Auto merge of #15303 - oxalica:fix/byte-escape-highlight, r=lowr

Fix highlighting of byte escape sequences

Currently non-UTF8 escape sequences in byte strings and any escape sequences in byte literals are ignored.

crates/hir-def/src/body/pretty.rs

@@ -634,7 +634,7 @@ impl Printer<'_> {
         match literal {
             Literal::String(it) => w!(self, "{:?}", it),
             Literal::ByteString(it) => w!(self, "\"{}\"", it.escape_ascii()),
-            Literal::CString(it) => w!(self, "\"{}\\0\"", it),
+            Literal::CString(it) => w!(self, "\"{}\\0\"", it.escape_ascii()),
             Literal::Char(it) => w!(self, "'{}'", it.escape_debug()),
             Literal::Bool(it) => w!(self, "{}", it),
             Literal::Int(i, suffix) => {

crates/hir-def/src/hir.rs

@@ -85,7 +85,7 @@ impl fmt::Display for FloatTypeWrapper {
 pub enum Literal {
     String(Box<str>),
     ByteString(Box<[u8]>),
-    CString(Box<str>),
+    CString(Box<[u8]>),
     Char(char),
     Bool(bool),
     Int(i128, Option<BuiltinInt>),

crates/hir-ty/src/mir/lower.rs

@@ -1355,7 +1355,6 @@ impl<'ctx> MirLowerCtx<'ctx> {
                 return Ok(Operand::from_concrete_const(data, mm, ty));
             }
             hir_def::hir::Literal::CString(b) => {
-                let b = b.as_bytes();
                 let bytes = b.iter().copied().chain(iter::once(0)).collect::<Vec<_>>();
 
                 let mut data = Vec::with_capacity(mem::size_of::<usize>() * 2);

crates/ide/src/syntax_highlighting.rs

@@ -24,7 +24,7 @@ use syntax::{
 
 use crate::{
     syntax_highlighting::{
-        escape::{highlight_escape_char, highlight_escape_string},
+        escape::{highlight_escape_byte, highlight_escape_char, highlight_escape_string},
         format::highlight_format_string,
         highlights::Highlights,
         macro_::MacroHighlighter,
@@ -471,6 +471,14 @@ fn traverse(
                 };
 
                 highlight_escape_char(hl, &char, range.start())
+            } else if ast::Byte::can_cast(token.kind())
+                && ast::Byte::can_cast(descended_token.kind())
+            {
+                let Some(byte) = ast::Byte::cast(token) else {
+                    continue;
+                };
+
+                highlight_escape_byte(hl, &byte, range.start())
             }
         }
 

crates/ide/src/syntax_highlighting/escape.rs

@@ -1,7 +1,7 @@
 //! Syntax highlighting for escape sequences
 use crate::syntax_highlighting::highlights::Highlights;
 use crate::{HlRange, HlTag};
-use syntax::ast::{Char, IsString};
+use syntax::ast::{Byte, Char, IsString};
 use syntax::{AstToken, TextRange, TextSize};
 
 pub(super) fn highlight_escape_string<T: IsString>(
@@ -10,14 +10,14 @@ pub(super) fn highlight_escape_string<T: IsString>(
     start: TextSize,
 ) {
     string.escaped_char_ranges(&mut |piece_range, char| {
-        if char.is_err() {
-            return;
-        }
-
         if string.text()[piece_range.start().into()..].starts_with('\\') {
+            let highlight = match char {
+                Ok(_) => HlTag::EscapeSequence,
+                Err(_) => HlTag::InvalidEscapeSequence,
+            };
             stack.add(HlRange {
                 range: piece_range + start,
-                highlight: HlTag::EscapeSequence.into(),
+                highlight: highlight.into(),
                 binding_hash: None,
             });
         }
@@ -26,6 +26,9 @@ pub(super) fn highlight_escape_string<T: IsString>(
 
 pub(super) fn highlight_escape_char(stack: &mut Highlights, char: &Char, start: TextSize) {
     if char.value().is_none() {
+        // We do not emit invalid escapes highlighting here. The lexer would likely be in a bad
+        // state and this token contains junks, since `'` is not a reliable delimiter (consider
+        // lifetimes). Nonetheless, parser errors should already be emitted.
         return;
     }
 
