LTR within RTL

Question

I am typesetting a Gospel and a Glossary. The language is RTL but in the Glossary there are some LTR embeddings. These embeddings work correctly in Paratext, but the PTXprint output has a problem as this picture will show:

Because I do not see most images in this forum, I am not confident you do either. So, here is a verbal explanation. The body text is in Arabic script, but inserted are the following items in Roman script (including the square brackets): [Bible], [Old Testament] and [New Testament], each surrounded by Arabic script text. The single word [Bible] displays correctly, but the other two display  Testament] [Old  and  Testament] [New

I have unsuccesfully tried inserting some RTL and LTR markers in the text, but wil not get the correct result. Does anybody have any advice how to do this?

Why does Paratext display this text correcty but PTXprint does not?

Answer 1

Yes, it is certainly getting muddled with the LTR text in the middle of an RTL block of text. Here is a way to solve that by using a changes.txt rule:

at GLO "(\([a-zA-Z\s]+\))" > "\\beginL \1 \\endL"

This is very specific to your situation, and may need to be tweaked to be more useful in other situations; but it says, (only in the GLO book) search for ( open parenthesis followed by one or more latin characters or space, followed by a closing parenthesis ). Wrap what has been found in these special markers \beginL ...\endL which will change the flow of the text.

Comments

Thank you Mark, your answer was extremely helpful!

---

Here is an improved version which might be more helpful in certain situations:
at GLO "(\([a-zA-Z]+?)\s([a-zA-Z]+?\))" > "\\beginL \1 \2 \\endL ~"