#!/bin/bash
# Check how much of an upload made it into the item yet (vs. how much is stuck in the S3 queue if you also watch the upload process)
# Usage: ia-upload-progress IA_IDENTIFIER [FILES]
# If no files are specified, all WARCs in the current directory are used.
identifier=$1
shift
files=("$@")
if [[ ${#files[@]} -eq 0 ]]
then
	files=(*.warc.gz)
fi
scriptpath="$(cd "$(dirname "$0")"; pwd -P)"

uploaded=$(ia metadata "${identifier}" | grep -Po '("source"\s*:\s*"original",[^}]*"size"\s*:\s*"\K\d+|"size"\s*:\s*"\K\d+(?="\s*,[^}]*"source"\s*:\s*"original"))' | awk '{sum+=$1} END {printf "%.0f\n", sum}')
localSize=$(du -bc "${files[@]}" | tail -1 | cut -f1)
echo "Uploaded $("${scriptpath}/format-size" <<<"${uploaded}") of $("${scriptpath}/format-size" <<<"${localSize}") ($(awk '{ printf "%.1f", 100 * ($1 / $2); }' <<<"${uploaded} ${localSize}") %)"