@@ -43,3 +46,24 @@ pub(super) fn highlight_escape_char(stack: &mut Highlights, char: &Char, start:
         TextRange::new(start + TextSize::from(1), start + TextSize::from(text.len() as u32 + 1));
     stack.add(HlRange { range, highlight: HlTag::EscapeSequence.into(), binding_hash: None })
 }
+
+pub(super) fn highlight_escape_byte(stack: &mut Highlights, byte: &Byte, start: TextSize) {
+    if byte.value().is_none() {
+        // See `highlight_escape_char` for why no error highlighting here.
+        return;
+    }
+
+    let text = byte.text();
+    if !text.starts_with("b'") || !text.ends_with('\'') {
+        return;
+    }
+
+    let text = &text[2..text.len() - 1];
+    if !text.starts_with('\\') {
+        return;
+    }
+
+    let range =
+        TextRange::new(start + TextSize::from(2), start + TextSize::from(text.len() as u32 + 2));
+    stack.add(HlRange { range, highlight: HlTag::EscapeSequence.into(), binding_hash: None })
+}

crates/ide/src/syntax_highlighting/html.rs

@@ -109,6 +109,7 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 ";

crates/ide/src/syntax_highlighting/tags.rs

@@ -29,6 +29,7 @@ pub enum HlTag {
     Comment,
     EscapeSequence,
     FormatSpecifier,
+    InvalidEscapeSequence,
     Keyword,
     NumericLiteral,
     Operator(HlOperator),
@@ -166,6 +167,7 @@ impl HlTag {
             HlTag::CharLiteral => "char_literal",
             HlTag::Comment => "comment",
             HlTag::EscapeSequence => "escape_sequence",
+            HlTag::InvalidEscapeSequence => "invalid_escape_sequence",
             HlTag::FormatSpecifier => "format_specifier",
             HlTag::Keyword => "keyword",
             HlTag::Punctuation(punct) => match punct {

crates/ide/src/syntax_highlighting/test_data/highlight_assoc_functions.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">fn</span> <span class="function declaration">not_static</span><span class="parenthesis">(</span><span class="parenthesis">)</span> <span class="brace">{</span><span class="brace">}</span>
 

crates/ide/src/syntax_highlighting/test_data/highlight_attributes.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="attribute_bracket attribute">#</span><span class="attribute_bracket attribute">[</span><span class="builtin_attr attribute library">allow</span><span class="parenthesis attribute">(</span><span class="none attribute">dead_code</span><span class="parenthesis attribute">)</span><span class="attribute_bracket attribute">]</span>
 <span class="attribute_bracket attribute">#</span><span class="attribute_bracket attribute">[</span><span class="tool_module attribute library">rustfmt</span><span class="operator attribute">::</span><span class="tool_module attribute library">skip</span><span class="attribute_bracket attribute">]</span>

crates/ide/src/syntax_highlighting/test_data/highlight_crate_root.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">extern</span> <span class="keyword">crate</span> <span class="module crate_root library">foo</span><span class="semicolon">;</span>
 <span class="keyword">use</span> <span class="module crate_root default_library library">core</span><span class="operator">::</span><span class="module default_library library">iter</span><span class="semicolon">;</span>

crates/ide/src/syntax_highlighting/test_data/highlight_default_library.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">use</span> <span class="module crate_root default_library library">core</span><span class="operator">::</span><span class="module default_library library">iter</span><span class="semicolon">;</span>
 

crates/ide/src/syntax_highlighting/test_data/highlight_doctest.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="comment documentation">//! This is a module to test doc injection.</span>
 <span class="comment documentation">//! ```</span>

