CPP: Improve accuracy further.

Author: geoffw0

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

@@ -6,30 +6,43 @@ import semmle.code.cpp.Preprocessor
  * Holds if comment `c` indicates that it might be in an auto-generated file, for
  * example because it contains the text "auto-generated by".
  */
-private predicate autogeneratedComment(Comment c) {
+private bindingset[comment] predicate autogeneratedComment(string comment) {
   // ?s = include newlines in anything (`.`)
   // ?i = ignore case
-  c.getContents().regexpMatch("(?si).*(" +
+  exists(string cond |
+    cond =
+      // generated by (not mid-sentence)
+      //"([^a-z\\s/\\*\\r\\n] generated by )|" +
+      "(^ generated by[^a-z])|" +
+      "(! generated by[^a-z])|" +
 
-    // auto-generated, automatically generated etc.
-    "(auto[\\w-]*\\s*?generated)|" +
+      // generated file
+      "(generated file)|" +
 
-    // generated by (at beginning of sentence)
-    "([^a-z\\s\\*\\r\\n][\\s\\*\\r\\n]*(generated by)[^a-z])|" +
+      // file [is/was/has been] generated
+      "(file( is| was| has been)? generated)|" +
 
-    // generated file
-    "(generated file)|" +
-
-    // file [is] generated
-    "(file( is)? generated)|" +
-
-    // changes made in this file will be lost
-    "(changes made in this file will be lost)|" +
+      // changes made in this file will be lost
+      "(changes made in this file will be lost)|" +
+  
+      // do not edit/modify
+      "(^ do(n't|nt| not) (hand-?)?(edit|modify))|" +
+      "(! do(n't|nt| not) (hand-?)?(edit|modify))" and
 
-    // do not edit/modify
-    "(do(n't|nt| not) (edit|modify))" +
+    comment.regexpMatch("(?si).*(" +
+      // replace `generated` with a regexp that also catches things like
+      // `auto-generated`.
+      cond.replaceAll("generated", "(auto[\\w-]*[\\s/\\*\\r\\n]*)?generated")
 
-  ").*")
+      // replace `!` with a regexp for end-of-sentence / separator characters.
+      .replaceAll("!", "[\\.\\?\\!\\-\\;\\,]")
+
+      // replace ` ` with a regexp for one or more whitespace characters
+      // (including newlines and `/*`).
+      .replaceAll(" ", "[\\s/\\*\\r\\n]+") +
+      ").*"
+    )
+  )
 }
 
 /**