Rollup merge of #134956 - compiler-errors:format-args-hidden-chars, r=jieyouxu

Account for C string literals and `format_args` in `HiddenUnicodeCodepoints` lint

This is stacked on #134955, and either that can land first or both of them can land together here. I split this out because this is a bit more involved of an impl.

Fixes #94945

compiler/rustc_ast/src/format.rs

@@ -4,6 +4,7 @@ use rustc_span::{Ident, Span, Symbol};
 
 use crate::Expr;
 use crate::ptr::P;
+use crate::token::LitKind;
 
 // Definitions:
 //
@@ -45,6 +46,10 @@ pub struct FormatArgs {
     pub span: Span,
     pub template: Vec<FormatArgsPiece>,
     pub arguments: FormatArguments,
+    /// The raw, un-split format string literal, with no escaping or processing.
+    ///
+    /// Generally only useful for lints that care about the raw bytes the user wrote.
+    pub uncooked_fmt_str: (LitKind, Symbol),
 }
 
 /// A piece of a format template string.

compiler/rustc_ast/src/mut_visit.rs

@@ -1596,7 +1596,7 @@ fn walk_inline_asm_sym<T: MutVisitor>(
 
 fn walk_format_args<T: MutVisitor>(vis: &mut T, fmt: &mut FormatArgs) {
     // FIXME: visit the template exhaustively.
-    let FormatArgs { span, template: _, arguments } = fmt;
+    let FormatArgs { span, template: _, arguments, uncooked_fmt_str: _ } = fmt;
     for FormatArgument { kind, expr } in arguments.all_args_mut() {
         match kind {
             FormatArgumentKind::Named(ident) | FormatArgumentKind::Captured(ident) => {

compiler/rustc_ast/src/visit.rs

@@ -1061,7 +1061,7 @@ pub fn walk_inline_asm_sym<'a, V: Visitor<'a>>(
 }
 
 pub fn walk_format_args<'a, V: Visitor<'a>>(visitor: &mut V, fmt: &'a FormatArgs) -> V::Result {
-    let FormatArgs { span: _, template: _, arguments } = fmt;
+    let FormatArgs { span: _, template: _, arguments, uncooked_fmt_str: _ } = fmt;
     for FormatArgument { kind, expr } in arguments.all_args() {
         match kind {
             FormatArgumentKind::Named(ident) | FormatArgumentKind::Captured(ident) => {

compiler/rustc_builtin_macros/src/asm.rs

@@ -16,7 +16,7 @@ use smallvec::smallvec;
 use {rustc_ast as ast, rustc_parse_format as parse};
 
 use crate::errors;
-use crate::util::expr_to_spanned_string;
+use crate::util::{ExprToSpannedString, expr_to_spanned_string};
 
 pub struct AsmArgs {
     pub templates: Vec<P<ast::Expr>>,
@@ -527,7 +527,12 @@ fn expand_preparsed_asm(
         let msg = "asm template must be a string literal";
         let template_sp = template_expr.span;
         let template_is_mac_call = matches!(template_expr.kind, ast::ExprKind::MacCall(_));
-        let (template_str, template_style, template_span) = {
+        let ExprToSpannedString {
+            symbol: template_str,
+            style: template_style,
+            span: template_span,
+            ..
+        } = {
             let ExpandResult::Ready(mac) = expr_to_spanned_string(ecx, template_expr, msg) else {
                 return ExpandResult::Retry(());
             };

compiler/rustc_builtin_macros/src/format.rs

@@ -17,7 +17,7 @@ use rustc_parse_format as parse;
 use rustc_span::{BytePos, ErrorGuaranteed, Ident, InnerSpan, Span, Symbol};
 
 use crate::errors;
-use crate::util::expr_to_spanned_string;
+use crate::util::{ExprToSpannedString, expr_to_spanned_string};
 
 // The format_args!() macro is expanded in three steps:
 //  1. First, `parse_args` will parse the `(literal, arg, arg, name=arg, name=arg)` syntax,
@@ -166,13 +166,18 @@ fn make_format_args(
 
     let MacroInput { fmtstr: efmt, mut args, is_direct_literal } = input;
 
-    let (fmt_str, fmt_style, fmt_span) = {
+    let ExprToSpannedString {
+        symbol: fmt_str,
+        span: fmt_span,
+        style: fmt_style,
+        uncooked_symbol: uncooked_fmt_str,
+    } = {
         let ExpandResult::Ready(mac) = expr_to_spanned_string(ecx, efmt.clone(), msg) else {
             return ExpandResult::Retry(());
         };
         match mac {
             Ok(mut fmt) if append_newline => {
-                fmt.0 = Symbol::intern(&format!("{}\n", fmt.0));
+                fmt.symbol = Symbol::intern(&format!("{}\n", fmt.symbol));
                 fmt
             }
             Ok(fmt) => fmt,
@@ -584,7 +589,12 @@ fn make_format_args(
         }
     }
 
