{"id":2000,"date":"2025-03-21T07:22:00","date_gmt":"2025-03-20T23:22:00","guid":{"rendered":"https:\/\/blog.laoyulaoyu.top\/?p=2000"},"modified":"2025-03-02T22:33:20","modified_gmt":"2025-03-02T14:33:20","slug":"%e5%b0%8f%e7%99%bd%e5%ad%a6ai%e9%87%8f%e5%8c%96%ef%bc%9adeepseekpython%e6%9e%84%e5%bb%ba%e5%bc%ba%e5%a4%a7%e7%9a%84%e9%87%91%e8%9e%8d%e6%95%b0%e6%8d%ae%e6%8c%96%e6%8e%98%e4%b8%8e%e5%a4%9a%e7%bb%b4","status":"publish","type":"post","link":"https:\/\/www.laoyulaoyu.com\/index.php\/2025\/03\/21\/%e5%b0%8f%e7%99%bd%e5%ad%a6ai%e9%87%8f%e5%8c%96%ef%bc%9adeepseekpython%e6%9e%84%e5%bb%ba%e5%bc%ba%e5%a4%a7%e7%9a%84%e9%87%91%e8%9e%8d%e6%95%b0%e6%8d%ae%e6%8c%96%e6%8e%98%e4%b8%8e%e5%a4%9a%e7%bb%b4\/","title":{"rendered":"\u5c0f\u767d\u5b66AI\u91cf\u5316\uff1aDeepSeek+Python\u6784\u5efa\u5f3a\u5927\u7684\u91d1\u878d\u6570\u636e\u6316\u6398\u4e0e\u591a\u7ef4\u5206\u6790\u673a\u5668\u4eba"},"content":{"rendered":"\n

\u4f5c\u8005\uff1a\u8001\u4f59\u635e\u9c7c<\/a><\/p>\n\n\n\n

\u539f\u521b\u4e0d\u6613\uff0c\u8f6c\u8f7d\u8bf7\u6807\u660e\u51fa\u5904\u53ca\u539f\u4f5c\u8005\u3002<\/mark><\/strong><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\n
\u5199\u5728\u524d\u9762\u7684\u8bdd\uff1a<\/strong>\u5728\u673a\u6784\u4e3b\u5bfc\u7684\u91cf\u5316\u4ea4\u6613\u65f6\u4ee3\uff0c\u666e\u901a\u6295\u8d44\u8005\u5982\u4f55\u7528\u4e00\u676f\u5976\u8336\u7684\u94b1\uff0815\u5143\/\u5929\uff09\u6253\u9020\u4e13\u4e1a\u7ea7\u667a\u80fd\u91cf\u5316\u4ea7\u54c1\uff1f<\/mark> \u672c\u6587\u5c06\u4e3a\u60a8\u63ed\u79d8\u4e00\u4e2a\u9769\u547d\u6027\u7684\u89e3\u51b3\u65b9\u6848\u2014\u2014\u57fa\u4e8e\u56fd\u4ea7\u5927\u6a21\u578bDeepSeek\u548cPython\u6784\u5efa\u7684\u667a\u80fd\u6570\u636e\u6316\u6398\u5206\u6790\u673a\u5668\u4eba\u3002<\/mark>\u5b83\u4e0d\u4ec5\u9002\u7528\u4e8e\u901a\u7528\u7f51\u9875\u6570\u636e\u6293\u53d6\uff0c\u66f4\u80fd\u6df1\u5ea6\u5e94\u7528\u4e8e\u91d1\u878d\u9886\u57df\uff0c\u7cbe\u51c6\u6355\u6349\u5e02\u573a\u4fe1\u53f7\u3002\u672c\u6587\u201c\u5e72\u8d27\u201d\u5f88\u591a\uff0c\u8bf7\u52a1\u5fc5\u8010\u5fc3\u8bfb\u5b8c\u3002<\/pre>\n<\/blockquote>\n\n\n\n
\u4e00\u3001\u98a0\u8986\u8ba4\u77e5\u7684\u6027\u4ef7\u6bd4\u9769\u547d<\/strong><\/h2>\n\n\n\n

1.1 GPT-4\u7684\u6559\u8bad<\/strong><\/h3>\n\n\n\n
\u53bb\u5e74\u65e9\u4e9b\u65f6\u5019\u6211\u4eec\u5728\u4e00\u4e2a\u201c\u540c\u4e1a\u7ade\u4e89\u60c5\u62a5\u5206\u6790\u201d\u9879\u76ee\u4e2d\u4f7f\u7528OpenAI\u7684GPT-4\u673a\u5668\u4eba\u6765\u6316\u6398\u548c\u5206\u6790\u6570\u636e\u3002\u5f53\u65f6\u60f3\uff1a\u4e0d\u5c31\u662f20\u4e2a\u5b9a\u5411\u7f51\u7ad9\u6bcf5\u5206\u949f\u641c\u7d22\u5206\u6790\u4e00\u6b21\u5417\uff0c\u80fd\u6709\u591a\u8d35\uff1f\u4f46\u4ec5\u8fc7\u4e86\u4e09\u5929\u5c31\u6536\u5230\u4e86OpenAI\u53d1\u6765\u7684email\uff1a”\u60a8\u7684\u7528\u91cf\u5df2\u8d85\u8fc71000\u7f8e\u5143\uff0c\u8bf7\u66f4\u65b0\u60a8\u7684\u8ba1\u8d39\u9650\u989d\u201c\u3002\u4e3b\u8981\u539f\u56e0\u5728\u4e8e\u5bf9\u539f\u59cbHTML\u7684\u5168\u91cf\u5904\u7406\u4e0e\u65e0\u8282\u5236\u7684token\u6d88\u8017\u3002\u6838\u5fc3\u75db\u70b9\u5728\u4e8e\uff1a<\/p>\n\n\n\n
    \n
  • \u65e0\u5dee\u522b\u5904\u7406\u539f\u59cbHTML\u5bfc\u81f4token\u7206\u70b8\uff1a\u6bcf\u4e2a\u9875\u9762\u90fd\u67095000 \u884cHTML\uff0c\u5176\u4e2d\u5145\u65a5\u7740\u811a\u672c\u548c\u5e7f\u544a\uff0c\u8ba924 \u5c0f\u65f6\u5185\u6d88\u8017\u4e00\u5343\u4e07\u4e2a token \u2248 288 \u7f8e\u5143\/\u5929\u3002<\/li>\n\n\n\n
  • \u590d\u6742\u7684\u91d1\u878d\u672f\u8bed\u95ee\u9898\uff0c\u800c\u4e14GPT-4 \u5904\u7406\u4e86\u6574\u4e2a\u539f\u59cb HTML\uff0c\u5c0610MB\u7684\u5783\u573e\u53d8\u6210\u4e86\u4ee4\u724c\uff0c\u9020\u6210\u7406\u89e3\u504f\u5dee\u7387\u8fbe38%\u3002<\/li>\n\n\n\n
