feat: highlight rust string interpolation macros that use format_args! (#13533)

Co-authored-by: Nik Revenco <[email protected]>

Author: nik-rev

book/src/generated/lang-support.md

@@ -199,6 +199,7 @@
 | rst | ✓ |  |  |  |
 | ruby | ✓ | ✓ | ✓ | `ruby-lsp`, `solargraph` |
 | rust | ✓ | ✓ | ✓ | `rust-analyzer` |
+| rust-format-args | ✓ |  |  |  |
 | sage | ✓ | ✓ |  |  |
 | scala | ✓ | ✓ | ✓ | `metals` |
 | scheme | ✓ |  | ✓ |  |

helix-term/tests/test/commands.rs

@@ -734,7 +734,7 @@ async fn surround_replace_ts() -> anyhow::Result<()> {
     const INPUT: &str = r#"\
 fn foo() {
     if let Some(_) = None {
-        todo!("f#[|o]#o)");
+        testing!("f#[|o]#o)");
     }
 }
 "#;
@@ -744,7 +744,7 @@ fn foo() {
         r#"\
 fn foo() {
     if let Some(_) = None {
-        todo!('f#[|o]#o)');
+        testing!('f#[|o]#o)');
     }
 }
 "#,
@@ -757,7 +757,7 @@ fn foo() {
         r#"\
 fn foo() {
     if let Some(_) = None [
-        todo!("f#[|o]#o)");
+        testing!("f#[|o]#o)");
     ]
 }
 "#,
@@ -770,7 +770,7 @@ fn foo() {
         r#"\
 fn foo() {
     if let Some(_) = None {
-        todo!{"f#[|o]#o)"};
+        testing!{"f#[|o]#o)"};
     }
 }
 "#,

helix-term/tests/test/movement.rs

@@ -379,9 +379,9 @@ async fn match_around_closest_ts() -> anyhow::Result<()> {
     test_with_config(
         AppBuilder::new().with_file("foo.rs", None),
         (
-            r#"fn main() {todo!{"f#[|oo]#)"};}"#,
+            r#"fn main() {testing!{"f#[|oo]#)"};}"#,
             "mam",
-            r#"fn main() {todo!{#[|"foo)"]#};}"#,
+            r#"fn main() {testing!{#[|"foo)"]#};}"#,
         ),
     )
     .await?;

languages.toml

@@ -4360,3 +4360,13 @@ file-types = [ { glob = "dunst/dunstrc" } ]
 [[grammar]]
 name = "dunstrc"
 source = { git = "https://github.com/rotmh/tree-sitter-dunstrc", rev = "9cb9d5cc51cf5e2a47bb2a0e2f2e519ff11c1431" }
+
+[[language]]
+name = "rust-format-args"
+scope = "source.rust-format-args"
+file-types = []
+injection-regex = "rust-format-args"
+
+[[grammar]]
+name = "rust-format-args"
+source = { git = "https://github.com/nik-rev/tree-sitter-rustfmt", rev = "2ca0bdd763d0c9dbb1d0bd14aea7544cbe81309c" }

runtime/queries/rust-format-args/highlights.scm

@@ -0,0 +1,30 @@
+; regular escapes like `\n` are detected using another grammar
+; Here, we only detect `{{` and `}}` as escapes for `{` and `}`
+(escaped) @constant.character.escape
+
+[
+  "#"
+  (type)
+] @special
+
+[
+  (sign)
+  (fill)
+  (align)
+  (width)
+] @operator
+
+(number) @constant.numeric
+
+(colon) @punctuation
+
+(identifier) @variable
+
+; SCREAMING_CASE is assumed to be constant
+((identifier) @constant
+ (#match? @constant "^[A-Z_]+$"))
+
+[
+  "{"
+  "}"
+] @punctuation.special

runtime/queries/rust/injections.scm

@@ -97,3 +97,127 @@
     ]
   )
   (#set! injection.language "sql"))
+
+; Special language `tree-sitter-rust-format-args` for Rust macros,
+; which use `format_args!` under the hood and therefore have
+; the `format_args!` syntax.
+;
+; This language is injected into a hard-coded set of macros.
+
+; 1st argument is `format_args!`
+(
+  (macro_invocation
+    macro:
+      [
+        (scoped_identifier
+          name: (_) @_macro_name)
+        (identifier) @_macro_name
+      ]
+    (token_tree
+      . (string_literal
+        (string_content) @injection.content
+      )
+    )
+  )
+  (#any-of? @_macro_name
+    ; std
+    "print" "println" "eprint" "eprintln"
+    "format" "format_args" "todo" "panic"
+    "unreachable" "unimplemented" "compile_error"
+    ; log
+    "crit" "trace" "debug" "info" "warn" "error"
+    ; anyhow
+    "anyhow" "bail"
+    ; syn
+    "format_ident"
+    ; indoc
+    "formatdoc" "printdoc" "eprintdoc" "writedoc"
+    ; iced
+    "text"
+    ; ratatui
+    "span"
+    ; eyre
+    "eyre"
+    ; miette
+    "miette"
+  )
+  (#set! injection.language "rust-format-args")
+  (#set! injection.include-children)
+)
+
+; 2nd argument is `format_args!`
+(
+  (macro_invocation
+    macro:
+      [
+        (scoped_identifier
+          name: (_) @_macro_name)
+        (identifier) @_macro_name
+      ]
+    (token_tree
+      . (_)
+      . (string_literal
+          (string_content) @injection.content
+      )
+    )
+  )
+  (#any-of? @_macro_name
+    ; std
+    "write" "writeln" "assert" "debug_assert"
+    ; defmt
+    "expect" "unwrap"
+    ; ratatui
+    "span"
+  )
+  (#set! injection.language "rust-format-args")
+  (#set! injection.include-children)
+)
+
+; 3rd argument is `format_args!`
+(
+  (macro_invocation
+    macro:
+      [
+        (scoped_identifier
+          name: (_) @_macro_name)
+        (identifier) @_macro_name
+      ]
+    (token_tree
+      . (_)
+      . (_)
+      . (string_literal
+          (string_content) @injection.content
+      )
+    )
+  )
+  (#any-of? @_macro_name
+    ; std
+    "assert_eq" "debug_assert_eq" "assert_ne" "debug_assert_ne"
+  )
+  (#set! injection.language "rust-format-args")
+  (#set! injection.include-children)
+)
+
+; Dioxus' "rsx!" macro relies heavily on string interpolation as well. The strings can be nested very deeply
+(
+  (macro_invocation
+    macro: [
+        (scoped_identifier
+          name: (_) @_macro_name)
+        (identifier) @_macro_name
+    ]
+    ; TODO: This only captures 1 level of string literals. But in dioxus you can have
+    ; nested string literals. For instance:
+    ; 
+    ; rsx! { "{hello} world" }:
+    ; -> (token_tree (string_literal))
+    ; rsx! { div { "{hello} world" } }
+    ; -> (token_tree (token_tree (string_literal)))
+    ; rsx! { div { div { "{hello} world" } } }
+    ; -> (token_tree (token_tree (token_tree (string_literal))))
+    (token_tree (string_literal) @injection.content)
+  )
+  (#eq? @_macro_name "rsx")
+  (#set! injection.language "rust-format-args")
+  (#set! injection.include-children)
+)