-    ExpandResult::Ready(Ok(FormatArgs { span: fmt_span, template, arguments: args }))
+    ExpandResult::Ready(Ok(FormatArgs {
+        span: fmt_span,
+        template,
+        arguments: args,
+        uncooked_fmt_str,
+    }))
 }
 
 fn invalid_placeholder_type_error(

compiler/rustc_builtin_macros/src/util.rs

@@ -57,7 +57,17 @@ pub(crate) fn warn_on_duplicate_attribute(ecx: &ExtCtxt<'_>, item: &Annotatable,
 
 /// `Ok` represents successfully retrieving the string literal at the correct
 /// position, e.g., `println("abc")`.
-type ExprToSpannedStringResult<'a> = Result<(Symbol, ast::StrStyle, Span), UnexpectedExprKind<'a>>;
+pub(crate) type ExprToSpannedStringResult<'a> = Result<ExprToSpannedString, UnexpectedExprKind<'a>>;
+
+pub(crate) struct ExprToSpannedString {
+    pub symbol: Symbol,
+    pub style: ast::StrStyle,
+    pub span: Span,
+    /// The raw string literal, with no escaping or processing.
+    ///
+    /// Generally only useful for lints that care about the raw bytes the user wrote.
+    pub uncooked_symbol: (ast::token::LitKind, Symbol),
+}
 
 /// - `Ok` is returned when the conversion to a string literal is unsuccessful,
 /// but another type of expression is obtained instead.
@@ -90,7 +100,12 @@ pub(crate) fn expr_to_spanned_string<'a>(
     ExpandResult::Ready(Err(match expr.kind {
         ast::ExprKind::Lit(token_lit) => match ast::LitKind::from_token_lit(token_lit) {
             Ok(ast::LitKind::Str(s, style)) => {
-                return ExpandResult::Ready(Ok((s, style, expr.span)));
+                return ExpandResult::Ready(Ok(ExprToSpannedString {
+                    symbol: s,
+                    style,
+                    span: expr.span,
+                    uncooked_symbol: (token_lit.kind, token_lit.symbol),
+                }));
             }
             Ok(ast::LitKind::ByteStr(..)) => {
                 let mut err = cx.dcx().struct_span_err(expr.span, err_msg);
@@ -128,7 +143,7 @@ pub(crate) fn expr_to_string(
             Ok((err, _)) => err.emit(),
             Err(guar) => guar,
         })
-        .map(|(symbol, style, _)| (symbol, style))
+        .map(|ExprToSpannedString { symbol, style, .. }| (symbol, style))
     })
 }
 
@@ -183,7 +198,7 @@ pub(crate) fn get_single_str_spanned_from_tts(
             Ok((err, _)) => err.emit(),
             Err(guar) => guar,
         })
-        .map(|(symbol, _style, span)| (symbol, span))
+        .map(|ExprToSpannedString { symbol, span, .. }| (symbol, span))
     })
 }
 

