feat: Add jsonnet highlighting support (#36134)

Use new sourcegraph tree-sitter parser for jsonnet and use it to do
highlighting when configured with tree-sitter.

cmd/frontend/internal/highlight/language.go

@@ -73,7 +73,8 @@ var engineConfig = syntaxEngineConfig{
 var baseEngineConfig = syntaxEngineConfig{
 	Default: EngineSyntect,
 	Overrides: map[string]EngineType{
-		"c#": EngineTreeSitter,
+		"c#":      EngineTreeSitter,
+		"jsonnet": EngineTreeSitter,
 	},
 }
 
@@ -107,6 +108,8 @@ func init() {
 	go func() {
 		conf.Watch(func() {
 			// Populate effective configuration with base configuration
+			//    We have to add here to make sure that even if there is no config,
+			//    we still update to use the defaults
 			engineConfig.Default = baseEngineConfig.Default
 			for name, engine := range baseEngineConfig.Overrides {
 				engineConfig.Overrides[name] = engine
@@ -126,7 +129,16 @@ func init() {
 			}
 
 			// Set overrides from configuration
+			//
+			// We populate the confuration with base again, because we need to
+			// create a brand new map to not take any values that were
+			// previously in the table from the last configuration.
+			//
+			// After that, we set the values from the new configuration
 			engineConfig.Overrides = map[string]EngineType{}
+			for name, engine := range baseEngineConfig.Overrides {
+				engineConfig.Overrides[name] = engine
+			}
 			for name, engine := range config.SyntaxHighlighting.Engine.Overrides {
 				if overrideEngine, ok := engineNameToEngineType(engine); ok {
 					engineConfig.Overrides[strings.ToLower(name)] = overrideEngine
@@ -180,9 +192,9 @@ func getLanguageFromConfig(config syntaxHighlightConfig, path string) (string, b
 // getLanguage will return the name of the language and default back to enry if
 // no language could be found.
 func getLanguage(path string, contents string) (string, bool) {
-	ft, found := getLanguageFromConfig(highlightConfig, path)
+	lang, found := getLanguageFromConfig(highlightConfig, path)
 	if found {
-		return ft, true
+		return lang, true
 	}
 
 	return enry.GetLanguage(path, []byte(contents)), false

docker-images/syntax-highlighter/Cargo.lock

@@ -1698,6 +1698,7 @@ dependencies = [
  "tree-sitter-c-sharp",
  "tree-sitter-go",
  "tree-sitter-highlight",
+ "tree-sitter-jsonnet",
  "tree-sitter-sql",
 ]
 
@@ -2206,6 +2207,15 @@ dependencies = [
  "tree-sitter",
 ]
 
+[[package]]
+name = "tree-sitter-jsonnet"
+version = "0.0.1"
+source = "git+https://github.com/sourcegraph/tree-sitter-jsonnet#009e6f06266f46ae07077dd6c8026ded56ab7dd8"
+dependencies = [
+ "cc",
+ "tree-sitter",
+]
+
 [[package]]
 name = "tree-sitter-sql"
 version = "0.0.2"

docker-images/syntax-highlighter/crates/sg-syntax/Cargo.toml

@@ -37,5 +37,7 @@ tree-sitter-sql = { git = "https://github.com/sourcegraph/tree-sitter-sql" }
 # is picked by tree-sitter-c-sharp, whereas tree-sitter-go picks 0.20.x
 tree-sitter-c-sharp = { git = "https://github.com/tree-sitter/tree-sitter-c-sharp" }
 
+tree-sitter-jsonnet = { git = "https://github.com/sourcegraph/tree-sitter-jsonnet" }
+
 [dev-dependencies]
 insta = "1.11.0"

docker-images/syntax-highlighter/crates/sg-syntax/queries/jsonnet/highlights.scm

@@ -0,0 +1,53 @@
+(id) @variable
+(param identifier: (id) @variable.parameter)
+(bind function: (id) @function)
+(fieldname) @string.special
+[
+  "["
+  "]"
+  "{"
+  "}"
+] @punctuation.bracket
+
+[ "error" "assert" ] @identifier.function
+
+; keyword style
+[
+  "if"
+  "then"
+  "else"
+] @conditional
+
+[
+  (local)
+  (super)
+  "function"
+  "for"
+  "in"
+  "import"
+  "importstr"
+] @keyword
+
+; Language basics
+(comment) @comment
+(number) @number
+[ (true) (false) ] @boolean
+[ (self) (dollar) ] @constant.builtin
+(binaryop) @operator
+(unaryop) @operator
+
+
+; It's possible for us to give a "special" highlight to
+; the triple ||| to start a string if we wanted with this query
+;; ((string_start) @comment
+;;  (#eq? @comment "|||"))
+
+[
+  (string_start)
+  (string_end)
+] @string.special
+
+(string_content) @string
+
+; Imports
+(import) @variable.module

docker-images/syntax-highlighter/crates/sg-syntax/src/sg_treesitter.rs

@@ -100,7 +100,7 @@ macro_rules! create_configurations {
 
 lazy_static::lazy_static! {
     static ref CONFIGURATIONS: HashMap<&'static str, HighlightConfiguration> = {
-        create_configurations!( go, sql, c_sharp )
+        create_configurations!( go, sql, c_sharp, jsonnet )
     };
 }
 

internal/gosyntect/gosyntect.go

@@ -110,6 +110,7 @@ var enryLanguageMappings = map[string]string{
 var supportedFiletypes = map[string]struct{}{
 	"go":      {},
 	"c_sharp": {},
+	"jsonnet": {},
 }
 
 // Client represents a client connection to a syntect_server.