我的自动化番剧工作流

啊B最近番剧越来越不能看了，加上鼓捣影音库已经上了Plex Pass，是时候鼓捣自动下番了！

Transmission

使用 Transmission 进行 BT下载

无脑 Docker

---
version: "2.1"
services:
  transmission:
    image: lscr.io/linuxserver/transmission:latest
    container_name: transmission
    environment:
      - PUID=1001
      - PGID=1001
      - TZ=Asia/Shanghai
      - TRANSMISSION_WEB_HOME= #optional
      - USER= #optional
      - PASS= #optional
      - WHITELIST= #optional
      - PEERPORT= #optional
      - HOST_WHITELIST= #optional
    volumes:
      - /path/to/config/folder:/config
      - /path/to/downloads/folder:/downloads
      - /path/to/watch/folder:/watch
    ports:
      - 9091:9091
      - 51413:51413
      - 51413:51413/udp
#    network_mode: host
    restart: unless-stopped

自动化下载

通过 FlexGet 定期向 acg.rip 爬取rss，解析想要的字幕组，推送到 Transmission 自动下载。

无脑安装命令

pip install flexget

config.yaml

templates:
  tr:
    transmission:
      host: localhost
      port: your_port
      username: your_username
      password: your_password
      # add_paused: yes  # 添加后不自动下载
variables:
  save_root: /mnt/Media/WEB-DL  # 保存的位置
  # 最终会保存到 /mnt/Media/WEB-DL/字幕组名称/番剧名称/S??/ 下

tasks:
  acg.rip:
    rss: https://acg.rip/.xml?term=
    seen:
      fields:
        - url
    manipulate:
    - full:
        from: title # 保存一份拷贝
    - full:
        from: full
        replace:
          regexp: '\[(Lilith-Raws)] (.* / )?(.*?)( S?(\d+))? - (\d+) \[([^]]+)].*?(1080|720)[pP].+'
          format: '\1#0#\3#1#\5#2#\6#3#\7#4#\8'
    - full:
        from: full
        replace:
          regexp: '【(喵萌.*?)】★.*?★\[.*? ?/ ?([^]]+)]\[(\d+)]\[(1080|720)[pP]?].*?(简).+'
          format: '\1#0#\2#1##2#\3#3#\4#4#'
    - full:
        from: full
        replace:
          regexp: '\[(ANi)] (.+?)( S(\d+))? - .+ - (\d+) \[(1080|720)[Pp]?]\[([^]]+)].+'
          format: '\1#0#\2#1#\4#2#\5#3#\7#4#\6'
    - full:
        from: full
        replace:
          regexp: '\[.*(LoliHouse).*] .+ / (.+?)( S(\d+))? - (\d+)(?:v\d+)? \[.*?(1080|720)[pP].*(简).+'
          format: '\1#0#\2#1#\4#2#\5#3##4#\6'
    - full:
        from: full
        replace:
          regexp: '\[(桜都字幕组|MingY|Oborozuki)] .*? / (.*?)( [Ss](\d+))? \[(\d+)]\[(1080|720)[pP]?].*简繁.*内封.+'
          format: '\1#0#\2#1#\4#2#\5#3##4#\6'
    - full:
        from: full
        replace:
          regexp: '\[(SweetSub)[^]]*]\[.+?]\[(.+?)([Ss](\d+))?]\[(\d+)].+?(1080|720)[pP].+?简.+'
          format: '\1#0#\2#1#\4#2#\5#3##4#\6'
    - group:  # 提取字幕组名称
        from: full
        replace:
          regexp: '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: '\1'
    - name:  # 提取番剧名
        from: full
        replace:
          regexp: '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: '\2'
    - session:    # 提取第几季
        from: full
        replace:
          regexp: '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: 'S\3'
    - session:    # 格式化 S1 改成 S01
        from: session
        replace:
          regexp: '^S(\d)$'
          format: 'S0\1'
    - session:    # 如果为空，默认S01
        from: session
        replace:
          regexp: '^S$'
          format: 'S01'
    - episode:    # 提取第几集
        from: full
        replace:
          regexp: '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: 'E\4'
    - episode:    # E1 改成 S01
        from: episode
        replace:
          regexp: '^E(\d)$'
          format: 'E0\1'
    - source:     # 来源，Baha 之类的
        from: full
        replace:
          regexp: '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: '\5'
    - resolution: # 分辨率
        from: full
        replace:
          regexp: '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: '\6'

