r4
/
southpark-de-downloader
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Rework scraping method

This commit is contained in:
xypwn 2020-12-15 14:59:14 +01:00
parent f7fc2ef0eb
commit dbf448bc77
2 changed files with 95 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
config.sh
View File

@ -2,8 +2,13 @@
# South Park Downloader Configuration File # # South Park Downloader Configuration File #
############################################ ############################################
# Relative or absolute paths are allowed. Please DON'T use ~/, but rather /$HOME or /home/$USER.
# youtube-dl (or youtube-dlc) executable path # youtube-dl (or youtube-dlc) executable path
YOUTUBE_DL="./yt-dlc/youtube-dlc" YOUTUBE_DL="./yt-dlc/youtube-dlc"
# Where the downloaded videos will get stored # Where the downloaded videos will get stored
OUTDIR="./downloads" OUTDIR="./downloads"
# Where cache files are stored
CACHEDIR="./cache"

151
southpark-downloader.sh
View File

@ -1,14 +1,17 @@
#!/usr/bin/env sh #!/usr/bin/env sh
source "$(dirname $0)/config.sh" source "$(dirname "$0")/config.sh"
[ ! -e "$(dirname $0)/$OUTDIR" ] && mkdir -p "$(dirname $0)/$OUTDIR" # Turn paths into absolute ones, if they aren't already. Will be necessary, since we'll change directories later.
[ ! "${CACHEDIR::1}" = "/" ] &&
# Turn paths into absolute ones, if they aren't already, as we will change directories later CACHEDIR="$(readlink -f "$(dirname "$0")/$CACHEDIR")"
[ ! "${OUTDIR::1}" = "/" ] && [ ! "${OUTDIR::1}" = "/" ] &&
OUTDIR="$(readlink -f $(dirname $0)/$OUTDIR)" OUTDIR="$(readlink -f "$(dirname "$0")/$OUTDIR")"
[ ! "${YOUTUBE_DL::1}" = "/" ] && [ ! "${YOUTUBE_DL::1}" = "/" ] &&
YOUTUBE_DL="$(readlink -f $(dirname $0)/$YOUTUBE_DL)" YOUTUBE_DL="$(readlink -f "$(dirname "$0")/$YOUTUBE_DL")"
[ ! -e "$OUTDIR" ] && mkdir -p "$OUTDIR"
[ ! -e "$CACHEDIR" ] && mkdir -p "$CACHEDIR"
p_info() { p_info() {
echo -e "\e[32m>>> $@\e[m" echo -e "\e[32m>>> $@\e[m"
@ -20,19 +23,25 @@ p_error() {
usage() { usage() {
echo "Usage:" echo "Usage:"
echo " $(basename $0) [OPTIONS] -a - Download all episodes" echo " $(basename "$0") [OPTIONS] -a - Download all episodes"
echo " $(basename $0) [OPTIONS] -s <season> - Download all episodes in the specified season" echo " $(basename "$0") [OPTIONS] -s <season> - Download all episodes of the specified season"
echo " $(basename $0) [OPTIONS] -s <season> -e <episode> - Download the specified episode" echo " $(basename "$0") [OPTIONS] -s <season> -e <episode> - Download the specified episode"
echo " $(basename "$0") -h - Show help page"
echo "Options:" echo "Options:"
echo " -p - Show progress (default)"
echo " -P - Hide progress"
echo " -E - Download episodes in English (default)" echo " -E - Download episodes in English (default)"
echo " -D - Download episodes in German" echo " -D - Download episodes in German"
echo " -p - Show progress" echo " -u - Update episode index (default)"
echo " -U - Skip episode index update"
} }
unset OPT_SEASON OPT_EPISODE OPT_ALL OPT_EN OPT_LANG OPT_PROGRESS unset OPT_SEASON OPT_EPISODE OPT_ALL OPT_EN OPT_LANG OPT_PROGRESS OPT_UPDATE_INDEX
OPT_LANG="EN" OPT_LANG="EN"
OPT_PROGRESS=true
OPT_UPDATE_INDEX=true
while getopts "haEDps:e:" arg; do while getopts "pPEDuUas:e:h" arg; do
case "$arg" in case "$arg" in
h) h)
usage usage
@ -56,66 +65,71 @@ while getopts "haEDps:e:" arg; do
p) p)
OPT_PROGRESS=true OPT_PROGRESS=true
;; ;;
P)
unset OPT_PROGRESS
echo hi
;;
u)
OPT_UPDATE_INDEX=true
;;
U)
unset OPT_UPDATE_INDEX
;;
?) ?)