  • \u56e0\u4e3a\u56fd\u5185\u5916\u7f51\u7edc\u9700\u8981\u4ee3\u7406\u7684\u539f\u56e0\uff0c\u4f7f\u5b9e\u65f6\u6570\u636e\u5904\u7406\u5ef6\u8fdf\u8d8515\u5206\u949f\u3002<\/li>\n<\/ul>\n\n\n\n
    1.2 DeepSeek\u7684\u7834\u5c40\u4f18\u52bf<\/strong><\/h3>\n\n\n\n
    DeepSeek V3\/R1\u7cfb\u5217\u7684\u51fa\u73b0\u5f7b\u5e95\u6539\u53d8\u4e86\u8fd9\u4e00\u5c40\u9762\uff1a\u4e0d\u4ec5API\u5b9a\u4ef7\u4f4e\u5ec9\uff0c\u800c\u4e14\u914d\u5408\u9884\u5904\u7406\u6280\u672f\u53ef\u4ee5\u5927\u5e45\u964d\u4f4etoken\u6d88\u8017\u3002<\/p>\n\n\n\n
    \u6210\u672c\u76f4\u964d99%<\/strong>\uff1a\u8f93\u51fatoken\u5355\u4ef7\u4ec5\u4e3aGPT-4\u76841\/100\u3002<\/p>\n\n\n\n
    \u91d1\u878d\u8bed\u4e49\u7406\u89e3\u4e13\u5bb6<\/strong>\uff1a\u7ecf30\u4e07\u4efd\u8d22\u62a5\u5fae\u8c03\uff0c\u6307\u6807\u8bc6\u522b\u51c6\u786e\u7387\u8fbe92%\u3002<\/p>\n\n\n\n
    \u6beb\u79d2\u7ea7\u54cd\u5e94<\/strong>\uff1a\u652f\u6301websocket\u5b9e\u65f6\u6570\u636e\u6d41\u5904\u7406\u3002<\/p>\n\n\n\n
    \u5728\u6211\u76ee\u524d\u7684\u9879\u76ee\u4e2d\uff0cDeepSeek\u5c06HTML\u6807\u8bb0\u7684\u6570\u91cf\u51cf\u5c11\u5230\u6bcf\u9875200\u4e2a\uff0c\u800cOpenAI \u5219\u9700\u8981 1500\u4e2a\u3002\u5176\u4f18\u52bf\u6211\u8ba4\u4e3a\u4e3b\u8981\u4e3a\u4ee5\u4e0b\u4e09\u4e2a\u65b9\u9762\uff1a<\/p>\n\n\n\n
    \u4f4e\u6210\u672c\u9ad8\u6548<\/strong>\uff1a
    \u6bcf\u767e\u4e07\u8f93\u51fa tokens \u7684\u8d39\u7528\u4ec5\u7ea6 2 \u5143\uff0c\u800c OpenAI \u7684\u540c\u7c7b\u4ea7\u54c1\u6210\u672c\u8fdc\u9ad8\u4e8e\u6b64\u3002<\/p>\n\n\n\n
    \u4f18\u79c0\u7684\u4e2d\u6587\u5904\u7406\u80fd\u529b<\/strong>\uff1a
    \u9488\u5bf9\u4e2d\u6587\u6587\u672c\u548c\u6280\u672f\u95ee\u9898\uff0cDeepSeek V3 \u80fd\u591f\u7ed9\u51fa\u81ea\u7136\u3001\u63a5\u5730\u6c14\u7684\u56de\u7b54\uff0c\u9002\u5408\u91d1\u878d\u6570\u636e\u4e0e\u62a5\u544a\u5206\u6790\u3002<\/p>\n\n\n\n
    \u5f00\u6e90\u5f00\u653e<\/strong>\uff1a
    \u6a21\u578b\u91c7\u7528 MIT \u534f\u8bae\u5f00\u6e90\uff0c\u7528\u6237\u53ef\u81ea\u7531\u4fee\u6539\u3001\u5b9a\u5236\u751a\u81f3\u672c\u5730\u90e8\u7f72\uff0c\u89c4\u907f\u7ebf\u4e0a\u63a5\u53e3\u7684\u90e8\u5206\u9650\u5236\u3002<\/p>\n\n\n\n
    1.3 <\/strong> \u6574\u4f53\u7ed3\u6784<\/strong><\/h3>\n\n\n\n
    \u672c\u6587\u5c06\u8981\u4ecb\u7ecd\u7684\u673a\u5668\u4eba\uff0c\u65e2\u9002\u7528\u4e8e\u901a\u7528\u7f51\u9875\u6570\u636e\u6293\u53d6\uff0c\u4e5f\u80fd\u6269\u5c55\u5230\u91d1\u878d\u9886\u57df\uff0c\u5b83\u80fd\u81ea\u52a8\u91c7\u96c6\u80a1\u7968\u7684\u5404\u79cd\u76f8\u5173\u4fe1\u606f\uff0c\u5e76\u7ed3\u5408AI\u667a\u80fd\u5206\u6790\u7ed9\u51fa\u89c1\u89e3\u3002\u901a\u8fc7\u672c\u6587\uff0c\u60a8\u5c06\u83b7\u5f97\u6210\u672c\u63a7\u5236\u4e0e\u6548\u7387\u4f18\u5316\u7b56\u7565\u3001\u673a\u6784\u7ea7\u6570\u636e\u5206\u6790\u7684\u5b9e\u9645\u6848\u4f8b\u89e3\u6790\u3001\u9ad8\u53ef\u7528\u7684\u6280\u672f\u67b6\u6784\u8bbe\u8ba1\u3002\u63a5\u4e0b\u6765\u6211\u4eec\u6309\u7167\u987a\u5e8f\u5206\u56db\u4e2a\u90e8\u5206\u8fdb\u884c\u4ecb\u7ecd\uff1a<\/p>\n\n\n\n
    1. \u6570\u636e\u6293\u53d6\u4e0e\u9884\u5904\u7406<\/strong>\uff1a
    \u9488\u5bf9\u91d1\u878d\u7f51\u7ad9\uff08\u5982\u65b0\u6d6a\u8d22\u7ecf\u3001\u4e1c\u65b9\u8d22\u5bcc\u7f51\u3001\u96ea\u7403\u7b49\uff09\u6293\u53d6\u80a1\u7968\u5bf9\u5e94\u7684\u4fe1\u606f\uff0c\u5e76\u5c06\u4fe1\u606f\u4f20\u5165 DeepSeek\u8fdb\u884c\u6295\u8d44\u4ef7\u503c\u5206\u6790\u3002\u4e3b\u8981\u4f7f\u7528Python\u722c\u866b\uff08\u5982 Crawl4AI\u3001BeautifulSoup\uff09\u5bf9\u76ee\u6807\u7f51\u9875\u8fdb\u884c\u8bf7\u6c42\u4e0e\u8fc7\u6ee4\uff0c\u5254\u9664\u65e0\u5173\u4fe1\u606f\uff0c\u964d\u4f4etoken\u6d88\u8017\u3002<\/p>\n\n\n\n
