Replace a letter with another from the last word from the last two lines of a text file

Question

How could I possibly replace a character with another, selecting the last word from the last two lines of a text file in shell, using only a single command? In my case, replacing every occurrence of a with E from the last word only.

Like, from a text file containing this:

tree;apple;another
mango.banana.half
monkey.shelf.karma

to this:

tree;apple;another
mango.banana.hElf
monkey.shelf.kErmE

I tried using sed -n 'tail -2 'mytext.txt' -r 's/[a]+/E/*$//' but it doesn't work (my error: sed expression #1, char 10: unknown option to 's).

Answer 1

Could you please try following, tac + awk solution. Completely based on OP's samples only.

tac Input_file | 
awk 'FNR<=2{if(/;/){FS=OFS=";"};if(/\./){FS=OFS="."};gsub(/a/,"E",$NF)} 1' | 
tac

Output with shown samples is:

tree;apple;another
mango.banana.hElf
monkey.shelf.kErmE

NOTE: Change gsub to sub in case you want to substitute only very first occurrence of character a in last field.

Answer 2

This might work for you (GNU sed):

sed -E 'N;${:a;s/a([^a.]*)$/E\1/mg;ta};P;D' file

Open a two line window throughout the length of the file by using the N to append the next line to the previous and the P and D commands to print then delete the first of these. Thus at the end of the file, signified by the $ address the last two lines will be present in the pattern space.

Using the m multiline flag on the substitution command, as well as the g global flag and a loop between :a and ta, replace any a in the last word (delimited by .) by an E.

Thus the first pass of the substitution command will replace the a in half and the last a in karma. The next pass will match nothing in the penultimate line and replace the a in karmE. The third pass will match nothing and thus the ta command will fail and the last two lines will printed with the required changes.