Ari’s solution was my next suggestion but i had to test a new possibility before i posted..

Although it is not possible to add characters automatically, this can still be automated using tracking. Again just as in my first post you will need to create a Character style. Though this time instead of creating a “no break” style, you will set only a tracking value of around 300 (this can be adjusted later to your liking.) In the paragraph styles options you will use the GREP Style to apply your tracking style to text: (?=\:).

You can do this with paragraph styles as a grep style.
You’ll need to create a character style with the “no break” setting.
Then in the paragraph style options under the GREP Styles add a New GREP Style
In the Apply Style: field you will select the “no break” character style. In the To Text: field you will put (?=\.\.)
This will create a nonbreaking space at every space followed by two periods.

Sorry i typed that on my phone last night without getting to verify the expression.. there should be parenthesis around (eg\.\,|eg\.|eg|e\.g\.) though if there are any instances where e.g. would not be enclosed by spaces on either side, (e.g.,, at the start or end of a paragraph) you may want to use (?<=[\L\U])(eg\.\,|eg\.|eg|e\.g\.)(?=[\L\U])

my first expression should be eg\.\,|eg\.|eg|e\.g\.

In my experiences, positive lookaheads do support wildcards; though they always search for the shortest match. If you used (?=a?c) “positive lookahead of zero or one a followed by a c” it will search instead for “positive lookahead of a c” since the shortest match of “a?c” is “c”.

That being said.. In dreerr’s original expression “(?<=~P)(.+?)~b(?=~b{2})” the “(.+?)~b(?=~b{2})” part of the expressions does what it should in selecting any text followed by a carriage return that comes before two carriage returns. The only thing throwing the expression off is the positive lookbehind of a page break. which goes back to Ari’s original reply as positive lookbehinds being unable to look to a previous text frame.

Though as a workaround you can apply the paragraph style to “(?<=~P)(.+?)~b(?=~b{2})” then go through and remove the paragraph style from “.+~P(?=.+~b{3})”. It’s a bit more work but it does work in the end.

David,
I’ve never come across a situation to where i needed to include the beginningnd of paragraph symbol for a grep search. Apart from changing “HY?” to “\u+?” for the obvious reason; could you tell me the difference in using, or not using, the additional beginningnd of paragraph from my expression to yours?

Riccardo,
A side note to David’s GREP expression.. If your codes use capital and lower case letters you may need to change “\u+?” to “[|\u]+?” otherwise leave his expression the way it is.