JavaScript: Don't extract files with TypeScript progenitors

Author: tausbn

javascript/extractor/src/com/semmle/js/extractor/AutoBuild.java

@@ -818,9 +818,19 @@ private CompletableFuture<?> extractFiles(
    */
   private boolean isFileDerivedFromTypeScriptFile(Path path, Set<Path> extractedFiles) {
     String name = path.getFileName().toString();
-    if (!name.endsWith(".js"))
+    // only skip JS variants when a corresponding TS/TSX file was already extracted
+    if (!(name.endsWith(".js")
+          || name.endsWith(".cjs")
+          || name.endsWith(".mjs")
+          || name.endsWith(".jsx")
+          || name.endsWith(".cjsx")
+          || name.endsWith(".mjsx"))) {
       return false;
-    String stem = name.substring(0, name.length() - ".js".length());
+    }
+    // strip off extension
+    int dot = name.lastIndexOf('.');
+    String stem = dot != -1 ? name.substring(0, dot) : name;
+    // if a TS/TSX file with same base name was extracted, skip this file
     for (String ext : FileType.TYPESCRIPT.getExtensions()) {
       if (extractedFiles.contains(path.getParent().resolve(stem + ext))) {
         return true;

javascript/extractor/test/com/semmle/js/extractor/test/AutoBuildTests.java

@@ -203,6 +203,25 @@ public void typescriptWrongConfig() throws IOException {
     runTest();
   }
 
+  @Test
+  public void skipJsFilesDerivedFromTypeScriptFiles() throws IOException {
+    // JS-derived files (.js, .cjs, .mjs, .jsx, .cjsx, .mjsx) should be skipped when TS indexing
+    envVars.put("LGTM_INDEX_TYPESCRIPT", "basic");
+    // Add TypeScript sources
+    addFile(true, LGTM_SRC, "foo.ts");
+    addFile(true, LGTM_SRC, "bar.tsx");
+    // Add derived JS variants (should be skipped)
+    addFile(false, LGTM_SRC, "foo.js");
+    addFile(false, LGTM_SRC, "bar.jsx");
+    addFile(false, LGTM_SRC, "foo.cjs");
+    addFile(false, LGTM_SRC, "foo.mjs");
+    addFile(false, LGTM_SRC, "bar.cjsx");
+    addFile(false, LGTM_SRC, "bar.mjsx");
+    // A normal JS file without TS counterpart should be extracted
+    addFile(true, LGTM_SRC, "normal.js");
+    runTest();
+  }
+
   @Test
   public void skipFilesInTsconfigOutDir() throws IOException {
     envVars.put("LGTM_INDEX_TYPESCRIPT", "basic");