JavaScript: Don't extract files in tsconfig.json outDir

Author: tausbn

javascript/extractor/src/com/semmle/js/dependencies/tsconfig/CompilerOptions.java

@@ -0,0 +1,13 @@
+package com.semmle.js.dependencies.tsconfig;
+
+public class CompilerOptions {
+    private String outDir;
+
+    public String getOutDir() {
+        return outDir;
+    }
+
+    public void setOutDir(String outDir) {
+        this.outDir = outDir;
+    }
+}

javascript/extractor/src/com/semmle/js/dependencies/tsconfig/TsConfigJson.java

@@ -0,0 +1,13 @@
+package com.semmle.js.dependencies.tsconfig;
+
+public class TsConfigJson {
+  private CompilerOptions compilerOptions;
+
+  public CompilerOptions getCompilerOptions() {
+    return compilerOptions;
+  }
+
+  public void setCompilerOptions(CompilerOptions compilerOptions) {
+    this.compilerOptions = compilerOptions;
+  }
+}

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

@@ -39,6 +39,8 @@
 
 import com.google.gson.Gson;
 import com.google.gson.JsonParseException;
+import com.semmle.js.dependencies.tsconfig.TsConfigJson;
+import com.semmle.js.dependencies.tsconfig.CompilerOptions;
 import com.semmle.js.dependencies.AsyncFetcher;
 import com.semmle.js.dependencies.DependencyResolver;
 import com.semmle.js.dependencies.packument.PackageJson;
@@ -745,6 +747,26 @@ private CompletableFuture<?> extractSource() throws IOException {
         .filter(p -> !isFileTooLarge(p))
         .sorted(PATH_ORDERING)
         .collect(Collectors.toCollection(() -> new LinkedHashSet<>()));
+    // exclude files in output directories as configured in tsconfig.json
+    final List<Path> outDirs = new ArrayList<>();
+    for (Path cfg : tsconfigFiles) {
+      try {
+        String txt = new WholeIO().read(cfg);
+        TsConfigJson root = new Gson().fromJson(txt, TsConfigJson.class);
+        if (root != null && root.getCompilerOptions() != null) {
+          if (root.getCompilerOptions().getOutDir() == null) {
+            // no outDir specified, so skip this tsconfig.json
+            continue;
+          }
+          Path odir = cfg.getParent().resolve(root.getCompilerOptions().getOutDir()).toAbsolutePath().normalize();
+          outDirs.add(odir);
+        }
+      } catch (Exception e) {
+        // ignore malformed tsconfig or missing fields
+      }
+    }
+    // exclude files in output directories as configured in tsconfig.json
+    filesToExtract.removeIf(f -> outDirs.stream().anyMatch(od -> f.startsWith(od)));
 
     DependencyInstallationResult dependencyInstallationResult = DependencyInstallationResult.empty;
     if (!tsconfigFiles.isEmpty()) {

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

@@ -203,6 +203,24 @@ public void typescriptWrongConfig() throws IOException {
     runTest();
   }
 
+  @Test
+  public void skipFilesInTsconfigOutDir() throws IOException {
+    envVars.put("LGTM_INDEX_TYPESCRIPT", "basic");
+    // Files under outDir in tsconfig.json should be excluded
+    // Create tsconfig.json with outDir set to "dist"
+    addFile(true, LGTM_SRC, "tsconfig.json");
+    Path config = Paths.get(LGTM_SRC.toString(), "tsconfig.json");
+    Files.write(config,
+        "{\"compilerOptions\":{\"outDir\":\"dist\"}}".getBytes(StandardCharsets.UTF_8));
+    // Add files outside outDir (should be extracted)
+    addFile(true, LGTM_SRC, "src", "app.ts");
+    addFile(true, LGTM_SRC, "main.js");
+    // Add files under dist/outDir (should be skipped)
+    addFile(false, LGTM_SRC, "dist", "generated.js");
+    addFile(false, LGTM_SRC, "dist", "sub", "x.js");
+    runTest();
+  }
+
   @Test
   public void includeFile() throws IOException {
     envVars.put("LGTM_INDEX_INCLUDE", "tst.js");