深入研究（Qwen-Deep-Research）

import os
import dashscope

# 以下为华北2（北京）地域的URL，各地域的URL不同。
dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'

# 配置API Key
# 若没有配置环境变量，请用百炼API Key将下行替换为：API_KEY = "sk-xxx"
API_KEY = os.getenv('DASHSCOPE_API_KEY')

def call_deep_research_model(messages, step_name):
    print(f"\n=== {step_name} ===")

    try:
        responses = dashscope.Generation.call(
            api_key=API_KEY,
            model="qwen-deep-research",
            messages=messages,
            # qwen-deep-research模型目前仅支持流式输出
            stream=True
            # incremental_output=True 使用增量输出请添加此参数
        )

        return process_responses(responses, step_name)

    except Exception as e:
        print(f"调用API时发生错误: {e}")
        return ""

# 显示阶段内容
def display_phase_content(phase, content, status):
    if content:
        print(f"\n[{phase}] {status}: {content}")
    else:
        print(f"\n[{phase}] {status}")

# 处理响应
def process_responses(responses, step_name):
    current_phase = None
    phase_content = ""
    research_goal = ""
    web_sites = []
    references = []
    keepalive_shown = False  # 标记是否已经显示过KeepAlive提示

    for response in responses:
        # 检查响应状态码
        if hasattr(response, 'status_code') and response.status_code != 200:
            print(f"HTTP返回码：{response.status_code}")
            if hasattr(response, 'code'):
                print(f"错误码：{response.code}")
            if hasattr(response, 'message'):
                print(f"错误信息：{response.message}")
            print("请参考文档：https://help.aliyun.com/zh/model-studio/developer-reference/error-code")
            continue

        if hasattr(response, 'output') and response.output:
            message = response.output.get('message', {})
            phase = message.get('phase')
            content = message.get('content', '')
            status = message.get('status')
            extra = message.get('extra', {})

            # 阶段变化检测
            if phase != current_phase:
                if current_phase and phase_content:
                    # 根据阶段名称和步骤名称来显示不同的完成描述
                    if step_name == "第一步：模型反问确认" and current_phase == "answer":
                        print(f"\n 模型反问阶段完成")
                    else:
                        print(f"\n {current_phase} 阶段完成")
                current_phase = phase
                phase_content = ""
                keepalive_shown = False  # 重置KeepAlive提示标记

                # 根据阶段名称和步骤名称来显示不同的描述
                if step_name == "第一步：模型反问确认" and phase == "answer":
                    print(f"\n 进入模型反问阶段")
                else:
                    print(f"\n 进入 {phase} 阶段")

            # 处理Answer阶段的references信息
            if phase == "answer":
                if extra.get('deep_research', {}).get('references'):
                    new_references = extra['deep_research']['references']
                    if new_references and new_references != references:  # 避免重复显示
                        references = new_references
                        print(f"\n   引用来源 ({len(references)} 个):")
                        for i, ref in enumerate(references, 1):
                            print(f"     {i}. {ref.get('title', '无标题')}")
                            if ref.get('url'):
                                print(f"        URL: {ref['url']}")
                            if ref.get('description'):
                                print(f"        描述: {ref['description'][:100]}...")
                            print()

            # 处理WebResearch阶段的特殊信息
            # 注意：qwen-deep-research-2025-12-15模型使用streamingThinking状态
            # 替代streamingQueries和streamingWebResult
            if phase == "WebResearch":
                if extra.get('deep_research', {}).get('research'):
                    research_info = extra['deep_research']['research']

                    # 处理streamingThinking（快照模型）或streamingQueries（主线模型）状态
                    if status in ("streamingThinking", "streamingQueries"):
                        if 'researchGoal' in research_info:
                            goal = research_info['researchGoal']
                            if goal:
                                research_goal += goal
                                print(f"\n   研究目标: {goal}", end='', flush=True)

                    # 处理streamingWebResult状态（主线模型）
                    # 快照模型使用streamingThinking合并了此状态
                    elif status == "streamingWebResult":
                        if 'webSites' in research_info:
                            sites = research_info['webSites']
                            if sites and sites != web_sites:  # 避免重复显示
                                web_sites = sites
                                print(f"\n   找到 {len(sites)} 个相关网站:")
                                for i, site in enumerate(sites, 1):
                                    print(f"     {i}. {site.get('title', '无标题')}")
                                    print(f"        描述: {site.get('description', '无描述')[:100]}...")
                                    print(f"        URL: {site.get('url', '无链接')}")
                                    if site.get('favicon'):
                                        print(f"        图标: {site['favicon']}")
                                    print()

