Update data in source file with random data from another file

I have data in source file as below (file.txt)

N4*WALTER*WHITE~DMG*D8*19630625~N4*JESSI*PINKMAN*15108~

Input command: (N4 = segment Identifier, 1= position , ref.txt=reference file)

N4*1*ref.txt

ref.txt has data as below

BILL
LEONARDO
BALE
BRAD
PITT

I have below code which displays the data in position x(input) for N4

identifier=N4
position=1
refile=ref.txt

awk -F[*~] -v id="$identifier" -v pos="$position" 
   'id { for (i=1; i<=NF; i++)
         if ($i == id) {
             if (i+pos <= NF)
                 print $(i+pos)
             else
                 print "invalid position"
         }
       }
    ' file.txt

WALTER
JESSI

identifier=N4
position=2
refile=ref.txt

awk -F[*~] -v id="$identifier" -v pos="$position" 
   'id { for (i=1; i<=NF; i++)
         if ($i == id) {
             if (i+pos <= NF)
                 print $(i+pos)
             else
                 print "invalid position"
         }
       }
    ' file.txt

WHITE
PINKMAN

Now how can i integrate ref.txt in above code to update WALTER and JESSI in file.txt with random text located in ref.txt file.

I know shuf command gives random data from ref.txt. but not sure how to integrate this in above awk command.

shuf -n -1 ref.txt

expected output: file.txt (Position 1 data for N4 segment) to get updated with random text from ref.txt

N4*BALE*WHITE~DMG*D8*19630625~N4*PITT*PINKMAN*15108~

Answer

Well, I can do it in bash, but it will be slow with while read loop:

# recreate the input files
cat <<EOF >file.txt
N4*WALTER*WHITE~DMG*D8*19630625~N4*JESSI*PINKMAN*15108~
EOF
cat <<EOF >input
N4*1*ref.txt
EOF
cat <<EOF >ref.txt
BILL
LEONARDO
BALE
BRAD
PITT
EOF

# read the input
IFS='*' read  -r segment position reference_file <input

{
  # for each line
  while IFS='*' read -r -d'~' -a data; do
    # if the segment id is the segmend
    if [ "${data[0]}" = "$segment" ]; then
        # update the data
        data[$position]=$(shuf -n1 "$reference_file")
    fi
    # and output the data
    ( IFS=*; printf "%s~" "${data[*]}"; )
  done
  # append a newline on the end
  echo
} < file.txt

I wanted to try it with sed to iterate over segments, but end up preprocessing seds inputs. Below it is, with comments:

IFS='*' read  -r segment position reference_file <input

# remove the nelwine from input
# and substitute each `~` with a newline
# so we can nicely process the file in sed
<file.txt tr -d 'n' | tr '~' 'n' >tmp.txt

# count of segments inside input we are interested in
segmentscnt=$(
  grep "^${segment}*" tmp.txt | wc -l
)

# generate single line with random words from reference_file
# words separated by `*`
# the count of words should that many as many are there
# segments we are interested in the input file
randoms=$(
  while shuf -n1 "$reference_file"; do :; done | 
  head -n"$segmentscnt" |
  tr 'n' '*'
)

sed -n "
  # the first line should be random words from referencefile
  # load it to hold space
  1{
    h
    d
  }
  # if this is our segment
  /^$segment*/{
    # append random words to our pattern space
    G

    # remember as many fields as the position we want to insert
    # each one word more
    # remember rest of line
    # remember first word from randoms that were inserted from hold space
    # then just substitute the words in proper order
    s/^(([^*]**){$position})[^*]*([^n]*)n([^*]*)*.*/143/

    # remove the first word from hold space
    x
    s/^([^*]*)*//
    x

  }
  p
# the first input are the random words separated by *
# the words are on a single line
# than the input file
" - <<<"$randoms" tmp.txt |
# then replace newlines with `~`. 
# also append a newline with echo
# as it will be missing
tr 'n' '~'; echo