From 5302f28e95e32f1458e6e18844f963ccaa2120fe Mon Sep 17 00:00:00 2001
From: Peter Wu <peter@lekensteyn.nl>
Date: Sat, 30 Apr 2016 00:11:42 +0200
Subject: Add Java comments check

---
 src/regex/RegexTest.java | 64 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 64 insertions(+)

diff --git a/src/regex/RegexTest.java b/src/regex/RegexTest.java
index fff0796..8345965 100644
--- a/src/regex/RegexTest.java
+++ b/src/regex/RegexTest.java
@@ -180,6 +180,69 @@ public class RegexTest {
         reMatlabComment.assertFail("%{\n%}x");
         reMatlabComment.assertFail("%{ \nx %}");
         reMatlabComment.assertFail("%{\n%}%}");
+        System.out.println();
+    }
+
+    static void checkJavaComment() {
+        /* CR                   ::= #0a
+         * LF                   ::= #0d
+         * InputCharacter       ::= Char - CR - LF
+         * LineTerminator       ::= LF | CR | CR LF
+         * NotStar              ::= InputCharacter - "*" | LineTerminator
+         * NotStarSlash         ::= InputCharacter - "*" - "/" | LineTerminator
+         *
+         * TraditionalComment   ::= "/" "*" CommentTail
+         * CommentTail          ::= "*" CommentTailStar | NotStar CommentTail
+         * CommentTailStar      ::= "/" | "*" CommentTailStar
+         * CommentTailStar      ::= NotStarSlash CommentTail
+         * EndOfLineComment     ::= "/" "/" InputCharacter*
+         * Comment              ::= TraditionalComment | EndOfLineComment
+         *
+         * Note: while the Java Language Specification dictates that
+         * UnicodeEscape tokens are recognized, here it is assumed that this
+         * step is already handled. If not, then all "/", "*", newlines and Char
+         * tokens should be either its respective literal or the UnicodeEscape
+         * that represents this literal.
+         *
+         * https://docs.oracle.com/javase/specs/jls/se8/html/jls-3.html#jls-3.7
+         */
+        String sl = "//[^\n\r]*";
+        // Match parts starting with a star, followed by CommentTailStar, but
+        // only where CommentTailStar = "*" CommentTailStar, terminated by
+        // NotStarSlash [CommentTail].  There is a special case in the "ml"
+        // regex for when CommentTailStar does not contain NotStarSlash.
+        String star = "\\*+[^/*]";
+        String nostar = "[^\\*]";
+        String ml = "/\\*" + "(" + star + "|" + nostar + ")*\\**" + "\\*/";
+        RegexTest reJavaComment = new RegexTest(sl + "|" + ml);
+
+        // Single-line comment
+        reJavaComment.assertOk("//");
+        reJavaComment.assertOk("// ok");
+        reJavaComment.assertOk("/// ok//");
+        reJavaComment.assertFail("//\n");
+        reJavaComment.assertFail("//\n//");
+
+        // Multi-line comments
+        reJavaComment.assertOk("/**/");
+        reJavaComment.assertOk("/***/");
+        reJavaComment.assertOk("/* xxx */");
+        reJavaComment.assertOk("/*\n*/");
+        reJavaComment.assertOk("/*\r\n*/");
+        reJavaComment.assertOk("/*/*/");
+        reJavaComment.assertOk("/*//*/");
+        reJavaComment.assertOk("/*///*/");
+        reJavaComment.assertOk("/** */");
+        reJavaComment.assertOk("/** /*/");
+        reJavaComment.assertOk("/*xxx\n*/");
+        reJavaComment.assertOk("/*\nxxx*/");
+        reJavaComment.assertOk("/*\nxxx\n*/");
+        reJavaComment.assertOk("/*\nxx\nx\n*/");
+        reJavaComment.assertFail("/**/ x");
+        reJavaComment.assertFail("/**///");
+        reJavaComment.assertFail("/**//*/");
+        reJavaComment.assertFail("/**//**/");
+        System.out.println();
     }
 
     public static void main(String[] args) {
@@ -188,6 +251,7 @@ public class RegexTest {
         checkFLOAT();
         checkString();
         checkMatlabComment();
+        checkJavaComment();
         System.out.println("Passed.");
     }
 }
-- 
cgit v1.2.1