joinlines.sh

joinlines.sh

#!/usr/bin/env bash

SCRIPT_PATH=${BASH_SOURCE[0]}
SCRIPT_NAME=${SCRIPT_PATH##*/}
SCRIPT_DIR="$(cd "$(dirname "${SCRIPT_PATH:-$PWD}")" 2>/dev/null 1>&2 && pwd)"
HELP=$(cat <<EOF
   Usage: $SCRIPT_NAME <file.csv>

   Joins lines in a csv file that are meant to be one line. If CRLF + LF
   terminated lines are present, this script will join all LF terminated lines.
   If CRLF + CR + LF lines are present, this script will join all CR terminated
   lines. This may or may not be the desired behavior, but has been the
   observed pattern so far.

   If $SCRIPT_NAME was unsuccessful in joining the lines, it will leave a copy
   of the modified file behind for inspection. Else, it will overwrite the
   original file with the modified copy.
EOF
)

argc="$#";: "$((i=0))"
while [ "$i" -lt "$argc" ]; do
  if [ "$1" = '--help' ] || [ "$1" = '-h' ] ; then
    echo "$HELP"
    exit 0
  elif [ "$1" = '--cr' ]; then
    JOIN_CR='true'
  elif [ "$1" = '--lf' ]; then
    JOIN_LF='true'
  else
    FILE="$1"
  fi
  shift;: "$((i=i+1))"
done

# printf "Checking for (CRLF + LF) or (CRLF + CR + LF) line terminators in $FILE... "
# if file "$FILE" | grep 'CRLF, LF' >/dev/null 2>&1; then
#   JOIN_LF='true'
#   printf "CRLF + LF found"
# elif file "$FILE" | grep 'CRLF, CR, LF' >/dev/null 2>&1; then
#   JOIN_CR='true'
#   printf "CRLF + CR + LF found"
# else
#   JOIN_NOTHING='true'
# fi
# printf "\n"
# if [ "$JOIN_NOTHING" ]; then
#   echo "$FILE does not have (CRLF + LF) nor (CRLF + CR + LF) line terminators present, exiting"
#   exit 0
# fi

if [ ! "$JOIN_CR" ] && [ ! "$JOIN_LF" ]; then
  echo 'please specify either join by --cr (^M) or --lf'
  exit 1
fi

RANDSTRING="$(date +%s)$(< /dev/urandom base64 | head -c10 | sed 's/\//+/')"
echo "making a copy of $FILE to $FILE.$RANDSTRING..."
cp "$FILE" "$FILE.$RANDSTRING" || exit 1
echo "modifying $FILE.$RANDSTRING..."
 if [ "$JOIN_LF" ]; then
   vim -Es \
     -c 'g/[^^M]$/.,/^M$/join' \
     -c 'g/[^^M]$/print' \
     -c '%s/^M$//g' \
     -c 'wq' \
     "$FILE.$RANDSTRING"
elif [ "$JOIN_CR" ]; then
   vim -Es \
     -c 'g/^M$/.,/[^^M]$/join' \
     -c 'g/^M$/print' \
     -c 'wq' \
     "$FILE.$RANDSTRING"
fi
 
if [ "$JOIN_LF" ] && file "$FILE.$RANDSTRING" | grep 'CRLF, LF' >/dev/null 2>&1; then
  echo "did not join lines successfully, leaving $FILE.$RANDSTRING behind"
  exit 1
fi
echo "moving $FILE.$RANDSTRING back into $FILE"
mv "$FILE.$RANDSTRING" "$FILE" || exit 1