UNIX for Dummies Questions & Answers

I have a data file that has 14 columns. I cannot use awk or perl but sed is installed on my host. I would like to delete a line if fields 10, 11 or twelve is greater than 999.99. How is this done using sed?

Code:

sed '/^[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,[^,]*,/d' infile
         1     2     3     4     5     6     7    8      9     10    11    12    13   14

Almost. ;-))

You can "multiply" regexp parts with the "\{...\}" operator. For instance: "x" means a single "x", but "x\{3\}" means "three x" and would be equivalent to "xxx". But it comes even better, the operator can be given two values, which will define a range: "x\{1,3\}" means "one, two or three x". Instead of "x" one could even put a complicated regexp in braces.

Now your problem: a single field in a comma-separated file would be matched by: "[^,]*,". You have already figured that out. Now we try to match something in the 11th field. This is: ten fields, followed by the field on question:

Code:

/^\([^,]*,\)\{10\}xxx/

This matches "xxx" in the eleventh field. You see you could shorten and simplify your regexp considerably using this device. Your search for "more than 999.9 in the 10th, 11th or 12th field" could look like this:

Code:

/^\([^,]*,\)\{9,11\}[0-9][0-9][0-9][0-9]/

9-11 fields, followed by a number of at least 4 digits.

I hope this helps.

bakunin

Of course, if any CSV commas are in double quotes, any simple sed is confused. PERL has CSV reading libraries.