AA_pbcorescript_with_checks

#!/bin/bash

# a version of the AA_PBCorescript that automatically checks each drive for derivatives and known failed files using stored csv reference files, and isolates them without processing them

# manual data input for metadata creation

echo "Drag and drop the drive to run the operation on"
read drive
echo "Drag and drop the directory where you would like to save your checksum files and instantiation records"
read dir
echo "Drag and drop the CSV file containing the AAPB GUIDs"
read AAPB
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

catfun ()
{ # create PBCore XML document for AAPB asset

	printf '%s\n' "MD5 ($i) = $c" >> "$dir"/"$barcode"_MD5_Source.csv
	printf '%s\n' "MD5 ($i) = $c" >> /Volumes/dept/MLA/Archives/LTO_Tape_Preservation/MD5/aapb_md5_total.csv
    guid=$( cat $AAPB | grep $artguid | cut -d , -f 2 )
    mediainfo --Output=PBCore2 "$i" >> "$dir"/mediainfo.xml
    driveid=$( pwd | cut -d / -f 3 )
    path=$( pwd | cut -c 10-)
    echo "<?xml version=\"1.0\" encoding=\"UTF-8\"?>" >> "$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>$LTO/$path/$i</instantiationLocation>" >> "$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}"_"${i%.*}".xml
    printf '\t%s\n\t%s\n' "<instantiationAnnotation annotationType=\"Drive\">Preservation copy on hard drive "$driveid", "$barcode"</instantiationAnnotation>" "<instantiationAnnotation annotationType=\"File Name\">$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
    cat "$dir"/PBCoreInstantiations/pbcore_instantiation_"${artguid}"_"${i%.*}".xml >> pbcore_instantiation_"${artguid}"_"${i%.*}".xml
    rm "$dir"/mediainfo.xml
}

# restructure drive for human-readability

cd "$drive"
mkdir files_needing_qc
for d in $(find "$drive" \( ! -regex '.*/\..*' \) -not -path . -type d -depth 1 -not -name *files_needing_qc); do
	mv "$d" "$drive"/files_needing_qc
done
mkdir confirmed_good_files
mkdir bad_files
mkdir derivatives

# check for derivatives and failed files before creating metadata

for d in $(find "$drive" \( ! -regex '.*/\..*' \) ! -path . -type d -depth 2); do
	cd "$d"
	artguid=$(echo "$d" | cut -d / -f 5)
	file=$( find . \( ! -regex '.*/\..*' \) ! -path . -type f -not -name *.xml)
	i=$(echo "$file" | cut -d '/' -f 2 )
	if grep $i "/Volumes/dept/MLA/Archives/LTO_Tape_Preservation/AA_Batch_IDs/derivatives.csv"
	then
		cd "$drive"
		mv "$d" "$drive"/derivatives
	elif grep "$i" "/Volumes/dept/MLA/Archives/LTO_Tape_Preservation/AA_Batch_IDs/md5_original_values.csv"; then
		c=$(md5 -q "$i")
		if grep "$i" "/Volumes/dept/MLA/Archives/LTO_Tape_Preservation/AA_Batch_IDs/md5_original_values.csv" | grep "$c"; then
			if grep "$c" /Volumes/dept/MLA/Archives/LTO_Tape_Preservation/MD5/aapb_md5_total.csv; then
				cd "$drive"
				mv "$d" "$drive"/derivatives
			else
			    cd "$drive"
				mv "$d" "$drive"/confirmed_good_files/
				cd "$drive"/confirmed_good_files/"$artguid"
				catfun
				cd "$drive"
			fi
		else 
			driveid=$( echo "$drive" | cut -d / -f 3 )
			cd "$drive"
			mv "$d" "$drive"/bad_files
			printf '\n%s' "$driveid,$artguid,$i" >> /Volumes/dept/MLA/Archives/LTO_Tape_Preservation/AA_Batch_IDs/corrupted_files.csv
		fi
	else
		c=$(md5 -q "$i")
		if grep "$c" /Volumes/dept/MLA/Archives/LTO_Tape_Preservation/MD5/aapb_md5_total.csv; then
				cd "$drive"
				mv "$d" "$drive"/derivatives
			else
				catfun
				cd "$drive"
			fi
	fi
done