agentic_media_pipeline.py 11.8 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432
#!/usr/bin/env python3
"""
agentic_media_pipeline.py - 完整的 Agentic Media 工作流

4-Agent Pipeline:
1. News Collector: 从微信群收集消息
2. News Editor: 用 Kimi 2.6 编辑内容,生成结构化日报
3. Layout Planner: 规划报纸版面布局
4. GPT Image 2: 生成维多利亚报纸风格信息图

Usage:
    python3 agentic_media_pipeline.py --group "8-TG0和朋友们的AGI" --output /sdcard/WeChatDigest/
"""

import argparse
import json
import os
import sys
import re
import xml.etree.ElementTree as ET
from datetime import datetime, timezone, timedelta

# Add skill paths
sys.path.insert(0, os.path.expanduser('~/.hermes/skills/research/kimi-api/scripts'))
sys.path.insert(0, os.path.expanduser('~/.hermes/skills/image-generation/scripts'))

from kimi_client import simple_chat
from generate_image import generate_image


def parse_args():
    parser = argparse.ArgumentParser(description='Agentic Media Pipeline - 4-Agent Daily Digest')
    parser.add_argument('--group', required=True, help='WeChat group name')
    parser.add_argument('--output', default='/sdcard/WeChatDigest/', help='Output directory')
    parser.add_argument('--date', help='Date (YYYYMMDD), default today')
    parser.add_argument('--skip-image', action='store_true', help='Skip image generation')
    return parser.parse_args()


def get_today_timestamp():
    """获取今天零点的时间戳(毫秒)"""
    beijing = timezone(timedelta(hours=8))
    now = datetime.now(beijing)
    today_start = now.replace(hour=0, minute=0, second=0, microsecond=0)
    return int(today_start.timestamp() * 1000)


def query_wechat_messages(group_name, date_str=None):
    """Agent 1: News Collector - 从微信群收集消息"""
    print("\n=== Agent 1: News Collector ===")
    
    # 查询群聊ID
    import subprocess
    result = subprocess.run([
        'sqlite3', os.path.expanduser('~/EnMicroMsg.db'),
        '-line',
        f"SELECT username, nickname FROM rcontact WHERE username LIKE '%chatroom%' AND nickname LIKE '%{group_name}%';"
    ], capture_output=True, text=True)
    
    # 解析结果
    chatroom_id = None
    for line in result.stdout.split('\n'):
        if 'username =' in line:
            chatroom_id = line.split('=')[1].strip()
        if 'nickname =' in line and group_name in line:
            break
    
    if not chatroom_id:
        print(f"❌ Group '{group_name}' not found")
        return []
    
    print(f"✅ Found group: {chatroom_id}")
    
    # 查询今天的消息
    today_ts = get_today_timestamp()
    
    result = subprocess.run([
        'sqlite3', os.path.expanduser('~/EnMicroMsg.db'),
        '-line',
        f"SELECT msgId, type, isSend, talker, content, createTime FROM message WHERE talker = '{chatroom_id}' AND createTime >= {today_ts} ORDER BY createTime DESC LIMIT 100;"
    ], capture_output=True, text=True)
    
    messages = []
    current_msg = {}
    
    for line in result.stdout.split('\n'):
        line = line.strip()
        if not line:
            if current_msg:
                messages.append(current_msg)
                current_msg = {}
            continue
        
        if '=' in line:
            key, value = line.split('=', 1)
            key = key.strip()
            value = value.strip()
            current_msg[key] = value
    
    if current_msg:
        messages.append(current_msg)
    
    print(f"✅ Collected {len(messages)} messages")
    return messages


def extract_articles(messages):
    """从消息中提取公众号文章"""
    articles = []
    discussions = []
    
    for msg in messages:
        msg_type = msg.get('type', '0')
        content = msg.get('content', '')
        
        if msg_type == '49':  # 公众号文章
            try:
                root = ET.fromstring(content)
                title = root.find('.//title')
                url = root.find('.//url')
                desc = root.find('.//des')
                
                if title is not None and url is not None:
                    articles.append({
                        'title': title.text or '',
                        'url': url.text or '',
                        'description': desc.text if desc is not None else '',
                        'source': 'WeChat'
                    })
            except:
                pass
        elif msg_type == '1':  # 文本消息
            if content and len(content) > 10:
                discussions.append({
                    'sender': msg.get('talker', 'Unknown'),
                    'content': content
                })
    
    return articles, discussions


def agent_news_editor(articles, discussions):
    """Agent 2: News Editor - 用 Kimi 2.6 编辑内容"""
    print("\n=== Agent 2: News Editor (Kimi 2.6) ===")
    
