Python: Detect a wider range of generated files.

Author: markshannon

python/ql/src/semmle/python/filters/GeneratedCode.qll

@@ -24,6 +24,8 @@ library class GenericGeneratedFile extends GeneratedFile {
             (lax_generated_by(this, _) or lax_generated_from(this, _)) and dont_modify(this)
             or
             strict_generated_by(this, _) or strict_generated_from(this, _)
+            or
+            auto_generated(this)
         )
     }
 
@@ -78,8 +80,15 @@ private predicate dont_modify(File f) {
      comment_or_docstring(f, _).regexpMatch("(?is).*\\b(Do not|Don't) (edit|modify|make changes)\\b.*")
 }
 
+private predicate auto_generated(File f) {
+    exists(Comment c |
+        c.getLocation().getFile() = f and
+        c.getText().regexpMatch("(?is)# *this +(code|file) +is +(auto(matically)?[ -]?generated|created automatically).*")
+    )
+}
+
 
-/** 
+/**
  * A file generated by a template engine 
  */
 abstract library class SpecificGeneratedFile extends GeneratedFile {

python/ql/test/library-tests/filters/generated/Filter.expected

@@ -1,3 +1,4 @@
+| antlr.py | none |
 | generic.py | tools/idna-data |
 | sphinx.py | Sphinx |
 | swig.py | SWIG |

python/ql/test/library-tests/filters/generated/Filter.ql

@@ -2,5 +2,9 @@
 import python
 import semmle.python.filters.GeneratedCode
 
-from GeneratedFile f
-select f.toString(), f.getTool()
+from GeneratedFile f, string tool
+where
+    tool = f.getTool()
+    or
+    not exists(f.getTool()) and tool = "none"
+select f.toString(), tool

python/ql/test/library-tests/filters/generated/antlr.py

@@ -0,0 +1,11 @@
+# $ANTLR 3.0.1 C.g 2010-02-23 09:58:53
+
+from antlr3 import *
+from antlr3.compat import set, frozenset
+
+## @file
+# The file defines the parser for C source files.
+#
+# THIS FILE IS AUTO-GENERATED.
+#
+##