crates/ide/src/syntax_highlighting/test_data/highlight_extern_crate.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">extern</span> <span class="keyword">crate</span> <span class="module crate_root default_library library">std</span><span class="semicolon">;</span>
 <span class="keyword">extern</span> <span class="keyword">crate</span> <span class="module crate_root default_library library">alloc</span> <span class="keyword">as</span> <span class="module crate_root default_library declaration library">abc</span><span class="semicolon">;</span>

crates/ide/src/syntax_highlighting/test_data/highlight_general.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">use</span> <span class="module">inner</span><span class="operator">::</span><span class="brace">{</span><span class="self_keyword">self</span> <span class="keyword">as</span> <span class="module declaration">inner_mod</span><span class="brace">}</span><span class="semicolon">;</span>
 <span class="keyword">mod</span> <span class="module declaration">inner</span> <span class="brace">{</span><span class="brace">}</span>

crates/ide/src/syntax_highlighting/test_data/highlight_injection.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">fn</span> <span class="function declaration">fixture</span><span class="parenthesis">(</span><span class="value_param declaration reference">ra_fixture</span><span class="colon">:</span> <span class="punctuation">&</span><span class="builtin_type">str</span><span class="parenthesis">)</span> <span class="brace">{</span><span class="brace">}</span>
 

crates/ide/src/syntax_highlighting/test_data/highlight_keywords.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">extern</span> <span class="keyword">crate</span> <span class="self_keyword crate_root public">self</span><span class="semicolon">;</span>
 

crates/ide/src/syntax_highlighting/test_data/highlight_lifetimes.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code>
 <span class="attribute_bracket attribute">#</span><span class="attribute_bracket attribute">[</span><span class="attribute attribute default_library library">derive</span><span class="parenthesis attribute">(</span><span class="parenthesis attribute">)</span><span class="attribute_bracket attribute">]</span>

crates/ide/src/syntax_highlighting/test_data/highlight_macros.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="module crate_root library">proc_macros</span><span class="operator">::</span><span class="macro library">mirror</span><span class="macro_bang">!</span> <span class="brace macro">{</span>
     <span class="brace macro">{</span>

crates/ide/src/syntax_highlighting/test_data/highlight_module_docs_inline.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="comment documentation">//! </span><span class="struct documentation injected intra_doc_link">[Struct]</span>
 <span class="comment documentation">//! This is an intra doc injection test for modules</span>

crates/ide/src/syntax_highlighting/test_data/highlight_module_docs_outline.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="comment documentation">/// </span><span class="struct documentation injected intra_doc_link">[crate::foo::Struct]</span>
 <span class="comment documentation">/// This is an intra doc injection test for modules</span>

crates/ide/src/syntax_highlighting/test_data/highlight_operators.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">fn</span> <span class="function declaration">main</span><span class="parenthesis">(</span><span class="parenthesis">)</span> <span class="brace">{</span>
     <span class="numeric_literal">1</span> <span class="arithmetic">+</span> <span class="numeric_literal">1</span> <span class="arithmetic">-</span> <span class="numeric_literal">1</span> <span class="arithmetic">*</span> <span class="numeric_literal">1</span> <span class="arithmetic">/</span> <span class="numeric_literal">1</span> <span class="arithmetic">%</span> <span class="numeric_literal">1</span> <span class="bitwise">|</span> <span class="numeric_literal">1</span> <span class="bitwise">&</span> <span class="numeric_literal">1</span> <span class="logical">!</span> <span class="numeric_literal">1</span> <span class="bitwise">^</span> <span class="numeric_literal">1</span> <span class="bitwise">&gt;&gt;</span> <span class="numeric_literal">1</span> <span class="bitwise">&lt;&lt;</span> <span class="numeric_literal">1</span><span class="semicolon">;</span>

