Detect NulInCStr error earlier.

compiler/rustc_ast/src/util/literal.rs

@@ -8,7 +8,6 @@ use rustc_lexer::unescape::{
 };
 use rustc_span::symbol::{kw, sym, Symbol};
 use rustc_span::Span;
-use std::ops::Range;
 use std::{ascii, fmt, str};
 
 // Escapes a string, represented as a symbol. Reuses the original symbol,
@@ -39,7 +38,6 @@ pub enum LitError {
     InvalidFloatSuffix,
     NonDecimalFloat(u32),
     IntTooLarge(u32),
-    NulInCStr(Range<usize>),
 }
 
 impl LitKind {
@@ -156,10 +154,7 @@ impl LitKind {
                 let s = symbol.as_str();
                 let mut buf = Vec::with_capacity(s.len());
                 let mut error = Ok(());
-                unescape_c_string(s, Mode::CStr, &mut |span, c| match c {
-                    Ok(CStrUnit::Byte(0) | CStrUnit::Char('\0')) => {
-                        error = Err(LitError::NulInCStr(span));
-                    }
+                unescape_c_string(s, Mode::CStr, &mut |_span, c| match c {
                     Ok(CStrUnit::Byte(b)) => buf.push(b),
                     Ok(CStrUnit::Char(c)) => {
                         buf.extend_from_slice(c.encode_utf8(&mut [0; 4]).as_bytes())
@@ -179,10 +174,7 @@ impl LitKind {
                 // can convert the symbol directly to a `Lrc<u8>` on success.
                 let s = symbol.as_str();
                 let mut error = Ok(());
-                unescape_c_string(s, Mode::RawCStr, &mut |span, c| match c {
-                    Ok(CStrUnit::Byte(0) | CStrUnit::Char('\0')) => {
-                        error = Err(LitError::NulInCStr(span));
-                    }
+                unescape_c_string(s, Mode::RawCStr, &mut |_, c| match c {
                     Ok(_) => {}
                     Err(err) => {
                         if err.is_fatal() {

compiler/rustc_lexer/src/unescape.rs

@@ -59,6 +59,9 @@ pub enum EscapeError {
     /// Non-ascii character in byte literal, byte string literal, or raw byte string literal.
     NonAsciiCharInByte,
 
+    // `\0` in a C string literal.
+    NulInCStr,
+
     /// After a line ending with '\', the next line contains whitespace
     /// characters that are not skipped.
     UnskippedWhitespaceWarning,
@@ -122,10 +125,20 @@ where
 {
     match mode {
         CStr => {
-            unescape_non_raw_common(src, mode, callback);
+            unescape_non_raw_common(src, mode, &mut |r, mut result| {
+                if let Ok(CStrUnit::Byte(0) | CStrUnit::Char('\0')) = result {
+                    result = Err(EscapeError::NulInCStr);
+                }
+                callback(r, result)
+            });
         }
         RawCStr => {
-            check_raw_common(src, mode, &mut |r, result| callback(r, result.map(CStrUnit::Char)));
+            check_raw_common(src, mode, &mut |r, mut result| {
+                if let Ok('\0') = result {
+                    result = Err(EscapeError::NulInCStr);
+                }
+                callback(r, result.map(CStrUnit::Char))
+            });
         }
         Char | Byte | Str | RawStr | ByteStr | RawByteStr => unreachable!(),
     }

compiler/rustc_parse/messages.ftl

@@ -613,6 +613,8 @@ parse_note_mut_pattern_usage = `mut` may be followed by `variable` and `variable
 
 parse_note_pattern_alternatives_use_single_vert = alternatives in or-patterns are separated with `|`, not `||`
 
+parse_nul_in_c_str = null characters in C string literals are not supported
+
 parse_or_pattern_not_allowed_in_fn_parameters = top-level or-patterns are not allowed in function parameters
 parse_or_pattern_not_allowed_in_let_binding = top-level or-patterns are not allowed in `let` bindings
 parse_out_of_range_hex_escape = out of range hex escape

compiler/rustc_parse/src/errors.rs

@@ -2144,6 +2144,11 @@ pub enum UnescapeError {
         #[subdiagnostic]
         suggestion: MoreThanOneCharSugg,
     },
+    #[diag(parse_nul_in_c_str)]
+    NulInCStr {
+        #[primary_span]
+        span: Span,
+    },
 }
 
