[arturobernalg]

https://issues.apache.org/jira/browse/JSPWIKI-492

TextUtil.cleanString keeps letters, digits and the characters of PUNCTUATION_CHARS_ALLOWED. The superscript digits are neither letters nor digits for Character.isLetterOrDigit, so MarkupParser.cleanLink removed them: Metre² became Metre when a page was referred to by a link, although a page saved under that name exists.

PUNCTUATION_CHARS_ALLOWED now also holds ¹, ² and ³ (as unicode escapes). LEGACY_CHARS_ALLOWED, used by wikifyLink, is not changed. The report is about ² and ³; I did not look at other characters.

LuceneSearchProvider also reads PUNCTUATION_CHARS_ALLOWED: it replaces those characters by spaces in the raw page name that it indexes. A page saved as Metre² is found by Metre, name:Metre* and Metre², on master and with this change; SearchManagerTest.testTitleSearchWithSuperscriptDigit covers it.

Tests: TextUtilTest.testCleanStringKeepsSuperscriptDigits and MarkupParserTest.testCleanLinkKeepsSuperscriptDigits fail on master (expected: <Metre²> but was: <Metre>) and pass with the change. TextUtilTest, MarkupParserTest, SearchManagerTest and JSPWikiMarkupParserTest pass on the branch. I did not run the whole test suite.