From 58ee6d89a6191d6ccae0d4394ca10a7f9c4deebd Mon Sep 17 00:00:00 2001
From: Peter Wu <peter@lekensteyn.nl>
Date: Fri, 29 Apr 2016 23:10:23 +0200
Subject: Implement Matlab comments

---
 src/regex/RegexTest.java | 58 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 58 insertions(+)

diff --git a/src/regex/RegexTest.java b/src/regex/RegexTest.java
index 56d49bb..fff0796 100644
--- a/src/regex/RegexTest.java
+++ b/src/regex/RegexTest.java
@@ -125,11 +125,69 @@ public class RegexTest {
         System.out.println();
     }
 
+    static void checkMatlabComment() {
+        /* EOL                  ::= #0a     (* LF character *)
+         * WSP                  ::= " " | #09   (* horizontal space and tab *)
+         * CharWithoutEOL       ::= Char - EOL
+         * SingleLineComment    ::= "%" CharWithoutEOL*
+         * MultiLineComment     ::= "%{" WSP* EOL (CharWithoutEOL* EOL)* WSP* "%}"
+         * Comment              ::= SingleLineComment | MultiLineComment
+         *
+         * According to the Mathlab docs[1], there can be no other data on the
+         * same line. According to a Wikibooks article[2] and by manual testing
+         * in Mathlab (actually Octave), it turns out that whitespace is allowed
+         * on the same line as "%{" and "%}".  We only care about the begin/end
+         * of comment, so excluding any leading and trailing whitespace.
+         *
+         * The EOL is not well-specified, assume LF as that is the format that
+         * seems to be used[3].
+         *
+         * [1]: http://nl.mathworks.com/help/matlab/matlab_prog/comments.html
+         * [2]: https://en.wikibooks.org/wiki/MATLAB_Programming/Comments
+         * [3]: https://stackoverflow.com/q/6823168
+         */
+        String sl = "%[^\n]*";
+        String ml = "%\\{[ \t]*\n" + "([^\n]*\n)*" +  "[ \t]*%\\}";
+        RegexTest reMatlabComment = new RegexTest("(" + sl + ")|(" + ml + ")");
+
+        // single line comments
+        reMatlabComment.assertOk("%");
+        reMatlabComment.assertOk("%abc");
+        reMatlabComment.assertOk("% abc");
+        reMatlabComment.assertOk("%%");
+        reMatlabComment.assertOk("%% abc");
+        reMatlabComment.assertFail("%\n");
+        reMatlabComment.assertFail("%a\n");
+        reMatlabComment.assertFail("%%\n");
+        reMatlabComment.assertFail("%a\n%b");
+
+        // Single-line comments that could be mistaken for multi-line comments
+        reMatlabComment.assertOk("%{%}");
+        reMatlabComment.assertOk("%{%}x");
+        reMatlabComment.assertOk("%{x%}");
+        reMatlabComment.assertOk("%{%}%}");
+
+        // Multi-line comments
+        reMatlabComment.assertOk("%{\n%}");
+        reMatlabComment.assertOk("%{\n\n%}");
+        reMatlabComment.assertOk("%{\nxxx\n%}");
+        reMatlabComment.assertOk("%{\n\nxxx\n%}");
+        reMatlabComment.assertOk("%{\nxxx\n\n%}");
+        reMatlabComment.assertOk("%{ \n %}");
+        reMatlabComment.assertOk("%{ \nx\n \t %}");
+        reMatlabComment.assertFail("%{x\n%}");
+        reMatlabComment.assertFail("%{\nx%}");
+        reMatlabComment.assertFail("%{\n%}x");
+        reMatlabComment.assertFail("%{ \nx %}");
+        reMatlabComment.assertFail("%{\n%}%}");
+    }
+
     public static void main(String[] args) {
         checkId();
         checkNAT();
         checkFLOAT();
         checkString();
+        checkMatlabComment();
         System.out.println("Passed.");
     }
 }
-- 
cgit v1.2.1