    2. \u6df1\u5ea6\u667a\u80fd\u5206\u6790<\/strong>\uff1a
    \u5229\u7528DeepSeek API\u5bf9\u6293\u53d6\u5230\u7684\u5173\u952e\u6570\u636e\u8fdb\u884c\u667a\u80fd\u89e3\u6790\u4e0e\u8bc4\u8bba\uff0c\u751f\u6210\u7ed3\u6784\u5316\u5206\u6790\u7ed3\u679c\u3002<\/p>\n\n\n\n
    3. \u6316\u6398\u91d1\u878d\u884c\u4e1a<\/strong>\u5e94\u7528<\/strong>\uff1a
    \u5c06\u901a\u8fc7\u673a\u6784\u7ea7\u7684\u771f\u5b9e\u6848\u4f8b\u6765\u5f15\u5bfc\u5927\u5bb6\u601d\u8003\u5982\u4f55\u5229\u7528\u6316\u6398\u6765\u7684\u6570\u636e\uff0c\u5e76\u5c06\u5176\u843d\u5730\u53d8\u73b0\u3002<\/p>\n\n\n\n
    4. \u6280\u672f\u67b6\u6784\u89e3\u6790\uff08\u91d1\u878d\u7ea7\u9ad8\u53ef\u7528\u8bbe\u8ba1\uff09<\/strong>\uff1a
    \u8fd9\u4e2a\u67b6\u6784\u5df2\u5728\u67d0\u5934\u90e8\u5238\u5546\u81ea\u8425\u4ea4\u6613\u7cfb\u7edf\u7a33\u5b9a\u8fd0\u884c14\u4e2a\u6708\uff0c\u65e5\u5747\u5904\u740623\u4ebf\u6761\u91d1\u878d\u6570\u636e\uff0c\u5f02\u5e38\u6062\u590d\u65f6\u95f4\u7f29\u77ed\u81f34.3\u79d2\uff08\u884c\u4e1a\u5e73\u5747\u4e3a58\u79d2\uff09\uff0c\u5177\u5907\u91d1\u878d\u7ea7\u9ad8\u53ef\u9760\u6027\u4e0e\u519b\u5de5\u7ea7\u5b89\u5168\u6027\u3002<\/p>\n\n\n\n
    \u4e8c\u3001\u6570\u636e\u6293\u53d6\u4e0e\u9884\u5904\u7406<\/strong>
    <\/h2>\n\n\n\n
    2.1 \u901a\u7528\u7f51\u9875\u6570\u636e\u9884\u5904\u7406<\/strong><\/h3>\n\n\n\n
    \u4e3a\u4e86\u9ad8\u6548\u5229\u7528 DeepSeek \uff0c\u6211\u4eec\u9700\u8981\u5148\u5bf9\u539f\u59cbHTML\u6570\u636e\u8fdb\u884c\u8fc7\u6ee4\uff0c\u53bb\u6389\u811a\u672c\u3001\u6837\u5f0f\u3001\u5bfc\u822a\u548c\u9875\u811a\u7b49\u65e0\u5173\u5185\u5bb9\u3002\u793a\u4f8b\u4ee3\u7801\u5982\u4e0b\uff1a<\/p>\n\n\n\n
    from crawl4ai import Crawler\nfrom bs4 import BeautifulSoup\n\n# \u521d\u59cb\u5316\u722c\u866b\u5e76\u83b7\u53d6\u7f51\u9875\u5185\u5bb9\ncrawler = Crawler()\nresult = crawler.warm_up(url=\"https:\/\/example.com\")\n\n# \u8fc7\u6ee4\u6389\u811a\u672c\u3001\u6837\u5f0f\u3001\u5bfc\u822a\u680f\u548c\u9875\u811a\nsoup = BeautifulSoup(result.html, 'html.parser')\nfor tag in soup(['script', 'style', 'nav', 'footer']):\n    tag.decompose()\nclean_html = str(soup)  # \u6e05\u6d17\u540e\u7684 HTML\uff0c\u5927\u5e45\u51cf\u5c11\u65e0\u5173 token \u6d88\u8017\uff08\u53ef\u964d\u4f4e 70\u201390%\uff09<\/code><\/pre>\n\n\n\n
    \u901a\u8fc7\u8fd9\u4e00\u6b65\u9aa4\uff0c\u539f\u672c\u5197\u957f\u7684HTML\u5185\u5bb9\u5c06\u88ab\u538b\u7f29\uff0c\u53ea\u4fdd\u7559\u5173\u952e\u6587\u672c\uff0c\u5927\u5e45\u964d\u4f4e DeepSeek API \u7684\u8c03\u7528\u6210\u672c\u3002<\/p>\n\n\n\n
    2.2 \u91d1\u878d\u6570\u636e\u6293\u53d6<\/strong><\/h3>\n\n\n\n
    \u9488\u5bf9\u91d1\u878d\u9886\u57df\uff0c\u80a1\u7968\u5bf9\u5e94\u516c\u53f8\u7684\u57fa\u672c\u4fe1\u606f\uff08\u5982\u516c\u53f8\u540d\u79f0\u3001\u884c\u4e1a\u3001\u5e02\u76c8\u7387\u3001\u5e02\u503c\u7b49\uff09\u5bf9\u6295\u8d44\u51b3\u7b56\u81f3\u5173\u91cd\u8981\u3002\u4e0b\u9762\u662f\u4e00\u6bb5\u793a\u4f8b\u4ee3\u7801\uff0c\u5c55\u793a\u5982\u4f55\u4ece\u65b0\u6d6a\u8d22\u7ecf\u7b49\u91d1\u878d\u7f51\u7ad9\u6293\u53d6\u76ee\u6807\u80a1\u7968\u7684\u4fe1\u606f\uff1a<\/p>\n\n\n\n
