【続編】「動いてたbot」が突然死んだ ── rss.app終了から、より良い構成に作り直すまでの全記録

name: MLBear2 AIニュース → Chatwork

on:
  schedule:
    - cron: '17 23 * * *'   # JST 8:00
    - cron: '17 11 * * *'   # JST 20:00
  workflow_dispatch:

permissions:
  contents: write          # last_guid.txt をコミットするのに必要

jobs:
  run:
    runs-on: ubuntu-latest
    steps:
    - uses: actions/checkout@v4

    - name: Python セットアップ
      uses: actions/setup-python@v5
      with:
        python-version: '3.11'

    - name: 依存インストール
      run: pip install feedparser requests

    - name: スクリプト実行
      env:
        CHATWORK_TOKEN: ${{ secrets.CHATWORK_TOKEN }}
        RSS_URL: ${{ secrets.RSS_URL }}
      run: python ml_bear_to_chatwork.py

    - name: 送信済みGUIDを保存
      run: |
        git config user.name "github-actions[bot]"
        git config user.email "github-actions[bot]@users.noreply.github.com"
        git add last_guid.txt
        git diff --staged --quiet || git commit -m "update last_guid [skip ci]"
        git push

import os
import re
import html
import requests
import feedparser
from html.parser import HTMLParser

# ================== 設定 ==================
CHATWORK_TOKEN = os.getenv("CHATWORK_TOKEN")
ROOM_ID = os.getenv("ROOM_ID", "********") # ←チャットワークROOMIDを入れてください
RSS_URL = os.getenv("RSS_URL")
LAST_GUID_FILE = "last_guid.txt"

# タイトルにこのいずれかが含まれていれば「まとめ投稿」として扱う
KEYWORDS = ["ニュースまとめ", "aiニュース"]

# リードとして採用しない定型文（ナビ・登録フォーム等）。部分一致で除外。
LEAD_NG = ["Home", "Archive", "About", "毎日AIニュース", "ご登録", "Subscribe", "Sign up"]


def get_last_guid():
    try:
        with open(LAST_GUID_FILE, encoding="utf-8") as f:
            return f.read().strip()
    except FileNotFoundError:
        return None


def save_last_guid(guid):
    with open(LAST_GUID_FILE, "w", encoding="utf-8") as f:
        f.write(guid)


def is_matome(entry):
    title = entry.title.lower()
    return any(k in title for k in KEYWORDS)


# ================== 記事ページから本文抽出 ==================
class _ArticleParser(HTMLParser):
    """記事ページ HTML から リード段落(<p>) と 目次(<ol><li>) を抽出する。"""

    def __init__(self):
        super().__init__()
        self.in_p = False
        self.in_li = False
        self.in_ol = False
        self.cur = ""
        self.paras = []
        self.toc_items = []
        self.seen_toc_heading = False
        self._in_heading = False
        self._heading_text = ""
        self._toc_capturing = False

    def handle_starttag(self, tag, attrs):
        if tag == "p":
            self.in_p = True
            self.cur = ""
        elif tag == "ol":
            self.in_ol = True
            if self.seen_toc_heading and not self.toc_items:
                self._toc_capturing = True
        elif tag == "li":
            self.in_li = True
            self.cur = ""
        elif tag in ("h1", "h2", "h3", "h4"):
            self._in_heading = True
            self._heading_text = ""

    def handle_endtag(self, tag):
        if tag == "p" and self.in_p:
            self.in_p = False
            text = self.cur.strip()
            if text:
                self.paras.append(text)
        elif tag == "ol":
            self.in_ol = False
            self._toc_capturing = False
        elif tag == "li" and self.in_li:
            self.in_li = False
            text = self.cur.strip()
            if text and self._toc_capturing:
                self.toc_items.append(text)
        elif tag in ("h1", "h2", "h3", "h4"):
            self._in_heading = False
            if "目次" in self._heading_text:
                self.seen_toc_heading = True

    def handle_data(self, data):
        if self.in_p or self.in_li:
            self.cur += data
        if self._in_heading:
            self._heading_text += data


def _is_ng_lead(text):
    return any(ng in text for ng in LEAD_NG)


