Regular expression - no capture for negative character class

Question

I am trying to write different sed scripts for beautifying ruby code. One of the cases i am trying to solve is to replace a=>b or a=> b strings with a => b. The regex for matching this condition is [^ ]=> but it also matches 1 character before =>. So, when i try to replace it is not giving me the desired result with s/[^ ]=>/ =>/g

Any suggestions?

Try replacing `\s*=>\s*` with `{space}=>{space}`. – Blender Apr 21 '13 at 02:44 — Blender, Apr 21 '13 at 02:44

score 1 · Accepted Answer · answered Apr 21 '13 at 02:48

1

You need to use a capture:

s/\([^ ]\)=>/\1 =>/g

answered Apr 21 '13 at 02:48

rici

234,347
28
237
341

That worked in Vim thanks! But the same dosent seem to be working in sed. – Pratik Khadloya Apr 21 '13 at 03:47
sed -i 's/\([^ ]\)=>/\1 =>/g' app/models/campaign.rb sed: 1: "app/models/campaign.rb": command a expects \ followed by text – Pratik Khadloya Apr 21 '13 at 03:51
This worked. sed -i '' 's/([^ ])=>/\1 =>/g' app/models/campaign.rb – Pratik Khadloya Apr 21 '13 at 03:56
@PratikKhadloya: I have no idea why that would work; I would have expected you to need to use the `-r` command-line flag with GNU sed in order to avoid having to use `\(` for captures. But I'm glad you got it working. – rici Apr 21 '13 at 04:19
@PratikKhadloya: Yes, the `-i` flag can be used to (appear to) do edits in place. But I don't think it changes the type of regex used by `sed`. Otherwise, you'd need to use `s/\(...`, but you seem to be saying that you used `s/(...`. – rici Apr 21 '13 at 06:21

score 0 · Answer 2 · answered Apr 21 '13 at 02:48

0

How about replacing both sides unconditionally?

s/ ?=> ?/ => /g

answered Apr 21 '13 at 02:48

Ry-

218,210
55
464
476

Regular expression - no capture for negative character class

2 Answers2