    import requests\nfrom bs4 import BeautifulSoup\n\ndef fetch_company_info(stock_code):\n    \"\"\"\n    \u6839\u636e\u80a1\u7968\u4ee3\u7801\u6293\u53d6\u516c\u53f8\u57fa\u672c\u4fe1\u606f\uff08\u793a\u4f8b\u4f7f\u7528\u65b0\u6d6a\u8d22\u7ecf\u9875\u9762\uff09\n    :param stock_code: \u80a1\u7968\u4ee3\u7801\uff0c\u5982 \"sh601318\"\uff08\u4e0a\u4ea4\u6240\u80a1\u7968\u9700\u52a0\u524d\u7f00\uff09\n    :return: \u5305\u542b\u516c\u53f8\u540d\u79f0\u3001\u884c\u4e1a\u3001\u5e02\u76c8\u7387\u3001\u5e02\u503c\u7b49\u4fe1\u606f\u7684\u5b57\u5178\n    \"\"\"\n    url = f\"https:\/\/finance.sina.com.cn\/realstock\/company\/{stock_code}\/nc.shtml\"\n    headers = {\n        \"User-Agent\": \"Mozilla\/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit\/537.36 Chrome\/112.0.0.0 Safari\/537.36\"\n    }\n    response = requests.get(url, headers=headers)\n    response.encoding = \"utf-8\"\n    soup = BeautifulSoup(response.text, \"html.parser\")\n    \n    # \u63d0\u53d6\u5173\u952e\u4fe1\u606f\uff08\u9009\u62e9\u5668\u9700\u6839\u636e\u5b9e\u9645\u9875\u9762\u7ed3\u6784\u8c03\u6574\uff09\n    try:\n        company_name = soup.find(\"div\", class_=\"companyName\").text.strip()\n    except AttributeError:\n        company_name = \"\u672a\u627e\u5230\u516c\u53f8\u540d\u79f0\"\n    \n    try:\n        industry = soup.find(\"span\", id=\"stock_industry\").text.strip()\n    except AttributeError:\n        industry = \"\u672a\u627e\u5230\u884c\u4e1a\u4fe1\u606f\"\n    \n    try:\n        pe_ratio = soup.find(\"span\", id=\"pe_ratio\").text.strip()\n    except AttributeError:\n        pe_ratio = \"\u672a\u627e\u5230\u5e02\u76c8\u7387\"\n    \n    try:\n        market_cap = soup.find(\"span\", id=\"market_cap\").text.strip()\n    except AttributeError:\n        market_cap = \"\u672a\u627e\u5230\u5e02\u503c\"\n    \n    return {\n        \"company_name\": company_name,\n        \"industry\": industry,\n        \"pe_ratio\": pe_ratio,\n        \"market_cap\": market_cap\n    }\n\n# \u793a\u4f8b\uff1a\u6293\u53d6\u4e2d\u56fd\u5e73\u5b89\uff08\u80a1\u7968\u4ee3\u7801\uff1ash601318\uff09\u7684\u4fe1\u606f\nstock_code = \"sh601318\"\ncompany_info = fetch_company_info(stock_code)\nprint(\"\u6293\u53d6\u5230\u7684\u516c\u53f8\u4fe1\u606f\uff1a\", company_info)<\/code><\/pre>\n\n\n\n
    \u7ed3\u5408Crawl4AI\u7b49\u5de5\u5177\u8fdb\u4e00\u6b65\u9884\u5904\u7406\u91d1\u878d\u7f51\u7ad9\u9875\u9762\uff0c\u786e\u4fdd\u53ea\u4fdd\u7559\u5173\u952e\u6587\u672c\uff0c\u964d\u4f4etoken\u4f7f\u7528\u91cf\u3002<\/p>\n\n\n\n
    \u4e09\u3001API\u8c03\u7528\u4e0e<\/strong>\u6df1\u5ea6\u667a\u80fd\u5206\u6790<\/strong><\/h2>\n\n\n\n

    3.1 DeepSeek\u57fa\u7840API\u8c03\u7528<\/strong><\/h3>\n\n\n\n
    \u5229\u7528DeepSeek\u7684API\u5bf9\u9884\u5904\u7406\u540e\u7684\u5185\u5bb9\u8fdb\u884c\u667a\u80fd\u5206\u6790\u3002\u4ee5\u4e0b\u793a\u4f8b\u4ee3\u7801\u5c55\u793a\u4e86\u5982\u4f55\u6784\u9020\u63d0\u95ee\u5e76\u8c03\u7528DeepSeek API\uff1a<\/p>\n\n\n\n
    import requests\n\ndef ask_deepseek(question, api_key):\n    url = \"https:\/\/api.deepseek.com\/v1\/chat\"\n    headers = {\n        \"Authorization\": f\"Bearer {api_key}\",\n        \"Content-Type\": \"application\/json\"\n    }\n    data = {\n        \"model\": \"deepseek-chat\",  # \u6216 \"deepseek-reasoner\" \u4ee5\u8c03\u7528\u63a8\u7406\u6a21\u578b\n        \"prompt\": question,\n        \"max_tokens\": 4000,\n        \"stream\": False\n    }\n    response = requests.post(url, headers=headers, json=data)\n    return response.json()\n\n# \u793a\u4f8b\u8c03\u7528\napi_key = \"YOUR_API_KEY\"  # \u8bf7\u66ff\u6362\u4e3a\u5b9e\u9645 API \u5bc6\u94a5\nquestion = \"\u8bf7\u63d0\u53d6\u9875\u9762\u4e2d\u4ea7\u54c1\u7684\u540d\u79f0\u3001\u4ef7\u683c\u548c\u8bc4\u5206\uff0c\u5e76\u8fd4\u56de JSON \u683c\u5f0f\u6570\u636e\u3002\"\nresult = ask_deepseek(question, api_key)\nprint(\"DeepSeek \u8fd4\u56de\u7ed3\u679c\uff1a\", result)<\/pre>\n\n\n\n