compiler/rustc_lint/src/hidden_unicode_codepoints.rs

@@ -82,7 +82,36 @@ impl HiddenUnicodeCodepoints {
             sub,
         });
     }
+
+    fn check_literal(
+        &mut self,
+        cx: &EarlyContext<'_>,
+        text: Symbol,
+        lit_kind: ast::token::LitKind,
+        span: Span,
+        label: &'static str,
+    ) {
+        if !contains_text_flow_control_chars(text.as_str()) {
+            return;
+        }
+        let (padding, point_at_inner_spans) = match lit_kind {
+            // account for `"` or `'`
+            ast::token::LitKind::Str | ast::token::LitKind::Char => (1, true),
+            // account for `c"`
+            ast::token::LitKind::CStr => (2, true),
+            // account for `r###"`
+            ast::token::LitKind::StrRaw(n) => (n as u32 + 2, true),
+            // account for `cr###"`
+            ast::token::LitKind::CStrRaw(n) => (n as u32 + 3, true),
+            // suppress bad literals.
+            ast::token::LitKind::Err(_) => return,
+            // Be conservative just in case new literals do support these.
+            _ => (0, false),
+        };
+        self.lint_text_direction_codepoint(cx, text, span, padding, point_at_inner_spans, label);
+    }
 }
+
 impl EarlyLintPass for HiddenUnicodeCodepoints {
     fn check_attribute(&mut self, cx: &EarlyContext<'_>, attr: &ast::Attribute) {
         if let ast::AttrKind::DocComment(_, comment) = attr.kind {
@@ -97,18 +126,11 @@ impl EarlyLintPass for HiddenUnicodeCodepoints {
         // byte strings are already handled well enough by `EscapeError::NonAsciiCharInByteString`
         match &expr.kind {
             ast::ExprKind::Lit(token_lit) => {
-                let text = token_lit.symbol;
+                self.check_literal(cx, token_lit.symbol, token_lit.kind, expr.span, "literal");
-                if !contains_text_flow_control_chars(text.as_str()) {
+            }
-                    return;
+            ast::ExprKind::FormatArgs(args) => {
-                }
+                let (lit_kind, text) = args.uncooked_fmt_str;
-                let padding = match token_lit.kind {
+                self.check_literal(cx, text, lit_kind, args.span, "format string");
-                    // account for `"` or `'`
-                    ast::token::LitKind::Str | ast::token::LitKind::Char => 1,
-                    // account for `r###"`
-                    ast::token::LitKind::StrRaw(n) => n as u32 + 2,
-                    _ => return,
-                };
-                self.lint_text_direction_codepoint(cx, text, expr.span, padding, true, "literal");
             }
             _ => {}
         };

tests/ui/parser/unicode-control-codepoints.rs

@@ -1,3 +1,5 @@
+//@ edition: 2021
+
 fn main() {
     // if access_level != "us‫e‪r" { // Check if admin
     //~^ ERROR unicode codepoint changing visible direction of text present in comment
@@ -25,6 +27,14 @@ fn main() {
     //~| ERROR non-ASCII character in raw byte string literal
     println!("{:?}", '‮');
     //~^ ERROR unicode codepoint changing visible direction of text present in literal
+
+    let _ = c"‮";
+    //~^ ERROR unicode codepoint changing visible direction of text present in literal
+    let _ = cr#"‮"#;
+    //~^ ERROR unicode codepoint changing visible direction of text present in literal
+
+    println!("{{‮}}");
+    //~^ ERROR unicode codepoint changing visible direction of text present in format string
 }
 
 //"/*‮ } ⁦if isAdmin⁩ ⁦ begin admins only */"

tests/ui/parser/unicode-control-codepoints.stderr

@@ -1,5 +1,5 @@
 error: unicode escape in byte string
-  --> $DIR/unicode-control-codepoints.rs:6:26
+  --> $DIR/unicode-control-codepoints.rs:8:26
    |
 LL |     println!("{:?}", b"us\u{202B}e\u{202A}r");
    |                          ^^^^^^^^ unicode escape in byte string
@@ -7,7 +7,7 @@ LL |     println!("{:?}", b"us\u{202B}e\u{202A}r");
    = help: unicode escape sequences cannot be used as a byte or in a byte string
 
 error: unicode escape in byte string
-  --> $DIR/unicode-control-codepoints.rs:6:35
+  --> $DIR/unicode-control-codepoints.rs:8:35
    |
 LL |     println!("{:?}", b"us\u{202B}e\u{202A}r");
    |                                   ^^^^^^^^ unicode escape in byte string
@@ -15,7 +15,7 @@ LL |     println!("{:?}", b"us\u{202B}e\u{202A}r");
    = help: unicode escape sequences cannot be used as a byte or in a byte string
 
 error: non-ASCII character in byte string literal
-  --> $DIR/unicode-control-codepoints.rs:16:26
+  --> $DIR/unicode-control-codepoints.rs:18:26
    |
 LL |     println!("{:?}", b"/*<2A> } <20>if isAdmin<69> <20> begin admins only ");
    |                          ^ must be ASCII but is '\u{202e}'
@@ -26,7 +26,7 @@ LL |     println!("{:?}", b"/*\xE2\x80\xAE } <20>if isAdmin<69> <20> begin admins o
    |                          ~~~~~~~~~~~~
 
 error: non-ASCII character in byte string literal
-  --> $DIR/unicode-control-codepoints.rs:16:30
+  --> $DIR/unicode-control-codepoints.rs:18:30
    |
 LL |     println!("{:?}", b"/*<2A> } <20>if isAdmin<69> <20> begin admins only ");
    |                              ^ must be ASCII but is '\u{2066}'
@@ -37,7 +37,7 @@ LL |     println!("{:?}", b"/*<2A> } \xE2\x81\xA6if isAdmin<69> <20> begin admins o
    |                              ~~~~~~~~~~~~
 
