AA_pbcorescript

#!/bin/sh

#  batch checksum and create PBCore instantiation records for AA files, using md5 and mediainfo

echo "Drag and drop the directory where you would like to save your checksum files and instantiation records and then hit ENTER"
read dir
echo "Drag and drop the CSV file containing the AAPB GUIDs"
read AAPB
echo "Type in the unique ID of the drive"
read drive
echo "Type in the barcode number of the drive, in the format barcode######"
read barcode
echo "Type in the barcode number of the designated LTO tape, in the format barcode######"
read LTO
mkdir "$dir"/PBCoreInstantiations

for d in $(find . \( ! -regex '.*/\..*' \) ! -path . -type d); do
    cd $d
    file=$( find . \( ! -regex '.*/\..*' \) ! -path . -type f -not -name *.xml )
    i=$( echo "$file" | cut -d '/' -f 2 )
    c=$(md5 -q "$i")
    printf '%s\n' "MD5 ($i) = $c" >> "$dir"/"$barcode"_MD5_Source.csv
    for i in "$i"; do
        artguid=${PWD##*/}
        guid=$( cat $AAPB | grep $artguid | cut -d , -f 2 )
        mediainfo --Output=PBCore2 "$i" >> "$dir"/mediainfo.xml
        echo "<?xml version=\"1.0\" encoding=\"UTF-8\"?>" >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        cat "$dir"/mediainfo.xml | grep '<pbcoreInstantiationDocument\|Generated' >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        printf '\t%s\n' "<instantiationIdentifier source=\"AAPB GUID\">$guid</instantiationIdentifier>" >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        cat "$dir"/mediainfo.xml | grep 'instantiationDate\|instantiationDigital' >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        printf '\t%s\n' "<instantiationLocation>$LTO/$drive/$artguid/$i</instantiationLocation>" >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        cat "$dir"/mediainfo.xml | grep 'instantiationMediaType\|instantiationFileSize\|instantiationTimeStart\|instantiationDuration\|instantiationDataRate\|instantiationTracks' >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        printf '\t%s\n' "<instantiationGenerations source=\"PBCore Controlled Vocabulary\">Master: preservation</instantiationGenerations>" >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        sed -n '/<instantiationEssenceTrack>/,/<\instantiationAnnotation/p' "$dir"/mediainfo.xml >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        sed -i '' '$d' "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        cat "$dir"/mediainfo.xml | grep 'instantiationAnnotation' >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        printf '\t%s\n\t%s\n' "<instantiationAnnotation annotationType=\"Drive\">Preservation copy on hard drive "$drive", "$barcode"</instantiationAnnotation>" "<instantiationAnnotation annotationType=\"File Name\">$i</instantiationAnnotation>" "<instantiationAnnotation annotationType=\"Artesia GUID\">$artguid</instantiationAnnotation>" "<instantiationAnnotation annotationType=\"MD5 Checksum\">$c</instantiationAnnotation>" >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
        printf '%s' "</pbcoreInstantiationDocument>" >> "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml
    done
    cat "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml >> pbcore_instantiation_"${artguid}"_"${i%.*}".xml
    rm "$dir"/mediainfo.xml
    cd ..
done
#
#  Created by MLA User on 12/18/14.
#