watcher.py

import os
import re
import time
import json
import requests
from datetime import datetime, timezone
from bs4 import BeautifulSoup
import ffmpeg
from tflink import TFLinkClient

# ===== Channel.io 設定 =====
GET_URL = "https://desk-api.channel.io/desk/channels/200605/groups/519217/messages"
POST_URL = GET_URL

PARAMS = {
    "sortOrder": "desc",
    "limit": 36,
    "logFolded": "false",
}

X_ACCOUNT = os.getenv("channeliotokenbot2")
if not X_ACCOUNT:
    raise RuntimeError("環境変数 channeliotokenbot2 が設定されていません")

HEADERS_GET = {
    "accept": "application/json",
    "accept-language": "ja",
    "x-account": X_ACCOUNT,
}

HEADERS_POST = {
    "accept": "application/json",
    "accept-language": "ja",
    "content-type": "application/json",
    "x-account": X_ACCOUNT,
}

# ===== ssyoutube =====
SSYOUTUBE_URL = "https://ssyoutube.online/yt-video-detail/"

# ===== Utils =====
def parse_updated_at(value):
    if isinstance(value, (int, float)):
        return datetime.fromtimestamp(value / 1000, tz=timezone.utc)
    elif isinstance(value, str):
        return datetime.fromisoformat(value.replace("Z", "+00:00"))
    return None

def extract_youtube_id(text):
    patterns = [
        r"v=([A-Za-z0-9_-]{11})",
        r"youtu\.be/([A-Za-z0-9_-]{11})",
    ]
    for p in patterns:
        m = re.search(p, text)
        if m:
            return m.group(1)
    return None

# ===== ssyoutube HTML 解析 =====
def fetch_download_links(youtube_url):
    res = requests.post(
        SSYOUTUBE_URL,
        data={"videoURL": youtube_url},
        timeout=30,
        headers={
            "User-Agent": "Mozilla/5.0",
            "Referer": "https://ssyoutube.online/",
        }
    )
    res.raise_for_status()

    soup = BeautifulSoup(res.text, "lxml")
    buttons = soup.select("button[data-url]")

    results = []
    for btn in buttons:
        url = btn.get("data-url")
        quality = btn.get("data-quality")  # 例: 1080p / 720p / None
        has_audio = btn.get("data-has-audio")  # "true" / "false" / None

        if not url:
            continue

        results.append({
            "url": url,
            "quality": quality or "audio",
            "has_audio": has_audio,
        })

    return results

def download_file(url, filename):
    with requests.get(url, stream=True) as r:
        r.raise_for_status()
        with open(filename, 'wb') as f:
            for chunk in r.iter_content(chunk_size=8192):
                f.write(chunk)
    return filename

def merge_video_audio(video_file, audio_file, output_file):
    (
        ffmpeg
        .input(video_file)
        .output(audio_file, output_file, vcodec='copy', acodec='aac', strict='experimental')
        .run(overwrite_output=True)
    )
    return output_file

def upload_to_tflink(file_path):
    client = TFLinkClient()
    result = client.upload(file_path)
    return result.download_link

def send_to_channel(text):
    payload = {
        "requestId": f"desk-web-{int(time.time() * 1000)}",
        "blocks": [
            {
                "type": "text",
                "value": text
            }
        ],
        "buttons": None,
        "form": None,
        "webPage": None,
        "files": None,
        "customPayload": None
    }

    res = requests.post(
        POST_URL,
        headers=HEADERS_POST,
        data=json.dumps(payload),
        timeout=30
    )
    res.raise_for_status()

def main():
    while True:
        try:
            res = requests.get(
                GET_URL,
                headers=HEADERS_GET,
                params=PARAMS,
                timeout=30,
            )
            res.raise_for_status()

            messages = res.json().get("messages", [])
            latest_msg = None
            latest_time = None

            for msg in messages:
                plain_text = msg.get("plainText")
                updated_at = msg.get("updatedAt")

                if not plain_text or updated_at is None:
                    continue

                t = parse_updated_at(updated_at)
                if not t:
                    continue

                if latest_time is None or t > latest_time:
                    latest_time = t
                    latest_msg = msg

            if not latest_msg:
                time.sleep(10)
                continue

            text = latest_msg["plainText"]
            youtube_id = extract_youtube_id(text)
            if not youtube_id:
                time.sleep(10)
                continue

            youtube_url = f"https://www.youtube.com/watch?v={youtube_id}"

            items = fetch_download_links(youtube_url)
            if not items:
                print("ダウンロードリンクが取得できませんでした")
                time.sleep(10)
                continue

            # ===== 最も高画質の動画と音声を選択 =====
            video_items = [i for i in items if i["has_audio"] == "false"]
            audio_items = [i for i in items if i["has_audio"] == "true" and i["quality"] == "audio"]

            if not video_items or not audio_items:
                print("動画または音声が見つかりません")
                time.sleep(10)
                continue

            # 解像度順にソートして最初を取得
            video_items.sort(key=lambda x: int(re.sub(r"\D", "", x["quality"] or "0")), reverse=True)
            video_url = video_items[0]["url"]
            audio_url = audio_items[0]["url"]

            video_file = download_file(video_url, "video.mp4")
            audio_file = download_file(audio_url, "audio.m4a")
            output_file = "merged.mp4"

            merge_video_audio(video_file, audio_file, output_file)

            # tflink にアップロード
            tflink_url = upload_to_tflink(output_file)

            message_text = f"結合動画のダウンロード: {tflink_url}"
            send_to_channel(message_text)
            print("送信完了:", tflink_url)

        except Exception as e:
            print("エラー:", e)

        time.sleep(15)

if __name__ == "__main__":
    main()