Rollup merge of #145747 - joshtriplett:builtin-diag-dyn, r=jdonszelmann

Refactor lint buffering to avoid requiring a giant enum

Lint buffering currently relies on a giant enum `BuiltinLintDiag` containing all the lints that might potentially get buffered. In addition to being an unwieldy enum in a central crate, this also makes `rustc_lint_defs` a build bottleneck: it depends on various types from various crates (with a steady pressure to add more), and many crates depend on it.

Having all of these variants in a separate crate also prevents detecting when a variant becomes unused, which we can do with a dedicated type defined and used in the same crate.

Refactor this to use a dyn trait, to allow using `LintDiagnostic` types directly.

Because the existing `BuiltinLintDiag` requires some additional types in order to decorate some variants, which are only available later in `rustc_lint`, use an enum `DecorateDiagCompat` to handle both the `dyn LintDiagnostic` case and the `BuiltinLintDiag` case.

---

With the infrastructure in place, use it to migrate three of the enum variants to use `LintDiagnostic` directly, as a proof of concept and to demonstrate that the net result is a reduction in code size and a removal of a boilerplate-heavy layer of indirection.

Also remove an unused `BuiltinLintDiag` variant.

3 files changed, 98 insertions, 10 deletions

diff --git a/compiler/rustc_parse/messages.ftl b/compiler/rustc_parse/messages.ftl
index 958bf1fde77..4ca2f57bd87 100644
--- a/compiler/rustc_parse/messages.ftl
+++ b/compiler/rustc_parse/messages.ftl
@@ -359,6 +359,20 @@ parse_generics_in_path = unexpected generic arguments in path
 