crates/ide/src/syntax_highlighting/test_data/highlight_rainbow.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">fn</span> <span class="function declaration">main</span><span class="parenthesis">(</span><span class="parenthesis">)</span> <span class="brace">{</span>
     <span class="keyword">let</span> <span class="variable declaration reference" data-binding-hash="8121853618659664005" style="color: hsl(273,88%,88%);">hello</span> <span class="operator">=</span> <span class="string_literal">"hello"</span><span class="semicolon">;</span>

crates/ide/src/syntax_highlighting/test_data/highlight_strings.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">macro_rules</span><span class="macro_bang">!</span> <span class="macro declaration">println</span> <span class="brace">{</span>
     <span class="parenthesis">(</span><span class="punctuation">$</span><span class="parenthesis">(</span><span class="punctuation">$</span>arg<span class="colon">:</span>tt<span class="parenthesis">)</span><span class="punctuation">*</span><span class="parenthesis">)</span> <span class="operator">=</span><span class="angle">&gt;</span> <span class="parenthesis">(</span><span class="brace">{</span>
@@ -105,6 +106,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
     <span class="keyword">let</span> <span class="variable declaration">a</span> <span class="operator">=</span> <span class="char_literal">'</span><span class="escape_sequence">\x65</span><span class="char_literal">'</span><span class="semicolon">;</span>
     <span class="keyword">let</span> <span class="variable declaration">a</span> <span class="operator">=</span> <span class="char_literal">'</span><span class="escape_sequence">\x00</span><span class="char_literal">'</span><span class="semicolon">;</span>
 
+    <span class="keyword">let</span> <span class="variable declaration">a</span> <span class="operator">=</span> <span class="byte_literal">b'</span><span class="escape_sequence">\xFF</span><span class="byte_literal">'</span><span class="semicolon">;</span>
+
     <span class="macro">println</span><span class="macro_bang">!</span><span class="parenthesis macro">(</span><span class="string_literal macro">"Hello </span><span class="escape_sequence">{{</span><span class="string_literal macro">Hello</span><span class="escape_sequence">}}</span><span class="string_literal macro">"</span><span class="parenthesis macro">)</span><span class="semicolon">;</span>
     <span class="comment">// from https://doc.rust-lang.org/std/fmt/index.html</span>
     <span class="macro">println</span><span class="macro_bang">!</span><span class="parenthesis macro">(</span><span class="string_literal macro">"Hello"</span><span class="parenthesis macro">)</span><span class="semicolon">;</span>                 <span class="comment">// =&gt; "Hello"</span>
@@ -159,8 +162,9 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
     <span class="macro">println</span><span class="macro_bang">!</span><span class="parenthesis macro">(</span><span class="string_literal macro">"Hello</span><span class="escape_sequence">\n</span><span class="string_literal macro">World"</span><span class="parenthesis macro">)</span><span class="semicolon">;</span>
     <span class="macro">println</span><span class="macro_bang">!</span><span class="parenthesis macro">(</span><span class="string_literal macro">"</span><span class="escape_sequence">\u{48}</span><span class="escape_sequence">\x65</span><span class="escape_sequence">\x6C</span><span class="escape_sequence">\x6C</span><span class="escape_sequence">\x6F</span><span class="string_literal macro"> World"</span><span class="parenthesis macro">)</span><span class="semicolon">;</span>
 
