Merge remote-tracking branch 'upstream/source/pr/777/head' into mergeback-20190121

Author: jbj

change-notes/1.20/analysis-cpp.md

@@ -2,7 +2,7 @@
 
 ## General improvements
 
-* The logic for identifying auto-generated files via `#line` directives has been improved.
+* The logic for identifying auto-generated files via comments and `#line` directives has been improved.
 
 ## New queries
 

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

@@ -3,11 +3,45 @@ 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 bindingset[comment] predicate autogeneratedComment(string comment) {
+  // ?s = include newlines in anything (`.`)
+  // ?i = ignore case
+  exists(string cond |
+    cond =
+      // generated by (not mid-sentence)
+      "(^ generated by[^a-z])|" +
+      "(! generated by[^a-z])|" +
+
+      // generated file
+      "(generated file)|" +
+
+      // file [is/was/has been] generated
+      "(file( is| was| has been)? generated)|" +
+
+      // 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
+  
+    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]+") +
+      ").*"
+    )
+  )
 }
 
 /**
@@ -25,6 +59,48 @@ predicate hasPragmaDifferentFile(File f) {
   )
 }
 
+/**
+ * The line where the first comment in file `f` begins (maximum of 5).  This allows
+ * us to skip past any preprocessor logic or similar code before the first comment.
+ */
+private int fileFirstComment(File f) {
+  result = min(int line |
+    exists(Comment c |
+      c.getFile() = f and
+      c.getLocation().getStartLine() = line and
+      line < 5
+    )
+  ).minimum(5)
+}
+
+/**
+ * The line where the initial comments of file `f` end.  This is just before the
+ * first bit of code, excluding anything skipped over by `fileFirstComment`.
+ */
+private int fileHeaderLimit(File f) {
+  exists(int fc |
+    fc = fileFirstComment(f) and
+    result = min(int line |
+      exists(DeclarationEntry de, Location l |
+        l = de.getLocation() and
+        l.getFile() = f and
+        line = l.getStartLine() - 1 and
+        line > fc
+      ) or exists(PreprocessorDirective pd, Location l |
+        l = pd.getLocation() and
+        l.getFile() = f and
+        line = l.getStartLine() - 1 and
+        line > fc
+      ) or exists(NamespaceDeclarationEntry nde, Location l |
+        l = nde.getLocation() and
+        l.getFile() = f and
+        line = l.getStartLine() - 1 and
+        line > fc
+      ) or line = f.getMetrics().getNumberOfLines()
+    )
+  )
+}
+
 /**
  * Holds if the file is probably an autogenerated file.
  *
@@ -36,12 +112,13 @@ 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)
+    autogeneratedComment(
+      strictconcat(Comment c |
+        c.getFile() = this and
+        c.getLocation().getStartLine() <= fileHeaderLimit(this) |
+        c.getContents() order by c.getLocation().getStartLine()
+      )
+    ) or
+    hasPragmaDifferentFile(this)
   }
 }