自動ニュース要約ボット | オリジナル、AI翻訳
この投稿では、Hacker News、GitHub Trending、NYTimes（中国語版）からトップニュースをスクレイピングし、Mistral APIを使用して要約するPythonベースのニュースボットを紹介します。Telegramを介して簡潔な日次レポートを送信し、自動実行のためにGitHub Actionsワークフローを利用しています。技術や世界のニュースを手軽に把握するのに最適です。
import requests
from bs4 import BeautifulSoup
import os
from dotenv import load_dotenv
import datetime
import sys
import re
import time
load_dotenv()
TELEGRAM_BOT_TOKEN = os.environ.get("TELEGRAM_BOT_API_KEY")
TELEGRAM_CHAT_ID = os.environ.get("TELEGRAM_CHAT_ID", "610574272")
MISTRAL_API_KEY = os.environ.get("MISTRAL_API_KEY")
TELEGRAM_MAX_LENGTH = 4096
def send_telegram_message(message):
if not TELEGRAM_BOT_TOKEN or not TELEGRAM_CHAT_ID:
print("Error: TELEGRAM_BOT_API_KEY or TELEGRAM_CHAT_ID not set.")
return False
url = f"https://api.telegram.org/bot{TELEGRAM_BOT_TOKEN}/sendMessage"
url_pattern = re.compile(r'(https?://[^\s]+)')
# メッセージからすべてのアスタリスク（太字/斜体用）を削除
message_no_stars = message.replace('*', '')
# メッセージからリンクを削除
message_no_links = url_pattern.sub('', message_no_stars)
messages = []
msg = message_no_links
while len(msg) > TELEGRAM_MAX_LENGTH:
split_idx = msg.rfind('\n', 0, TELEGRAM_MAX_LENGTH)
if split_idx == -1 or split_idx < TELEGRAM_MAX_LENGTH // 2:
split_idx = TELEGRAM_MAX_LENGTH
messages.append(msg[:split_idx])
msg = msg[split_idx:]
messages.append(msg)
success = True
for part in messages:
params = {
"chat_id": TELEGRAM_CHAT_ID,
"text": part,
}
try:
response = requests.post(url, params=params)
response.raise_for_status()
print(f"Successfully sent Telegram message part ({len(part)} chars).")
except requests.exceptions.RequestException as e:
print(f"Error sending Telegram message: {e}")
success = False
return success
def fetch_html_content(url):
try:
print(f"Fetching HTML content from: {url}")
response = requests.get(url, timeout=15, verify=False)
response.raise_for_status()
print(f"Successfully fetched HTML content from: {url}")
return response.text
except requests.exceptions.RequestException as e:
print(f"Could not fetch URL: {url} - {e}")
return None
def extract_hacker_news_links(html, max_links=5):
soup = BeautifulSoup(html, 'html.parser')
links = []
seen = set()
for item in soup.select('.titleline > a'):
url = item['href']
title = item.text.strip()
if url.startswith('item?id='):
url = f"https://news.ycombinator.com/{url}"
if url not in seen and title:
links.append({'url': url, 'text': title})
seen.add(url)
if len(links) >= max_links:
break
print(f"Extracted {len(links)} links from Hacker News.")
return links
def extract_github_trending(html, max_links=5):
soup = BeautifulSoup(html, 'html.parser')
links = []
for repo in soup.select('article.Box-row h2 a'):
url = f"https://github.com{repo['href']}"
title = re.sub(r'\s+', ' ', repo.text).strip()
if title and url:
links.append({'url': url, 'text': title})
if len(links) >= max_links:
break
print(f"Extracted {len(links)} trending repositories from GitHub.")
return links
def call_mistral_api(prompt, model="mistral-small-latest"):
api_key = MISTRAL_API_KEY
if not api_key:
print("Error: MISTRAL_API_KEY environment variable not set.")
return None
url = "https://api.mistral.ai/v1/chat/completions"
headers = {
"Content-Type": "application/json",
"Accept": "application/json",
"Authorization": f"Bearer {api_key}"
}
data = {
"model": model,
"messages": [
{
"role": "user",
"content": prompt
}
]
}
try:
print(f"Calling Mistral API with model: {model}")
print(f"Prompt being sent: {prompt[:1000]}...")
