Define special sorting for Lualatex glossaries in Japanese?

Question

I am currently struggling with creating a Japanese glossary in Japanese alphabetical order. I've checked and tried this: glossarieswithcjkcharacters however it seems not to work with Lualatex (which I was using for my entire paper), and -- on a sidenote -- when I tried to test, it gave me an error

(! Package inputenc Error: Unicode character 繝 (U+7E5D)
(inputenc)                not set up for use with LaTeX.)

So I tried to work with Lualatex and created a rather simple glossary. However the sorting neither follows the alphabetic order in english or japanese, nor the order of "first appearance" or the order of the list where they were created. Therefore I do not know how to bring them in an order. Most preferably would be the japanese alphabet, which is different from the english one - just a quick example for the first ones: (a, i, u, e, o, ka, ki, ku, ke, ko, sa, shi, su, se, so...)

so this is my code example:

    \documentclass[12pt,a4paper,headings=standardclasses,numbers=noenddot]{scrreprt}
\usepackage[left=2.50cm, right=2.50cm, top=2.50cm, bottom=2.00cm]{geometry}

\usepackage[main=ngerman, english]{babel}
\usepackage[babel, german=quotes]{csquotes}

\usepackage[no-math]{fontspec}
\newfontface\cjkfont{MogaHMinEx90}
\usepackage[no-math,match]{luatexja-preset}
\setmainjfont{MogaHMinEx90}

\usepackage{luatexja-ruby}
\ltjsetparameter{jacharrange={-3}}

\usepackage[record]{glossaries-extra}

\newrobustcmd{\cjkname}[1]{\begin{CJK}{UTF8}{min}#1\end{CJK}}
\glsnoexpandfields
\GlsXtrLoadResources[
  src={entries},% data in entries.bib
  sort=ja,% sort according to Japanese alphabet
  category=japanese
]

\glssetcategoryattribute{japanese}{glossnamefont}{textcjk}
\glssetcategoryattribute{japanese}{glossdescfont}{textcjk}
\glssetcategoryattribute{japanese}{textformat}{textcjk}

\begin{document}
オイルショック
\gls{oilshock} abc 
\printunsrtglossary
\end{document}

This is my glossary entry

    \newglossaryentry{oilshock}{name={\cjkname{オイルショック}},
  description={\cjkname{《和 oil＋shock》アラブ産油国の原油生産削減と価格の大幅引き上げが、石油を主なエネルギー資源とする先進工業諸国に与えた深刻な経済的混乱のこと。第一次は昭和48年（1973）、第二次は昭和54年（1979）。石油危機。}}}

\newglossaryentry{fms}{name={cjkname{\ltjruby[]{多|品|種少|生産}{た|ひんしゅ|しょう|せいさん}}},
  description={\cjkname{工場における生産様式の一つで，同一の工場において類似性の低い製品を多品種にしかも1品目あたりは少量生産するもの。}}}  

Is there a neat solution for this or is this completely over-engineering an only 15 page long paper? (I actually really liked the example I linked above, where you could also add dual languages with different ordering)

Sorry, if there are any open questions, as I am still pretty new to LaTeX. Please feel free to ask :) And thank you in advance!

Answer 1

Since you're using LuaLaTeX rather than PDFLaTeX (which I think the linked question is using), I don't think you need CJKutf8 (which loads inputenc). As I mentioned in the comments, makeindex isn't suitable for UTF-8 and xindy doesn't provide Japanese support, but bib2gls does.

Instead of your glossaries.tex file containing \newglossaryentry, you now need a .bib file containing the glossary data. For example, the file entries.bib contains:

% Encoding: UTF-8

@preamble{"\providecommand{\ltjruby}[3][]{#2#3}"}

@entry{oilshock,
 name={オイルショック},
 description={《和 oil＋shock》アラブ産油国の原油生産削減と価格の大幅引き上げが、石油を主なエネルギー資源とする先進工業諸国に与えた深刻な経済的混乱のこと。第一次は昭和48年（1973）、第二次は昭和54年（1979）。石油危機。}
}

@entry{fms,
 name={\ltjruby[]{多|品|種少|生産}{た|ひんしゅ|しょう|せいさん}},
 description={工場における生産様式の一つで，同一の工場において類似性の低い製品を多品種にしかも1品目あたりは少量生産するもの。}
}  

@entry{afghanistanintervention,
 name={アフガン\ltjruby{侵|攻}{しん|こう}},
 description={1979年12月の、ソ連軍によるアフガニスタンへの侵攻。ソ連は1978年調印の友好協力善隣条約に基づく、アフガニスタン政府の要請によるとしたが、反政府派の激しい抵抗と、強い国際的非難にあい、1988年に全面撤退した。}
}  

@entry{communism,
 name={ソ\ltjruby{連|共産|党}{れん| きょうさん|とう}},
 description={1991年8月24日、ゴルバチョフ・ソ連大統領は、ソ連共産党の解散を勧告する声明を出し、自らも共産党書記長の辞任を表明した。事実上の共産党解体宣言であった。}
}

@entry{keizai,
 name={\ltjruby{社会|主義|経済}{しゃかい|しゅぎ|けいざい}},
 description={マルクス主義の未来社会観（共産主義思想）に導かれて樹立されたもので、生産手段の社会的所有に基づいて、社会全体の生産と消費が計画によって調整され、各人の労働に応じて所得分配が行われる経済をいう。}
}

I've trimmed your MWE and used fontspec to set up the fonts (instead of using CJKutf8). I don't have luatexja-ruby installed, and don't know the correct syntax for \ltjruby. I've provided a definition of \ltjruby in the .bib file using \providecommand. This will be ignored if it's already defined earlier in the document (when you uncomment the luatexja-ruby related lines), but the provided definition in @preamble gives bib2gls a guide to sorting. You can modify this definition according to how you want bib2gls to sort that command. For example:

@preamble{"\providecommand{\ltjruby}[3][]{#2}"}

will instruct bib2gls to ignore the first (optional) and third argument, whereas:

@preamble{"\providecommand{\ltjruby}[3][]{#3}"}

will instruct bib2gls to ignore the first and second arguments.

Here's the modified MWE (you may need to change the font names, as appropriate):

\documentclass[12pt,a4paper,headings=standardclasses,numbers=noenddot]{scrreprt}
\usepackage[left=2.50cm, right=2.50cm, top=2.50cm, bottom=2.00cm]{geometry}

\usepackage[main=ngerman, english]{babel}
\usepackage[babel, german=quotes]{csquotes}

%\usepackage{luatexja-ruby}
%\ltjsetparameter{jacharrange={-3}} 

\usepackage{fontspec}
\setmainfont{FreeSerif}
\newfontface\cjkfont{Source Han Sans CN}

\usepackage[record]{glossaries-extra}

\newrobustcmd{\textcjk}[1]{{\cjkfont #1}}

\GlsXtrLoadResources[
  src={entries},% data in entries.bib
  sort=ja,% sort according to Japanese alphabet
  category=japanese
]

\glssetcategoryattribute{japanese}{glossnamefont}{textcjk}
\glssetcategoryattribute{japanese}{glossdescfont}{textcjk}
\glssetcategoryattribute{japanese}{textformat}{textcjk}

\begin{document}
\gls{oilshock} abc \gls{fms} def \gls{afghanistanintervention} ghi \gls{communism} jkl \gls{keizai}
\printunsrtglossary
\end{document}

I don't know any Japanese, but this should order the entries according to the locale provider's collation rules for Japanese. (See the bib2gls manual for more details about the locale provider.) You may get a different glossary list when you uncomment the luatexja-ruby lines.