Look through a file and print text from specific lines

Here's a sed approach:

$ sed -nE '1s/.{11}(.{8}).*/\1/p; 3s/.{3}(.{4}).*/\1/p' file  
Ethernet
t6 a

Explanation

The -n suppresses normal output (normal is to print every input line) so that it only prints when told to. The -E enables extended regular expressions.

The sed script has two commands, both using the substitution operator (s/original/replacement/). The 1s/.{11}(.{8}).*/\1/p will only run on the 1st line (that's what the 1s does), and will match the 1st 11 characters of the line (.{11}), then it captures the next 8 ((.{8}), the parentheses are a "capture group") and then everything else until the end of the line (.*). All this is replaced with whatever was in the capture group (\1; if there were a second capture group, it would be \2 etc.). Finally, the p at the end (s/foo/bar/p) causes the line to be printed after the substitution has been made. This results in only the target 8 characters being output.

The second command is the same general idea except that it will only run on the 3rd line (3s) and will keep the 4 characters starting from the 4th.

You could also do the same thing with perl:

$ perl -ne 'if($.==1){s/.{11}(.{8}).*/\1/}
            elsif($.==3){s/.{3}(.{4}).*/\1/}
            else{next}; print; ' file 
Ethernet
t6 a

Explanation

The -ne means "read the input file line by line and apply the script given by -e to each line. The script is the same basic idea as before. The $. variable holds the current line number so we check if the line number is either 1 or 3 and, if so, run the substitution, else skip. Therefore the print will only be run for those two lines since all others will be skipped.

Of course, this is Perl, so TIMTOWTDI:

$ perl -F"" -lane '$. == 1 && print @F[11..19]; $.==3 && print @F[3..6]' file 
Ethernet 
t6 a

Explanation

Here, the -a means "split each input line on the character given by -F and save as the array @F. Since the character given is empty, this will save each character of the input line as an element in @F. Then, we print elements 11-19 (arrays start counting at 0) for the 1st line and 3-7 for the 3rd.

awk approach:

$ awk 'NR==1{print substr($0,12,8)};NR==3{print substr($0,4,4)}' input.txt  
Ethernet
t6 a

Uses NR for determining line (in awk terminology - record) number, and accordingly print substring of the line. substr() function is in format

substr(string,starting position,how much offset) 

Python

$ python -c 'import sys                                                                                                                                                
> for index,line in enumerate(sys.stdin,1):                                                                                                                            
>     if index == 1:
>          print line[11:19]
>     if index == 3:
>          print line[3:7]' < input.txt
Ethernet
t6 a

This uses < shell operator to redirect input stream to python process from the input file. Note that strings in python are 0-indexed, hence you need to shift your desired character numbers all by 1.

portable shell way

This works in ksh, dash, bash. Relies only on shell utilities, nothing external.

#!/bin/sh

rsubstr(){
    i=0;
    while [ $i -lt  $2 ];
    do
        rmcount="${rmcount}?"
        i=$(($i+1))
    done;
    echo "${1#$rmcount}"
}

lsubstr(){
    printf "%.${2}s\n" "$1"
}

line_handler(){
    case $2 in
        1) lsubstr "$(rsubstr "$1" 11)" 8 ;;
        3) lsubstr "$(rsubstr "$1" 3)" 5 ;;
    esac
}

readlines(){
    line_count=1
    while IFS= read -r line;
    do
        line_handler "$line" "$line_count"
        line_count=$(($line_count+1))
    done < $1
}

readlines "$1"

And it works like so:

$ ./get_line_substrings.sh input.txt                                                                                                                                   
Ethernet
t6 ad