response = requests.post(url, headers=headers, json=data)
response.raise_for_status()
response_json = response.json()
print(f"Mistral API Response: {response_json}")
if response_json and response_json.get('choices'):
content = response_json['choices'][0]['message']['content']
print(f"Mistral API Content: {content}")
return content
else:
print(f"Mistral API Error: Invalid response format: {response_json}")
return None
except requests.exceptions.RequestException as e:
print(f"Mistral API Error: {e}")
if hasattr(e, "response") and e.response is not None:
print(f"Response status code: {e.response.status_code}")
print(f"Response content: {e.response.text}")
return None
def fetch_and_summarize(url, fallback_title=None):
print(f"Summarizing: {url}")
html = fetch_html_content(url)
if not html:
return {"url": url, "summary": "Could not fetch content.", "title": fallback_title or url}
soup = BeautifulSoup(html, 'html.parser')
title = soup.title.text.strip() if soup.title else (fallback_title or url)
paragraphs = soup.find_all('p')
text_content = "\n".join(p.get_text() for p in paragraphs)
if not text_content or len(text_content) < 100:
text_content = soup.get_text(separator="\n")
text_content = text_content.strip()
if len(text_content) > 3000:
text_content = text_content[:3000]
summary = ai_summarize(text_content, url, title)
return {"url": url, "summary": summary, "title": title}
def limit_to_n_words(text, n):
words = text.strip().split()
if len(words) <= n:
return text.strip()
return ' '.join(words[:n]) + "..."
def ai_summarize(text, url=None, title=None):
if not MISTRAL_API_KEY:
print("No MISTRAL_API_KEY set. Returning first 15 words as summary.")
return limit_to_n_words(text, 15)
prompt = (
"If the original text is in Chinese, summarize it in English. "
"Summarize the following web page content in clear, concise English. "
"Focus on the single most important point or insight. "
"Your summary should be around 300 characters. "
"Output only the summary sentence:\n"
f"Title: {title if title else ''}\n"
f"{text}\n"
f"{'Original link: ' + url if url else ''}"
)
summary = call_mistral_api(prompt)
if summary is None:
return limit_to_n_words(text, 15)
# 最終手段として300文字に切り詰め
return summary.strip()[:300]
def generate_summarized_report(summaries, source_name):
text = f"{source_name}\n"
text += "-" * len(source_name) + "\n"
if not summaries:
text += "No items found.\n\n"
return text
url_pattern = re.compile(r'(https?://[^\s]+)')
for idx, item in enumerate(summaries, 1):
summary = item.get('summary', '').replace('\n', ' ').replace('\r', ' ').strip()
summary = summary.replace('*', '')
summary = url_pattern.sub('', summary)
# 最終手段として各要約を300文字に切り詰め
summary = summary[:300]
text += f"{idx}. {summary}\n\n" # 要約間に余分な改行を追加
text += "\n"
return text
# --- NYTimes (m.cn.nytimes.com) integration ---
def extract_nytimes_links(html, max_links=5):
"""
cn.nytimes.comのメインページからリンクを抽出します。
'https://cn.nytimes.com/'で始まるリンクのみを含めます。
"""
soup = BeautifulSoup(html, 'html.parser')
links = []
for a in soup.find_all('a', href=True):
url = a['href']
if url.startswith('https://cn.nytimes.com/'):
links.append({
'url': url,
'text': a.text.strip()
})
if len(links) >= max_links:
break
print(f"Extracted {len(links)} links from main page.")
