[SOLVED] How do I break a tabular list into pairs?

Jykke · 04-02-2013, 07:38 AM

I am looking for a sed or awk one liner for following.
I have a table (usually) 8 numbers/line, separated with comma.
The format in this special case is, for example
0, 10., 0.0001, 50., 0.0002, 84.0, 0.0003, 11.6,
0.0004, 15., 0.0005, 18.5, 0.0006, 28.7, 0.0007, 25.4,
0.0008, 28., 0.0009, 39., 0.001, 353.4, 0.0011, 60.8,
0.0012, 89., 0.0013, 114., 0.0014, 146.1, 0.0015, 17.9,

I would like to pass this through sed and get:
0, 10.
0.0001, 50.
0.0002, 84.0
0.0003, 11.6
0.0004, 15.
0.0005, 18.5
0.0006, 28.7
0.0007, 25.4
0.0008, 28.
0.0009, 39.
0.001, 353.4
0.0011, 60.8
0.0012, 89.
0.0013, 114.
0.0014, 146.1
0.0015, 17.9

I try sed 's/,/\n/2' but it only replaces the second occurrence on each line, not every second occurrence. Any ideas how to perfect it?

millgates · 04-02-2013, 07:59 AM

sed:

Code:

sed -r 's/([^,]*,[^,]*),\s*/\1\n/g' inFile

awk:

Code:

awk 'BEGIN{RS="[,\n ]+"}ORS=NR%2?",":"\n"' inFile

Jykke · 04-02-2013, 08:24 AM

Thanks, seems to give extra linefeed in the end of the line but that I can pipe through a second sed

danielbmartin · 04-02-2013, 08:30 AM

Quote:

Originally Posted by millgates

sed:

Code:

sed -r 's/([^,]*,[^,]*),\s*/\1\n/g' inFile

awk:

Code:

awk 'BEGIN{RS="[,\n ]+"}ORS=NR%2?",":"\n"' inFile

Something puzzling... the awk works perfectly but the sed OutFile contains unexpected blank lines.

Daniel B. Martin

millgates · 04-02-2013, 08:38 AM

Quote:

Originally Posted by danielbmartin

Something puzzling... the awk works perfectly but the sed OutFile contains unexpected blank lines.
Daniel B. Martin

Yes, the sed takes characters up to the second comma encountered, and puts a newline after that. But if the inFile already contains a newline there, there will be two newlines then.

This one is a bit longer, but should take care of the spaces:

Code:

sed -r 's/([^,]*,[^,]*),(\s*\S)/\1\2\n/g' inFile

What it does is that it checks that there is at least one non-whitespace character after the pattern (= the pattern is not at the end of the line) before the substitution.
Maybe someone can find a simpler solution...

Of course, the sed examples assume there will be an even number of numbers in each line.

danielbmartin · 04-02-2013, 10:20 AM

Quote:

Originally Posted by millgates

Code:

awk 'BEGIN{RS="[,\n ]+"}ORS=NR%2?",":"\n"' inFile

OP asked for a blank following each comma. Your excellent awk is easily changed to provide this.

Code:

awk 'BEGIN{RS="[,\n ]+"}ORS=NR%2?", ":"\n"' $InFile

Daniel B. Martin

danielbmartin · 04-02-2013, 10:27 AM

Quote:

Originally Posted by millgates

Code:

sed -r 's/([^,]*,[^,]*),(\s*\S)/\1\2\n/g' inFile

What it does is that it checks that there is at least one non-whitespace character after the pattern (= the pattern is not at the end of the line) before the substitution.

Please elaborate. The (\s*\S) part is way over my head.

Daniel B. Martin

David the H. · 04-02-2013, 10:35 AM

"\s" is simply a shortcut for "[[:space:]]", supported in some flavors of regex. "\S" is the inverse.

The regular expression section of the grep info page has a good breakdown of all the supported expressions in the regex supported by these programs.