Need to format text file and update based of key value format

Question

I have a file having the below values:

cat data.txt

server1: calv
server2: anot
log: /u/log/1
server3: calv
server4: anot
server5: secd
server6: calv
LIB_TRGT_calv,anot: /tmp/hello.txt
LIB_TRGT_secd: /var/del.tmp

I get the variables starting with LIB_TRGT i.e LIB_TRGT_calv,anot & LIB_TRGT_secd

I need to get the names after TRGT_ from the variable above i.e calv,anot.

considering we got calv & anot; I need to get all the entries having calv & anot and add the entries found in the data.txt as below:

Desired Output:

server1: calv
server2: anot
log: /u/log/1
server3: calv
server4: anot
server5: secd
server6: calv
LIB_TRGT_calv,anot: /tmp/hello.txt
LIB_TRGT_secd: /var/del.tmp
LIB_server1:  /tmp/hello.txt
LIB_server2:  /tmp/hello.txt
LIB_server3:  /tmp/hello.txt
LIB_server4:  /tmp/hello.txt
LIB_server6:  /tmp/hello.txt
LIB_server5: /var/del.tmp

Likewise for LIB_TRGT_secd

Below is what I did so far:

grep TRGT* data.txt | cut -d: -f1 |cut -d_ -f3
calv,anot
secd

further

grep TRGT* test.txt | cut -d: -f1 |cut -d_ -f3 | sed -n 1'p' | tr ',' '\n'
calv
anot

However, secd is missing and i m not sure how to use xargs and further go about it.

Tried the solution by user @Kusalananda but it does not work. See output below:

Answer 1

An awk program to do this, assuming the LIB_TRGT_ lines always occur after the other lines that these refer to:

BEGIN {
        FS = ":[[:blank:]]*"
        OFS = ": "
}

{ print }

$1 !~ /^LIB_TRGT_/ {
        # "list" is an associative array that uses the keys "calv",
        # "anot", and "secd", and for each of these stores the relevant
        # servers as a SUBSEP-delimited string.

        list[$2] = (list[$2] == "" ? $1 : list[$2] SUBSEP $1)
        next
}

{
        path = $2

        # Pick out the comma-delimited list of keys after "TRGT_", and
        # split these up into elements of the array "k".

        sub("^.*TRGT_", "", $1)
        nk = split($1, k, ",")

        $0 = ""
        $2 = path

        # For each key, get the corresponding list from "list" and
        # output the path for each of its values.

        for (i = 1; i <= nk; ++i) {
                nv = split(list[k[i]], v, SUBSEP)
                for (j = 1; j <= nv; ++j) {
                        $1 = "LIB_" v[j]
                        print
                }
        }
}

Testing, with the above script in the file script, and with your data in the Unix text file called file:

$ awk -f script file
server1: calv
server2: anot
log: /u/log/1
server3: calv
server4: anot
server5: secd
server6: calv
LIB_TRGT_calv,anot: /tmp/hello.txt
LIB_server1: /tmp/hello.txt
LIB_server3: /tmp/hello.txt
LIB_server6: /tmp/hello.txt
LIB_server2: /tmp/hello.txt
LIB_server4: /tmp/hello.txt
LIB_TRGT_secd: /var/del.tmp
LIB_server5: /var/del.tmp

The ordering is slightly different from what you show in the question due to reacting to the LIB_TRGT_ lines as soon as they appear rather than saving them and handling them at the end.

Answer 2

We can solve this problem by means of generating code from the data and then applying that generated code to the data itself.

GNU sed with extended regex mode used.

inputf='./data.txt'

< "$inputf" \
sed -E '
  s/^\s+|\s+$//g
  s/\s+/\t/g
' |
sed -E '
  /_TRGT_[^:\t]/{
    s/^(([^_]+_)+TRGT_)([^,:]+)[,:](.*(\t\S+))$/\3\5\n\1\4/
    h;s/\n.*//;s/.*\t//
    s:[\/&]:\\&:g;G
    s/(.*)\n(.*\t).*(\n.*)/\2\1\3/
    P
  }
  D
' |
sed -En '
  1i\
p

  s#(.*)\t(.*)#/:\\t\1$/{s/^/LIB_/;s/:.*/:\\t\2/;H;ba;}#p

  $a\
:a\
$!d;g;s/.//
' |
sed -Ef - "$inputf"

server1:    calv
server2:    anot
log:    /u/log/1
server3:    calv
server4:    anot
server5:    secd
server6:    calv
LIB_TRGT_calv,anot: /tmp/hello.txt
LIB_TRGT_secd:  /var/del.tmp
LIB_server1:    /tmp/hello.txt
LIB_server2:    /tmp/hello.txt
LIB_server3:    /tmp/hello.txt
LIB_server4:    /tmp/hello.txt
LIB_server5:    /var/del.tmp
LIB_server6:    /tmp/hello.txt

---

• Step0, clean the data by removing leading trailing whitespace. Monospace to TABs.
• Step1 we isolate the key value pairs, put them in one per line.
• Step2 armed with these kv tuples, we generate sed code.
• Step3, we apply the code to data.

Answer 3

The below is a unified solution to both of your scenarios, viz., when the second fields are quoted or not.

awk -v q=\' '1
NF<2{next}
!/^([^_]+_)+TRGT_/ {
  a[FNR] = $2
  b[FNR,$2] = $1
  if (index($2,q) == 1) quoted++
  next
}
{
  n = split($1, temp, /[_,:]/)
  s=0
  for (i=1; i<n; i++) {
    t = temp[i]
    s += length(t)+1
    ch = substr($1, s-1, 1)
    if (ch == "_") continue
    key = (!quoted) ? t : q t q
    c[key] = $2
  }
  split("", temp, ":")
}
BEGIN { OFS="\t" }
END {
  for (i=1; i in a; i++)
    if ( (a[i] in c) && ((i SUBSEP a[i]) in b))
      print "LIB_"b[i,a[i]], c[a[i]]
}' data.txt

server1:    'calv'
server2:    'anot'
log:    '/u/log/1'
server3:    'calv'
server4:    'anot'
server5:    'secd'
server6:    'calv'
LIB_TRGT_calv,anot: '/tmp/hello.txt'
LIB_TRGT_secd:  '/var/del.tmp'
LIB_server1:    '/tmp/hello.txt'
LIB_server2:    '/tmp/hello.txt'
LIB_server3:    '/tmp/hello.txt'
LIB_server4:    '/tmp/hello.txt'
LIB_server5:    '/var/del.tmp'
LIB_server6:    '/tmp/hello.txt'

---