Merge pull request github#5107 from asgerf/js/json-in-script-tag

codeql-ci · web-flow · commit 8a2e063af7c4 · 2021-02-08T09:52:53.000Z
Approved by erik-krogh
diff --git a/javascript/extractor/src/com/semmle/js/extractor/HTMLExtractor.java b/javascript/extractor/src/com/semmle/js/extractor/HTMLExtractor.java
@@ -331,7 +331,10 @@ private void extractSnippet(
               textualExtractor.getMetrics(),
               textualExtractor.getExtractedFile());
       Pair<Label, LoCInfo> result = extractor.extract(tx, source, toplevelKind, scopeManager);
-      emitTopLevelXmlNodeBinding(parentHtmlNode, result.fst(), context, trapWriter);
+      Label toplevelLabel = result.fst();
+      if (toplevelLabel != null) { // can be null when script ends up being parsed as JSON
+        emitTopLevelXmlNodeBinding(parentHtmlNode, toplevelLabel, context, trapWriter);
+      }
       locInfo.add(result.snd());
     } catch (ParseError e) {
       e.setPosition(scriptLocationManager.translatePosition(e.getPosition()));
diff --git a/javascript/extractor/src/com/semmle/js/extractor/Main.java b/javascript/extractor/src/com/semmle/js/extractor/Main.java
@@ -43,7 +43,7 @@ public class Main {
    * A version identifier that should be updated every time the extractor changes in such a way that
    * it may produce different tuples for the same file under the same {@link ExtractorConfig}.
    */
-  public static final String EXTRACTOR_VERSION = "2020-12-11";
+  public static final String EXTRACTOR_VERSION = "2021-02-05";
 
   public static final Pattern NEWLINE = Pattern.compile("\n");
 
diff --git a/javascript/extractor/tests/html/input/json_in_script.html b/javascript/extractor/tests/html/input/json_in_script.html
@@ -0,0 +1,5 @@
+<html>
+<script>
+{"Hello": 123}
+</script>
+</html>
diff --git a/javascript/extractor/tests/html/output/trap/json_in_script.html.trap b/javascript/extractor/tests/html/output/trap/json_in_script.html.trap
@@ -0,0 +1,34 @@
+#10000=@"/json_in_script.html;sourcefile"
+files(#10000,"/json_in_script.html","json_in_script","html",0)
+#10001=@"/;folder"
+folders(#10001,"/","")
+containerparent(#10001,#10000)
+#10002=@"loc,{#10000},0,0,0,0"
+locations_default(#10002,#10000,0,0,0,0)
+hasLocation(#10000,#10002)
+#20000=@"global_scope"
+scopes(#20000,0)
+#20001=*
+json(#20001,5,#10000,0,"{""Hello"": 123}")
+#20002=@"loc,{#10000},3,1,3,14"
+locations_default(#20002,#10000,3,1,3,14)
+json_locations(#20001,#20002)
+#20003=*
+json(#20003,2,#20001,0,"123")
+#20004=@"loc,{#10000},3,11,3,13"
+locations_default(#20004,#10000,3,11,3,13)
+json_locations(#20003,#20004)
+json_literals("123","123",#20003)
+json_properties(#20001,"Hello",#20003)
+#20005=*
+xmlElements(#20005,"html",#10000,0,#10000)
+#20006=@"loc,{#10000},1,1,5,7"
+locations_default(#20006,#10000,1,1,5,7)
+xmllocations(#20005,#20006)
+#20007=*
+xmlElements(#20007,"script",#20005,0,#10000)
+#20008=@"loc,{#10000},2,1,4,9"
+locations_default(#20008,#10000,2,1,4,9)
+xmllocations(#20007,#20008)
+numlines(#10000,5,0,0)
+filetype(#10000,"html")
