Problem with lua pattern matching - TeX

TopAnswers TeX

Meta

Databases

TeX

Code Golf

APL

C++

.net

db<>fiddle

Java

*nix

PHP

PowerShell

Python

Rust

टेक्-मराठी

Typst

Web Client Dev

Web Server Dev

Problem with lua pattern matching

add tag

निरंजन

In this code there are two lines which ideally should appear identical, but apparently there is some problem. I tried `{%1}͡{%2}`, `%1{͡}%2` and also `%1{}͡{}%2` instead of `%1\\relax ͡%2`. None of these seem to work. Is there any way to process such sequences correctly?

```
\documentclass{article}
\usepackage{luacode}
\usepackage{fontspec}
\setlength{\parindent}{0pt}
\setmainfont{CharisSIL}

\begin{luacode}
  function tiebarup ( s )
  s = unicode.utf8.gsub
  (s,'\\texttoptiebar{(.*)(.*)}','%1\\relax ͡%2')
  return s
  end
\end{luacode}
\directlua{%
  luatexbase.add_to_callback("process_input_buffer",
  tiebarup, "tiebarup")
}%

\begin{document}
\texttoptiebar{\oe i}

œ͡i
\end{document}
```

comments

transcript

निरंजन replying to Jairo A. del Rio

Thanks for the explanations :)

Jairo A. del Rio replying to निरंजन

The LaTeX version seems to be cleverer in the sense it allows you to define a special case and a fallback (e.g. for each vowel, do something special; otherwise, do the following...). In your case, you only need a general case (`\DeclareTextComposite` would do the corresponding for exceptions and special cases)

Jairo A. del Rio replying to निरंजन

`\newcommand` and friends are just wrappers for `\def`. If you want to include a definition inside another, you need to double the number of hashes so you can distinguish arguments of inner and outer definitions. Test the following:  
```    
\def\hello#1{\def\Z##1{Hello ##1!}\Z{#1}}%
\hello{world}
```   
If you need to nest another `\def`, you will have to use four hashes and so on. As for `\begingroup` and `\endgroup`, I keep the helper macro defined only inside a group to avoid redefinitions... If simpler to you, replace them with a pair of braces.

Jairo A. del Rio replying to निरंजन

Yes, it works (and you don't need `\relax` then), but I prefer to write by Unicode numbers not to mess up with text editors :)

निरंजन replying to Jairo A. del Rio

This is pretty simple. I like it. Can I use `#1\relax ͡\relax #2`  instead of `\char"361` in this syntax?

निरंजन replying to Jairo A. del Rio

Okay. Thanks

निरंजन replying to Jairo A. del Rio

Thanks, this works, but can you please explain your code a bit? I don't understand it fully. What are two "##" used for? What is `\begingroup`, `\endgroup`? All of this is a bit alien for me. Is it any different from what you have done in the more LaTeXy version given below?

निरंजन replying to samcarter

No, it only connects two characters

Jairo A. del Rio replying to user 3.14159

`%s` only captures one space. One is able to add `*` for at least zero space characters and `+` for a minimum of one character. In this case it would be `(.*)%s+(.*)`  as `(.*)%s*(.*)`would capture only one string. A brief explanation is here: https://www.lua.org/manual/5.3/manual.html#6.4.1

user 3.14159 replying to Jairo A. del Rio

Oh, thanks, I did not know this. But maybe one then has to match `(.*)%s(.*)`?

Jairo A. del Rio replying to user 3.14159

In Lua, spaces and related characters are captured with `%s`, but it looks ugly from a TeX perspective (you have to include at least one space between vowels)

samcarter replying to joulev

Oh, good to know!

joulev replying to samcarter

Trick: if you also select text that is not in a code block (for example, from "correctly?"), you can copy all code blocks at one. Tested with Chrome and Firefox, however in FF somehow there is a blank line between every line.

user 3.14159

I never used LuaLaTeX pattern matching, but use normal pattern matching in my editor and in Perl. In those `(.*).(.*)` is ambiguous because `(.*)` stands for an arbitrary long sequence. Maybe you should match against `(.*)\s(.*)`?

Jairo A. del Rio

I'm not pretty sure I'm very fluent in LaTeX, but it seems to be more "LaTeXy" IMO:  
```    
%It works in XeLaTeX, too
\documentclass{article}
\DeclareTextCommand{\dotexttoptiebar}{TU}[2]{#1\char"361\relax#2}  
%Since there's a lot of packages, I wouldn't be surprised if another package defined this command
\providecommand\texttoptiebar[1]{\dotexttoptiebar#1}
\usepackage{fontspec}
\setlength{\parindent}{0pt}
\setmainfont{CharisSIL}
\begin{document}
\texttoptiebar{\oe i}
\texttoptiebar{ai}
\end{document}  
```

Jairo A. del Rio

As I said (and not only me, other LuaTeX users), you should be careful when using the input processing callback.

Jairo A. del Rio

As for your question, it won't work because your pattern isn't taking account of spaces. See, for example, the case below:  
```  
\documentclass{article}
\usepackage{luacode}
\usepackage{fontspec}
\setlength{\parindent}{0pt}
\setmainfont{CharisSIL}
\begin{luacode*}
	function tiebarup (s)
	s = unicode.utf8.gsub
		--This will ignore spaces in the middle
		(s,'\\texttoptiebar{(.*)%s+(.*)}','%1 ͡%2')
		return s
	end
  luatexbase.add_to_callback("process_input_buffer",
  tiebarup, "tiebarup")
\end{luacode*}
\begin{document}
%\texttoptiebar{ai} It won't work
\texttoptiebar{a i}
\texttoptiebar{\oe i}
\end{document}  
```

Jairo A. del Rio

```  
\documentclass{article}
\usepackage{luacode}
\usepackage{fontspec}
\setlength{\parindent}{0pt}
\setmainfont{CharisSIL}
\providecommand\texttoptiebar[1]%
	{\begingroup%
	\def\dotexttoptiebar##1##2{##1\char"361\relax##2}%
	\dotexttoptiebar#1%
	\endgroup}
\begin{document}
\texttoptiebar{\oe i}
\texttoptiebar{ai}
\end{document}  
```

Jairo A. del Rio

What about...?

samcarter

@निरंजन, re: [your question](#question), btw I removed the additional lua code block just because I could not copy the document in one go otherwise. Not sure if this is my browser or this site, but I can't select more than one codeblock at once, but being able to copy the code seemed more useful than syntax highlighting

samcarter

Is there any rule to where the arc thingy should be placed if you pass more than one letter to your macro?

0 Answers