Shell Programming and Scripting

Hi,

I need to do a sum of two columns in a file where delimiter is |^

input

Code:

1|^2|^3|^4|^50|^2|^3|^100
2|^3|^4|^6|^100|^7|^2|^50
3|^4|^2|^3|^50|^6|^3|^50
4|^2|^5|^7|^25|^2|^8|^25

Output required:

Code:

 Sum of 2 columns 5 & 8 which is 450

Try

Code:

awk 'BEGIN {FS="\|\^"} {sum+=$5+$8} END {print sum}' infile

Hi RudiC,

Tried as suggested..

following is the issue

Code:

[root@host-6-113 ~]# awk 'BEGIN {FS="\|\^"} {sum+=$5+$8} END {print sum}' infile
awk: warning: escape sequence `\|' treated as plain `|'
awk: warning: escape sequence `\^' treated as plain `^'
0

---------- Post updated at 11:24 PM ---------- Previous update was at 10:50 PM ----------

Rudic,

tweaked the code you've provided.. It is giving me the desired result. Thanks !!

Code:

[root@host-6-113 ~]# awk 'BEGIN {FS="[|^]"} {sum+=$9+$15} END {print sum}' infile
450

But it looks like it is considering the delimiter |^ as two strings. Please suggest alternatives if any..

Try

Code:

awk -F'[.|.][.^.]' ' {sum+=$5+$8} END {print sum}' infile

The '[!^]' is rejected by some versions of awk, but when it is accepted it is saying the field separator is a "|" or a "^". You want it to be a "|" immediately followed by a "^".

Given awk's ability to evaluate a number up to the first non-digit, the carrot (^) can be used as the simple field separator:

Code:

awk -F "^" '{s += $5 + $8;} END { print s; }' input-file

There does not seem to be an issue for me - awk gave warnings but treated the FS as requested. The 0 result I guess was due to 0 input, sum (0)=0; in your second post you suddenly summed columns (9 & 15), totally different to your first example (5 & 8).

According to POSIX, -F "|^" or any other mechanism that sets FS to the two characters <vertical-line><circumflex> produces undefined results because <vertical-line> is special in an extended regular expression (ERE) except when it appears in a bracket expression. Also in an ERE ^ is an anchor (even when it is not the first character in the ERE) except when it appears in a bracket expression, in a collating symbol ([.^.]), or equivalence class ([=^=]). So, any of the solutions here that set FS to a string starting with "|" is not portable. And any of the solution here that set FS to a string ending with "^" is not portable. Setting FS to "[|^]" is valid, but makes both <vertical-line> and <circumflex> field separators, not the two character string "|^". (This is why the solutions that set FS to "[|^]" had to sum fields 9 and 15 instead of fields 5 and 8. Note that this solution won't work if there are any <vertical-line> or <circumflex> characters in the input that aren't adjacent pairs and it that order).

Since "^" is an anchor in an ERE, setting FS to "^" produces undefined results. (It is anchoring the empty expression following the <circumflex> to the beginning of the string. But since this isn't a replacement [where an empty expression is replaced by the previously matched expression], the standards don't specify what this means and the results are not portable.)

I'm sure the solutions given all work on one or more implementations. Several of them fail with syntax errors when using Mac OS X's awk.

To portably search for "|^" as a field delimiter and sum the numeric values in the 5th and 8th columns, I think you need something like:

Code:

awk -F "[|][[.^.]]" ' {sum+=$5+$8}
END {print "Sum of 2 columns 5 & 8 which is",sum}' input

which should work on any system that meets the POSIX awk utility requirements.