Rollup merge of #101966 - crlf0710:emoji_unittest, r=Mark-Simulacrum

Add unit test for identifier Unicode emoji diagnostics

Seems current diagnostics has some support for emoji usages, however it seems outdated and incomplete. This adds a simple unit test to showcase the status quo.

Author: matthiaskrgr

src/test/ui/lexer/lex-emoji-identifiers.rs

@@ -0,0 +1,17 @@
+fn invalid_emoji_usages() {
+    let arrow↔️ = "basic emoji"; //~ ERROR: identifiers cannot contain emoji
+    // FIXME
+    let planet🪐 = "basic emoji"; //~ ERROR: unknown start of token
+    // FIXME
+    let wireless🛜 = "basic emoji"; //~ ERROR: unknown start of token
+    // FIXME
+    let key1️⃣ = "keycap sequence"; //~ ERROR: unknown start of token
+                                    //~^ WARN: identifier contains uncommon Unicode codepoints
+    let flag🇺🇳 = "flag sequence"; //~ ERROR: identifiers cannot contain emoji
+    let wales🏴 = "tag sequence"; //~ ERROR: identifiers cannot contain emoji
+    let folded🙏🏿 = "modifier sequence"; //~ ERROR: identifiers cannot contain emoji
+}
+
+fn main() {
+    invalid_emoji_usages();
+}

src/test/ui/lexer/lex-emoji-identifiers.stderr

@@ -0,0 +1,52 @@
+error: unknown start of token: \u{1fa90}
+  --> $DIR/lex-emoji-identifiers.rs:4:15
+   |
+LL |     let planet🪐 = "basic emoji";
+   |               ^^
+
+error: unknown start of token: \u{1f6dc}
+  --> $DIR/lex-emoji-identifiers.rs:6:17
+   |
+LL |     let wireless🛜 = "basic emoji";
+   |                 ^^
+
+error: unknown start of token: \u{20e3}
+  --> $DIR/lex-emoji-identifiers.rs:8:14
+   |
+LL |     let key1️⃣ = "keycap sequence";
+   |             ^
+
+error: identifiers cannot contain emoji: `arrow↔️`
+  --> $DIR/lex-emoji-identifiers.rs:2:9
+   |
+LL |     let arrow↔️ = "basic emoji";
+   |         ^^^^^^
+
+error: identifiers cannot contain emoji: `flag🇺🇳`
+  --> $DIR/lex-emoji-identifiers.rs:10:9
+   |
+LL |     let flag🇺🇳 = "flag sequence";
+   |         ^^^^^^
+
+error: identifiers cannot contain emoji: `wales🏴`
+  --> $DIR/lex-emoji-identifiers.rs:11:9
+   |
+LL |     let wales🏴 = "tag sequence";
+   |         ^^^^^^^
+
+error: identifiers cannot contain emoji: `folded🙏🏿`
+  --> $DIR/lex-emoji-identifiers.rs:12:9
+   |
+LL |     let folded🙏🏿 = "modifier sequence";
+   |         ^^^^^^^^^^
+
+warning: identifier contains uncommon Unicode codepoints
+  --> $DIR/lex-emoji-identifiers.rs:8:9
+   |
+LL |     let key1️⃣ = "keycap sequence";
+   |         ^^^^
+   |
+   = note: `#[warn(uncommon_codepoints)]` on by default
+
+error: aborting due to 7 previous errors; 1 warning emitted
+