Unicode chars in URL with XeLaTeX

Question

I want to print an URL with greek chars (compiling with XeLaTeX), but the below script fails to display the greek chars in the URL (they are displayed as underscores), while if I enclose the greek chars of URL in \textgreek I get an error.

\documentclass{book}
\usepackage{fontspec}
\setmainfont{Times New Roman}
\usepackage{polyglossia}
\setdefaultlanguage{english}
\setotherlanguage{greek}
%\newfontfamily\greekfont[Script=Greek, Scale=MatchUppercase, Ligatures=TeX]{Gentium Plus}
\newfontfamily\greekfont{CMU Serif}
\usepackage{hyperref,xurl}
\begin{document}
δημοτικο-κεντρο-εικαστικων
\textgreek{δημοτικο-κεντρο-εικαστικων}
\url{http://www.elemesos.com/index.php/municipalities/2018-12-20-10-21-59/item/14463-δημοτικο-κεντρο-εικαστικων-τεχνων-αποθηκεσ-παπαδακη-3-9-2014.html}
\end{document}

Greek chars outside \url are displayed correctly, either inside or outside \textgreek.

Am I doing something wrong?

P.S. I use package xurl because it breaks the links at every character.

Answer 1

As Don wrote you need at first a font with greek chars. But beside this the url must also be percent encoded in the PDF to work correctly.

This means you should use \href:

\href{http://www.dezemeros.com/index.php/2018-12-20-10-21-59/14463-%CE%B4%CE%B7%CE%BC%CE%BF%CF%84%CE%B9%CE%BA%CE%BF-%CE%BA%CE%B5%CE%BD%CF%84%CF%81%CE%BF-%CE%B5%CE%B9%CE%BA%CE%B1%CF%83%CF%84%CE%B9%CE%BA%CF%89%CE%BD-%CF%84%CE%B5%CF%87%CE%BD%CF%89%CE%BD-%CE%B4%CE%B7%CE%BC%CE%BF%CF%84%CE%B9%CE%BA%CE%BF-%CE%BA%CE%B5%CE%BD%CF%84%CF%81%CE%BF-%CE%B5%CE%B9%CE%BA%CE%B1%CF%83%CF%84%CE%B9%CE%BA%CF%89%CE%BD-%CF%84%CE%B5%CF%87%CE%BD%CF%89%CE%BD.html}
{http://www.dezemeros.com/index.php/2018-12-20-10-21-59/14463-δημοτικο-κεντρο-εικαστικων-τεχνων-δημοτικο-κεντρο-εικαστικων-τεχνων.html}

As the second argument is simply text you can use there various formatting commands, e.g. \texttt or \textgreek or \nolinkurl.

With the new pdfmanagement of LaTeX, which is currently in the testphase you can let url do the percent encoding with the option urlencode:

\DocumentMetadata{pdfversion=2.0}
\documentclass{book}
\usepackage{fontspec}
\setmonofont{CMU Typewriter Text}
\usepackage{hyperref,xurl}
\begin{document}
\url[urlencode]{http://www.dezemeros.com/index.php/2018-12-20-10-21-59/14463-δημοτικο-κεντρο-εικαστικων-τεχνων-δημοτικο-κεντρο-εικαστικων-τεχνων.html}
\end{document}

Answer 2

The problem is that there's not really a fallback mechanism for the Unicode characters to be output in another font if the current font lacks them. lmtt, which is the default typewriter font in XeLaTeX, does not have the Greek characters. There is, however, CMU Typewriter Text which does have them. Adding the line \setmonofont{CMU Typewriter Text}¹ to your document preamble will switch to that variation of Computer Modern and you'll get your Greek characters.

I'd also note that in your document, the first instance of δημοτικο-κεντρο-εικαστικων is set in Times New Roman (which has been extended to cover Greek in Unicode). Your second instance, enclosed in \textgreek uses the designated Greek font of CMU Serif, although I would be inclined to omit the \newfontfamily\greekfont to go for that font (although it appears that you may have done this to make the MWE manageable for those without Gentium Plus).

---

1. Or any monospace font that has Unicode coverage for Greek.