-    <span class="keyword">let</span> <span class="punctuation">_</span> <span class="operator">=</span> <span class="string_literal">"</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x00</span><span class="escape_sequence">\x63</span><span class="escape_sequence">\n</span><span class="string_literal">"</span><span class="semicolon">;</span>
+    <span class="keyword">let</span> <span class="punctuation">_</span> <span class="operator">=</span> <span class="string_literal">"</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x00</span><span class="escape_sequence">\x63</span><span class="invalid_escape_sequence">\xFF</span><span class="escape_sequence">\u{FF}</span><span class="escape_sequence">\n</span><span class="string_literal">"</span><span class="semicolon">;</span> <span class="comment">// invalid non-UTF8 escape sequences</span>
-    <span class="keyword">let</span> <span class="punctuation">_</span> <span class="operator">=</span> <span class="string_literal">b"</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x00</span><span class="escape_sequence">\x63</span><span class="escape_sequence">\n</span><span class="string_literal">"</span><span class="semicolon">;</span>
+    <span class="keyword">let</span> <span class="punctuation">_</span> <span class="operator">=</span> <span class="string_literal">b"</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x28</span><span class="escape_sequence">\x00</span><span class="escape_sequence">\x63</span><span class="escape_sequence">\xFF</span><span class="invalid_escape_sequence">\u{FF}</span><span class="escape_sequence">\n</span><span class="string_literal">"</span><span class="semicolon">;</span> <span class="comment">// valid bytes, invalid unicodes</span>
+    <span class="keyword">let</span> <span class="punctuation">_</span> <span class="operator">=</span> <span class="string_literal">c"</span><span class="escape_sequence">\u{FF}</span><span class="escape_sequence">\xFF</span><span class="string_literal">"</span><span class="semicolon">;</span> <span class="comment">// valid bytes, valid unicodes</span>
     <span class="keyword">let</span> <span class="variable declaration reference">backslash</span> <span class="operator">=</span> <span class="string_literal">r"\\"</span><span class="semicolon">;</span>
 
     <span class="macro">println</span><span class="macro_bang">!</span><span class="parenthesis macro">(</span><span class="string_literal macro">"</span><span class="format_specifier">{</span><span class="escape_sequence">\x41</span><span class="format_specifier">}</span><span class="string_literal macro">"</span><span class="comma macro">,</span> <span class="none macro">A</span> <span class="operator macro">=</span> <span class="numeric_literal macro">92</span><span class="parenthesis macro">)</span><span class="semicolon">;</span>

crates/ide/src/syntax_highlighting/test_data/highlight_unsafe.html

@@ -40,7 +40,8 @@ pre                 { color: #DCDCCC; background: #3F3F3F; font-size: 22px; padd
 .control            { font-style: italic; }
 .reference          { font-style: italic; font-weight: bold; }
 
-.unresolved_reference { color: #FC5555; text-decoration: wavy underline; }
+.invalid_escape_sequence { color: #FC5555; text-decoration: wavy underline; }
+.unresolved_reference    { color: #FC5555; text-decoration: wavy underline; }
 </style>
 <pre><code><span class="keyword">macro_rules</span><span class="macro_bang">!</span> <span class="macro declaration">id</span> <span class="brace">{</span>
     <span class="parenthesis">(</span><span class="punctuation">$</span><span class="parenthesis">(</span><span class="punctuation">$</span>tt<span class="colon">:</span>tt<span class="parenthesis">)</span><span class="punctuation">*</span><span class="parenthesis">)</span> <span class="operator">=</span><span class="angle">&gt;</span> <span class="brace">{</span>

crates/ide/src/syntax_highlighting/tests.rs

@@ -451,6 +451,8 @@ fn main() {
     let a = '\x65';
     let a = '\x00';
 
+    let a = b'\xFF';
+
     println!("Hello {{Hello}}");
     // from https://doc.rust-lang.org/std/fmt/index.html
     println!("Hello");                 // => "Hello"
@@ -505,8 +507,9 @@ fn main() {
     println!("Hello\nWorld");
     println!("\u{48}\x65\x6C\x6C\x6F World");
 