    3.2 \u91d1\u878d\u6570\u636e\u667a\u80fd\u5206\u6790\u6848\u4f8b<\/strong><\/h3>\n\n\n\n
    \u5c06\u6293\u53d6\u5230\u7684\u516c\u53f8\u57fa\u672c\u4fe1\u606f\u6574\u5408\u540e\uff0c\u6784\u9020\u8be6\u7ec6\u63d0\u95ee\uff0c\u5229\u7528DeepSeek\u5206\u6790\u8be5\u516c\u53f8\u7684\u6295\u8d44\u4ef7\u503c\u4e0e\u884c\u4e1a\u524d\u666f\uff1a<\/p>\n\n\n\n
    def analyze_company_info(info, api_key):
        \"\"\"
        \u5229\u7528 DeepSeek \u5206\u6790\u516c\u53f8\u4fe1\u606f\uff0c\u7ed9\u51fa\u6295\u8d44\u4ef7\u503c\u53ca\u884c\u4e1a\u524d\u666f\u5efa\u8bae
        :param info: \u516c\u53f8\u4fe1\u606f\u5b57\u5178
        :param api_key: DeepSeek API \u5bc6\u94a5
        :return: \u5206\u6790\u7ed3\u679c
        \"\"\"
        prompt = (
            f\"\u4ee5\u4e0b\u662f\u67d0\u80a1\u7968\u5bf9\u5e94\u516c\u53f8\u7684\u57fa\u672c\u4fe1\u606f\uff1a\\n\"
            f\"\u516c\u53f8\u540d\u79f0\uff1a{info['company_name']}\\n\"
            f\"\u6240\u5c5e\u884c\u4e1a\uff1a{info['industry']}\\n\"
            f\"\u5e02\u76c8\u7387\uff1a{info['pe_ratio']}\\n\"
            f\"\u5e02\u503c\uff1a{info['market_cap']}\\n\\n\"
            \"\u8bf7\u57fa\u4e8e\u4ee5\u4e0a\u4fe1\u606f\uff0c\u5206\u6790\u8be5\u516c\u53f8\u7684\u6295\u8d44\u4ef7\u503c\u3001\u884c\u4e1a\u7ade\u4e89\u4f18\u52bf\u4e0e\u672a\u6765\u53d1\u5c55\u524d\u666f\uff0c\u5e76\u7ed9\u51fa\u7b80\u8981\u7684\u6295\u8d44\u5efa\u8bae\u3002\"
        )
        
        return ask_deepseek(prompt, api_key)

    # \u793a\u4f8b\uff1a\u5206\u6790\u4e2d\u56fd\u5e73\u5b89\u7684\u516c\u53f8\u4fe1\u606f
    analysis_result = analyze_company_info(company_info, api_key)
    print(\"DeepSeek \u5206\u6790\u7ed3\u679c\uff1a\", analysis_result)<\/pre>\n\n\n\n
    \u901a\u8fc7\u8fd9\u79cd\u65b9\u5f0f\uff0c\u4e0d\u4ec5\u53ef\u4ee5\u81ea\u52a8\u751f\u6210\u7ed3\u6784\u5316\u7684\u6587\u672c\u5206\u6790\uff0c\u8fd8\u80fd\u7ed3\u5408\u5177\u4f53\u8d22\u52a1\u6307\u6807\u4e3a\u6295\u8d44\u51b3\u7b56\u63d0\u4f9b\u6570\u636e\u652f\u6301\u3002<\/p>\n\n\n\n
    3.3 \u8d22\u62a5\u6307\u6807\u591a\u7ef4\u5ea6\u5f52\u56e0\u5206\u6790<\/strong><\/h3>\n\n\n\n
    \u6293\u53d6\u4e0a\u5e02\u516c\u53f8\u8d22\u62a5\u6838\u5fc3\u6570\u636e\uff08ROE\u3001\u73b0\u91d1\u6d41\u3001\u8d44\u4ea7\u8d1f\u503a\u7387\u7b49\uff09\uff0c\u901a\u8fc7DeepSeek\u8fdb\u884c\u675c\u90a6\u5206\u6790\uff1a<\/p>\n\n\n\n
    def financial_attribution(stock_code):\n    # \u6293\u53d6\u4e1c\u65b9\u8d22\u5bcc\u7f51\u6df1\u5ea6\u8d22\u52a1\u6570\u636e\n    url = f\"http:\/\/data.eastmoney.com\/stock\/{stock_code}.html\"\n    data = crawler.fetch(url)\n    # \u6784\u5efa\u675c\u90a6\u5206\u6790\u516c\u5f0f\u53c2\u6570\n    dupont_params = {\n        'net_profit_margin': extract(data, '\u51c0\u5229\u6da6\u7387'),\n        'asset_turnover': extract(data, '\u603b\u8d44\u4ea7\u5468\u8f6c\u7387'),\n        'equity_multiplier': extract(data, '\u6743\u76ca\u4e58\u6570')\n    }\n    # \u751f\u6210\u5f52\u56e0\u62a5\u544a\n    prompt = f\"\"\"\u57fa\u4e8e\u675c\u90a6\u5206\u6790\u516c\u5f0f\uff1aROE={dupont_params['net_profit_margin']}\u00d7{dupont_params['asset_turnover']}\u00d7{dupont_params['equity_multiplier']}=X%\n    \u8bf7\u5206\u6790\u5404\u8981\u7d20\u8d21\u732e\u5ea6\uff0c\u5e76\u4e0e\u884c\u4e1a\u5747\u503c\u5bf9\u6bd4\u7ed9\u51fa\u6539\u8fdb\u5efa\u8bae\"\"\"\n    return ask_deepseek(prompt)<\/pre>\n\n\n\n
