Fix unescaping of C string literals

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/test_data/highlight_strings.html

@@ -161,8 +161,9 @@ 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="string_literal">\xFF</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">"</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="string_literal">\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">\xFF</span><span class="escape_sequence">\n</span><span class="string_literal">"</span><span class="semicolon">;</span> <span class="comment">// valid bytes</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="string_literal">\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/tests.rs

@@ -507,8 +507,9 @@ fn main() {
     println!("Hello\nWorld");
     println!("\u{48}\x65\x6C\x6C\x6F World");
 
-    let _ = "\x28\x28\x00\x63\xFF\n"; // invalid non-UTF8 escape sequences
+    let _ = "\x28\x28\x00\x63\xFF\u{FF}\n"; // invalid non-UTF8 escape sequences
-    let _ = b"\x28\x28\x00\x63\xFF\n"; // valid bytes
+    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/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},
@@ -285,45 +287,70 @@ impl ast::ByteString {
 
 impl IsString for ast::CString {
     const RAW_PREFIX: &'static str = "cr";
-    // XXX: `Mode::CStr` is not supported by `unescape_literal` of ra-ap-rustc_lexer yet.
+    const MODE: Mode = Mode::CStr;
-    // Here we pretend it to be a byte string.
+
-    const MODE: Mode = Mode::ByteStr;
+    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, Self::MODE, &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)),
         }
     }