sapo-videos/saca-sapos

#!/usr/bin/env bash
# SPDX-License-Identifier: AGPL-3.0-only
# SPDX-FileCopyrightText: 2023 Hugo Peixoto <hugo.peixoto@gmail.com>

set -ueo pipefail

test_command() {
  command -v "$1" > /dev/null || (
    echo "$1: comando não instalado"
    exit 1;
  )
}

test_command curl
test_command jq
test_command xargs
test_command sha256sum
test_command stat

PROCS=${PROCS:=4}
SERVER="https://sapo.pxto.pt/config.json"
export SAPO_EXEC="$0"

get_config() {
  curl -s "$SERVER"
}

download_many() {
  AMOUNT="$1"
  export SAPO_CONTACT="$2"

  CONFIG="$(get_config)"
  export SAPO_CONFIG_VIDEOS_URL="$(echo "$CONFIG" | jq -r .videos_url)"
  export SAPO_CONFIG_UPLOAD_URL="$(echo "$CONFIG" | jq -r .upload_url)"
  export SAPO_CONFIG_MAX="$(echo "$CONFIG" | jq -r .max_amount)"

  if [ "$AMOUNT" -gt "$SAPO_CONFIG_MAX" ]; then
    echo "Não podes pedir mais do que $SAPO_CONFIG_MAX vídeos!" >&2
    exit 1;
  fi

  mkdir -p videos
  while [ "$AMOUNT" -gt 0 ]; do
    if [ "$AMOUNT" -ge 1000 ]; then
      N=1000
    else
      N=$AMOUNT
    fi

    echo "another part ($N)" >&2
    curl -s "$SAPO_CONFIG_VIDEOS_URL?amount=$N" |
      jq -r '.videos[]' |
      xargs "-P$PROCS" -I {} "$SAPO_EXEC" single {}

    AMOUNT=$(( $AMOUNT - $N ))
  done
}

download_single() {
  VIDEO_ID="$1"
  VIDEO_URL="https://rd.videos.sapo.pt/$VIDEO_ID/mov/1"
  FILENAME="videos/$VIDEO_ID.mp4"

  if [ -f "$FILENAME" ]; then
    SHA=$(sha256sum "$FILENAME" | awk '{print $1}')
    SIZE=$(stat --format=%s "$FILENAME")
    MB=$(( $SIZE / 1024 / 1024 ))
    echo "repetido: $VIDEO_ID / sha256:$SHA / ${MB} MiB"
  else
    curl -Ls "$VIDEO_URL" -o "$FILENAME"
    SHA=$(sha256sum "$FILENAME" | awk '{print $1}')
    SIZE=$(stat --format=%s "$FILENAME")
    MB=$(( $SIZE / 1024 / 1024 ))

    PAYLOAD="$(jq -rc --null-input \
      --arg video_id "$VIDEO_ID" \
      --arg sha "$SHA" \
      --arg email "$SAPO_CONTACT" \
      --argjson size "$SIZE" \
      '{video_id: $video_id, sha256: $sha, size: $size, email: $email}')"
    curl -s "$SAPO_CONFIG_UPLOAD_URL" -H 'content-type: application/json' -d "$PAYLOAD" > /dev/null

    echo "descarregado: $VIDEO_ID / sha256:$SHA / ${MB} MiB"
  fi
}

unknown() {
  CONFIG="$(get_config)"
  SAPO_CHECK_VIDEOS_URL="$(echo "$CONFIG" | jq -r .check_url)"

  find videos/ -name "*.mp4" |
    sed -e 's|videos/\(.*\).mp4|\1|g' |
    xargs -n1000 |
    sed -e 's/ /","/g' -e 's/^/["/' -e 's/$/"]/' |
    while read body; do
      echo "$body" | curl --silent --json "@-" "$SAPO_CHECK_VIDEOS_URL"
    done |
    jq -rc '.videos[]|select(.sha256==null)|("videos/" + .randname + ".mp4")'
}

case "$1" in
  "download")
    download_many "$2" "$3"
    ;;
  "single")
    download_single "$2"
    ;;
  "unknown")
    unknown
    ;;
  "stats")
    BYTES="$(du -b videos/ | awk '{print $1}')"
    NUM="$(ls -1 videos/ | wc -l)"
    MB=$(( $BYTES / 1024 / 1024 ))
    MB_PER_VIDEO=$(( $MB / $NUM ))

    echo "Número de vídeos: $NUM"
    echo "Tamanho total: $MB MiB"
    echo "Tamanho médio: $MB_PER_VIDEO MiB/vídeo"
    ;;
  "test")
    get_config > /dev/null || (
      echo "A comunicação com o servidor falhou: $SERVER"
      exit 1;
    )
    echo "Tudo pronto!"
esac