topic Re: sed how to in Operating System - HP-UX

sed how to

Roger G. Vincent — Mon, 20 Nov 2000 15:37:54 GMT

I need to modify a file in the following manner:
- remove all blank lines
- change the last occurance of a comma (,) to a left paren

any ideas

Re: sed how to

Lasse Knudsen — Mon, 20 Nov 2000 15:48:36 GMT

cat file | grep -v '^$' | sed -e 's/,$/(/'

Re: sed how to

Rainer_1 — Mon, 20 Nov 2000 15:50:57 GMT

sed -e '/^$/d' -e 's/,$/(/'

Re: sed how to

Bruce Regittko — Mon, 20 Nov 2000 16:07:54 GMT

Hi,

Is the comma the last character of the line? If so, then the first two suggestions will work fine. If you mean the last comma, but not necessarily the last character, try the following:

sed -e '/^$/d' -e 's/,[^,]*$/(/'

Hope this helps,

Bruce

Re: sed how to

Lasse Knudsen — Mon, 20 Nov 2000 16:24:30 GMT

Roger, the other 2 posts definetely desirve 10pts for their smooth and "true hack" solutions.

Re: sed how to

Roger G. Vincent — Mon, 20 Nov 2000 16:26:04 GMT

This was close however, I want to change the last occurance of a comma (,) in the file to a paren NOT each occurance of a comma to a paren.

Re: sed how to

John Palmer — Mon, 20 Nov 2000 16:36:35 GMT

Hi,

To change the last occurrence of ',' try the following:-

sed -e '/^$/d' -e 's/$.*$,$.*$/\1(\2/'

Hope the backslashes appear! there's enough of them.

Regards,
John

Re: sed how to

CHRIS_ANORUO — Mon, 20 Nov 2000 16:46:50 GMT

Try: cat file |sed -e '/^$/d' -e 's/$.*$,$.*$/\1(\2/' > newfile

Re: sed how to

Roger G. Vincent — Mon, 20 Nov 2000 16:50:57 GMT

The commas are killing me. I have a file with multiple lines which end with a comma (,). I need to change the very last comma of the file to a right paren ")". I am sorry for the confusion, however, I stated the question improperly (or at least it was unclear). I am new to this point value stuff and I appreciate all responses and will get the points right soon.

Re: sed how to

CHRIS_ANORUO — Mon, 20 Nov 2000 16:58:13 GMT

Hi Rogers,

Check this very well and run into another file as indicated: cat filename |sed -e '/^$/d' -e 's/$.*$,$.*$/\1)\2/' > newfilename

Re: sed how to

Roger G. Vincent — Mon, 20 Nov 2000 17:45:08 GMT

the command cat filename | sed -e '/^$/d' -e 's/$.*$,$.*$/\1)\2/' > newfilename appears to change ALL occurances of the , at the end of a line to a right paren. I want to change the LAST comma of the file to a right paren and preserve the remaining commas on all other lines in the file.

Re: sed how to

Dan Hetzel — Mon, 20 Nov 2000 17:50:56 GMT

Hi Roger,

This one works for sure:

sed -e '/^$/d' -e 's/$,$$[^,]*$$/)\2/' < infile

Explanation:

/^$/d deletes empty lines
s/$,$$[^,]*$$/)\2/ substitutes a comma (1st matching item) followed by 0 or more non-commas up to the end of line (2nd matching item) by a right paren followed by the 2nd match item

Matching Items are delimited by $ and $ in the search string

Best regards,

Dan

Re: sed how to

Roger G. Vincent — Mon, 20 Nov 2000 19:08:45 GMT

Maybe it is me, however the last command did NOT work...It changed each line with a trailing comma to a right paren. I have the following input data:

This is first line,
This is second line,
This is the last line,

I want to leave commas on first and second line alone and change the comma on the last line to a right paren ).

Solution should be:
This is first line,
This is second line,
This is the last line)

Re: sed how to

Shannon Petry — Mon, 20 Nov 2000 19:19:01 GMT

You may need to use awk instead. Are all of the lines 3 long?
line1,
line2,
line3)
or is there a variety? Many files, or just one short one?
awk can use the same sed commands, but you have the ability to count, track record numbers, etc....

Re: sed how to

Rainer_1 — Mon, 20 Nov 2000 19:38:38 GMT

think now it's time for awk.
try this

awk 'BEGIN{
line="";}
{
if(length($0)==0)continue;
if(length(line)>0)print line;
line=$0;
}
END{
sub(",$",")",line);
print line;
}'

Re: sed how to

Tony Constantine_1 — Mon, 20 Nov 2000 19:47:29 GMT

Heres a good link for some sed 1 liners

http://www.cornerstonemag.com/sed/sed1line.txt

Re: sed how to

Kevin Ernst — Mon, 20 Nov 2000 22:02:39 GMT

As long as there aren't any blank lines at the end of the file, you got it, Ranier. I thought about using 'tac' ('cat' the file backwards) to reverse the input, search for and replace the FIRST occurrence of ",$", then 'tac' the file again to get it back in the right order--but HP-UX doesn't have a 'tac.' So much for quick-and-dirty.

Re: sed how to

Kevin Ernst — Mon, 20 Nov 2000 22:05:46 GMT

Doh! RaInEr--sorry.

Re: sed how to

Dan Hetzel — Tue, 21 Nov 2000 06:35:26 GMT

Kevin, Roger,

We don't have 'tac' but we have 'rev' which prints every line backwards

one solution could be:

rev infile | sed -e '/^$/d -e '$s/,/)/' | rev > outputfile

or

sed -e '/^$/d' < infile | sed -e '$s/$,$$[^,]*$$/)\2/'

will both replace the last comma of the last non blank line by a right paren, even if it's not the last character on the line,providing there is a comma on that line.

Dan