UNIX for Dummies Questions & Answers

Hi folks. I've tried to research this, but haven't found a good answer (could be my harried state).

At any rate, I have records that end with

two commas, a number, two commas

this could be anywhere from

,,01,,

to

,,09875953,,

I need to remove the last two fields (the number and last double comma) from these records.

However, I also have records that end in a number and two commas. I do not want to change these.

Also, the number of fields varies from record to record.

sed 's/,,*,,/,,/g' file > newfile

does not work.

What am I missing here?

Thanks in advance,
Kristy

sed 's/,,[0-9]*,,$/,,/'

That worked! Thanks so much.

Now, why did that work and mine did not?

Because I did it correctly and you didn't.

Seriously, you left out [0-9]

I would think that would stand out like a sore thumb. That $ anchors the regular expression to the end. It seemed like a good idea. [0-9]
matches a digit. [0-9]* matches a string of digits. The RE you used ,* matches a string of commas.