fix-pdf-s3.sh

#!/bin/bash

# usage
# ./fix-pdf-s3.sh bucket path/to/folder/to/scan
# No trailing slash on the bucket or path

bucket=$1
path=$2
origin=$1

fixfile() {
    echo "Fixing file $file"
    aws s3 cp --content-type "application/pdf" --metadata-directive REPLACE s3://$bucket/$path/$file s3://$bucket/$path/$file
}

checkpdffile() {
    fileInfo=`aws s3api get-object --bucket $bucket --key $path/$file /dev/null`
    mime=`echo $fileInfo | python3 -c "import sys, json; print(json.load(sys.stdin)['ContentType'])"`
    if [ $mime != "application/pdf" ]
    then
        fixfile
    else
  	echo $file is the correct mime type.
    fi
}

checkfile() {
    if [[ $file == *.pdf ]]
    then
	checkpdffile
    fi
}

for file in $(aws s3 ls $bucket/$path/ | awk '{$1=$2=$3=""; print $0}' | sed 's/^[ \t]*//');
do
    checkfile
done