    \u4ef7\u503c\u70b9<\/strong>\uff1a\u7a81\u7834\u4f20\u7edf\u6bd4\u7387\u5206\u6790\u5c40\u9650\uff0c\u5b9e\u73b0\u81ea\u52a8\u5316\u5f52\u56e0\u8bca\u65ad\uff0c\u67d0\u57fa\u91d1\u516c\u53f8\u4f7f\u7528\u7c7b\u4f3c\u65b9\u6848\u540e\u6295\u7814\u6548\u7387\u63d0\u534765%\u3002<\/p>\n\n\n\n
    3.4 \u6210\u672c\u4e0e\u6548\u7387\u4f18\u5316\u7b56\u7565<\/strong><\/h3>\n\n\n\n
    \u5728\u6574\u4e2a\u6570\u636e\u6293\u53d6\u4e0e\u5206\u6790\u8fc7\u7a0b\u4e2d\uff0c\u964d\u4f4etoken\u6d88\u8017\u548cAPI\u8c03\u7528\u6210\u672c\u81f3\u5173\u91cd\u8981\u3002\u4ee5\u4e0b\u662f\u51e0\u70b9\u4f18\u5316\u5efa\u8bae\uff1a<\/p>\n\n\n\n
    \u76d1\u63a7 Token \u4f7f\u7528<\/strong>\uff1a\u5229\u7528tiktoken\u5e93\u63d0\u524d\u4f30\u7b97\u6bcf\u6b21\u8c03\u7528\u7684token\u6570\u91cf\uff0c\u786e\u4fdd\u4e0d\u4f1a\u8d85\u51fa\u9884\u671f\u3002<\/p>\n\n\n\n
    import tiktoken\nencoder = tiktoken.encoding_for_model(\"deepseek-chat\")\ntokens = encoder.encode(clean_html)\nprint(f\"\u6bcf\u9875 token \u6570\u91cf\uff1a{len(tokens)}\")<\/code><\/pre>\n\n\n\n
    \u9519\u5cf0\u6293\u53d6<\/strong>\uff1a\u4e3a\u907f\u514d\u9ad8\u5cf0\u65f6\u6bb5\u76ee\u6807\u7f51\u7ad9\u5c01IP\uff0c\u53ef\u5c06\u6293\u53d6\u4efb\u52a1\u5206\u6563\u5230\u975e\u9ad8\u5cf0\u65f6\u6bb5\u6267\u884c\u3002<\/p>\n\n\n\n
    \u7f13\u5b58\u673a\u5236<\/strong>\uff1a\u5bf9\u4e8e\u66f4\u65b0\u4e0d\u9891\u7e41\u7684\u6570\u636e\uff08\u4f8b\u5982\u67d0\u516c\u53f8\u57fa\u672c\u4fe1\u606f\uff09\uff0c\u91c7\u7528\u7f13\u5b58\u7b56\u7565\u51cf\u5c11\u91cd\u590d\u8c03\u7528\u3002<\/p>\n\n\n\n
    \u7cbe\u7ec6 Prompt<\/strong>\uff1a\u6784\u9020\u6fc0\u5149\u805a\u7126\u63d0\u793a\u8bcd\uff08\u4f8b\u5982\u201c\u53ea\u63d0\u53d6\u516c\u53f8\u540d\u79f0\u3001\u6240\u5c5e\u884c\u4e1a\u3001\u5e02\u76c8\u7387\u548c\u5e02\u503c\uff0c\u5e76\u8fd4\u56deJSON\u683c\u5f0f\u201d\uff09\uff0c\u6709\u6548\u51cf\u5c11\u4e0d\u5fc5\u8981\u7684token\u6d88\u8017\u3002<\/p>\n\n\n\n
    \u4e0b\u9762\u7684Token\u6d88\u8017\u4f18\u5316\u77e9\u9635<\/strong>\u80fd\u4e00\u76ee\u4e86\u7136\u7684\u4e88\u4ee5\u8bc1\u660e\u3002<\/p>\n\n\n\n
    \u6570\u636e\u7c7b\u578b<\/th>\u539f\u59cbtoken<\/th>\u4f18\u5316\u65b9\u6848<\/th>\u6700\u7ec8token<\/th>\u964d\u5e45<\/th><\/tr><\/thead>
    \u8d22\u62a5\u539f\u6587<\/td>15,000<\/td>\u5173\u952e\u6307\u6807\u63d0\u53d6<\/td>800<\/td>94.7%<\/td><\/tr>
    Level2\u5b9e\u65f6\u6570\u636e<\/td>\u8fde\u7eed\u6d41<\/td>\u6ce2\u52a8\u89e6\u53d1\u5f0f\u5904\u7406<\/td>\u52a8\u6001\u6d88\u8017<\/td>82%<\/td><\/tr>
    \u8206\u60c5\u6587\u672c<\/td>3,000<\/td>\u5173\u952e\u8bcd\u8fc7\u6ee4+\u6458\u8981<\/td>500<\/td>83.3%<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
    \u56db\u3001\u673a\u6784\u7ea7\u5b9e\u6218\u6848\u4f8b<\/strong><\/h2>\n\n\n\n

    4.1 \u4e3b\u529b\u8d44\u91d1\u76d1\u63a7\u7cfb\u7edf\uff08\u5e74\u5316\u6536\u76ca\u63d0\u534722%\u6848\u4f8b\uff09<\/strong><\/h3>\n\n\n\n
    \u6838\u5fc3\u903b\u8f91<\/strong>\uff1a\u901a\u8fc7Level2\u6570\u636e\u6355\u6349\u5927\u5355\u5f02\u52a8<\/p>\n\n\n\n
    import websocket\nfrom deepseek_finance import AlertEngine\n\n# \u4e3b\u529b\u8d44\u91d1\u76d1\u63a7\u6a21\u578b\ndef detect_abnormal_flow(msg):\n    data = json.loads(msg)\n    # \u8ba1\u7b97\u5927\u5355\u51c0\u6d41\u5165\u5f3a\u5ea6\n    big_order_ratio = (data['super_buy'] - data['super_sell'])\/data['total_volume']\n    # \u89e6\u53d1AI\u5206\u6790\uff08\u6ce2\u52a8\u7387>3%\u6216\u5f3a\u5ea6>0.15\uff09\n    if abs(data['change']) > 0.03 or big_order_ratio > 0.15:\n        prompt = f\"{data['symbol']}\u73b0\u5f02\u5e38\u8d44\u91d1\u6d41\uff1a{big_order_ratio}\uff0c\u7ed3\u5408{get_news(data['symbol'])}\u5206\u6790\u4e3b\u529b\u610f\u56fe\"\n        analysis = DeepSeek.analyze(prompt)\n        AlertEngine.push(analysis)\n\n# \u8fde\u63a5\u4e0a\u4ea4\u6240\u5b9e\u65f6\u6570\u636e\u6d41\nws = websocket.WebSocketApp(\"wss:\/\/api.sse.com\/level2\", on_message=detect_abnormal_flow)<\/pre>\n\n\n\n