    - subpath: # 可以用黑名单来拦截
        from: full
        replace:
          regexp:  '^(.+?)#0#(.+?)#1#(\d*)#2#(\d+)#3#(\S*)#4#(\S*)$'
          format: 'Group: \1; Name: \2; Session: S\3'
    - subpath:
        from: subpath
        replace:
          regexp:  '^Group: (.+); Name: (.*); Session: S$'
          format: 'Group: \1; Name: \2; Session: S1'
    - subpath:
        from: subpath
        replace:
          regexp:  '^Group: (.+); Name: (.*); Session: S(\d)$'
          format: 'Group: \1; Name: \2; Session: S0\3'
    - subpath:
        from: subpath
        replace:
          regexp:  '^Group: (.+); Name: (.*); Session: (.*)$'
          format: '\1/\2/\3'
    regexp:
      reject:
        - ANi/Mononogatari/S01  # 黑名单，这样不会下载这个番，支持正则
      accept:
        - .+\/.+\/S\d+
      from: subpath
    template: tr
    transmission:
      path: '{? save_root ?}/{{ subpath }}/'

Plex 识别优化 和 自动清空旧种

Plex 有一个bug，如果文件名为 [Nekomoe kissaten][Undead Girl Murder Farce][02][1080p][JPTC].mp4 是无法识别的，
需要替换为 [Nekomoe kissaten] Undead Girl Murder Farce - 02 [1080p][JPTC].mp4，把名称和集数给分开。
所以写一个简单的脚本来修改名称，也顺便清理旧种子。

flexget-auto.py

import transmission_rpc
import sys
import time
import re

TRANS_CLIENT_CONF = {
    "host": "localhost",
    "port": "9091",
    "username": "username",
    "password": "password",
}

def Airota(client: transmission_rpc.Client, torrent: transmission_rpc.Torrent):
    r = re.compile(r'^(\[Airota\])\[([^]]+)\]\[(\d+)\]')
    new_file_name = r.sub(r"\1 \2 - \3 ", torrent.name, 1)
    if new_file_name == torrent.name:
        return
    ret = client.rename_torrent_path(torrent.id, torrent.name, new_file_name)
    print(f'Airota Rename: {torrent.name} -> {new_file_name}: {ret}')

def NekomoeKissaten(client: transmission_rpc.Client, torrent: transmission_rpc.Torrent):
    r = re.compile(r'^(\[Nekomoe kissaten\])\[([^]]+)\]\[(\d+)\]')
    new_file_name = r.sub(r"\1 \2 - \3 ", torrent.name, 1)
    if new_file_name == torrent.name:
        return
    ret = client.rename_torrent_path(torrent.id, torrent.name, new_file_name)
    print(f'NekomoeKissaten Rename: {torrent.name} -> {new_file_name}: {ret}')

def DeleteOldTorrent(client: transmission_rpc.Client, torrent: transmission_rpc.Torrent) -> bool:
    if torrent.is_finished is False:
        return False
    if torrent.done_date.timestamp() + 3600 * 24 * 7 > time.time():
        return False
    ret = client.remove_torrent(torrent.id, delete_data=False)
    print(f'Delete: {torrent.name}: {ret}')
    return True

def main():
    try:
        trans_client = transmission_rpc.Client(**TRANS_CLIENT_CONF)
    except transmission_rpc.error.TransmissionAuthError as e:
        print(e)
        sys.exit()

    torrents = trans_client.get_torrents()
    for torrent in torrents:
        print(f'{torrent.hashString[-8:]}: {torrent.name} ({torrent.status})', 'in transmission.')
        if DeleteOldTorrent(trans_client, torrent): # 删除老种子
            continue
        NekomoeKissaten(trans_client, torrent)
        Airota(trans_client, torrent)
main()

