ont-open-data-dataset/s3://ont-open-data/gm24385_2020.09/config/rationalize_fastq

#!/bin/bash

usage="
$(basename "$0") [-h] [-i -o]

    Consolidate fastq(.gz) files.

    -h  show this help text.
    -i  input file or directory.
    -o  output file. If input file is a single file a symlink is made
        otherwise the output is the concatenation of all files found.
        If the output file has the extension '.gz' the final output
        will be compressed, else it will be plain text."

iflag=false
oflag=false
while getopts ':hi:o:' option; do
  case "$option" in
    h) echo "$usage"; exit;;
    i) INPUT=$OPTARG; iflag=true;;
    o) OUTPUT=$OPTARG oflag=true;;
    :) printf "missing argument for -%s\n" "$OPTARG" >&2
       echo "$usage" >&2
       exit 1;;
   \?) printf "illegal option: -%s\n" "$OPTARG" >&2
       echo "$usage" >&2
       exit 1;;
  esac
done
shift $((OPTIND - 1))

if ! $iflag || ! $oflag ; then
    echo "$usage" >&2
    echo "" >&2
    echo "Both -i and -o must be specified."
    exit 1
fi

function abspath {
    echo "$(cd "$(dirname "$1")"; pwd -P)/$(basename "$1")"
}

function extension {
    filename=$(basename -- "$1")
    echo "${filename##*.}"
}
export -f extension

INPUT=$(abspath $INPUT)
OUTPUT=$(abspath $OUTPUT)
OUTEXT=$(extension $OUTPUT)

echo "Input: ${INPUT}"
echo "Output: ${OUTPUT}"

if [[ -e "${OUTPUT}" ]]; then
    echo "Output exists, exiting"
    exit 1
fi

if [[ ! -e "${INPUT}" ]]; then
    echo "Error: ${INPUT} does not exist."
    exit 1
fi

if [[ -f "${INPUT}" ]]; then
    echo "Single file detected."
    INEXT=$(extension $INPUT)
    if [[ "$OUTEXT" == "$INEXT" ]]; then
        echo "- Creating symlink."
        ln -s ${INPUT} ${OUTPUT} \
            && echo "Created symlink ${OUTPUT}" \
            || (cecho "Failed to create symlink ${OUTPUT}" && exit 1)
    elif [[ "$OUTEXT" == "gz" ]]; then
        echo -n " - Compressing input..."
        gzip ${INPUT} -c > ${OUTPUT}
        echo "done."
    else
        echo -n " - Decompressing input..."
        gzip -d -c ${INPUT} > ${OUTPUT}
        echo "done."
    fi

elif [[ -d "${INPUT}" ]]; then
    echo "Directory detected, concatenating found files."
    flist=$(mktemp)
    nfiles=$(find ${INPUT} -type f \( \
        -iname '*.fastq' -o -iname '*.fastq.gz' -o -iname '*.fq' -o -iname '*.fq.gz' \
        ! -wholename ${OUTPUT} \) | tee ${flist} | wc -l)
    echo " - Found ${nfiles} files."
    ext=""
    for fname in $(cat ${flist}); do
        newext=$(extension ${fname})
        if [[ -z "${ext}" ]]; then
            ext=${newext}
        elif [[ ${ext} != ${newext} ]]; then
            echo " - Multiple file extensions found, exiting"
            exit 1
        fi
    done
    echo -n " - Concatenating files..."
    if [[ "$ext" == "gz" ]] && [[ "$OUTEXT" != "gz" ]]; then
        cat ${flist} | xargs cat | gzip -d > ${OUTPUT}
    elif [[ "$ext" != "gz" ]] && [[ "$OUTEXT" == "gz" ]]; then
        cat ${flist} | xargs cat | gzip > ${OUTPUT}
    else
        cat ${flist} | xargs cat > ${OUTPUT}
    fi
    echo "done."
    rm ${flist}
fi