Fix/tree sitter parsing expression failure

CHANGELOG.md

@@ -10,6 +10,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 - (Rust) Support for r# raw strings with step definition patterns ([#176](https://github.com/cucumber/language-service/pull/176))
 - (Rust) Line continuation characters in rust step definition patterns ([#179](https://github.com/cucumber/language-service/pull/179))
 - (Python) Unexpected spaces and commas in generated step definitions [#160](https://github.com/cucumber/language-service/issues/160)
+- (TypeScript) Tree sitter parser failing on class decorators ([#186](https://github.com/cucumber/language-service/pull/186))
 
 ### Added
 - (Python) Support for u-strings with step definition patterns ([#173](https://github.com/cucumber/language-service/pull/173))

src/language/SourceAnalyzer.ts

@@ -1,7 +1,12 @@
 import { RegExps, StringOrRegExp } from '@cucumber/cucumber-expressions'
 import { LocationLink } from 'vscode-languageserver-types'
 
-import { createLocationLink, makeParameterType, syntaxNode } from './helpers.js'
+import {
+  createLocationLink,
+  makeParameterType,
+  stripBlacklistedExpressions,
+  syntaxNode,
+} from './helpers.js'
 import { getLanguage } from './languages.js'
 import {
   Language,
@@ -136,7 +141,10 @@ language: ${source.languageName}
   private parse(source: Source<LanguageName>): TreeSitterTree {
     let tree: TreeSitterTree | undefined = this.treeByContent.get(source)
     if (!tree) {
-      this.treeByContent.set(source, (tree = this.parserAdapter.parser.parse(source.content)))
+      // This is currently necessary since the tree-sitter parser currently errors on certain expressions
+      const content = stripBlacklistedExpressions(source.content, source.languageName)
+      tree = this.parserAdapter.parser.parse(content)
+      this.treeByContent.set(source, tree)
     }
     return tree
   }

src/language/helpers.ts

@@ -1,7 +1,13 @@
 import { ParameterType, RegExps } from '@cucumber/cucumber-expressions'
 import { DocumentUri, LocationLink, Range } from 'vscode-languageserver-types'
 
-import { Link, NodePredicate, TreeSitterQueryMatch, TreeSitterSyntaxNode } from './types'
+import {
+  LanguageName,
+  Link,
+  NodePredicate,
+  TreeSitterQueryMatch,
+  TreeSitterSyntaxNode,
+} from './types'
 
 export function syntaxNode(match: TreeSitterQueryMatch, name: string): TreeSitterSyntaxNode | null {
   const nodes = syntaxNodes(match, name)
@@ -70,3 +76,66 @@ export function filter(
 function flatten(node: TreeSitterSyntaxNode): TreeSitterSyntaxNode[] {
   return node.children.reduce((r, o) => [...r, ...flatten(o)], [node])
 }
+
+/**
+ *
+ * This constant represents a record of language names that contain lists
+ * of regular expressions that should be stripped from the content of a file.
+ */
+export const BLACKLISTED_EXPRESSIONS: {
+  [key in LanguageName]: RegExp[]
+} = {
+  tsx: [
+    /*
+     * This regular expression matches sequences of decorators applied to a class,
+     * potentially including type parameters and arguments.
+     * The regex supports matching these patterns preceding
+     * an optionally exported class definition.
+     */
+    /(@(\w+)(?:<[^>]+>)?\s*(?:\([^)]*\))?\s*)*(?=\s*(export)*\s+class)/g,
+  ],
+  java: [],
+  c_sharp: [],
+  php: [],
+  python: [],
+  ruby: [],
+  rust: [],
+  javascript: [],
+}
+
+/**
+ *
+ * Strips blacklisted expressions from the given content.
+ *
+ * @param content The content to strip blacklisted expressions from.
+ * @param languageName The name of the language to use for stripping.
+ *
+ * @returns The content with blacklisted expressions stripped.
+ *
+ * @example
+ *
+ * ```typescript
+ * const content =
+ *  "@decorator\n" ++
+ *  "export class Foo {\n" ++
+ *  "@decorator\n" ++
+ *  "public bar() { }\n" ++
+ *  "}"
+ *
+ * const strippedContent = stripBlacklistedExpressions(content, 'tsx')
+ * console.log(strippedContent)
+ *
+ * // Output:
+ * "export class Foo {\n" ++
+ * "@decorator\n" ++
+ * "public bar() { }\n" ++
+ * "}"
+ *
+ * ```
+ */
+export function stripBlacklistedExpressions(content: string, languageName: LanguageName): string {
+  return BLACKLISTED_EXPRESSIONS[languageName].reduce(
+    (acc, regExp) => acc.replace(regExp, ''),
+    content
+  )
+}