Version: (using KDE KDE 3.5.5) Installed from: Ubuntu Packages When I generate manpages from .po files, created manpages contain Polish letters in the form of &#number; instead of proper unicode characters. Example (generated from stable/l10n/pl/docs/kdeedu/keduca/man-keduca.1.docbook): KEDUCA(1) Podręcznik użytkownika KDE KEDUCA(1) NAME keduca - Program do przeprowadzanie testów i egzaminów SYNOPSIS keduca [plik] [Standardowe opcje KDE] [Standardowe opcje Qt] OPIS KEduca jest przydatnym programem dzięki któremu można przeprowadzać testy w formie interaktywnych formularzy. OPCJE plik Nazwa pliku do załadowania ZOBACZ TAKżE Szczegółowa dokumentacja jest dostępna pod adresem help:/keduca (należy go wprowadzić jako URL w programie Konqueror, albo wykonać polecenie: khelpcenter help:/keduca). Więcej informacji na temat programu dostępne jest na stronach środowiska KDE: http://edu.kde.org/keduca/ AUTORZY Autorem programu KEduca jest Javier Campos K Desktop Environment 16 marzec 2005 KEDUCA(1)
I think this is a fairly common problem with XSLT and non-XML target formats - there's a similar problem in the DocBook->PDF stuff I've been working on. I don't know of a proper fix, although perhaps there's an xsltproc option somewhere
If it's an XSLT issue, make sure you have the following in your XSLT: <xsl:output method="text" encoding="UTF-8"/> If you get numeric entities with that xsl:output declaration, there's a bug in your XSLT processor.
Ah, thanks Shaun. I'll look into that.
And the result of looking into is?
Note that the way we deal with this in the DocBook Project XSL stylesheets is to use a character map to convert the Unicode characters to whatever equivalent roff escapes they might have. https://docbook.svn.sourceforge.net/svnroot/docbook/trunk/xsl/manpages/charmap.groff.xsl That file contains mapping for more than 800 Unicode characters.
Yeah, but it doesn't contain, as far as I can see, ISO-8859-2 and, in turn, Polish letters (for example s-acute). Can you provide some instructions how to create such mapping for ISO-8859-2? I would like to add this mapping and commit it, so that in KDE 3.5.8 the problem is solved.
> Yeah, but it doesn't contain, as far as I can see, ISO-8859-2 and, > in turn, Polish letters (for example s-acute). I don't think there are any groff escapes for those characters. I'm not sure but I think the way to get those characters is to author/encode your source in ISO-8859-2, and set the stylesheet output encoding to ISO-8859-2 also, which will generate a man page that is viewable by users who have their locale set to ISO-8859-2. > Can you provide some instructions how to create such mapping for > ISO-8859-2? I would like to add this mapping and commit it, so > that in KDE 3.5.8 the problem is solved. There is no way to create such a mapping if groff does not already provide groff escapes for those characters. Information about the available groff escapes is in the groff_char(7) man page, so if you do "man 7 groff_char" you can look through the list there. I don't know if that list is complete, but if it's not, I'm not sure where else to look.
Can you please check if this bug is still valid for KDE SC >=4.6.x? Manpage generation now relies on upstream stylesheets, so the bug should be (hopefully) fixed.
Checked with master/4.7 compiled from sources, bug is no longer valid.
NEEDINFO for 7 years, I suspect it has been really fixed for a while.