If an empty line is touching another, remove it, otherwise leave it

Question

Regarding is this a duplicate: There are similarly worded questions such as https://unix.stackexchange.com/questions/76061/can-sed-remove-double-newline-characters or https://stackoverflow.com/questions/27510462/how-can-i-remove-double-line-breaks-with-sed - on the popular first, although the original question arguably is the same as mine, its accepted and most upvoted question removes all empty lines, not just "when there are 2 or more together" like the question asked. Some comments complain that that answer and others behave that way, but no answers are given to leave a single empty line be. Some other answers turn duplicate empty lines into a single empty line (squeezing), rather than removing them entirely.

---

I'm looking for a scriptable way to remove back to back empty lines, but leave single empty lines there.

I'm looking to automatically clean up .srt (subtitle) files. The format requires newlines to be between subtitle sections (what to display at a particular amount of time.) Usually, if there's 2 lines to be displayed at once, the subtitle author just has the 2 lines. There's another style that some authors use of placing 2 empty lines between the lines to be displayed. On my device, this has the effect of displaying the first line only, and presumably rendering the second line off the TV.

So, I'd like to change this:

1
00:00:01,800 --> 00:00:03,802
    First line is here


    Second line is here

2
...

Into this:

1
00:00:01,800 --> 00:00:03,802
    First line is here
    Second line is here

2
...

Not that it probably needs to be handled differently, but the file format requires there be an empty line at the bottom of the file, which must be left there.

I want this to work probably by first removing trailing whitespace, then only removing all empty lines that touch another empty line. I don't want it to be anchored based off the rest of the format of a .srt, like having to do with how many lines are between numbered sections. (I've thought that all empty lines could be removed, and newlines could be added back in on lines containing only numerical characters, but I'm hoping to keep it more generic than that, ignoring the actual .srt format.)

Also, if for some reason a .srt has more than 2 lines of text, I'd like it left that way.

So, perhaps something along the lines of:

cat some.srt | sed 's/[ \t]*$//' | SOMETHING_ELSE

I'd prefer a bash, sed, or awk solution over a perl one. If I understand right, I think awk will be easier to implement it in rather than sed, being multi-line.

Answer 1

If the rest of the adjacent lines in your files are unique, and it's just the adjacent blank lines you want to remove, you could just use uniq:

uniq - report or omit repeated lines

Filter adjacent matching lines from INPUT (or standard input), writing to OUTPUT (or standard output).

With no options, matching lines are merged to the first occurrence.

Running your example file through it returns:

$ uniq testfile
1
00:00:01,800 --> 00:00:03,802
    First line is here

    Second line is here

2
...

PS. your example does not do what the subject seems to request, it deletes all the blank lines between First & Second - it doesn't leave a single empty line.

Interestingly, using uniq -u (only print unique lines) on your example file gives the results in your example output (it deletes the two blank lines, leaving none between First & Second):

$ uniq -u testfile
1
00:00:01,800 --> 00:00:03,802
    First line is here
    Second line is here

2
...