Plex 去重

之后，会有部分剧集多次重复下载，这里使用 plex dupefinder 来去重。
它是通过 文件名，音频编码格式，视频编码格式，分辨率来给资源算分，自动算出最高的进行保留。

无脑安装命令

git clone https://github.com/l3uddz/plex_dupefinder.git
cd plex_dupefinder
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

再创建 plex_dupefinder.sh 作为启动入口

#! /bin/bash
BASE_PATH=`dirname $0`
source $BASE_PATH/venv/bin/activate
python $BASE_PATH/plex_dupefinder.py

由于我的播放器支持 hevc 编码和我喜欢mkv格式，所以微调了下config。一般用默认即可。
FILENAME_SCORES 也能主动给一些喜欢的字幕组加分。
config.json

{
  "AUDIO_CODEC_SCORES": {
    "Unknown": 0,
    "aac": 1000,
    "ac3": 1000,
    "dca": 2000,
    "dca-ma": 4000,
    "eac3": 1250,
    "flac": 2500,
    "mp2": 500,
    "mp3": 1000,
    "pcm": 2500,
    "truehd": 4500,
    "wmapro": 200
  },
  "AUTO_DELETE": true,
  "FIND_DUPLICATE_FILEPATHS_ONLY": false,
  "FILENAME_SCORES": {
    "*.avi": -1000,
    "*.ts": -1000,
    "*.vob": -5000,
    "*.mp4": -500,
    "*.mkv": 500,
    "*1080p*BluRay*": 15000,
    "*720p*BluRay*": 10000,
    "*HDTV*": -1000,
    "*PROPER*": 1500,
    "*REPACK*": 1500,
    "*Remux*": 20000,
    "*WEB*CasStudio*": 5000,
    "*WEB*KINGS*": 5000,
    "*WEB*NTB*": 5000,
    "*WEB*QOQ*": 5000,
    "*WEB*SiGMA*": 5000,
    "*WEB*TBS*": -1000,
    "*WEB*TROLLHD*": 2500,
    "*WEB*VISUM*": 5000,
    "*dvd*": -1000
  },
  "PLEX_LIBRARIES": [
    "Bangumi"
  ],
  "PLEX_SERVER": "https://your.plex/",
  "PLEX_TOKEN": "",
  "SCORE_FILESIZE": true,
  "SKIP_LIST": [],
  "VIDEO_CODEC_SCORES": {
    "Unknown": 0,
    "h264": 8000,
    "h265": 10000,
    "hevc": 10000,
    "mpeg1video": 250,
    "mpeg2video": 250,
    "mpeg4": 500,
    "msmpeg4": 100,
    "msmpeg4v2": 100,
    "msmpeg4v3": 100,
    "vc1": 3000,
    "vp9": 1000,
    "wmv2": 250,
    "wmv3": 250
  },
  "VIDEO_RESOLUTION_SCORES": {
    "1080": 10000,
    "480": 3000,
    "4k": 20000,
    "720": 5000,
    "Unknown": 0,
    "sd": 1000
  }
}

统一启动 + 定时任务

这里需要再 flexget 执行后，需要跑一些脚本，不用 flexget 的 schedule，用 crontab 触发

run.sh

# ! /bin/bash
find /mnt/WEB-DL/ -type d -empty  # 自动清空目录为空的文件夹
source /flexget/bin/activate
cd /config/flexget  # 配置在这里
flexget --cron execute
python flexget-auto.py > flexget-auto.log
/plex_dupefinder/plex_dupefinder.sh > plex_dupefinder.log

再加入 crontab 就搞定啦~

*/10 * * * * "/flexget/run.sh"