#!/bin/bash
# wut-audio-sha1
#
# Usage:
# wut-audio-sha1
# Example:
# wut-audio-sha1
#
# This script verifies sha1 checksums of files downloaded from archive.org.
# If the checksum doesn't match, the file is deleted and the download is
# re-tried.
# 
# This script only downloads audio files from archive.org, not satnogs.org.

cd download || exit
# Compile a list of ogg files (NOTE: THIS WILL BREAK AS ARCHIVE GROWS XXX)
echo "Total audio files: `ls -1 */satnogs_*.ogg | wc -l`"
for i in */satnogs_*.ogg
	do OBSID=`dirname $i`
	# Go into directories with audiofiles
	echo $OBSID
	cd $OBSID
	# See if there is an archive.org XML file, if not, download it.
	XMLURL="https://archive.org/download/satnogs-observation-$OBSID/satnogs-observation-$OBSID""_files.xml"
	XMLFILE=`basename "$XMLURL"`
	[ ! -f "$XMLFILE" ] &&						\
	curl                                                            \
       		--location                                              \
		--silent                                                \
		--http2 --ipv4                                          \
		--remote-time                                           \
		--output $XMLFILE                                       \
		$XMLURL                                                 \
		&& sleep `echo $((0 + RANDOM % 1))`
	# Get name of audio file.
	AUDIOURL=`cat $OBSID.json | jq --compact-output '.[0] | {archive_url}' | grep ogg | cut -f 2- -d : | sed -e 's/}//g' -e 's/http:/https:/g' -e 's/"//g'`
	AUDIOFILE=`basename "$AUDIOURL"`
	# Get sha1 for audio file. XXX sgrep dependency XXX
	AUDIOXMLSHA1=`sgrep -g xml							\
	'"<file name=\"'$AUDIOFILE'\" source=\"original\"" .. "/file>"' \
	"$XMLFILE" |				\
	grep "<sha1>" |							\
	sed -e 's/.*<sha1>//' -e 's/<\/sha1>//'`
	AUDIOFILESHA1=`sha1sum $AUDIOFILE | cut -f 1 -d " "`
	echo -e -n "XML:  $AUDIOXMLSHA1\nFile: $AUDIOFILESHA1 "
	if [ "$AUDIOXMLSHA1" = "$AUDIOFILESHA1" ] ; then
		echo "Good"
	else
		echo "Bad, re-downloading $AUDIOURL"
		rm "$AUDIOFILE"
		curl							\
		--location						\
		--silent						\
		--http2 --ipv4						\
		--remote-time						\
		--output $AUDIOFILE					\
		$AUDIOURL						\
                && sleep `echo $((0 + RANDOM % 1))`
		AUDIOFILESHA1=`sha1sum $AUDIOFILE | cut -f 1 -d " "`
		echo -e -n "XML:  $AUDIOXMLSHA1\nFile: $AUDIOFILESHA1 "
		if [ "$AUDIOXMLSHA1" = "$AUDIOFILESHA1" ] ; then
			echo "Re-download is good"
		else
			echo "Still bad after re-downloading"
			rm "$AUDIOFILE"
		fi
	fi
	cd ..
done