JS: classify single-line JSON files as generated

Author: Esben Sparre Andreasen

javascript/ql/src/semmle/javascript/GeneratedCode.qll

@@ -112,6 +112,21 @@ private predicate isData(File f) {
   )
 }
 
+/**
+ * Holds if `f` is a single line that looks like a non-trivial amount of JSON data, which is often a sign of generated data code.
+ */
+private predicate isJsonLine(File f) {
+  f.getNumberOfLines() = 1 and
+  count(Expr e | e.getFile() = f) > 100 and
+  forall(Expr e | e.getFile() = f |
+    e instanceof ObjectExpr or
+    e instanceof ArrayExpr or
+    e instanceof NumberLiteral or
+    e instanceof StringLiteral or
+    e instanceof BooleanLiteral
+  )
+}
+
 /**
  * Holds if `f` is a generated HTML file.
  */
@@ -132,7 +147,6 @@ private HTML::Element getAStartingElement(File f, int l) {
   result.getFile() = f and result.getLocation().getStartLine() = l
 }
 
-
 /**
  * Gets the number of HTML elements that start at line `l` in file `f`.
  */
@@ -151,6 +165,7 @@ predicate isGenerated(TopLevel tl) {
   exists(GeneratedCodeMarkerComment gcmc | tl = gcmc.getTopLevel()) or
   hasManyInvocations(tl) or
   isData(tl.getFile()) or
+  isJsonLine(tl.getFile()) or
   isGeneratedHtml(tl.getFile())
 }
 

javascript/ql/test/query-tests/filters/ClassifyFiles/ClassifyFiles.expected

@@ -11,6 +11,7 @@
 | jquery-datatables.js:0:0:0:0 | jquery-datatables.js | library |
 | jquery-jstree.js:0:0:0:0 | jquery-jstree.js | library |
 | jquery-snippet.js:0:0:0:0 | jquery-snippet.js | library |
+| json-like.js:0:0:0:0 | json-like.js | generated |
 | jsx-old.js:0:0:0:0 | jsx-old.js | generated |
 | jsx.js:0:0:0:0 | jsx.js | generated |
 | multi-part-bundle.html:0:0:0:0 | multi-part-bundle.html | generated |