CPP: Improve accuracy of AutogeneratedFile.qll.

Author: geoffw0

cpp/ql/src/semmle/code/cpp/AutogeneratedFile.qll

@@ -3,11 +3,30 @@ import semmle.code.cpp.File
 import semmle.code.cpp.Preprocessor
 
 /**
- * Holds if `c` is a comment which is usually seen in autogenerated files.
- * For example, comments containing 'autogenerated' or 'generated by'.
+ * Holds if comment `c` indicates that it might be in an auto-generated file, for
+ * example because it contains the text "auto-generated by".
  */
-predicate isAutogeneratedComment(Comment c) {
-  c.getContents().regexpMatch("(?si).*(?:auto[ -]?generated|generated (?:by|file)|changes made in this file will be lost).*")
+private predicate autogeneratedComment(Comment c) {
+  // ?s = include newlines in anything (`.`)
+  // ?i = ignore case
+
+  // auto-generated, automatically generated etc.
+  c.getContents().regexpMatch("(?si).*(auto[\\w-]*\\s*?generated).*") or
+
+  // generated by (not used mid-sentence)
+  c.getContents().regexpMatch("(?si).*[^a-zA-Z\\s\\*\\r\\n][\\s\\*\\r\\n]*(generated by).*")  or
+
+  // generated file
+  c.getContents().regexpMatch("(?si).*(generated file).*")  or
+
+  // file [is] generated
+  c.getContents().regexpMatch("(?si).*(file( is)? generated).*")  or
+
+  // changes made in this file will be lost
+  c.getContents().regexpMatch("(?si).*(changes made in this file will be lost).*") or
+  
+  // do not edit/modify
+  c.getContents().regexpMatch("(?si).*(do(n't|nt| not) (edit|modify)).*")
 }
 
 /**
@@ -25,6 +44,17 @@ predicate hasPragmaDifferentFile(File f) {
   )
 }
 
+/**
+ * The maximum line number we consider to be "near the beginning" of file `f`.
+ */
+private int fileHeaderLimit(File f) {
+  result = max(int head, int line |
+    head <= 5 and
+    locations_default(_, underlyingElement(f), head, _, line, _) |
+    line
+  ) + 5
+}
+
 /**
  * Holds if the file is probably an autogenerated file.
  *
@@ -36,12 +66,10 @@ predicate hasPragmaDifferentFile(File f) {
  */
 class AutogeneratedFile extends File {
   cached AutogeneratedFile() {
-    exists(int limit, int head |
-      head <= 5 and
-      limit = max(int line | locations_default(_, underlyingElement(this), head, _, line, _)) + 5
-    |
-      exists (Comment c | c.getFile() = this and c.getLocation().getStartLine() <= limit and isAutogeneratedComment(c))
-    )
-    or hasPragmaDifferentFile(this)
+    exists(Comment c |
+      c.getFile() = this and
+      c.getLocation().getStartLine() <= fileHeaderLimit(this) and
+      autogeneratedComment(c)
+    ) or hasPragmaDifferentFile(this)
   }
 }