echo "Invalid option: -$OPTARG"
usage usage
exit 1 exit 1
;; ;;
esac esac
done done
# Parts of the URL differ depending on the language of the website
if [ "$OPT_LANG" = "DE" ]; then if [ "$OPT_LANG" = "DE" ]; then
SEASON_1_URL="https://www.southpark.de/seasons/south-park/yjy8n9/staffel-1" INDEX_FILENAME="$CACHEDIR/_episode_index_DE_"
REGEX_SEASON_URL="\"/seasons/south-park/[0-9a-z]\+/staffel-[0-9]\+\"" INDEX_INITIAL_URL="https://www.southpark.de/folgen/940f8z/south-park-cartman-und-die-analsonde-staffel-1-ep-1"
REGEX_EPISODE_URL="\"/folgen/[0-9a-z]\+/south-park-[0-9a-z-]\+-staffel-[0-9]\+-ep-[0-9]\+\"" REGEX_EPISODE_URL="\"/folgen/[0-9a-z]\+/south-park-[0-9a-z-]\+-staffel-[0-9]\+-ep-[0-9]\+\""
elif [ "$OPT_LANG" = "EN" ]; then elif [ "$OPT_LANG" = "EN" ]; then
SEASON_1_URL="https://www.southpark.de/en/seasons/south-park/yjy8n9/season-1" INDEX_FILENAME="$CACHEDIR/_episode_index_EN_"
REGEX_SEASON_URL="\"/en/seasons/south-park/[0-9a-z]\+/season-[0-9]\+\"" INDEX_INITIAL_URL="https://www.southpark.de/en/episodes/940f8z/south-park-cartman-gets-an-anal-probe-season-1-ep-1"
REGEX_EPISODE_URL="\"/en/episodes/[0-9a-z]\+/south-park-[0-9a-z-]\+-season-[0-9]\+-ep-[0-9]\+\"" REGEX_EPISODE_URL="\"/en/episodes/[0-9a-z]\+/south-park-[0-9a-z-]\+-season-[0-9]\+-ep-[0-9]\+\""
fi fi
# Indexes all season page URLs update_index() {
index_seasons() { [ ! -e "$INDEX_FILENAME" ] && echo "$INDEX_INITIAL_URL" > "$INDEX_FILENAME"
# Get all season URLs by matching the regex echo -ne "\e[32m>>> Updating episode index\e[m"
SEASON_URLS=$(curl -s "$SEASON_1_URL" | grep -o "$REGEX_SEASON_URL" | tr -d "\"" | sed -E "s/^/https:\/\/www.southpark.de/g") while true; do
local URL=$(tail -n1 "$INDEX_FILENAME")
local NEWURLS=$(curl -s "$URL" | grep -o "$REGEX_EPISODE_URL" | tr -d "\"" | sed -E "s/^/https:\/\/www.southpark.de/g")
[ "$URL" = $(printf "$NEWURLS" | tail -n1) ] && break
echo "$NEWURLS" >> "$INDEX_FILENAME"
echo -ne "\e[32m.\e[m"
done
# The awk command removes duplicate lines
local NEW_INDEX=$(awk '!x[$0]++' "$INDEX_FILENAME")
printf "$NEW_INDEX" > "$INDEX_FILENAME"
echo
} }
# Indexes all episode URLs of the currently indexed season (can only index 1 season at once, for now) # Returns all episode URLs in the specified season
index_episodes() { get_season() {
local SEASON_NUMBER="$1" local SEASON_NUMBER="$1"
get_season_url "$SEASON_NUMBER" grep "\-${SEASON_NUMBER}-ep-[0-9]\+$" "$INDEX_FILENAME"
local SEASON_URL="$RES"
EPISODE_URLS=$(curl -s "$SEASON_URL" | grep -o "$REGEX_EPISODE_URL" | tr -d "\"" | sed -E "s/^/https:\/\/www.southpark.de/g")
INDEXED_SEASON="$SEASON_NUMBER"
