travel-agent-cli 0.2.0 → 0.2.2

package/python/config/models.py

@@ -0,0 +1,119 @@
+"""数据模型定义"""
+from pydantic import BaseModel, Field
+from typing import Optional, List
+from datetime import datetime
+from enum import Enum
+
+
+class SourceType(str, Enum):
+    """数据来源类型"""
+    XIAOHONGSHU = "xiaohongshu"
+    WEIBO = "weibo"
+    WENLV = "wenlv"
+    OTA = "ota"
+
+
+class SocialPost(BaseModel):
+    """社交媒体帖子"""
+    id: str
+    source: SourceType
+    title: str
+    content: str
+    author: str
+    author_id: str
+    images: List[str] = Field(default_factory=list)
+    likes: int = 0
+    comments: int = 0
+    shares: int = 0
+    tags: List[str] = Field(default_factory=list)
+    url: str
+    published_at: Optional[datetime] = None
+    collected_at: datetime = Field(default_factory=datetime.now)
+
+    @property
+    def engagement_score(self) -> float:
+        """计算互动得分"""
+        return self.likes * 1 + self.comments * 2 + self.shares * 3
+
+
+class WenlvInfo(BaseModel):
+    """文旅局信息"""
+    id: str
+    source: SourceType
+    title: str
+    content: str
+    region: str
+    url: str
+    info_type: str  # 政策/活动/推荐路线
+    published_at: Optional[datetime] = None
+    collected_at: datetime = Field(default_factory=datetime.now)
+
+
+class FlightInfo(BaseModel):
+    """航班信息"""
+    id: str
+    departure_city: str
+    arrival_city: str
+    price: float
+    currency: str = "CNY"
+    airline: Optional[str] = None
+    departure_time: Optional[str] = None
+    travel_time: Optional[str] = None
+    platform: str = "fliggy"  # fliggy/ctrip
+    collected_at: datetime = Field(default_factory=datetime.now)
+
+
+class HotelInfo(BaseModel):
+    """酒店信息"""
+    id: str
+    name: str
+    destination: str
+    price_per_night: float
+    currency: str = "CNY"
+    rating: Optional[float] = None
+    stars: Optional[int] = None
+    platform: str = "fliggy"  # fliggy/ctrip
+    url: Optional[str] = None
+    collected_at: datetime = Field(default_factory=datetime.now)
+
+
+class Destination(BaseModel):
+    """目的地信息"""
+    name: str
+    region: str  # 省份/国家
+    city: Optional[str] = None
+    tags: List[str] = Field(default_factory=list)
+    description: Optional[str] = None
+
+
+class DestinationRecommendation(BaseModel):
+    """目的地推荐结果"""
+    name: str
+    rank: int
+    score: float
+    reason: str
+    estimated_cost: str
+    suggested_days: int
+    best_time: str
+    highlights: List[str]
+    flight_info: Optional[FlightInfo] = None
+    hotel_info: Optional[HotelInfo] = None
+    related_posts: List[SocialPost] = Field(default_factory=list)
+
+
+class TravelRoute(BaseModel):
+    """旅行路线"""
+    destination: str
+    duration_days: int
+    daily_plan: List[dict]
+    budget_estimate: dict
+
+
+class Report(BaseModel):
+    """生成的报告"""
+    id: str
+    generated_at: datetime
+    destinations: List[DestinationRecommendation]
+    routes: List[TravelRoute]
+    summary: str
+    markdown_content: str