return links
def summarize_nytimes_article(url):
html = fetch_html_content(url)
if not html:
return {"url": url, "summary": "Could not fetch content.", "title": url}
soup = BeautifulSoup(html, 'html.parser')
# メイン記事のタイトルを抽出
title_element = soup.select_one('.article-area .article-content .article-header header h1')
title = title_element.text.strip() if title_element else (soup.title.text.strip() if soup.title else url)
# メイン記事のテキストを抽出
article_area = soup.find('section', class_='article-body')
if article_area:
article_text = article_area.get_text(separator='\n', strip=True)
else:
article_text = soup.get_text(separator='\n', strip=True)
if not article_text or len(article_text) < 100:
article_text = soup.get_text(separator='\n', strip=True)
if len(article_text) > 3000:
article_text = article_text[:3000]
summary = ai_summarize(article_text, url, title)
return {"url": url, "summary": summary, "title": title}
def main():
# --test引数をチェック
is_test = "--test" in sys.argv
today = datetime.datetime.now().strftime("%Y-%m-%d")
report = f"Daily News Summary - {today}\n\n"
if is_test:
# 1つのリンクのみスクレイピングし、1つの要約を送信（NYTimes中国語）
ny_html = fetch_html_content('https://m.cn.nytimes.com')
ny_links = []
ny_summaries = []
if ny_html:
ny_links = extract_nytimes_links(ny_html, max_links=1)
if ny_links:
link = ny_links[0]
summary = summarize_nytimes_article(link['url'])
ny_summaries.append(summary)
report = generate_summarized_report(ny_summaries, "NYTimes (Chinese)")
if ny_summaries:
if send_telegram_message(report):
print("Test summary sent to Telegram successfully.")
sys.exit(0)
else:
print("Failed to send test summary to Telegram.")
sys.exit(1)
else:
print("No news collected, nothing sent to Telegram.")
sys.exit(1)
else:
# --- Hacker News ---
hn_html = fetch_html_content('https://news.ycombinator.com')
hn_links = []
hn_summaries = []
if hn_html:
hn_links = extract_hacker_news_links(hn_html)
for link in hn_links:
summary = fetch_and_summarize(link['url'], fallback_title=link['text'])
hn_summaries.append(summary)
time.sleep(2)
report += generate_summarized_report(hn_summaries, "Hacker News")
# --- GitHub Trending ---
gh_html = fetch_html_content('https://github.com/trending')
gh_links = []
gh_summaries = []
if gh_html:
gh_links = extract_github_trending(gh_html)
for link in gh_links:
summary = fetch_and_summarize(link['url'], fallback_title=link['text'])
gh_summaries.append(summary)
time.sleep(2)
report += generate_summarized_report(gh_summaries, "GitHub Trending")
# --- NYTimes (cn.nytimes.com) ---
ny_html = fetch_html_content('https://m.cn.nytimes.com')
ny_links = []
ny_summaries = []
if ny_html:
ny_links = extract_nytimes_links(ny_html, max_links=5)
for link in ny_links:
summary = summarize_nytimes_article(link['url'])
ny_summaries.append(summary)
time.sleep(2)
report += generate_summarized_report(ny_summaries, "NYTimes (Chinese)")
if any([hn_summaries, gh_summaries, ny_summaries]):
if len(report) > TELEGRAM_MAX_LENGTH:
print(f"Report exceeds {TELEGRAM_MAX_LENGTH} chars, will be split into multiple messages.")
if send_telegram_message(report):
print("Daily news report sent to Telegram successfully.")
sys.exit(0)
else:
print("Failed to send daily news report to Telegram.")
sys.exit(1)
else:
print("No news collected, nothing sent to Telegram.")
sys.exit(1)
if __name__ == "__main__":
main()
name: News Bot
on:
schedule:
# 毎日北京時間午前9時（UTC午前1時）に実行
- cron: '0 1 * * *'
workflow_dispatch: # 手動トリガーを許可
push:
# 両方のファイルが同じコミット/プッシュで変更された場合のみトリガー
# これには以下のフィルタージョブで両方のファイルをチェックする必要がある
paths:
- scripts/nytimes/news_bot.py
- .github/workflows/news.yml
concurrency:
group: 'news'
cancel-in-progress: false
jobs:
send-news:
runs-on: ubuntu-latest
environment: github-pages
env:
TELEGRAM_BOT_API_KEY: $
MISTRAL_API_KEY: $
steps:
- name: Checkout repository
uses: actions/checkout@v4
with:
fetch-depth: 5
- name: Set up Python 3.10.x
uses: actions/setup-python@v4
with:
python-version: "3.10.x"
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install -r requirements.simple.txt
- name: Run news bot script
run: python scripts/nytimes/news_bot.py