} }
################ # Returns the URL of the specified episode
# All functions named get_<something> store their result in the RES variable. get_episode() {
# We're not using command substitution, because then these functions couldn't set variables.
################
get_season_url() {
local SEASON_NUMBER="$1"
[ -z "$SEASON_URLS" ] && index_seasons
RES=$(echo "$SEASON_URLS" | grep "\-${SEASON_NUMBER}$")
}
get_episode_url() {
local SEASON_NUMBER="$1" local SEASON_NUMBER="$1"
local EPISODE_NUMBER="$2" local EPISODE_NUMBER="$2"
[ ! "$INDEXED_SEASON" = "$SEASON_NUMBER" ] && index_episodes "$SEASON_NUMBER" grep "\-${SEASON_NUMBER}-ep-${EPISODE_NUMBER}$" "$INDEX_FILENAME"
RES=$(echo "$EPISODE_URLS" | grep "ep-${EPISODE_NUMBER}$")
} }
get_num_seasons() { get_num_seasons() {
[ -z "$SEASON_URLS" ] && index_seasons # Effectively searches, how many "episode 1s" there are in the index
RES=$(echo "$SEASON_URLS" | wc -l) grep "\-[0-9]\+-ep-1$" "$INDEX_FILENAME" | wc -l
} }
# Returns the number of episodes in the specified season
get_num_episodes() { get_num_episodes() {
local SEASON_NUMBER="$1" local SEASON_NUMBER="$1"
[ ! "$INDEXED_SEASON" = "$SEASON_NUMBER" ] && index_episodes "$SEASON_NUMBER" get_season "$SEASON_NUMBER" | wc -l
RES=$(echo "$EPISODE_URLS" | wc -l)
} }
tmp_cleanup() { tmp_cleanup() {
@ -133,17 +147,32 @@ monitor_progress() {
done done
} }
download_interrupt() {
p_info "User interrupt received"
tmp_cleanup
exit 0
}
merge_interrupt() {
p_info "User interrupt received"
tmp_cleanup
p_info "Cleaning up corrupted output file"
rm -rf "$1"
exit 0
}
# Takes season and episode number as arguments # Takes season and episode number as arguments
download_episode() { download_episode() {
local SEASON_NUMBER="$1" local SEASON_NUMBER="$1"
local EPISODE_NUMBER="$2" local EPISODE_NUMBER="$2"
get_episode_url "$SEASON_NUMBER" "$EPISODE_NUMBER"
local URL="$RES"
local OUTFILE="${OUTDIR}/South_Park_S${SEASON_NUMBER}_E${EPISODE_NUMBER}_${OPT_LANG}.mp4" local OUTFILE="${OUTDIR}/South_Park_S${SEASON_NUMBER}_E${EPISODE_NUMBER}_${OPT_LANG}.mp4"
[ -e "$OUTFILE" ] && echo "Already downloaded Season ${SEASON_NUMBER} Episode ${EPISODE_NUMBER}" && return [ -e "$OUTFILE" ] && echo "Already downloaded Season ${SEASON_NUMBER} Episode ${EPISODE_NUMBER}" && return
local URL=$(get_episode "$SEASON_NUMBER" "$EPISODE_NUMBER")
[ -z "$URL" ] && echo "Unable to download Season ${SEASON_NUMBER} Episode ${EPISODE_NUMBER}; skipping" && return
p_info "Downloading Season $SEASON_NUMBER Episode $EPISODE_NUMBER ($URL)" p_info "Downloading Season $SEASON_NUMBER Episode $EPISODE_NUMBER ($URL)"
trap download_interrupt SIGINT
TMPDIR=$(mktemp -d "/tmp/southparkdownloader.XXXXXXXXXX") TMPDIR=$(mktemp -d "/tmp/southparkdownloader.XXXXXXXXXX")
[ -n "OPT_PROGRESS" ] && monitor_progress "$TMPDIR"& [ -n "$OPT_PROGRESS" ] && monitor_progress "$TMPDIR"&
pushd "$TMPDIR" > /dev/null pushd "$TMPDIR" > /dev/null