 #[derive(Subdiagnostic)]

compiler/rustc_parse/src/lexer/unescape_error_reporting.rs

@@ -262,6 +262,9 @@ pub(crate) fn emit_unescape_error(
         EscapeError::LoneSlash => {
             dcx.emit_err(UnescapeError::LoneSlash(err_span));
         }
+        EscapeError::NulInCStr => {
+            dcx.emit_err(UnescapeError::NulInCStr { span: err_span });
+        }
         EscapeError::UnskippedWhitespaceWarning => {
             let (c, char_span) = last_char();
             dcx.emit_warn(UnescapeError::UnskippedWhitespace {

compiler/rustc_session/messages.ftl

@@ -72,8 +72,6 @@ session_not_circumvent_feature = `-Zunleash-the-miri-inside-of-you` may not be u
 
 session_not_supported = not supported
 
-session_nul_in_c_str = null characters in C string literals are not supported
-
 session_octal_float_literal_not_supported = octal float literal is not supported
 
 session_optimization_fuel_exhausted = optimization-fuel-exhausted: {$msg}

compiler/rustc_session/src/errors.rs

@@ -7,7 +7,7 @@ use rustc_errors::{
     error_code, DiagCtxt, DiagnosticBuilder, DiagnosticMessage, IntoDiagnostic, Level, MultiSpan,
 };
 use rustc_macros::Diagnostic;
-use rustc_span::{BytePos, Span, Symbol};
+use rustc_span::{Span, Symbol};
 use rustc_target::spec::{SplitDebuginfo, StackProtector, TargetTriple};
 
 pub struct FeatureGateError {
@@ -327,13 +327,6 @@ pub(crate) struct BinaryFloatLiteralNotSupported {
     pub span: Span,
 }
 
-#[derive(Diagnostic)]
-#[diag(session_nul_in_c_str)]
-pub(crate) struct NulInCStr {
-    #[primary_span]
-    pub span: Span,
-}
-
 pub fn report_lit_error(sess: &ParseSess, err: LitError, lit: token::Lit, span: Span) {
     // Checks if `s` looks like i32 or u1234 etc.
     fn looks_like_width_suffix(first_chars: &[char], s: &str) -> bool {
@@ -413,12 +406,6 @@ pub fn report_lit_error(sess: &ParseSess, err: LitError, lit: token::Lit, span:
             };
             dcx.emit_err(IntLiteralTooLarge { span, limit });
         }
-        LitError::NulInCStr(range) => {
-            let lo = BytePos(span.lo().0 + range.start as u32 + 2);
-            let hi = BytePos(span.lo().0 + range.end as u32 + 2);
-            let span = span.with_lo(lo).with_hi(hi);
-            dcx.emit_err(NulInCStr { span });
-        }
     }
 }
 

src/tools/rust-analyzer/crates/parser/src/lexed_str.rs

@@ -371,6 +371,7 @@ fn error_to_diagnostic_message(error: EscapeError, mode: Mode) -> &'static str {
             "non-ASCII character in byte string literal"
         }
         EscapeError::NonAsciiCharInByte => "non-ASCII character in raw byte string literal",
+        EscapeError::NulInCStr => "null character in C string literal",
         EscapeError::UnskippedWhitespaceWarning => "",
         EscapeError::MultipleSkippedLinesWarning => "",
     }

src/tools/rust-analyzer/crates/syntax/src/validation.rs

@@ -106,6 +106,9 @@ fn rustc_unescape_error_to_string(err: unescape::EscapeError) -> (&'static str,
         EE::NonAsciiCharInByte  => {
             "Byte literals must not contain non-ASCII characters"
         }
+        EE::NulInCStr  => {
+            "C strings literals must not contain null characters"
+        }
         EE::UnskippedWhitespaceWarning => "Whitespace after this escape is not skipped",
         EE::MultipleSkippedLinesWarning => "Multiple lines are skipped by this escape",