Closed Freed-Wu closed 1 year ago
Please refer to our wiki: https://github.com/TeX-Live/texdoc/wiki/Tab-completion
Oh, I thought the code stores the option names and option help text in a data structure, then use the data structure to parse command line input and generate
help text from a template like https://github.com/latex3/l3build/blob/main/l3build-help.lua#L41-L93. If so, we can generate completion script from
a template and add a new option to print the script like
--help
to print help text. However, there doesn't exist this data structure.
This code just use a written manually help text
https://github.com/TeX-Live/texdoc/blob/master/script/texdoclib-const.tlu#L38-L58
And parse command line input by many if-clauses in
https://github.com/TeX-Live/texdoc/blob/master/script/texdoclib-cli.tlu#L101-L146
So I think if it is hard to generate shell scripts except there are some one would rather to write one manually like https://gist.github.com/wtsnjp/07d2b4acb04d5286d12d10d6021c2716. And this written manually script must be updated when the new option be added, which is usually out-of-time, compared to the script generated from a template.
https://github.com/latex3/l3build/blob/main/l3build-arguments.lua#L37-L163 is the data structure I mentioned. https://github.com/latex3/l3build/blob/main/l3build-help.lua#L41-L93 is using the data structure to generate help text. https://github.com/latex3/l3build/pull/267/files#diff-6539b5c3e409d80383dc21dd22e22250445f066d4faf9f6deb8514e3c5bbc714R37-R133 is using the data structure to generate shell completion.
Thanks for providing the details of your request and also pointers to the existing project, l3build.
Having a data structure of the command-line options is feasible. It is also possible to support a new option to generate the shell completion script. However, supporting multiple shells such as bash, csh, zsh, fish, and so on can be costly. Choosing one or two shells with a good reason might be difficult. Since most of the tools in TeX Live do not have such a feature, I wonder if only a few TeX Live users will use the feature. I don't want to have that burden directly in this project unless many people request it. Besides, this can be just done outside of this project.
(Hereafter, off-topic for this project)
The zsh completion function that I personally maintain is currently hosted at https://github.com/wtsnjp/texlive-completions.
As a maintainer of Texdoc, I am maintaining the script according to Texdoc updates for the time being, so please do not worry too much about that script will be outdated.
Having a data structure of the command-line options is feasible.
:+1:
Choosing one or two shells with a good reason might be difficult.
I think it should be: if a shell user wants to support a shell, he will contribute code to this project. Not the maintainers support all shells by themselves. Because:
The zsh completion function that I personally maintain is currently hosted
I suggest pushing it to zsh-completions.
Anyway, I don't have the time and motivation to implement and maintain this feature. I will only review codes if you send pull requests to us. In this case, please make sure to include the corresponding description in the documentation (texdoc.tex and texdoc.1.md) and test cases under the spec directory.
Is there any project about a data structure of the command-line?
No, there isn't. You can include that part also to the pull request if you have a plan.
That is, user can
eval $(texdoc --print-completion XXsh)
in their bashrc/zshrc thenTIA!