if ! "$YOUTUBE_DL" "$URL" 2>/dev/null | grep --line-buffered "^\[download\]" | grep -v --line-buffered "^\[download\] Destination:"; then if ! "$YOUTUBE_DL" "$URL" 2>/dev/null | grep --line-buffered "^\[download\]" | grep -v --line-buffered "^\[download\] Destination:"; then
p_info "possible youtube-dl \e[1;31mERROR\e[m" p_info "possible youtube-dl \e[1;31mERROR\e[m"
@ -151,43 +180,42 @@ download_episode() {
exit 1 exit 1
fi fi
echo "[download] Merging video files" echo "[download] Merging video files"
# Remove all single quotes from video files, as they cause problems trap "merge_interrupt \"$OUTFILE\"" SIGINT
for i in ./*.mp4; do mv -n "$i" "$(echo $i | tr -d \')"; done # Remove all single quotes and dashes from video files, as they cause problems
for i in ./*.mp4; do mv -n "$i" "$(echo $i | tr -d \'-)"; done
# Find all video files and write them into the list # Find all video files and write them into the list
printf "file '%s'\n" ./*.mp4 > list.txt printf "file '%s'\n" ./*.mp4 > list.txt
# Merge video files # Merge video files
ffmpeg -safe 0 -f concat -i "list.txt" -c copy "$OUTFILE" 2>/dev/null ffmpeg -safe 0 -f concat -i "list.txt" -c copy "$OUTFILE" 2>/dev/null
popd > /dev/null popd > /dev/null
trap - SIGINT
tmp_cleanup tmp_cleanup
} }
# Takes season number as an argument # Takes season number as an argument
download_season() { download_season() {
local SEASON_NUMBER="$1" local SEASON_NUMBER="$1"
get_num_episodes "$SEASON_NUMBER" local NUM_EPISODES=$(get_num_episodes "$SEASON_NUMBER")
local NUM_EPISODES="$RES"
for i in $(seq "$NUM_EPISODES"); do for i in $(seq "$NUM_EPISODES"); do
download_episode "$SEASON_NUMBER" "$i" download_episode "$SEASON_NUMBER" "$i"
done done
} }
download_all() { download_all() {
get_num_seasons local NUM_SEASONS=$(get_num_seasons)
local NUM_SEASONS="$RES"
for i in $(seq "$NUM_SEASONS"); do for i in $(seq "$NUM_SEASONS"); do
download_season "$i" download_season "$i"
done done
} }
if [ -n "$OPT_SEASON" ]; then if [ -n "$OPT_SEASON" ]; then
get_season_url "$OPT_SEASON" [ -n "$OPT_UPDATE_INDEX" ] && update_index
[ -z "$RES" ] && [ -z "$(get_season $OPT_SEASON)" ] &&
p_error "Unable to open Season $OPT_SEASON" && p_error "Unable to find Season $OPT_SEASON" &&
exit 1 exit 1
if [ -n "$OPT_EPISODE" ]; then if [ -n "$OPT_EPISODE" ]; then
get_episode_url "$OPT_SEASON" "$OPT_EPISODE" [ -z "$(get_episode $OPT_SEASON $OPT_EPISODE)" ] &&
[ -z "$RES" ] && p_error "Unable to find Season $OPT_SEASON Episode $OPT_EPISODE" &&
p_error "Unable to open Season $OPT_SEASON Episode $OPT_EPISODE" &&
exit 1 exit 1
p_info "Going to download Season $OPT_SEASON Episode $OPT_EPISODE" p_info "Going to download Season $OPT_SEASON Episode $OPT_EPISODE"
download_episode "$OPT_SEASON" "$OPT_EPISODE" download_episode "$OPT_SEASON" "$OPT_EPISODE"
@ -196,6 +224,7 @@ if [ -n "$OPT_SEASON" ]; then
download_season "$OPT_SEASON" download_season "$OPT_SEASON"
fi fi
elif [ -n "$OPT_ALL" ]; then elif [ -n "$OPT_ALL" ]; then
[ -n "$OPT_UPDATE_INDEX" ] && update_index
p_info "Going to download ALL episodes" p_info "Going to download ALL episodes"
download_all download_all
else else