                    # 处理WebResultFinished状态
                    elif status == "WebResultFinished":
                        print(f"\n   网络搜索完成，共找到 {len(web_sites)} 个参考信息源")
                        if research_goal:
                            print(f"   研究目标: {research_goal}")

            # 累积内容并显示
            if content:
                phase_content += content
                # 实时显示内容
                print(content, end='', flush=True)

            # 显示阶段状态变化
            if status and status != "typing":
                print(f"\n   状态: {status}")

                # 显示状态说明
                if status == "streamingThinking":
                    print("   → 正在拆解研究任务并总结网页内容（WebResearch阶段）")
                elif status == "streamingQueries":
                    print("   → 正在生成研究目标和搜索查询（WebResearch阶段）")
                elif status == "streamingWebResult":
                    print("   → 正在执行搜索、网页阅读和代码执行（WebResearch阶段）")
                elif status == "WebResultFinished":
                    print("   → 网络搜索阶段完成（WebResearch阶段）")

            # 当状态为finished时，显示token消耗情况
            if status == "finished":
                if hasattr(response, 'usage') and response.usage:
                    usage = response.usage
                    print(f"\n    Token消耗统计:")
                    print(f"      输入tokens: {usage.get('input_tokens', 0)}")
                    print(f"      输出tokens: {usage.get('output_tokens', 0)}")
                    print(f"      请求ID: {response.get('request_id', '未知')}")

            if phase == "KeepAlive":
                # 只在第一次进入KeepAlive阶段时显示提示
                if not keepalive_shown:
                    print("当前步骤已经完成，准备开始下一步骤工作")
                    keepalive_shown = True
                continue

    if current_phase and phase_content:
        if step_name == "第一步：模型反问确认" and current_phase == "answer":
            print(f"\n 模型反问阶段完成")
        else:
            print(f"\n {current_phase} 阶段完成")

    return phase_content

def main():
    # 检查API Key
    if not API_KEY:
        print("错误：未设置 DASHSCOPE_API_KEY 环境变量")
        print("请设置环境变量或直接在代码中修改 API_KEY 变量")
        return

    print("用户发起对话：研究一下人工智能在教育中的应用")

    # 第一步：模型反问确认
    # 模型会分析用户问题，提出细化问题来明确研究方向
    messages = [{'role': 'user', 'content': '研究一下人工智能在教育中的应用'}]
    step1_content = call_deep_research_model(messages, "第一步：模型反问确认")

    # 第二步：深入研究
    # 基于第一步的反问内容，模型会执行完整的研究流程
    messages = [
        {'role': 'user', 'content': '研究一下人工智能在教育中的应用'},
        {'role': 'assistant', 'content': step1_content},  # 包含模型的反问内容
        {'role': 'user', 'content': '我主要关注个性化学习和智能评估这两个方面'}
    ]

    call_deep_research_model(messages, "第二步：深入研究")
    print("\n 研究完成！")

if __name__ == "__main__":
    main()

import os
import dashscope

# 以下为华北2（北京）地域的URL，各地域的URL不同。
dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'

API_KEY = os.getenv('DASHSCOPE_API_KEY')

messages = [
    {
        "role": "user",
        "content": [
            {"image": "https://example.aliyuncs.com/example.png"},
            {"text": "分析这张图表中的数据趋势，并对关键发现进行深入研究"}
        ]
    }
]

responses = dashscope.Generation.call(
    api_key=API_KEY,
    model="qwen-deep-research",
    messages=messages,
    stream=True
)

for response in responses:
    if hasattr(response, 'output') and response.output:
        message = response.output.get('message', {})
        content = message.get('content', '')
        if content:
            print(content, end='', flush=True)

import os
import dashscope

# 以下为华北2（北京）地域的URL，各地域的URL不同。
dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'

API_KEY = os.getenv('DASHSCOPE_API_KEY')

messages = [
    {
        "role": "user",
        "content": "使用知识库搜索近期发布的产品更新公告，并整理成研究报告"
    }
]

responses = dashscope.Generation.call(
    api_key=API_KEY,
    model="qwen-deep-research-2025-12-15",
    messages=messages,
    stream=True,
    enable_feedback=False,
    research_tools=[{
        "type": "mcp",
        "server_label": "my-server",
        "server_url": "https://your-mcp-server.example.com/sse",
        "allowed_tools": ["search", "fetch"],
        "authentication": {
            "bearer": "your_jwt_token_here"
        }
    }]
)

for response in responses:
    if hasattr(response, 'output') and response.output:
        message = response.output.get('message', {})
        content = message.get('content', '')
        if content:
            print(content, end='', flush=True)