How to make awk only run after certain headings? - awk

Question

I have a file with tables of values under certain headings. Like so:
Series A
a 01 20
b 21 43
c 44 59
d 60 72
Series B
a 01 25
b 26 48
c 49 70
d 71 90
My question is, how can I make awk only draw information from the specified series? So if awk is told to look at "Series B", column 2, it will output that column.

Answer 1

If the data is separated by a blank line then you can use something like this -
[jaypal:~/Temp] cat file
Series A
a 01 20
b 21 43
c 44 59
d 60 72
Series B
a 01 25
b 26 48
c 49 70
d 71 90
[jaypal:~/Temp] awk '/Series B/,/^$/ { if (NF==3) print $2 }' file
01
26
49
71
Or you can do something similar in sed:
[jaypal:~/Temp] sed -n '/Series B/,/^$/s/\(.*[^ ]\) \(.*[^ ]\) \(.*\)/\2/p' file
01
26
49
71

Answer 2

Assuming the tables are separated by empty line(s):
awk '$0 ~ p, !NF {
if (!($0 ~ p) && NF) print $c
}' p='Series B' c=2 infile
$0 ~ p, !NF - the pair of patterns separated by a comma specify a range of records.
The first one evaluates to true if the current record $0 matches the pattern indicated by
the value of the variable p. The second one is true when the value of NF (Number of Fields)
is 0 (empty or blank line).
if (!($0 ~ p) && NF) print $c - if the current record doesn't match the pattern p
(i.e. skip the header), and (&&) the current record contains at least one column (NF != 0)
print the desired column.

Answer 3

I prefer this way:
awk -v sect="$1" -v col="$2" '
/Series/ {header = $0; next}
NF > 1 {
set[header] = set[header]$col"|";
}
END {print set[sect];}
' $3 | tr '|' '\n'
I introduce associate array and temp seperator(not needed if you don't need sort on "Series", just use "\n" and omit tr), don't rely on blank line, use "|" as temp sep, use tr to transfer it back.
This is my common solution since in real life I can create custom "sorting column", then pipe to unix sort, and cut them back. Nawk(alas AIX) don't have sort and unix sort is far more superior then any hand-made sort in awk.