pdownyq/pdown.sh

#!/usr/bin/bash

#Dependencies:
# yq https://github.com/mikefarah/yq
# curl

YQ=yq

CONFIG=${1:-./config.yaml}
CACHE_DIR=./cache
STATE_DIR=./state
PODCASTS_DIR=./podcasts

yqf() {
		$YQ "$1" "$CONFIG"
}

make_fname() {
		local name ftype title
		ftype=$1; name=$2; title=$3
		case $ftype in
				"date")
						echo "-o $name-$(date --iso-8601).mp3"
						;;
				"default")
						echo "-O"
						;;
				"title")
						sanitized=$(echo "$title" | sed 's/  */-/g' | tr -d -c 'A-Za-z-0-9\n' | grep -o '[^-].*' | cut -c-20)
						echo "-o $name-${sanitized}.mp3"
						;;
		esac
}

download() {
		local e name url limit mp3url fname ftype title
		e=$1
		name=$(yqf "${e}.name");
		url=$(yqf "${e}.url")
		limit=$(yqf "${e}.limit // 1")
		filter=$(yqf "${e}.filter // true")
		ftype=$(yqf "${e}.filename_type // \"default\"")

		local rss_path="$CACHE_DIR/${name}.rss" 
		local downloads_path="$STATE_DIR/${name}.downloads"

		curl -s -L -o "$rss_path" "$url"
		items=".rss.channel.item"

		ids=$($YQ -p xml "$items | .[] | select($filter) | path | .[-1]" "$rss_path" | 
				head -n "$limit")

		for id in $ids
		do
				title=$($YQ -p xml "$items.$id | (.title |= .[0]) | .title" "$rss_path")
				mp3url=$($YQ -p xml "$items.$id | .enclosure.+@url" "$rss_path")
				grep -qF "$mp3url" "$downloads_path" && continue

				echo "Downloading: $title - $mp3url"
				pushd "$CACHE_DIR" || return
				read -ra title_options <  <(make_fname "$ftype" "$name" "$title")
				fname=$(curl -s -w "%{filename_effective}" -f -L "${title_options[@]}" "$mp3url")
				popd || return

				mv "$CACHE_DIR/$fname" "$PODCASTS_DIR"
				echo "$mp3url" >> "$downloads_path"
		done
}

set_proxy() {
		local proxy
		proxy="$(yqf .proxy)"
		if [[ -n "$proxy" ]]; then
				export ALL_PROXY="$proxy"
		fi
}

set_proxy

set_directories() {
		local base_dir cache_dir state_dir podcasts_dir
		base_dir="$(yqf '.base_dir // ""')"
		cache_dir="$(yqf '.cache_dir // ""')"
		state_dir="$(yqf '.state_dir // ""')"
		podcasts_dir="$(yqf '.podcasts_dir // ""')"
		base_dir=${base_dir:-.}
		CACHE_DIR=${cache_dir:-$base_dir/cache}
		STATE_DIR=${state_dir:-$base_dir/state}
		PODCASTS_DIR=${podcasts_dir:-$base_dir/podcasts}
}

set_directories

mkdir -p "$CACHE_DIR" "$STATE_DIR" "$PODCASTS_DIR"

entries=".entries"
for id in $(yqf "$entries | .[] | path | .[-1]")
do
		download "${entries}.$id" &
done
wait