Script to generate a flat Kiji schema shell DDL based on the headers for a CSV

generate-ddl.sh

#!/bin/bash
# Generates a flat Kiji schema shell DDL based on the headers for a CSV

FILENAME=$1

if [ ! -e "$FILENAME" ]; then
  echo "Could not find file: $FILENAME"
  exit 1
fi

echo "CREATE TABLE train WITH DESCRIPTION 'Training data'"
echo "ROW KEY FORMAT HASHED"
echo "  WITH LOCALITY GROUP default"
echo "  WITH DESCRIPTION 'Main locality group' ("
echo "  MAXVERSIONS = 10,"
echo "  TTL = FOREVER,"
echo "  INMEMORY = false,"
echo "  COMPRESSED WITH NONE,"
echo "  FAMILY info WITH DESCRIPTION 'basic information' ("

# Loop over fields and put in the relevant description lines
FIRST=true
for x in `head -1 $FILENAME | tr -d '\r' | tr , '\n'`; do
  if [ "$FIRST" != "true" ]; then
    echo ,
  fi
  echo -n "    $x \"string\" WITH DESCRIPTION 'Imported $x'"
  FIRST=false
done
echo
echo "  )"
echo ");"