Delete specific character on previous line

Question

I deleted lines in a file with

:g/^\s*key3/d

and now have a file like this:

entry{entryname1,
    key1 = {text},
    key2 = {text},
}

entry{entryname2,
    anotherkey = {text},
    yetanotherkey = {text}
}
...

I want to delete the trailing comma after key2 (the name of the last key can change) on multiple lines in the file.

Is there a command (or multiple commands) that does the following:

for each line with the pattern ^}$:
  if the last character in the previous line is a comma:
    delete that comma

Answer 1

You can use :g again with a regular expression, but use a line offset of -1 to have the global command act on the lines above the match.

Then you can use that to apply a :s command to remove any trailing commas.

Putting it all together:

:g/^{$/-1s/,$//

These offsets are explained in section 10.3 of the Vim manual:

ADD AND SUBTRACT

There is a slight error in the above command: If the title of the next chapter had included "grey" it would be replaced as well. Maybe that's what you wanted, but what if you didn't? Then you can specify an offset.

To search for a pattern and then use the line above it:

/Chapter/-1

etc.

Documentation under :help cmdline-ranges has more advanced examples. Even though the discussion is mainly around ranges preceding a command, many of those expressions will work after :g as well.

Answer 2

I can think of a number of ways to make this work:

First would be only the key2 case:

global/key2/substitute/,$//

Now, in the general case, I believe you want to do

global/^}$/-substitute/,$//

:global can be abbreviated to :g and :substitute to :s. - in the range here is equivalent to -1, as explained at :help [range].

There's one other way to note: :g/^}$/-normal! 0f,x should work if there's only one comma in the lines you want to modify, and none in the lines you want to leave alone—this works because normal! will stop if any of the given commands fail. Since f, will fail when there's no comma, x will never be executed.

Answer 3

I interpret your question thus: wherever in the file you have comma followed by new-line that contains only }, delete all such commas.

:%s/,\n}$/\r}/

\n in the pattern identifies a line break, \r in replacement inserts a line break. I've no idea why they should be different.