package/python/config/prompts.py

@@ -0,0 +1,105 @@
+"""LLM Prompt 模板"""
+
+# 目的地分析 Prompt
+DESTINATION_ANALYSIS_PROMPT = """
+你是一个专业的旅行目的地分析师。请根据以下采集到的数据，分析并推荐最佳旅行目的地。
+
+## 输入数据
+{context_data}
+
+## 任务要求
+1. 从数据中识别出有潜力的旅行目的地
+2. 综合考虑以下维度进行评分（1-10 分）：
+   - 热度：社交媒体讨论度、点赞评论数
+   - 季节性：当前季节是否适合前往
+   - 性价比：机酒价格水平
+   - 独特性：是否有特色活动/景观
+   - 政策利好：是否有免签、补贴等优惠
+
+3. 推荐 10-15 个目的地，按综合得分排序
+4. 为每个目的地提供：
+   - 目的地名称
+   - 推荐理由（100-200 字）
+   - 预估机酒价格范围
+   - 建议游玩天数
+   - 最佳旅行时间
+
+## 输出格式
+请严格按照以下 JSON 格式输出：
+{{
+    "destinations": [
+        {{
+            "name": "目的地名称",
+            "rank": 排名，
+            "score": 综合得分，
+            "reason": "推荐理由",
+            "estimated_cost": "机酒价格范围",
+            "suggested_days": 建议天数，
+            "best_time": "最佳旅行时间",
+            "highlights": ["亮点 1", "亮点 2"]
+        }}
+    ],
+    "summary": "整体分析总结"
+}}
+"""
+
+# 路线规划 Prompt
+ROUTE_PLANNING_PROMPT = """
+你是一个专业的旅行规划师。请为以下目的地设计详细的旅行路线。
+
+## 目的地信息
+- 名称：{destination_name}
+- 建议游玩天数：{days}
+- 亮点：{highlights}
+
+## 任务要求
+设计一个详细的 {days} 天行程，包括：
+1. 每日行程安排（景点、活动、用餐建议）
+2. 交通建议（市内交通、景点间接驳）
+3. 住宿区域推荐
+4. 预算估算
+
+## 输出格式
+{{
+    "destination": "目的地名称",
+    "duration_days": {days},
+    "daily_plan": [
+        {{
+            "day": 1,
+            "theme": "今日主题",
+            "activities": [
+                {{"time": "上午", "item": "活动/景点", "duration": "时长"}},
+                {{"time": "下午", "item": "活动/景点", "duration": "时长"}},
+                {{"time": "晚上", "item": "活动/景点", "duration": "时长"}}
+            ],
+            "food_recommendation": "用餐建议",
+            "accommodation_area": "住宿区域建议"
+        }}
+    ],
+    "budget_estimate": {{
+        "accommodation": "住宿预算",
+        "food": "餐饮预算",
+        "transport": "交通预算",
+        "activities": "活动预算",
+        "total": "总计"
+    }}
+}}
+"""
+
+# 报告总结 Prompt
+REPORT_SUMMARY_PROMPT = """
+你是一个旅行专栏作家。请根据以下目的地推荐数据，撰写一篇吸引人的旅行推荐报告。
+
+## 数据
+{destinations_data}
+
+## 要求
+1. 标题要有吸引力，体现"热点"和"推荐"
+2. 开篇有一段引人入胜的导语
+3. 对 TOP10 目的地分别进行生动描述
+4. 加入实用建议（预订时机、注意事项等）
+5. 语言风格轻松有趣，适合社交媒体传播
+
+## 输出
+直接输出 Markdown 格式的报告正文。
+"""

