tukuppt-downloader.sh

#!/usr/local/bin/bash
#
# Background Music / Sound Effects Downloader
#

BASE_URL=https://www.tukuppt.com
IDX_URL=$BASE_URL/peiyueso/shipin5097/__zonghe_0_0_0_0_0_0_%d.html
#IDX_URL=$BASE_URL/yinxiaomuban/shipin/__zonghe_0_0_0_0_0_0_%d.html
API_URL=$BASE_URL/api/audio
TOTAL_PAGES=46
#TOTAL_PAGES=50
HTTP_PROXY=
OUTPUT_FILE=tukuppt-background-music.jsonl
#OUTPUT_FILE=tukuppt-sound-effects.jsonl

declare -a SEEN
readarray -t SEEN < <(jq -r '.id' $OUTPUT_FILE)
echo "### Previous Downloaded: ${#SEEN[@]} ###" >&2

for ((page=1;page<=$TOTAL_PAGES;page++)); do
  url=$(printf "$IDX_URL" $page)
  html=$(curl -s "$url")
  count=0
  paste <(echo "$html" | pup 'dl i attr{value}') \
        <(echo "$html" | pup 'dl dt .title text{}') \
        <(echo "$html" | pup 'dl dt .title attr{href}') \
        <(echo "$html" | pup 'dl .end-time text{}') |
  while IFS=$'\t' read id title url time; do
    echo -n "#$page.$((++count)): $title ($time) " >&2
    if printf '%s\0' "${SEEN[@]}" | grep -qFxz -- "$id"; then
      echo "☑️" >&2
      continue
    fi
    duration=$(date -d "1970-01-01T00:${time} UTC" "+%s")
    download=$(curl -s -m5 ${HTTP_PROXY:+-x $HTTP_PROXY} -e $BASE_URL $API_URL -d value=$id | jq -r '.data.swf')
    if (($? == 0)); then
      jq -n -c --arg id "$id" \
               --arg title "$title" \
               --argjson duration "$duration" \
               --arg url "$BASE_URL$url" \
               --arg download "$download" '{$id, $title, $duration, $url, $download}' >> $OUTPUT_FILE
      echo "✅" >&2
    else
      echo "❌" >&2
    fi
  done
done