(lang_opt, Span::new(lang_start, lang_end)))
-                },
+            let terminated = backticks == 3;
+            let end = self.index() - if terminated { 3 } else { 0 };
+
+            Code {
+                lang,
                 raw: &self.src[start..end],
                 terminated
             }
+        } else {
+            Raw { raw, terminated }
         }
-        Raw { raw, terminated }
     }
 
     fn read_until_unescaped(&mut self, c: char) -> (&'s str, bool) {
diff --git a/src/syntax/tree.rs b/src/syntax/tree.rs
index 313e76a4..44acd023 100644
--- a/src/syntax/tree.rs
+++ b/src/syntax/tree.rs
@@ -33,8 +33,8 @@ pub enum SyntaxNode {
     Text(String),
     /// Lines of raw text.
     Raw(Vec<String>),
-    /// An optionally highlighted multi-line code block.
-    CodeBlock(CodeBlockExpr),
+    /// An optionally highlighted (multi-line) code block.
+    Code(Code),
     /// A function call.
     Call(CallExpr),
 }
@@ -201,9 +201,10 @@ impl CallExpr {
         }
     }
 }
-/// An code block.
+/// A code block.
 #[derive(Debug, Clone, PartialEq)]
-pub struct CodeBlockExpr {
+pub struct Code {
     pub lang: Option<Spanned<Ident>>,
-    pub raw: Vec<String>,
+    pub lines: Vec<String>,
+    pub block: bool,
 }