Inspect parquet files stored in S3 using `parquet-tools`

parquet-s3

#!/usr/bin/env bash

set -o errexit

### ------ Preparation ------ ###

# brew tap adoptopenjdk/openjdk && brew cask install adoptopenjdk8
# command -v mvn > /dev/null || brew install maven
# command -v parquet-tools > /dev/null || brew install parquet-tools

# brew cask list --versions osxfuse > /dev/null || brew cask install osxfuse
# command -v goofys > /dev/null || brew install goofys

### ------ -------------- --- ###


profile=default
cmd=head

while [ -n "$1" ]; do
    case $1 in
        --profile )
            shift
            profile=$1
            ;;

        --command )
            shift
            cmd=$1
            ;;

        * )
            inputfile=$1
            ;;
    esac
    shift
done

inputfile=${inputfile/s3\:\/\//}
key=${inputfile#*/}
bucket=${inputfile%%/*}

if [[ -z "$key" ]] || [[ -z "$bucket" ]]
then
    echo "
        Invalid S3 file path
    "
    exit 1
fi

prefix=`dirname ${key}`
file=`basename ${key}`
mountingpoint=~/.mounting-points/${prefix}

mkdir -p ${mountingpoint}

function _umount() {
    mount | grep ${mountingpoint} > /dev/null && umount ${mountingpoint} || true
}

trap _umount EXIT

goofys \
    --profile ${profile} \
    ${bucket}:${prefix} \
    ${mountingpoint}

parquet-tools ${cmd} ${mountingpoint}/${file} 2> /dev/null