Add ClassifyFiles.ql

Author: aibaars

ql/lib/codeql/ruby/filters/GeneratedCode.qll

@@ -0,0 +1,43 @@
+/** Provides classes for detecting generated code. */
+
+private import ruby
+private import codeql.ruby.ast.internal.TreeSitter
+
+/** A source file that contains generated code. */
+abstract class GeneratedCodeFile extends RubyFile { }
+
+/** A file contining comments suggesting it contains generated code. */
+class GeneratedCommentFile extends GeneratedCodeFile {
+  GeneratedCommentFile() { this = any(GeneratedCodeComment c).getLocation().getFile() }
+}
+
+/** A comment line that indicates generated code. */
+abstract class GeneratedCodeComment extends Ruby::Comment { }
+
+/**
+ * A generic comment line that suggests that the file is generated.
+ */
+class GenericGeneratedCodeComment extends GeneratedCodeComment {
+  GenericGeneratedCodeComment() {
+    exists(string line, string entity, string was, string automatically | line = getValue() |
+      entity = "file|class|art[ei]fact|module|script" and
+      was = "was|is|has been" and
+      automatically = "automatically |mechanically |auto[- ]?" and
+      line.regexpMatch("(?i).*\\bThis (" + entity + ") (" + was + ") (" + automatically +
+          ")?generated\\b.*")
+    )
+  }
+}
+
+/** A comment warning against modifications. */
+class DontModifyMarkerComment extends GeneratedCodeComment {
+  DontModifyMarkerComment() {
+    exists(string line | line = getValue() |
+      line.regexpMatch("(?i).*\\bGenerated by\\b.*\\bDo not edit\\b.*") or
+      line.regexpMatch("(?i).*\\bAny modifications to this file will be lost\\b.*")
+    )
+  }
+}
+
+/** Holds if `file` looks like it contains generated code. */
+predicate isGeneratedCode(GeneratedCodeFile file) { any() }

ql/src/filters/ClassifyFiles.ql

@@ -0,0 +1,20 @@
+/**
+ * @name Classify files
+ * @description This query produces a list of all files in a database
+ *              that are classified as generated code or test code.
+ *
+ *              Used by LGTM.
+ * @kind file-classifier
+ * @id rb/file-classifier
+ */
+
+import ruby
+import codeql.ruby.filters.GeneratedCode
+
+predicate classify(File f, string category) {
+  f instanceof GeneratedCodeFile and category = "generated"
+}
+
+from File f, string category
+where classify(f, category)
+select f, category