 parse_help_set_edition_cargo = set `edition = "{$edition}"` in `Cargo.toml`
 parse_help_set_edition_standalone = pass `--edition {$edition}` to `rustc`
+
+parse_hidden_unicode_codepoints = unicode codepoint changing visible direction of text present in {$label}
+    .label = this {$label} contains {$count ->
+        [one] an invisible
+        *[other] invisible
+    } unicode text flow control {$count ->
+        [one] codepoint
+        *[other] codepoints
+    }
+    .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
+    .suggestion_remove = if their presence wasn't intentional, you can remove them
+    .suggestion_escape = if you want to keep them but make them visible in your source code, you can escape them
+    .no_suggestion_note_escape = if you want to keep them but make them visible in your source code, you can escape them: {$escaped}
+
 parse_if_expression_missing_condition = missing condition for `if` expression
     .condition_label = expected condition here
     .block_label = if this block is the condition of the `if` expression, then it must be followed by another block
diff --git a/compiler/rustc_parse/src/errors.rs b/compiler/rustc_parse/src/errors.rs
index 7f32f1cf030..797d4830c2f 100644
--- a/compiler/rustc_parse/src/errors.rs
+++ b/compiler/rustc_parse/src/errors.rs
@@ -8,8 +8,9 @@ use rustc_ast::{Path, Visibility};
 use rustc_errors::codes::*;
 use rustc_errors::{
     Applicability, Diag, DiagCtxtHandle, Diagnostic, EmissionGuarantee, Level, Subdiagnostic,
+    SuggestionStyle,
 };
-use rustc_macros::{Diagnostic, Subdiagnostic};
+use rustc_macros::{Diagnostic, LintDiagnostic, Subdiagnostic};
 use rustc_session::errors::ExprParenthesesNeeded;
 use rustc_span::edition::{Edition, LATEST_STABLE_EDITION};
 use rustc_span::{Ident, Span, Symbol};
@@ -3601,3 +3602,76 @@ pub(crate) struct ExpectedRegisterClassOrExplicitRegister {
     #[primary_span]
     pub(crate) span: Span,
 }
+
+#[derive(LintDiagnostic)]
+#[diag(parse_hidden_unicode_codepoints)]
+#[note]
+pub(crate) struct HiddenUnicodeCodepointsDiag {
+    pub label: String,
+    pub count: usize,
+    #[label]
+    pub span_label: Span,
+    #[subdiagnostic]
+    pub labels: Option<HiddenUnicodeCodepointsDiagLabels>,
+    #[subdiagnostic]
+    pub sub: HiddenUnicodeCodepointsDiagSub,
+}
+
+pub(crate) struct HiddenUnicodeCodepointsDiagLabels {
+    pub spans: Vec<(char, Span)>,
+}
+
+impl Subdiagnostic for HiddenUnicodeCodepointsDiagLabels {
+    fn add_to_diag<G: EmissionGuarantee>(self, diag: &mut Diag<'_, G>) {
+        for (c, span) in self.spans {
+            diag.span_label(span, format!("{c:?}"));
+        }
+    }
+}
+
+pub(crate) enum HiddenUnicodeCodepointsDiagSub {
+    Escape { spans: Vec<(char, Span)> },
+    NoEscape { spans: Vec<(char, Span)> },
+}
+
+// Used because of multiple multipart_suggestion and note
+impl Subdiagnostic for HiddenUnicodeCodepointsDiagSub {
+    fn add_to_diag<G: EmissionGuarantee>(self, diag: &mut Diag<'_, G>) {
+        match self {
+            HiddenUnicodeCodepointsDiagSub::Escape { spans } => {
+                diag.multipart_suggestion_with_style(
+                    fluent::parse_suggestion_remove,
+                    spans.iter().map(|(_, span)| (*span, "".to_string())).collect(),
+                    Applicability::MachineApplicable,
+                    SuggestionStyle::HideCodeAlways,
+                );
+                diag.multipart_suggestion(
+                    fluent::parse_suggestion_escape,
+                    spans
+                        .into_iter()
+                        .map(|(c, span)| {
+                            let c = format!("{c:?}");
+                            (span, c[1..c.len() - 1].to_string())
+                        })
+                        .collect(),
+                    Applicability::MachineApplicable,
+                );
+            }
+            HiddenUnicodeCodepointsDiagSub::NoEscape { spans } => {
+                // FIXME: in other suggestions we've reversed the inner spans of doc comments. We
+                // should do the same here to provide the same good suggestions as we do for
+                // literals above.
+                diag.arg(
+                    "escaped",
+                    spans
+                        .into_iter()
+                        .map(|(c, _)| format!("{c:?}"))
+                        .collect::<Vec<String>>()
+                        .join(", "),
+                );
+                diag.note(fluent::parse_suggestion_remove);
+                diag.note(fluent::parse_no_suggestion_note_escape);
+            }
+        }
+    }
+}
diff --git a/compiler/rustc_parse/src/lexer/mod.rs b/compiler/rustc_parse/src/lexer/mod.rs
index 7c7e7e50b27..9792240a548 100644
--- a/compiler/rustc_parse/src/lexer/mod.rs
+++ b/compiler/rustc_parse/src/lexer/mod.rs
@@ -543,21 +543,21 @@ impl<'psess, 'src> Lexer<'psess, 'src> {
             })
             .collect();
 
+        let label = label.to_string();
         let count = spans.len();
-        let labels = point_at_inner_spans.then_some(spans.clone());
+        let labels = point_at_inner_spans
+            .then_some(errors::HiddenUnicodeCodepointsDiagLabels { spans: spans.clone() });
+        let sub = if point_at_inner_spans && !spans.is_empty() {
+            errors::HiddenUnicodeCodepointsDiagSub::Escape { spans }
+        } else {
+            errors::HiddenUnicodeCodepointsDiagSub::NoEscape { spans }
+        };
 
         self.psess.buffer_lint(
             TEXT_DIRECTION_CODEPOINT_IN_LITERAL,
             span,
             ast::CRATE_NODE_ID,
-            BuiltinLintDiag::HiddenUnicodeCodepoints {
-                label: label.to_string(),
-                count,
-                span_label: span,
-                labels,
-                escape: point_at_inner_spans && !spans.is_empty(),
-                spans,
-            },
+            errors::HiddenUnicodeCodepointsDiag { label, count, span_label: span, labels, sub },
         );
     }