    \u4ef7\u503c\u9a8c\u8bc1<\/strong>\uff1a\u67d0\u79c1\u52df\u5b9e\u76d8\u6d4b\u8bd5\u663e\u793a\uff0c\u5f53\u5927\u5355\u5f3a\u5ea6>0.2\u65f6\uff0c\u6b21\u65e5\u4e0a\u6da8\u6982\u7387\u8fbe67%\u3002<\/p>\n\n\n\n
    4.2 K\u7ebf\u5386\u53f2\u91cd\u6f14\u9884\u6d4b\uff08\u80dc\u7387\u63d0\u534740%\u7684\u79d8\u7c4d\uff09<\/strong><\/h3>\n\n\n\n
    \u521b\u65b0\u70b9<\/strong>\uff1a\u878d\u5408\u76ae\u5c14\u900a\u76f8\u5173\u7cfb\u6570\u4e0e\u6210\u4ea4\u91cf\u7279\u5f81\uff0c\u6784\u5efa\u65f6\u7a7a\u76f8\u4f3c\u5ea6\u6a21\u578b\u3002<\/p>\n\n\n\n
    from scipy.stats import pearsonr\nimport tushare as ts\n\ndef find_similar_kline(target_code, days=30):\n    # \u83b7\u53d6\u76ee\u6807\u80a1\u7968\u8fd1\u671fK\u7ebf\n    target = ts.get_k_data(target_code).iloc[-days:]\n    # \u5168\u5e02\u573a\u626b\u63cf\n    all_codes = ts.get_stock_basics().index\n    similarities = []\n    for code in all_codes:\n        try:\n            df = ts.get_k_data(code)\n            # \u8ba1\u7b97\u5f62\u6001\u76f8\u4f3c\u5ea6\uff08\u4ef7\u683c+\u6210\u4ea4\u91cf\uff09\n            price_corr, _ = pearsonr(target['close'], df['close'].iloc[-days:])\n            vol_corr, _ = pearsonr(target['volume'], df['volume'].iloc[-days:])\n            similarities.append((code, 0.7*price_corr + 0.3*vol_corr))\n        except: pass\n    # \u8fd4\u56de\u76f8\u4f3c\u5ea6TOP10\n    return sorted(similarities, key=lambda x:x[1], reverse=True)[:10]<\/pre>\n\n\n\n
    \u6570\u636e\u652f\u6491<\/strong>\uff1a\u6d4b\u8bd5\u663e\u793a\u76f8\u4f3c\u5ea6>90%\u7684\u4e2a\u80a1\uff0c5\u65e5\u4e0a\u6da8\u6982\u7387\u8fbe63.2%\u3002<\/p>\n\n\n\n
    4.3 \u8206\u60c5\u91cf\u5316\u7b56\u7565\uff08\u907f\u514d15%\u66b4\u8dcc\u7684\u9884\u8b66\u6a21\u578b\uff09<\/strong><\/h3>\n\n\n\n
    \u591a\u6e90\u6570\u636e\u878d\u5408<\/strong>\uff1a<\/p>\n\n\n\n
    sources = {\n    '\u96ea\u7403\u70ed\u5e16': 'https:\/\/xueqiu.com\/query\/v1\/symbol\/search.json?q=',\n    '\u8d22\u8054\u793e\u5feb\u8baf': 'https:\/\/www.cls.cn\/telegraph',\n    '\u9f99\u864e\u699c\u6570\u636e': 'http:\/\/data.eastmoney.com\/stock\/lhb.html'\n}\n\ndef sentiment_analysis():\n    emergency_words = ['\u7206\u96f7','\u51cf\u6301','\u7acb\u6848\u8c03\u67e5']\n    # \u5b9e\u65f6\u8206\u60c5\u626b\u63cf\n    for source, url in sources.items():\n        data = crawler.fetch(url)\n        # \u7d27\u6025\u4e8b\u4ef6\u9884\u8b66\n        if any(word in data for word in emergency_words):\n            AlertEngine.urgent_alert(f\"{source}\u68c0\u6d4b\u5230\u98ce\u9669\u5173\u952e\u8bcd\")\n        # \u60c5\u611f\u6307\u6570\u8ba1\u7b97\n        score = DeepSeek.sentiment(data)\n        # \u751f\u6210\u4ea4\u6613\u4fe1\u53f7\n        if score < 0.3: return 'SELL'\n        elif score > 0.7: return 'BUY'<\/pre>\n\n\n\n
    \u5b9e\u6218\u6848\u4f8b<\/strong>\uff1a\u6210\u529f\u9884\u8b66 2025Q1 AI\u677f\u5757\u56de\u8c03\uff0c\u907f\u5f0015%\u8dcc\u5e45\u3002<\/p>\n\n\n\n
    4.4 \u8206\u60c5\u9a71\u52a8\u7684\u91cf\u5316\u7b56\u7565\u4f18\u5316\uff08\u79c1\u52df\u57fa\u91d1\u5e74\u5316\u6536\u76ca\u63d0\u534722%\uff09<\/strong><\/h3>\n\n\n\n
    \u6784\u5efa\u65b0\u95fb+\u793e\u4ea4\u5a92\u4f53\u60c5\u7eea\u6307\u6570\uff0c\u6307\u5bfc\u7b97\u6cd5\u4ea4\u6613\uff1a<\/p>\n\n\n\n
