This occurred when I copied some text from Word
Some of the returns / line feeds are replaced with invisible character - ASCII “226 128 168”
I found this on the Internet as explanation:
“226 128 168”, or more precisely e2 80 a8, is the UTF-8 encoding of the unicode symbol U+2028, the “Line Separator”.
Is there a way in TBX to replace it via Regex?
I tried … but it didnt work
$Text.replace(“\x226\x128\x168”,“\n”)
$Text.replace(“\u2028”,“\n”)
At this point, would it not be easier to just fix the Word source to use a proper line break?
So the ‘bad’ character you have is:
So the decimal version would, I believe, be \x8232, vs. Unicode of \u2028. It’s hard to test as there is no test example and trying to make one’s own tests using non-printing characters is hard as how do we know we are testing the same character as you. 
This might give some indication as to why the source doc uses a 'line separator` but it has the smell of old/badly configured MS Word docs.
Thanks Mark
I will check it in my word config.
I have the same issues as well when I copy text from PDF files. In my example the PDF was created in FinalDraft.
When I use a regex engine link https://regexr.com for testing, I can replace u2028 by ([\u2028]). But this doesn’t work in TBX.
It seems most Tinderbox regex is BOOST/PCRE; some exceptions are listed here. Under the former, U+2028 is defined in regex as \x{2028}. Thus:
$Text = $Text.replace("\x{2028}","\n");
See this demo file: unicode-regex-test.tbx (82.4 KB)
I’ve offered two test stamps. The first uses a # as the replace character, the second \n. This is because \n is an invisible character like the line separator it replaces. The first test stamp gives you a more positive feedback that a replacement has actually happened.
Thanks Mark, much appreciated, this works perfectly.