def extract_from_article(article_url):
    """記事ページを取得し (lead[:3], toc[:10]) を返す。失敗時は ([], [])。"""
    try:
        r = requests.get(article_url, timeout=20,
                         headers={"User-Agent": "Mozilla/5.0 (ML_Bear bot)"})
        r.raise_for_status()
    except Exception as e:
        print(f"⚠️ 記事ページ取得失敗: {e}")
        return [], []

    parser = _ArticleParser()
    try:
        parser.feed(r.text)
    except Exception:
        pass

    lead = []
    for p in parser.paras:
        t = html.unescape(p).strip()
        if not t or _is_ng_lead(t):
            continue
        lead.append(t)
        if len(lead) >= 3:
            break

    toc = [html.unescape(t).strip() for t in parser.toc_items[:10]]
    return lead, toc


def _fallback_clean_text(raw_html):
    text = re.sub(r"<script.*?</script>", "", raw_html, flags=re.DOTALL | re.IGNORECASE)
    text = re.sub(r"<br\s*/?>", "\n", text, flags=re.IGNORECASE)
    text = re.sub(r"</p\s*>", "\n", text, flags=re.IGNORECASE)
    text = re.sub(r"<[^>]+>", "", text)
    text = html.unescape(text)
    text = re.sub(r"https?://\S+", "", text)
    text = re.sub(r"[ \t\u3000]+", " ", text)
    text = re.sub(r"\n{3,}", "\n\n", text)
    return "\n".join(ln.rstrip() for ln in text.splitlines()).strip()


def format_for_chatwork(entry):
    title = entry.get("title", "AIニュースまとめ").strip()
    link = entry.get("link", "")

    lead, toc = ([], [])
    if link:
        lead, toc = extract_from_article(link)

    parts = []
    if lead:
        parts.append("\n".join(lead))

    if toc:
        numbered = []
        for i, item in enumerate(toc, 1):
            item = item.strip()
            head = item.split(".", 1)[0].strip()
            if head.isdigit():
                numbered.append(item)
            else:
                numbered.append(f"{i}. {item}")
        parts.append("【目次】\n" + "\n".join(numbered))

    if parts:
        body = "\n\n".join(parts)
    else:
        raw = entry.get("description") or entry.get("summary") or ""
        body = _fallback_clean_text(raw) or "(本文を取得できませんでした)"

    MAX_BODY = 4000
    if len(body) > MAX_BODY:
        body = body[:MAX_BODY].rstrip() + "\n…(以下略)"

    link_line = f"\n\n▼ 全文はこちら\n{link}" if link else ""
    return f"[info][title]{title}[/title]\n{body}{link_line}[/info]"


# ================== 送信 ==================
def send_to_chatwork(message):
    url = f"https://api.chatwork.com/v2/rooms/{ROOM_ID}/messages"
    headers = {"X-ChatWorkToken": CHATWORK_TOKEN}
    r = requests.post(url, headers=headers, data={"body": message})
    if r.status_code == 200:
        print("✅ 送信成功")
        return True
    print(f"❌ 送信失敗 {r.status_code} - {r.text[:300]}")
    return False


def main():
    if not CHATWORK_TOKEN or not RSS_URL:
        print("⚠️ CHATWORK_TOKEN または RSS_URL が未設定です")
        return

    print("🔍 RSS取得開始")
    feed = feedparser.parse(RSS_URL)
    if not feed.entries:
        print("⚠️ RSSエントリがありません（フィードURLを確認してください）")
        return

    matched = [e for e in feed.entries if is_matome(e)]
    if not matched:
        print("該当するニュースまとめ投稿が見つかりませんでした")
        return

    last_guid = get_last_guid()

    if last_guid is None:
        new_entries = matched[:1]
    else:
        new_entries = []
        for e in matched:
            if e.guid == last_guid:
                break
            new_entries.append(e)

    if not new_entries:
        print("新しいまとめはありません（送信済み）")
        return

    for e in reversed(new_entries):
        print(f"📰 送信: {e.title[:50]} ...")
        if send_to_chatwork(format_for_chatwork(e)):
            save_last_guid(e.guid)


if __name__ == "__main__":
    main()