Rollup merge of #120259 - HTGAzureX1212:HTGAzureX1212/split-diagnostics-uncommon-codepoints, r=Manishearth

Split Diagnostics for Uncommon Codepoints: Add List to Display Characters Involved

This Pull Request adds a list of the uncommon codepoints involved in the `uncommon_codepoints` lint, as outlined as a first step in #120228.

Example rendered diagnostic:
```
error: identifier contains an uncommon Unicode codepoint: 'µ'
  --> $DIR/lint-uncommon-codepoints.rs:3:7
   |
LL | const µ: f64 = 0.000001;
   |       ^
   |
note: the lint level is defined here
  --> $DIR/lint-uncommon-codepoints.rs:1:9
   |
LL | #![deny(uncommon_codepoints)]
   |         ^^^^^^^^^^^^^^^^^^^
```

(Retrying #120258.)

4 files changed, 27 insertions, 3 deletions

diff --git a/compiler/rustc_errors/src/diagnostic_impls.rs b/compiler/rustc_errors/src/diagnostic_impls.rs
index 39252dea283..f6679ae9bb3 100644
--- a/compiler/rustc_errors/src/diagnostic_impls.rs
+++ b/compiler/rustc_errors/src/diagnostic_impls.rs
@@ -110,6 +110,14 @@ impl IntoDiagnosticArg for char {
     }
 }
 
+impl IntoDiagnosticArg for Vec<char> {
+    fn into_diagnostic_arg(self) -> DiagnosticArgValue<'static> {
+        DiagnosticArgValue::StrListSepByAnd(
+            self.into_iter().map(|c| Cow::Owned(format!("{c:?}"))).collect(),
+        )
+    }
+}
+
 impl IntoDiagnosticArg for Symbol {
     fn into_diagnostic_arg(self) -> DiagnosticArgValue<'static> {
         self.to_ident_string().into_diagnostic_arg()
diff --git a/compiler/rustc_lint/messages.ftl b/compiler/rustc_lint/messages.ftl
index 33f96139f20..5652a34103b 100644
--- a/compiler/rustc_lint/messages.ftl
+++ b/compiler/rustc_lint/messages.ftl
@@ -240,7 +240,10 @@ lint_hidden_unicode_codepoints = unicode codepoint changing visible direction of
 
 lint_identifier_non_ascii_char = identifier contains non-ASCII characters
 
-lint_identifier_uncommon_codepoints = identifier contains uncommon Unicode codepoints
+lint_identifier_uncommon_codepoints = identifier contains {$codepoints_len ->
+    [one] an uncommon Unicode codepoint
+    *[other] uncommon Unicode codepoints
+}: {$codepoints}
 
 lint_ignored_unless_crate_specified = {$level}({$name}) is ignored unless specified at crate level
 
diff --git a/compiler/rustc_lint/src/lints.rs b/compiler/rustc_lint/src/lints.rs
index 94ecc7d9587..e19bb1cb62f 100644
--- a/compiler/rustc_lint/src/lints.rs
+++ b/compiler/rustc_lint/src/lints.rs
@@ -1107,7 +1107,10 @@ pub struct IdentifierNonAsciiChar;
 
 #[derive(LintDiagnostic)]
 #[diag(lint_identifier_uncommon_codepoints)]
-pub struct IdentifierUncommonCodepoints;
+pub struct IdentifierUncommonCodepoints {
+    pub codepoints: Vec<char>,
+    pub codepoints_len: usize,
+}
 
 #[derive(LintDiagnostic)]
 #[diag(lint_confusable_identifier_pair)]
diff --git a/compiler/rustc_lint/src/non_ascii_idents.rs b/compiler/rustc_lint/src/non_ascii_idents.rs
index 00f87a5af80..f78b32ce5e7 100644
--- a/compiler/rustc_lint/src/non_ascii_idents.rs
+++ b/compiler/rustc_lint/src/non_ascii_idents.rs
@@ -190,7 +190,17 @@ impl EarlyLintPass for NonAsciiIdents {
             if check_uncommon_codepoints
                 && !symbol_str.chars().all(GeneralSecurityProfile::identifier_allowed)
             {
-                cx.emit_span_lint(UNCOMMON_CODEPOINTS, sp, IdentifierUncommonCodepoints);
+                let codepoints: Vec<_> = symbol_str
+                    .chars()
+                    .filter(|c| !GeneralSecurityProfile::identifier_allowed(*c))
+                    .collect();
+                let codepoints_len = codepoints.len();
+
+                cx.emit_span_lint(
+                    UNCOMMON_CODEPOINTS,
+                    sp,
+                    IdentifierUncommonCodepoints { codepoints, codepoints_len },
+                );
             }
         }