I have Unix scripting question

Question

Input----

System:root,bin,user,saaa

Displayed output----

System,root
System,bin
System,user
System,says

How to get this output??

Answer 1

I would recommend that you use perl, but since you specified shell scripting...

Step 1: split your line into two parts, based on the : character. Use the cut command or the ${parameter#word} and ${parameter%word} constructs.

Step 2: split the second part of your line into multiple parts, based on the , character. Use the awk command -- it should tell you how many pieces you will have (I'm not an awk expert, so I'm not sure that this will work the way I envision it.)

Step 3: cycle through the various parts you get from Step 2, attach them to the first part from Step 1, and print.

Answer 2

If we can assume that your input lines contain exactly one colon (:), that no comma (,) may appear before the colon and that neither commas nor colons may be part of the extracted substrings (not even escaped), then a simple awk script may be enough:

$ printf '%s\n' 'System:one,two,three' |
    awk -v FS=':|,' '{ for (i=2;i<=NF;i++) { print $1","$i } }'

Output:

System,one
System,two
System,three

The field separator FS is an extended regular expression that will split on every character that is : or ,.

---

If, instead, you want to pick everything up to the first colon (possibly including commas) as the first output field and split the remainder of any input line at any comma (assuming that no comma is intended to be preserved as part of any of the substrings (not even escaped)), you can resort to shell features, as suggested in hymie's answer:

$ printf '%s\n' 'System:one,two,three' |
    while IFS= read -r rem; do      # IFS= to preserve blank characters
      first=${rem%%:*}              # Remove from the first ':' on
      rem=${rem#"$first"}           # Remove first from the beginning of rem
      rem=${rem#:}                  # Strip the remaining ':' at rem's beginning
      while test "$rem"; do         # Exit when rem is empty
          second=${rem%%,*}         # Remove from the first ',' on
          rem=${rem#"$second"}      # Remove second from the beginning of rem
          rem=${rem#,}              # Strip the remaining ',' at rem's beginning
          printf '%s\n' "$first,$second"
      done
    done

Just make sure you understand the caveats of using shell loops to process text.

Alternatively, with GNU sed:

$ printf '%s\n' 'System:one,two,three' |
sed -n '
  :l1
    s/^[^:]*:\n//g;
    t l2;
  s/^\([^:]\{1,\}\):\([^,\n]\{1,\}\)*,\{0,1\}\(.*\)$/\1:\3\n\1,\2/;
    t l1;
  q;
  :l2 p;
'

Here, branching (t) to a label (l1) is used to process each line of input with a loop. One at a time, the substrings between the first : and the first following , are appended as new lines to the pattern space, concatenated after the substring that comes before the first :. When there is no further substring to extract, what remains of the original string is removed, the pattern space is printed and the program exits.

(With GNU sed version >= 4.6 you can look at what is happening, step by step, by invoking it with the --debug option).

Note that the use of \n inside bracket expressions to match (here, negate a match) a <newline> character is non-standard: POSIX states that <backslash> shall lose any special meaning in that context.