JS: Set SourceType correctly

Author: asgerf

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

@@ -2,6 +2,8 @@
 
 import java.nio.file.Path;
 
+import com.semmle.js.extractor.ExtractorConfig.SourceType;
+
 /**
  * Denotes where a code snippet originated from within a file.
  */
@@ -10,12 +12,14 @@ public class FileSnippet {
   private int line;
   private int column;
   private int topLevelKind;
+  private SourceType sourceType;
 
-  public FileSnippet(Path originalFile, int line, int column, int topLevelKind) {
+  public FileSnippet(Path originalFile, int line, int column, int topLevelKind, SourceType sourceType) {
     this.originalFile = originalFile;
     this.line = line;
     this.column = column;
     this.topLevelKind = topLevelKind;
+    this.sourceType = sourceType;
   }
 
   public Path getOriginalFile() {
@@ -33,4 +37,8 @@ public int getColumn() {
   public int getTopLevelKind() {
     return topLevelKind;
   }
+
+  public SourceType getSourceType() {
+    return sourceType;
+  }
 }

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

@@ -1,5 +1,6 @@
 package com.semmle.js.extractor;
 
+import java.io.File;
 import java.nio.file.Path;
 import java.util.regex.Pattern;
 
@@ -55,7 +56,7 @@ public LoCInfo extract(TextualExtractor textualExtractor) {
     for (Element elt : src.getAllElements()) {
       LoCInfo snippetLoC = null;
       if (elt.getName().equals(HTMLElementName.SCRIPT)) {
-        SourceType sourceType = getScriptSourceType(elt);
+        SourceType sourceType = getScriptSourceType(elt, textualExtractor.getExtractedFile());
         if (sourceType != null) {
           // Jericho sometimes misparses empty elements, which will show up as start tags
           // ending in "/"; we manually exclude these cases to avoid spurious syntax errors
@@ -149,26 +150,31 @@ public LoCInfo extract(TextualExtractor textualExtractor) {
    * Deduce the {@link SourceType} with which the given <code>script</code> element should be
    * extracted, returning <code>null</code> if it cannot be determined.
    */
-  private SourceType getScriptSourceType(Element script) {
+  private SourceType getScriptSourceType(Element script, File file) {
     String scriptType = getAttributeValueLC(script, "type");
     String scriptLanguage = getScriptLanguage(script);
+    
+    SourceType fallbackSourceType = config.getSourceType();
+    if (file.getName().endsWith(".vue")) {
+      fallbackSourceType = SourceType.MODULE;
+    }
 
-    if (isTypeScriptTag(script)) return config.getSourceType();
+    if (isTypeScriptTag(script)) return fallbackSourceType;
 
     // if `type` and `language` are both either missing, contain the
     // string "javascript", or if `type` is the string "text/jsx", this is a plain script
     if ((scriptType == null || scriptType.contains("javascript") || "text/jsx".equals(scriptType))
         && (scriptLanguage == null || scriptLanguage.contains("javascript")))
       // use default source type
-      return config.getSourceType();
+      return fallbackSourceType;
 
     // if `type` is "text/babel", the source type depends on the `data-plugins` attribute
     if ("text/babel".equals(scriptType)) {
       String plugins = getAttributeValueLC(script, "data-plugins");
       if (plugins != null && plugins.contains("transform-es2015-modules-umd")) {
         return SourceType.MODULE;
       }
-      return config.getSourceType();
+      return fallbackSourceType;
     }
 
     // if `type` is "module", extract as module
@@ -214,7 +220,7 @@ private LoCInfo extractSnippet(
       boolean isTypeScript) {
     if (isTypeScript) {
       Path file = textualExtractor.getExtractedFile().toPath();
-      FileSnippet snippet = new FileSnippet(file, line, column, toplevelKind);
+      FileSnippet snippet = new FileSnippet(file, line, column, toplevelKind, config.getSourceType());
       VirtualSourceRoot vroot = config.getVirtualSourceRoot();
       // Vue files are special in that they can be imported as modules, and may only contain one <script> tag.
       // For .vue files we omit the usual snippet decoration to ensure the TypeScript compiler can find it.

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

@@ -25,9 +25,9 @@ public LoCInfo extract(TextualExtractor textualExtractor) {
     ScopeManager scopeManager =
         new ScopeManager(textualExtractor.getTrapwriter(), ECMAVersion.ECMA2017);
     try {
-      SourceType sourceType = jsExtractor.establishSourceType(source, false);
       FileSnippet snippet = state.getSnippets().get(sourceFile.toPath());
-      int toplevelKind = snippet == null ? 0 : snippet.getTopLevelKind();
+      SourceType sourceType = snippet != null ? snippet.getSourceType() : jsExtractor.establishSourceType(source, false);
+      int toplevelKind = snippet != null ? snippet.getTopLevelKind() : 0;
       return jsExtractor.extract(textualExtractor, source, toplevelKind, scopeManager, sourceType, res).snd();
     } catch (ParseError e) {
       e.setPosition(locationManager.translatePosition(e.getPosition()));