如何使用 awk/sed 仅替换某些行中特定列中的数据

Question

I have a space-delimited text file where the bulk of the rows look like this:

--- 23:123456_A_B 123456 A B [and ~600K more columns after that]

However, there are some rows where the first 5 columns look like this:

23 . 234567 C D

Can anyone suggest an awk or sed command (or any alternative method) that would change the first column from 23 to --- and the second column from . to 23:234567_C_D (using the data from the first five columns) while leaving the other columns untouched, but only in those rows that start with 23 ?

Answer 1

Assuming the field separator is just a single space,

sed -r 's/^23 \. ([^ ]+) ([^ ]+) ([^ ]+)/--- 23:\1_\2_\3 \1 \2 \3/' file

More readable with awk

awk '$1 == 23 && $2 == "." {$1 = "---"; $2 = "23:" $3 "_" $4 "_" $5} 1' file

With sed, you can save the changes in-place with the -i option.
With awk, you may have to explicitly write to a temp file:

tmp=$(mktemp)
awk '....' file > "$tmp" && mv "$tmp" file

Answer 2

Is this what you want?

$ cat file
--- 23:123456_A_B 123456 A B [and ~600K more columns after that]
23 . 234567 C D

$ awk '$1==23 { $2=$1":"$3"_"$4"_"$5; $1="---" }1' file
--- 23:123456_A_B 123456 A B [and ~600K more columns after that]
--- 23:234567_C_D 234567 C D