dhrb

#!/bin/bash

set -u

VIDEO_EXTENSION=""
# nhrb = NOID, Handle, Runtime, BSN
#
# This script is meant to be run on HIDVL video objects in the R* repository
# the script assumes the standard "wip" directory structure
#   foo/
#       aux/
#       data/6465441_marcxml.xml
#       data/6465441_mods.xml
#       data/foo_m.mov
#       handle
#
# the script outputs the following information:
# <digitization id>,<handle URL>,<runtime>,<BSN>
#
# foo,http://hdl.handle.net/2333.1/foo,00:00:05.015,6465441
#
# this is initially designed for HIDVL objects.
err_exit() {
    exit 1
}

print_usage() {
    echo "usage: $0 <path to HIDVL WIP>"
    echo "       e.g., "
    echo "       $0 fasdlkf"
    echo "       $0 /path/to/wip/root/fasdlkf"
    echo "  "
    echo "CSV header:"
    echo "noid,handle,duration (hh:mm:ss.msec),bsn"
}

MEDIAINFO=$(which mediainfo 2>/dev/null)
if [[ "$?" != 0 ]]; then
    echo "ERROR: mediainfo not found" >&2
    err_exit
fi

init_runtime() {
    runtime=$($MEDIAINFO -f ${wip_path}/data/*.${VIDEO_EXTENSION} | grep uration | head -5 | tail -1 | cut -d':' -f2- | sed -e 's/ //')
}

init_handle() {
    handle="https://hdl.handle.net/$(cat $wip_path/handle)"
}

init_digi_id() {
    digi_id=$(basename $wip_path)
}

init_bsn() {
    local marcxml_path
    local marcxml

    marcxml_path=$(ls -1 ${wip_path}/data/*_marcxml.xml | head -1)
    marcxml=$(basename $marcxml_path)
    bsn=$(echo $marcxml | rev | cut -d'_' -f2 | rev)
}


assert_dir() {
    if [[ ! -d "$1" ]]; then
	echo "ERROR: $1 must be a directory"
	print_usage
	err_exit
    fi
}


assert_handle() {
    if [[ ! -f "${wip_path}/handle" ]]; then
	echo "ERROR: handle file missing"
	err_exit
    fi
}
	
# assert that there is at least one .mov file in WIP
init_video_extension() {
    VIDEO_EXTENSION=""
    (ls -1 ${wip_path}/data/*.mov &>/dev/null)
    if [[ "$?" == 0 ]]; then
	VIDEO_EXTENSION="mov"
    fi

    (ls -1 ${wip_path}/data/*.mp4 &>/dev/null)
    if [[ "$?" == 0 ]]; then
	VIDEO_EXTENSION="mp4"
    fi

    if [[ "$VIDEO_EXTENSION" == "" ]]; then
	echo "ERROR: no video (.mov, .mp4) files found in ${wip_path}/data" >&2
	err_exit
    fi
}

# assert that there is one _marcxml.xml file in WIP
assert_marcxml() {
    local file_list
    local file_count

    file_list=$(ls -1 ${wip_path}/data/*_marcxml.xml 2>/dev/null)
    if [[ "$?" != 0 ]]; then
	echo "ERROR: no _marcxml.xml files found in ${wip_path}/data" >&2
	err_exit
    fi

    if [[ $(echo "$file_list" | wc -l) -ne 1 ]]; then
	echo "ERROR: too many _marcxml.xml files found in ${wip_path}/data" >&2
	err_exit
    fi
}


main() {

    if [[ "$#" -ne 1 ]]; then
	echo "ERROR: incorrect argument count"
	print_usage
	err_exit
    fi

    wip_path="$1"

    assert_dir "$wip_path"
    assert_handle
    init_video_extension
    assert_marcxml

    digi_id=''
    handle=''
    runtime=''
    bsn=''

    init_digi_id
    init_handle
    init_runtime
    init_bsn

    echo "${digi_id},${handle},${runtime},${bsn}"
    exit 0
}

main "$@"