Hatuka*nezumi - IKEDA Soji writes:

I don't think that's the right approach. First of all, you may have just a partial text fragment, and all you want to know just where the potential line breaks are.

Unless I'm misreading UAX#14, given an arbitrary text fragment, you should be able to determine where all the valid linebreaks are, and whether the linebreaking character should be removed, or not.

So that's your first step. Just identify where all the potential line breaks are, verbatim as per UAX #14.

This can be used for purposes other than computing actual line breaks. Since potential linebreaks are, essentially, word terminators, this function can also be used to do other things, such as returning the individual words from a text fragment.

The second step is to take the list of allowed line breaks, together with the desired line length, and, using wcwidth(), compute the optimum locations of all the line breaks, given the list of valid locations for linebreaks that was computed by the first function.

There are two separate functions here: 1) Compute the list of all possible line break positions, and 2) Actually pick the right line break positions, out of the potential linebreaks computed by the first function, given the desired line length.