-    let _ = "\x28\x28\x00\x63\n";
+    let _ = "\x28\x28\x00\x63\xFF\u{FF}\n"; // invalid non-UTF8 escape sequences
-    let _ = b"\x28\x28\x00\x63\n";
+    let _ = b"\x28\x28\x00\x63\xFF\u{FF}\n"; // valid bytes, invalid unicodes
+    let _ = c"\u{FF}\xFF"; // valid bytes, valid unicodes
     let backslash = r"\\";
 
     println!("{\x41}", A = 92);

crates/rust-analyzer/src/semantic_tokens.rs

@@ -78,6 +78,7 @@ define_semantic_token_types![
         (DERIVE_HELPER, "deriveHelper") => DECORATOR,
         (DOT, "dot"),
         (ESCAPE_SEQUENCE, "escapeSequence") => STRING,
+        (INVALID_ESCAPE_SEQUENCE, "invalidEscapeSequence") => STRING,
         (FORMAT_SPECIFIER, "formatSpecifier") => STRING,
         (GENERIC, "generic") => TYPE_PARAMETER,
         (LABEL, "label"),

crates/rust-analyzer/src/to_proto.rs

@@ -640,6 +640,7 @@ fn semantic_token_type_and_modifiers(
         HlTag::CharLiteral => semantic_tokens::CHAR,
         HlTag::Comment => semantic_tokens::COMMENT,
         HlTag::EscapeSequence => semantic_tokens::ESCAPE_SEQUENCE,
+        HlTag::InvalidEscapeSequence => semantic_tokens::INVALID_ESCAPE_SEQUENCE,
         HlTag::FormatSpecifier => semantic_tokens::FORMAT_SPECIFIER,
         HlTag::Keyword => semantic_tokens::KEYWORD,
         HlTag::None => semantic_tokens::GENERIC,

crates/syntax/src/ast/token_ext.rs

@@ -2,7 +2,9 @@
 
 use std::borrow::Cow;
 
-use rustc_lexer::unescape::{unescape_byte, unescape_char, unescape_literal, Mode};
+use rustc_lexer::unescape::{
+    unescape_byte, unescape_c_string, unescape_char, unescape_literal, CStrUnit, Mode,
+};
 
 use crate::{
     ast::{self, AstToken},
@@ -146,6 +148,7 @@ impl QuoteOffsets {
 
 pub trait IsString: AstToken {
     const RAW_PREFIX: &'static str;
+    const MODE: Mode;
     fn is_raw(&self) -> bool {
         self.text().starts_with(Self::RAW_PREFIX)
     }
@@ -181,7 +184,7 @@ pub trait IsString: AstToken {
         let text = &self.text()[text_range_no_quotes - start];
         let offset = text_range_no_quotes.start() - start;
 
-        unescape_literal(text, Mode::Str, &mut |range, unescaped_char| {
+        unescape_literal(text, Self::MODE, &mut |range, unescaped_char| {
             let text_range =
                 TextRange::new(range.start.try_into().unwrap(), range.end.try_into().unwrap());
             cb(text_range + offset, unescaped_char);
@@ -196,6 +199,7 @@ pub trait IsString: AstToken {
 
 impl IsString for ast::String {
     const RAW_PREFIX: &'static str = "r";
+    const MODE: Mode = Mode::Str;
 }
 
 impl ast::String {
@@ -213,7 +217,7 @@ impl ast::String {
         let mut buf = String::new();
         let mut prev_end = 0;
         let mut has_error = false;
-        unescape_literal(text, Mode::Str, &mut |char_range, unescaped_char| match (
+        unescape_literal(text, Self::MODE, &mut |char_range, unescaped_char| match (
             unescaped_char,
             buf.capacity() == 0,
         ) {
@@ -239,6 +243,7 @@ impl ast::String {
 
 impl IsString for ast::ByteString {
     const RAW_PREFIX: &'static str = "br";
+    const MODE: Mode = Mode::ByteStr;
 }
 
 impl ast::ByteString {
@@ -256,7 +261,7 @@ impl ast::ByteString {
         let mut buf: Vec<u8> = Vec::new();
         let mut prev_end = 0;
         let mut has_error = false;
-        unescape_literal(text, Mode::ByteStr, &mut |char_range, unescaped_char| match (
+        unescape_literal(text, Self::MODE, &mut |char_range, unescaped_char| match (
             unescaped_char,
             buf.capacity() == 0,
         ) {
@@ -282,42 +287,70 @@ impl ast::ByteString {
 