    # 准备输入内容
    articles_text = "\n\n".join([
        f"标题: {a['title']}\n来源: {a['source']}\n链接: {a['url']}\n描述: {a['description']}"
        for a in articles[:20]  # 限制数量
    ])
    
    discussions_text = "\n".join([
        f"{d['sender']}: {d['content'][:100]}"
        for d in discussions[:50]
    ])
    
    # Prompt V1: 提取结构化信息
    prompt_v1 = f"""你是一名《AI时代日报》的主编。

请阅读以下内容:

【公众号文章】
{articles_text}

【群聊讨论】
{discussions_text}

任务:

从所有内容中提取:

1. 今日最重要的4条新闻
2. 今日最有价值的5条观点
3. 今日最重要的人物
4. 今日最重要的项目
5. 今日最重要的趋势

要求:

不要简单摘要。

要按照科技日报编辑部的方式:

- 去重
- 聚类
- 提炼主题
- 总结行业意义

输出JSON:

{{
  "date":"{datetime.now().strftime('%Y-%m-%d')}",

  "headline":"一句话总标题",

  "hero_story":{{
      "title":"主标题",
      "summary":"100字摘要"
  }},

  "news":[
      {{
        "title":"",
        "summary":"",
        "source":""
      }}
  ],

  "quotes":[
      {{
         "author":"",
         "quote":""
      }}
  ],

  "trend":{{
      "title":"",
      "summary":""
  }}
}}"""
    
    # 调用 Kimi 2.6
    result = simple_chat(prompt_v1, timeout=120)
    
    # 提取 JSON
    try:
        json_match = re.search(r'\{.*\}', result, re.DOTALL)
        if json_match:
            daily_json = json.loads(json_match.group())
        else:
            daily_json = json.loads(result)
    except:
        print(f"⚠️ JSON parse failed, using raw text")
        daily_json = {
            "date": datetime.now().strftime('%Y-%m-%d'),
            "headline": "AI行业日报",
            "news": [],
            "quotes": [],
            "trend": {"title": "", "summary": ""}
        }
    
    print(f"✅ Generated structured report with {len(daily_json.get('news', []))} news items")
    
    # Prompt V2: 深度编辑
    prompt_v2 = f"""你是一名拥有30年经验的科技媒体总编辑。

请根据以下内容:

{json.dumps(daily_json, ensure_ascii=False, indent=2)}

完成以下工作:

1. 找出今天最大的行业主线
2. 生成报纸头版标题
3. 生成副标题
4. 给每条新闻重新命名
5. 给每条新闻写50字导语
6. 提炼5条最有洞察的观点

要求:

风格参考:

- 华尔街日报
- The Economist
- Wired
- MIT Technology Review

输出Markdown格式日报。
"""
    
    markdown_report = simple_chat(prompt_v2, timeout=120)
    
    print(f"✅ Generated Markdown report ({len(markdown_report)} chars)")
    
    return daily_json, markdown_report


def agent_layout_planner(daily_json):
    """Agent 3: Layout Planner - 规划报纸版面"""
    print("\n=== Agent 3: Layout Planner ===")
    
    layout = {
        "newspaper_name": "TGO和朋友们的AGI",
        "headline": daily_json.get("headline", "AI行业日报"),
        "hero_image": {
            "concept": "server rack shaking hands with printing press"
        },
        "columns": [],
        "quotes": daily_json.get("quotes", [])
    }
    