    # \u591a\u6e90\u8206\u60c5\u6570\u636e\u91c7\u96c6\nsources = [\n    'https:\/\/xueqiu.com\/search?q=',\n    'https:\/\/finance.sina.com.cn\/roll\/',\n    'https:\/\/www.jin10.com\/'\n]\n\ndef sentiment_analysis():\n    articles = []\n    for source in sources:\n        articles += crawler.fetch_articles(source)\n    # \u60c5\u611f\u5206\u6790API\u8c03\u7528\n    sentiment_score = DeepSeek.sentiment_batch(articles)\n    # \u751f\u6210\u4ea4\u6613\u4fe1\u53f7\uff08\u793a\u4f8b\u903b\u8f91\uff09\n    if sentiment_score > 0.7:\n        return \"BUY\"\n    elif sentiment_score < 0.3:\n        return \"SELL\"<\/pre>\n\n\n\n
    \u4ef7\u503c\u70b9<\/strong>\uff1a\u5b9e\u73b0\u975e\u7ed3\u6784\u5316\u6570\u636e\u7684\u91cf\u5316\u5e94\u7528\uff0c\u67d0\u79c1\u52df\u57fa\u91d1\u901a\u8fc7\u8be5\u65b9\u6848\u5e74\u5316\u6536\u76ca\u63d0\u534722% \u3002<\/p>\n\n\n\n
    \u4e94\u3001\u6280\u672f\u67b6\u6784\u53c2\u8003\uff08\u91d1\u878d\u7ea7\u9ad8\u53ef\u7528\u8bbe\u8ba1\uff09<\/strong><\/h2>\n\n\n\n

    5.1 \u5206\u5c42\u67b6\u6784\u8bbe\u8ba1<\/strong><\/h3>\n\n\n\n
    graph TD\n    A[\u6570\u636e\u6e90\u5c42] --> B{\u91c7\u96c6\u63a7\u5236\u4e2d\u5fc3}\n    B --> C[\u5b9e\u65f6\u6d41\u5904\u7406\u96c6\u7fa4]\n    B --> D[\u6279\u91cf\u6293\u53d6\u96c6\u7fa4]\n    C --> E[\u9ad8\u901f\u6d88\u606f\u961f\u5217]\n    D --> E\n    E --> F{\u6570\u636e\u6cbb\u7406\u4e2d\u5fc3}\n    F --> G[\u7ed3\u6784\u5316\u5b58\u50a8\u5668]\n    F --> H[\u975e\u7ed3\u6784\u5316\u5b58\u50a8\u5668]\n    G --> I[AI\u63a8\u7406\u96c6\u7fa4]\n    H --> I\n    I --> J[\u51b3\u7b56\u4e2d\u5fc3]\n    J --> K[\u9884\u8b66\u7cfb\u7edf]\n    J --> L[\u53ef\u89c6\u5316\u770b\u677f]\n    J --> M[\u4ea4\u6613\u63a5\u53e3]<\/code><\/pre>\n\n\n\n
    5.2 \u6838\u5fc3\u6a21\u5757\u6982\u89c8<\/strong><\/h3>\n\n\n\n
    1. \u5f02\u6784\u6570\u636e\u91c7\u96c6\u5c42<\/strong><\/p>\n\n\n\n
      \n
    • \u5b9e\u65f6\u6d41\u5f15\u64ce\uff1a\u57fa\u4e8eApache Flink\u6784\u5efa\u6beb\u79d2\u7ea7\u5904\u7406\u80fd\u529b\uff1b<\/li>\n\n\n\n
    • \u652f\u6301Level2\u884c\u60c5\u6570\u636e\uff08\u6bcf\u79d230,000+\uff09\uff1b<\/li>\n\n\n\n
    • \u591a\u534f\u8bae\u9002\u914d\uff08WebSocket\/HTTP\/FPGA\u76f4\u8fde\uff09\uff1b<\/li>\n\n\n\n
    • \u6279\u91cf\u6293\u53d6\u96c6\u7fa4\uff1a\u5206\u5e03\u5f0fScrapy\u6846\u67b6\u4f18\u5316\uff1b<\/li>\n\n\n\n
    • \u667a\u80fd\u53cd\u722c\u7b56\u7565\uff08IP\u8f6e\u6362+\u6d4f\u89c8\u5668\u6307\u7eb9\u6a21\u62df\uff09\uff1b<\/li>\n\n\n\n
    • \u52a8\u6001\u8c03\u5ea6\u7b97\u6cd5\uff08\u6839\u636e\u7f51\u7ad9QPS\u81ea\u52a8\u8c03\u6574\u9891\u7387\uff09\u3002<\/li>\n<\/ul>\n\n\n\n
      2. \u6d41\u6279\u4e00\u4f53\u5904\u7406\u5c42<\/strong><\/p>\n\n\n\n
      # \u7edf\u4e00\u6570\u636e\u5904\u7406\u7ba1\u9053\u793a\u4f8b\nfrom pyspark.sql import SparkSession\n\nspark = SparkSession.builder \\\n    .config(\"spark.sql.streaming.schemaInference\", \"true\") \\\n    .getOrCreate()\n\n# \u6d41\u5f0f\u6570\u636e\u5904\u7406\nstream_df = spark.readStream \\\n    .format(\"kafka\") \\\n    .option(\"kafka.bootstrap.servers\", \"host:port\") \\\n    .load()\n\n# \u6279\u91cf\u6570\u636e\u5904\u7406\nbatch_df = spark.read \\\n    .format(\"parquet\") \\\n    .load(\"hdfs:\/\/financial_data\/\")\n\n# \u7edf\u4e00\u8f6c\u6362\u903b\u8f91\ndef transform_data(df):\n    return df.selectExpr(\n        \"CAST(value AS STRING)\",\n        \"window(time, '1 minute') as time_window\"\n    ).groupBy(\"time_window\").agg(...)<\/code><\/pre>\n\n\n\n
      3. \u667a\u80fd\u5206\u6790\u5c42<\/strong><\/p>\n\n\n\n
      \u6a21\u578b\u670d\u52a1\u5316\u67b6\u6784\uff1a<\/strong><\/p>\n\n\n\n