package/python/config/settings.py

@@ -0,0 +1,172 @@
+"""配置管理模块 - 支持多 LLM 提供商"""
+from pydantic_settings import BaseSettings
+from pydantic import Field
+from pathlib import Path
+from functools import lru_cache
+from typing import Optional, Literal
+import json
+
+
+# 支持的 LLM 提供商
+LLM_PROVIDERS = {
+    "anthropic": {
+        "models": ["claude-sonnet-4-5-20250929", "claude-opus-4-0-20250514", "claude-haiku-4-5-20250417"],
+        "env_prefix": "ANTHROPIC",
+        "key_name": "ANTHROPIC_API_KEY",
+    },
+    "openai": {
+        "models": ["gpt-4o", "gpt-4-turbo", "gpt-4", "gpt-3.5-turbo"],
+        "env_prefix": "OPENAI",
+        "key_name": "OPENAI_API_KEY",
+    },
+    "deepseek": {
+        "models": ["deepseek-chat", "deepseek-coder"],
+        "env_prefix": "DEEPSEEK",
+        "key_name": "DEEPSEEK_API_KEY",
+    },
+    "azure": {
+        "models": ["gpt-4", "gpt-35-turbo"],
+        "env_prefix": "AZURE",
+        "key_name": "AZURE_OPENAI_API_KEY",
+        "extra": ["AZURE_OPENAI_ENDPOINT", "AZURE_OPENAI_API_VERSION"],
+    },
+    "ollama": {
+        "models": ["llama3", "mistral", "qwen2"],
+        "env_prefix": "OLLAMA",
+        "key_name": "",  # 本地部署，不需要 API Key
+        "extra": ["OLLAMA_BASE_URL"],
+    },
+    "qwen": {
+        "models": ["qwen-max", "qwen-plus", "qwen-turbo", "qwen-long"],
+        "env_prefix": "DASHSCOPE",
+        "key_name": "DASHSCOPE_API_KEY",
+    },
+}
+
+
+class Settings(BaseSettings):
+    """应用配置 - 支持多 LLM 提供商"""
+
+    # ========== LLM 配置 ==========
+    # 当前使用的 LLM 提供商：anthropic, openai, deepseek, azure, ollama
+    llm_provider: str = Field(default="anthropic", description="LLM 提供商")
+
+    # 使用的模型名称（留空则使用默认模型）
+    llm_model: str = Field(default="", description="模型名称，留空则使用默认")
+
+    # 各厂商 API Key（可选，至少配置一个）
+    anthropic_api_key: str = Field(default="", description="Anthropic API Key")
+    openai_api_key: str = Field(default="", description="OpenAI API Key")
+    deepseek_api_key: str = Field(default="", description="DeepSeek API Key")
+    azure_openai_api_key: str = Field(default="", description="Azure OpenAI API Key")
+    azure_openai_endpoint: str = Field(default="", description="Azure OpenAI Endpoint")
+    azure_openai_api_version: str = Field(default="2024-02-15-preview", description="Azure API 版本")
+    ollama_base_url: str = Field(default="http://localhost:11434", description="Ollama 服务地址")
+    dashscope_api_key: str = Field(default="", description="阿里云 DashScope API Key")
+
+    # ========== 其他配置 ==========
+    # 微博 API
+    weibo_app_key: str = Field(default="", description="微博 App Key")
+    weibo_app_secret: str = Field(default="", description="微博 App Secret")
+
+    # 代理配置
+    http_proxy: str = Field(default="", description="HTTP 代理")
+    https_proxy: str = Field(default="", description="HTTPS 代理")
+
+    # 路径配置
+    database_path: str = Field(default="./data/travel_agent.db", description="数据库路径")
+    output_dir: str = Field(default="./output", description="报告输出目录")
+
+    # 爬取配置
+    request_timeout: int = Field(default=30, description="HTTP 请求超时 (秒)")
+    max_retries: int = Field(default=3, description="最大重试次数")
+    request_delay: float = Field(default=1.0, description="请求间隔 (秒)")
+
+    @property
+    def data_dir(self) -> Path:
+        return Path(self.database_path).parent
+
+    @property
+    def output_path(self) -> Path:
+        return Path(self.output_dir)
+
+    def ensure_dirs(self):
+        """确保所需目录存在"""
+        self.data_dir.mkdir(parents=True, exist_ok=True)
+        self.output_path.mkdir(parents=True, exist_ok=True)
+
+    def get_active_api_key(self) -> str:
+        """获取当前激活的 API Key"""
+        key_map = {
+            "anthropic": self.anthropic_api_key,
+            "openai": self.openai_api_key,
+            "deepseek": self.deepseek_api_key,
+            "azure": self.azure_openai_api_key,
+            "ollama": "local",  # Ollama 不需要 Key
+            "qwen": self.dashscope_api_key,
+        }
+        return key_map.get(self.llm_provider, "")
+
+    def get_active_model(self) -> str:
+        """获取当前使用的模型"""
+        if self.llm_model:
+            return self.llm_model
+
+        # 返回各提供商的默认模型
+        defaults = {
+            "anthropic": "claude-sonnet-4-5-20250929",
+            "openai": "gpt-4o",
+            "deepseek": "deepseek-chat",
+            "azure": "gpt-4",
+            "ollama": "llama3",
+            "qwen": "qwen-max",
+        }
+        return defaults.get(self.llm_provider, "claude-sonnet-4-5-20250929")
+
+    def is_provider_configured(self, provider: str) -> bool:
+        """检查指定提供商是否已配置"""
+        if provider == "ollama":
+            return True  # 本地部署，始终可用
+
+        key_map = {
+            "anthropic": bool(self.anthropic_api_key),
+            "openai": bool(self.openai_api_key),
+            "deepseek": bool(self.deepseek_api_key),
+            "azure": bool(self.azure_openai_api_key) and bool(self.azure_openai_endpoint),
+            "qwen": bool(self.dashscope_api_key),
+        }
+        return key_map.get(provider, False)
+
+    def get_configured_providers(self) -> list:
+        """获取已配置的提供商列表"""
+        return [p for p in LLM_PROVIDERS.keys() if self.is_provider_configured(p)]
+
+    def get_llm_config(self) -> dict:
+        """获取当前 LLM 配置"""
+        return {
+            "provider": self.llm_provider,
+            "model": self.get_active_model(),
+            "api_key_configured": bool(self.get_active_api_key()),
+            "available_providers": self.get_configured_providers(),
+        }
+
+    class Config:
+        env_file = ".env"
+        env_file_encoding = "utf-8"
+        extra = "ignore"
+
+
+@lru_cache()
+def get_settings() -> Settings:
+    """获取配置单例"""
+    return Settings()
+
+
+def list_providers() -> str:
+    """列出所有支持的提供商和模型"""
+    lines = ["支持的 LLM 提供商和模型：", ""]
+    for provider, info in LLM_PROVIDERS.items():
+        lines.append(f"  {provider}:")
+        for model in info["models"]:
+            lines.append(f"    - {model}")
+    return "\n".join(lines)