    # 为每条新闻创建版面
    for news in daily_json.get("news", []):
        layout["columns"].append({
            "title": news.get("title", ""),
            "content": news.get("summary", "")[:100]
        })
    
    print(f"✅ Planned layout with {len(layout['columns'])} columns")
    
    return layout


def agent_gpt_image2(layout, output_path):
    """Agent 4: GPT Image 2 - 生成维多利亚报纸风格信息图"""
    print("\n=== Agent 4: GPT Image 2 (Victorian Newspaper) ===")
    
    # 构建提示词
    headline = layout["headline"]
    columns_text = "\n".join([
        f"{i+1}. {col['title']}: {col['content'][:50]}"
        for i, col in enumerate(layout["columns"][:6])
    ])
    
    prompt = f"""Create a front-page Victorian newspaper poster.

Style:
1880 Victorian newspaper
19th century financial times
The Economist cover
engraving illustration
woodcut etching
vintage printing press
old yellow paper
aged newspaper texture
ink engraving
black and white line art

Layout:
large newspaper masthead "AI AGI 日报" on top
main headline: "{headline}"
multiple newspaper columns with Chinese text
{columns_text}

Main Illustration:
server rack shaking hands with printing press
engraving style
US dollar bill engraving
extremely detailed line work
woodcut illustration
banknote etching
centerpiece illustration

Mood:
historical
authoritative
intellectual
industrial revolution

Color:
sepia paper
black ink only

Aspect ratio: 1:1"""
    
    # 生成图片
    result = generate_image(prompt, model="GPT Image 2")
    
    if result.get("image_path") and os.path.exists(result["image_path"]):
        import shutil
        shutil.copy(result["image_path"], output_path)
        print(f"✅ Generated Victorian newspaper: {output_path}")
        return output_path
    else:
        print(f"❌ Image generation failed")
        return None


def main():
    args = parse_args()
    
    # 确保输出目录存在
    os.makedirs(args.output, exist_ok=True)
    
    date_str = args.date or datetime.now().strftime('%Y%m%d')
    
    # Agent 1: News Collector
    messages = query_wechat_messages(args.group, date_str)
    
    if not messages:
        print("❌ No messages found")
        return
    
    # 提取文章和讨论
    articles, discussions = extract_articles(messages)
    print(f"✅ Extracted {len(articles)} articles, {len(discussions)} discussions")
    
    # Agent 2: News Editor
    daily_json, markdown_report = agent_news_editor(articles, discussions)
    
    # 保存 Markdown 报告
    report_path = os.path.join(args.output, f"report_{date_str}.md")
    with open(report_path, 'w', encoding='utf-8') as f:
        f.write(markdown_report)
    print(f"✅ Saved report: {report_path}")
    
    # 保存 JSON
    json_path = os.path.join(args.output, f"report_{date_str}.json")
    with open(json_path, 'w', encoding='utf-8') as f:
        json.dump(daily_json, f, ensure_ascii=False, indent=2)
    print(f"✅ Saved JSON: {json_path}")
    
    # Agent 3: Layout Planner
    layout = agent_layout_planner(daily_json)
    
    # 保存 Layout JSON
    layout_path = os.path.join(args.output, f"layout_{date_str}.json")
    with open(layout_path, 'w', encoding='utf-8') as f:
        json.dump(layout, f, ensure_ascii=False, indent=2)
    print(f"✅ Saved layout: {layout_path}")
    
    # Agent 4: GPT Image 2
    if not args.skip_image:
        image_path = os.path.join(args.output, f"infographic_{date_str}.png")
        image_result = agent_gpt_image2(layout, image_path)
        
        if image_result:
            print(f"\n🎉 Complete! Files generated:")
            print(f"  - Report: {report_path}")
            print(f"  - JSON: {json_path}")
            print(f"  - Layout: {layout_path}")
            print(f"  - Image: {image_result}")
    else:
        print(f"\n✅ Pipeline complete (image generation skipped)")
        print(f"  - Report: {report_path}")
        print(f"  - JSON: {json_path}")
        print(f"  - Layout: {layout_path}")


if __name__ == '__main__':
    main()