 error: non-ASCII character in byte string literal
-  --> $DIR/unicode-control-codepoints.rs:16:41
+  --> $DIR/unicode-control-codepoints.rs:18:41
    |
 LL |     println!("{:?}", b"/*<2A> } <20>if isAdmin<69> <20> begin admins only ");
    |                                         ^ must be ASCII but is '\u{2069}'
@@ -48,7 +48,7 @@ LL |     println!("{:?}", b"/*<2A> } <20>if isAdmin\xE2\x81\xA9 <20> begin admins o
    |                                         ~~~~~~~~~~~~
 
 error: non-ASCII character in byte string literal
-  --> $DIR/unicode-control-codepoints.rs:16:43
+  --> $DIR/unicode-control-codepoints.rs:18:43
    |
 LL |     println!("{:?}", b"/*<2A> } <20>if isAdmin<69> <20> begin admins only ");
    |                                           ^ must be ASCII but is '\u{2066}'
@@ -59,31 +59,31 @@ LL |     println!("{:?}", b"/*<2A> } <20>if isAdmin<69> \xE2\x81\xA6 begin admins o
    |                                           ~~~~~~~~~~~~
 
 error: non-ASCII character in raw byte string literal
-  --> $DIR/unicode-control-codepoints.rs:21:29
+  --> $DIR/unicode-control-codepoints.rs:23:29
    |
 LL |     println!("{:?}", br##"/*<2A> } <20>if isAdmin<69> <20> begin admins only "##);
    |                             ^ must be ASCII but is '\u{202e}'
 
 error: non-ASCII character in raw byte string literal
-  --> $DIR/unicode-control-codepoints.rs:21:33
+  --> $DIR/unicode-control-codepoints.rs:23:33
    |
 LL |     println!("{:?}", br##"/*<2A> } <20>if isAdmin<69> <20> begin admins only "##);
    |                                 ^ must be ASCII but is '\u{2066}'
 
 error: non-ASCII character in raw byte string literal
-  --> $DIR/unicode-control-codepoints.rs:21:44
+  --> $DIR/unicode-control-codepoints.rs:23:44
    |
 LL |     println!("{:?}", br##"/*<2A> } <20>if isAdmin<69> <20> begin admins only "##);
    |                                            ^ must be ASCII but is '\u{2069}'
 
 error: non-ASCII character in raw byte string literal
-  --> $DIR/unicode-control-codepoints.rs:21:46
+  --> $DIR/unicode-control-codepoints.rs:23:46
    |
 LL |     println!("{:?}", br##"/*<2A> } <20>if isAdmin<69> <20> begin admins only "##);
    |                                              ^ must be ASCII but is '\u{2066}'
 
 error: unicode codepoint changing visible direction of text present in comment
-  --> $DIR/unicode-control-codepoints.rs:2:5
+  --> $DIR/unicode-control-codepoints.rs:4:5
    |
 LL |     // if access_level != "us<75>e<EFBFBD>r" { // Check if admin
    |     ^^^^^^^^^^^^^^^^^^^^^^^^^-^-^^^^^^^^^^^^^^^^^^^^^^
@@ -97,7 +97,7 @@ LL |     // if access_level != "us<75>e<EFBFBD>r" { // Check if admin
    = help: if their presence wasn't intentional, you can remove them
 
 error: unicode codepoint changing visible direction of text present in comment
-  --> $DIR/unicode-control-codepoints.rs:30:1
+  --> $DIR/unicode-control-codepoints.rs:40:1
    |
 LL | //"/*<2A> } <20>if isAdmin<69> <20> begin admins only */"
    | ^^^^^-^^^-^^^^^^^^^^-^-^^^^^^^^^^^^^^^^^^^^^^
@@ -112,7 +112,7 @@ LL | //"/*<2A> } <20>if isAdmin<69> <20> begin admins only */"
    = help: if their presence wasn't intentional, you can remove them
 
 error: unicode codepoint changing visible direction of text present in literal
-  --> $DIR/unicode-control-codepoints.rs:11:22
+  --> $DIR/unicode-control-codepoints.rs:13:22
    |
 LL |     println!("{:?}", "/*<2A> } <20>if isAdmin<69> <20> begin admins only ");
    |                      ^^^-^^^-^^^^^^^^^^-^-^^^^^^^^^^^^^^^^^^^^
@@ -132,7 +132,7 @@ LL |     println!("{:?}", "/*\u{202e} } \u{2066}if isAdmin\u{2069} \u{2066} begi
    |                         ~~~~~~~~   ~~~~~~~~          ~~~~~~~~ ~~~~~~~~
 
