Rust: Add initial support for string interpolation

alecthomas · Jan 31, 2023 · 4bfa1bb · 4bfa1bb
1 parent f5f48e2
commit 4bfa1bb
Show file tree

Hide file tree

Showing 4 changed files with 186 additions and 1 deletion.
diff --git a/lexers/embedded/rust.xml b/lexers/embedded/rust.xml
@@ -105,6 +105,24 @@
         <pop depth="1"/>
       </rule>
     </state>
+    <state name="formatted_string">
+      <rule pattern="&#34;">
+        <token type="LiteralString"/>
+        <pop depth="1"/>
+      </rule>
+      <rule pattern="\\[&#39;&#34;\\nrt]|\\(?=\n)|\\x[0-7][0-9a-fA-F]|\\0|\\u\{[0-9a-fA-F]{1,6}\}|\{\{|\}\}">
+        <token type="LiteralStringEscape"/>
+      </rule>
+      <rule pattern="\{[^}]*\}">
+        <token type="LiteralStringInterpol"/>
+      </rule>
+      <rule pattern="[^\\&#34;\{\}]+">
+        <token type="LiteralString"/>
+      </rule>
+      <rule pattern="\\">
+        <token type="LiteralString"/>
+      </rule>
+    </state>
     <state name="string">
       <rule pattern="&#34;">
         <token type="LiteralString"/>
@@ -294,6 +312,16 @@
       <rule pattern="\b(r#)?_?([A-Z][A-Z0-9_]*){2,}\b">
         <token type="NameConstant"/>
       </rule>
+      <rule pattern="((?:e?print(?:ln)?|format(?:_args)?|panic|todo|un(?:reachable|implemented))!)(\s*)(\()(\s*)(&#34;)">
+        <bygroups>
+          <token type="NameFunctionMagic"/>
+          <token type="TextWhitespace"/>
+          <token type="Punctuation"/>
+          <token type="TextWhitespace"/>
+          <token type="LiteralString"/>
+        </bygroups>
+        <push state="formatted_string"/>
+      </rule>
       <rule pattern="([a-zA-Z_]\w*!)(\s*)(\(|\[|\{)">
         <bygroups>
           <token type="NameFunctionMagic"/>

diff --git a/lexers/testdata/rust/rust.expected b/lexers/testdata/rust/rust.expected
@@ -60,7 +60,9 @@
   {"type":"TextWhitespace","value":"    "},
   {"type":"NameFunctionMagic","value":"println!"},
   {"type":"Punctuation","value":"("},
-  {"type":"LiteralString","value":"\"rect1 is {:?}\""},
+  {"type":"LiteralString","value":"\"rect1 is "},
+  {"type":"LiteralStringInterpol","value":"{:?}"},
+  {"type":"LiteralString","value":"\""},
   {"type":"Punctuation","value":","},
   {"type":"TextWhitespace","value":" "},
   {"type":"Name","value":"rect1"},

diff --git a/lexers/testdata/rust/string_interpol.actual b/lexers/testdata/rust/string_interpol.actual
@@ -0,0 +1,29 @@
+fn main() {
+    let foo = 'x';
+
+    println!(
+        "
+        Kitchen sink
+        {{ }} - Escaped
+        {{{}}} - Escaped with inner interpol
+        {foo:#?} - Ident with pretty debug format
+        {0:-<5} - Positional with complex fill/alignment
+        ",
+        foo,
+    );
+
+    // Unconventional formatting
+    println!	(	"Hello, {foo}!");
+
+    // Exhausting the supported macros
+    eprintln!("{foo}");
+    eprint!("{foo}");
+    println!("{foo}");
+    print!("{foo}");
+    format!("{foo}");
+    format_args!("{foo}");
+    panic!("{foo}");
+    todo!("{foo}");
+    unreachable!("{foo}");
+    unimplemented!("{foo}");
+}
diff --git a/lexers/testdata/rust/string_interpol.expected b/lexers/testdata/rust/string_interpol.expected
@@ -0,0 +1,126 @@
+[
+  {"type":"Keyword","value":"fn"},
+  {"type":"Text","value":" "},
+  {"type":"NameFunction","value":"main"},
+  {"type":"Punctuation","value":"()"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"KeywordDeclaration","value":"let"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"foo"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Operator","value":"="},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralStringChar","value":"'x'"},
+  {"type":"Punctuation","value":";"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"NameFunctionMagic","value":"println!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"LiteralString","value":"\"\n        Kitchen sink\n        "},
+  {"type":"LiteralStringEscape","value":"{{"},
+  {"type":"LiteralString","value":" "},
+  {"type":"LiteralStringEscape","value":"}}"},
+  {"type":"LiteralString","value":" - Escaped\n        "},
+  {"type":"LiteralStringEscape","value":"{{"},
+  {"type":"LiteralStringInterpol","value":"{}"},
+  {"type":"LiteralStringEscape","value":"}}"},
+  {"type":"LiteralString","value":" - Escaped with inner interpol\n        "},
+  {"type":"LiteralStringInterpol","value":"{foo:#?}"},
+  {"type":"LiteralString","value":" - Ident with pretty debug format\n        "},
+  {"type":"LiteralStringInterpol","value":"{0:-\u003c5}"},
+  {"type":"LiteralString","value":" - Positional with complex fill/alignment\n        \""},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Name","value":"foo"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"CommentSingle","value":"// Unconventional formatting\n"},
+  {"type":"TextWhitespace","value":"    "},
+  {"type":"NameFunctionMagic","value":"println!"},
+  {"type":"TextWhitespace","value":"\t"},
+  {"type":"Punctuation","value":"("},
+  {"type":"TextWhitespace","value":"\t"},
+  {"type":"LiteralString","value":"\"Hello, "},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"!\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"CommentSingle","value":"// Exhausting the supported macros\n"},
+  {"type":"TextWhitespace","value":"    "},
+  {"type":"NameFunctionMagic","value":"eprintln!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"eprint!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"println!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"print!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"format!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"format_args!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"panic!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"todo!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"unreachable!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"NameFunctionMagic","value":"unimplemented!"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\""},
+  {"type":"LiteralStringInterpol","value":"{foo}"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n"}
+]