UNIX for Dummies Questions & Answers

An extension from an earlier question. Now need a sed script to delete columns 7,15 and 16 from an example txt below..
Again, thanks in advance.

98M-01.WAV,98M,01,00:00:49,01:07:36:00,"MIX",,"BOOM-MKH50",,,,,,,,,,"",
98L-01.WAV,98L,01,00:00:51,01:01:45:00,"MIX",,"BOOM-MKH50",,,,,,,,,,"",

Code:

 cut -d, -f 7,15,16 --complement < infile >outfile

Jim,
Thx for the reply but it needs to be sed. I'm running the script on an ipad which doesn't appear support cut

Code:

$ cat test.x
00000000011111111112222222222
12345678901234567890123456789

Using GNU sed:

Code:

$ sed -r "s/^(.{6}).(.{7})..(.*)$/\1\2\3/" test.x
00000000111111112222222222
12345689012347890123456789

Your version of sed may require some other variation of -r and \( \) and \{ \} depending on whether it uses "modern" or "basic" regular expressions. Also, does not require ^ and $ as part of regular expression, if you want to leave those anchor characters out.

Heres th version Gnu sed version is 4.1.5
I ran the script and got unexpected results

It should be easy: according to your sample a "column" is: zero or more non-","-characters followed by a ",". The regexp for this is:

Code:

/[^,]*,/

In sed you can use numbers to make the substitution-rule chang only the n-th occurrence, therefore:

Code:

sed 's/[^,]*,//16' /path/to/input

will delete only the 16th column. Analogously for the other columns, but note that deleting a column changes the numbers of the others in subsequent deletions. So either start with the last and work leftwards or calculate.

I hope this helps.

bakunin

That last info worked. For anyone interested here's the full script. This runs on a jail broke ipad. I download a .csv sound report file from a Sound Devices 788 recorder. Part 1 is a workaround for Numbers inability to handle timecode correctly. Part 2 strips out tracks that aren't used. Part 3 changes header names..thx to you guys for the help

sed -e 's/\(..:..:..\):/\1./g' -e 's/[^,]*,//7' -e 's/[^,]*,//15' -e 's/[^,]*,//15' -e 's/\T3/\T2/g' -e 's/\T4/\T3/g' -e 's/\T5/\T4/g' -e 's/\T6/\T5/g' -e 's/\T7/\T6/g' -e 's/\T8/\T7/g' -e 's/\T9/\T8/g' -e 's/\T10/\T9/g ' /original.CSV>/new file.csv