Shell Programming and Scripting

I have a file containing rows with the following format.

Field1|Field2|Field3|data1:data data2:data data3:"dataA:data dataB:data" data4:data:data (and so on)

I need to format the above row into multiple rows that look like this:

Field1|Field2|Field3|data1|data

Field1|Field2|Field3|data2|data

Field1|Field2|Field3|data3|dataA:data dataB:data

Field1|Field2|Field3|data4|data:data

Here are the steps in my mind it would take to accomplish this.
1. Store everything before the last | on each row into a string variable, lets call this $header
2. use sed or awk to replace every space that does not appear between quotations " " with \n$header
3. Use sed to turn the first : on every row into a | in and output into the results file.

I performed a similar task with the help of some people here using awk to deliniate a row into rows, but the originating row was purely space delimited. With the added criteria of not delimiting if the space appears between " " I am a bit stumped.

I believe a workaround to the problem with the spaces between " " is to deliniate by spaces, and then if the string has a " then it needs to be concatinated with all subsequent strings until another " is found. I plan on trying this solution tonight, but I believe that there is likely much more elegant ways to do this.

Additionally, I still need a hand in creating the header string. I could possibly check to see if the deliniated string for | and if they exist, write the values before the last | into $header.

Any help here would be appreciated. Thanks!

See if this works for you:

Code:

#!/bin/ksh
typeset -i mCnt=0
typeset -i mTot
IFS='|'
while read mFld1 mFld2 mFld3 mFld4; do
  mFldOut=${mFld1}'|'${mFld2}'|'${mFld3}'|'
  IFS=' '
  for mEach in ${mFld4}; do
    mTot=$(echo ${mEach} | grep -c '"')
    if [[ ${mTot} -eq 0 ]]; then
      mOut=$(echo ${mEach} | sed 's/:/|/')
      echo "${mFldOut}${mOut}"
    else
      if [[ ${mCnt} -eq 1 ]]; then
        mOut=${mOut}' '${mEach}
        mOut=$(echo ${mOut} | sed -e 's/"//'g -e 's/:/|/')
        echo "${mFldOut}${mOut}"
        mCnt=0
      else
        mOut=${mEach}
        mCnt=${mCnt}+1
      fi
    fi
  done
done < Input_File

POSIX shell:

Code:

while read line; do
  start=${line%|*}
  eval set -- ${line##*|}
  for data; do
    echo "$start|$data" | sed 's/:/|/'
  done
done  <INPUTFILE