 impl IsString for ast::CString {
     const RAW_PREFIX: &'static str = "cr";
+    const MODE: Mode = Mode::CStr;
+
+    fn escaped_char_ranges(
+        &self,
+        cb: &mut dyn FnMut(TextRange, Result<char, rustc_lexer::unescape::EscapeError>),
+    ) {
+        let text_range_no_quotes = match self.text_range_between_quotes() {
+            Some(it) => it,
+            None => return,
+        };
+
+        let start = self.syntax().text_range().start();
+        let text = &self.text()[text_range_no_quotes - start];
+        let offset = text_range_no_quotes.start() - start;
+
+        unescape_c_string(text, Self::MODE, &mut |range, unescaped_char| {
+            let text_range =
+                TextRange::new(range.start.try_into().unwrap(), range.end.try_into().unwrap());
+            // XXX: This method should only be used for highlighting ranges. The unescaped
+            // char/byte is not used. For simplicity, we return an arbitrary placeholder char.
+            cb(text_range + offset, unescaped_char.map(|_| ' '));
+        });
+    }
 }
 
 impl ast::CString {
-    pub fn value(&self) -> Option<Cow<'_, str>> {
+    pub fn value(&self) -> Option<Cow<'_, [u8]>> {
         if self.is_raw() {
             let text = self.text();
             let text =
                 &text[self.text_range_between_quotes()? - self.syntax().text_range().start()];
-            return Some(Cow::Borrowed(text));
+            return Some(Cow::Borrowed(text.as_bytes()));
         }
 
         let text = self.text();
         let text = &text[self.text_range_between_quotes()? - self.syntax().text_range().start()];
 
-        let mut buf = String::new();
+        let mut buf = Vec::new();
         let mut prev_end = 0;
         let mut has_error = false;
-        unescape_literal(text, Mode::Str, &mut |char_range, unescaped_char| match (
+        let mut char_buf = [0u8; 4];
-            unescaped_char,
+        let mut extend_unit = |buf: &mut Vec<u8>, unit: CStrUnit| match unit {
+            CStrUnit::Byte(b) => buf.push(b),
+            CStrUnit::Char(c) => buf.extend(c.encode_utf8(&mut char_buf).as_bytes()),
+        };
+        unescape_c_string(text, Self::MODE, &mut |char_range, unescaped| match (
+            unescaped,
             buf.capacity() == 0,
         ) {
-            (Ok(c), false) => buf.push(c),
+            (Ok(u), false) => extend_unit(&mut buf, u),
             (Ok(_), true) if char_range.len() == 1 && char_range.start == prev_end => {
                 prev_end = char_range.end
             }
-            (Ok(c), true) => {
+            (Ok(u), true) => {
                 buf.reserve_exact(text.len());
-                buf.push_str(&text[..prev_end]);
+                buf.extend(text[..prev_end].as_bytes());
-                buf.push(c);
+                extend_unit(&mut buf, u);
             }
             (Err(_), _) => has_error = true,
         });
 
         match (has_error, buf.capacity() == 0) {
             (true, _) => None,
-            (false, true) => Some(Cow::Borrowed(text)),
+            (false, true) => Some(Cow::Borrowed(text.as_bytes())),
             (false, false) => Some(Cow::Owned(buf)),
         }
     }

editors/code/package.json

@@ -1804,12 +1804,16 @@
             },
             {
                 "id": "escapeSequence",
-                "description": "Style for char escapes in strings"
+                "description": "Style for char or byte escapes in strings"
             },
             {
                 "id": "formatSpecifier",
                 "description": "Style for {} placeholders in format strings"
             },
+            {
+                "id": "invalidEscapeSequence",
+                "description": "Style for invalid char or byte escapes in strings"
+            },
             {
                 "id": "label",
                 "description": "Style for labels"