 error: unicode codepoint changing visible direction of text present in literal
-  --> $DIR/unicode-control-codepoints.rs:14:22
+  --> $DIR/unicode-control-codepoints.rs:16:22
    |
 LL |     println!("{:?}", r##"/*<2A> } <20>if isAdmin<69> <20> begin admins only "##);
    |                      ^^^^^^-^^^-^^^^^^^^^^-^-^^^^^^^^^^^^^^^^^^^^^^
@@ -151,7 +151,7 @@ LL |     println!("{:?}", r##"/*\u{202e} } \u{2066}if isAdmin\u{2069} \u{2066} b
    |                            ~~~~~~~~   ~~~~~~~~          ~~~~~~~~ ~~~~~~~~
 
 error: unicode codepoint changing visible direction of text present in literal
-  --> $DIR/unicode-control-codepoints.rs:26:22
+  --> $DIR/unicode-control-codepoints.rs:28:22
    |
 LL |     println!("{:?}", '<27>');
    |                      ^-^
@@ -166,8 +166,56 @@ help: if you want to keep them but make them visible in your source code, you ca
 LL |     println!("{:?}", '\u{202e}');
    |                       ~~~~~~~~
 
+error: unicode codepoint changing visible direction of text present in literal
+  --> $DIR/unicode-control-codepoints.rs:31:13
+   |
+LL |     let _ = c"<22>";
+   |             ^^-^
+   |             | |
+   |             | '\u{202e}'
+   |             this literal contains an invisible unicode text flow control codepoint
+   |
+   = note: these kind of unicode codepoints change the way text flows on applications that support them, but can cause confusion because they change the order of characters on the screen
+   = help: if their presence wasn't intentional, you can remove them
+help: if you want to keep them but make them visible in your source code, you can escape them
+   |
+LL |     let _ = c"\u{202e}";
+   |               ~~~~~~~~
+
+error: unicode codepoint changing visible direction of text present in literal
+  --> $DIR/unicode-control-codepoints.rs:33:13
+   |
+LL |     let _ = cr#"<22>"#;
+   |             ^^^^-^^
+   |             |   |
+   |             |   '\u{202e}'
+   |             this literal contains an invisible unicode text flow control codepoint
+   |
+   = note: these kind of unicode codepoints change the way text flows on applications that support them, but can cause confusion because they change the order of characters on the screen
+   = help: if their presence wasn't intentional, you can remove them
+help: if you want to keep them but make them visible in your source code, you can escape them
+   |
+LL |     let _ = cr#"\u{202e}"#;
+   |                 ~~~~~~~~
+
+error: unicode codepoint changing visible direction of text present in format string
+  --> $DIR/unicode-control-codepoints.rs:36:14
+   |
+LL |     println!("{{<7B>}}");
+   |              ^^^-^^^
+   |              |  |
+   |              |  '\u{202e}'
+   |              this format string contains an invisible unicode text flow control codepoint
+   |
+   = note: these kind of unicode codepoints change the way text flows on applications that support them, but can cause confusion because they change the order of characters on the screen
+   = help: if their presence wasn't intentional, you can remove them
+help: if you want to keep them but make them visible in your source code, you can escape them
+   |
+LL |     println!("{{\u{202e}}}");
+   |                 ~~~~~~~~
+
 error: unicode codepoint changing visible direction of text present in doc comment
-  --> $DIR/unicode-control-codepoints.rs:33:1
+  --> $DIR/unicode-control-codepoints.rs:43:1
    |
 LL | /**  '<27>'); */fn foo() {}
    | ^^^^^^^^^^^^^ this doc comment contains an invisible unicode text flow control codepoint
@@ -177,7 +225,7 @@ LL | /**  '<27>'); */fn foo() {}
    = note: if you want to keep them but make them visible in your source code, you can escape them: '\u{202e}'
 
 error: unicode codepoint changing visible direction of text present in doc comment
-  --> $DIR/unicode-control-codepoints.rs:36:1
+  --> $DIR/unicode-control-codepoints.rs:46:1
    |
 LL | / /**
 LL | |  *
@@ -188,5 +236,5 @@ LL | |  *  '<27>'); */fn bar() {}
    = note: if their presence wasn't intentional, you can remove them
    = note: if you want to keep them but make them visible in your source code, you can escape them: '\u{202e}'
 
-error: aborting due to 17 previous errors
+error: aborting due to 20 previous errors