package/python/export/__init__.py

@@ -0,0 +1,6 @@
+"""报告导出模块"""
+from export.report import ReportGenerator
+
+__all__ = [
+    "ReportGenerator",
+]

package/python/export/report.py

@@ -0,0 +1,192 @@
+"""报告生成模块"""
+import asyncio
+from typing import List, Optional
+from datetime import datetime
+from pathlib import Path
+import json
+
+from jinja2 import Template
+from config.models import DestinationRecommendation, TravelRoute, Report
+from config.settings import get_settings
+from config.prompts import REPORT_SUMMARY_PROMPT
+from analyzers.ranker import DestinationRanker
+from analyzers.route_planner import RoutePlanner
+
+
+# Markdown 报告模板
+REPORT_TEMPLATE = """
+# 旅行目的地推荐报告
+
+> 生成时间：{{ generated_at }}
+
+{{ summary }}
+
+---
+
+## 🏆 TOP 目的地推荐
+
+{% for dest in destinations %}
+### {{ dest.rank }}. {{ dest.name }}
+
+**综合得分：** {{ dest.score }}/10
+
+**推荐理由：** {{ dest.reason }}
+
+- 💰 **预估费用：** {{ dest.estimated_cost }}
+- 📅 **建议天数：** {{ dest.suggested_days }}天
+- 🕐 **最佳时间：** {{ dest.best_time }}
+- ✨ **亮点：** {{ dest.highlights | join(', ') }}
+
+{% endfor %}
+
+---
+
+## 🗺️ 推荐路线
+
+{% for route in routes %}
+### {{ route.destination }} - {{ route.duration_days }}天行程
+
+{% for day in route.daily_plan %}
+#### Day {{ day.day }}：{{ day.theme }}
+
+| 时间 | 活动 | 时长 |
+|------|------|------|
+{% for activity in day.activities %}| {{ activity.time }} | {{ activity.item }} | {{ activity.duration }} |
+{% endfor %}
+
+**用餐建议：** {{ day.food_recommendation }}
+
+**住宿区域：** {{ day.accommodation_area }}
+
+{% endfor %}
+
+**预算估算：**
+- 住宿：{{ route.budget_estimate.get('accommodation', 'N/A') }}
+- 餐饮：{{ route.budget_estimate.get('food', 'N/A') }}
+- 交通：{{ route.budget_estimate.get('transport', 'N/A') }}
+- 活动：{{ route.budget_estimate.get('activities', 'N/A') }}
+- **总计：** {{ route.budget_estimate.get('total', 'N/A') }}
+
+---
+
+{% endfor %}
+
+## 📌 出行提示
+
+1. **预订时机：** 建议提前 2-4 周预订机票酒店
+2. **证件准备：** 检查身份证/护照有效期
+3. **天气查询：** 出行前关注目的地天气预报
+4. **保险建议：** 建议购买旅行意外险
+
+---
+
+*本报告由 AI 驱动生成，数据来源于社交媒体热点和 OTA 平台*
+"""
+
+
+class ReportGenerator:
+    """报告生成器"""
+
+    def __init__(self):
+        self.settings = get_settings()
+        self.settings.ensure_dirs()
+        self.ranker = DestinationRanker()
+        self.route_planner = RoutePlanner()
+        self.template = Template(REPORT_TEMPLATE)
+
+    async def generate(
+        self,
+        format: str = "markdown",
+    ) -> Report:
+        """生成完整报告"""
+        # 1. 获取目的地推荐
+        destinations = await self.ranker.generate_recommendations(limit=12)
+
+        # 2. 为 TOP 目的地生成路线
+        routes = []
+        for dest in destinations[:5]:  # 为 TOP5 生成路线
+            route = await self.route_planner.plan_route(dest)
+            if route:
+                routes.append(route)
+
+        # 3. 生成报告摘要
+        summary = await self._generate_summary(destinations)
+
+        # 4. 渲染 Markdown
+        markdown_content = self.template.render(
+            generated_at=datetime.now().strftime("%Y-%m-%d %H:%M"),
+            summary=summary,
+            destinations=destinations,
+            routes=routes,
+        )
+
+        report = Report(
+            id=f"report_{datetime.now().strftime('%Y%m%d%H%M%S')}",
+            generated_at=datetime.now(),
+            destinations=destinations,
+            routes=routes,
+            summary=summary,
+            markdown_content=markdown_content,
+        )
+
+        return report
+
+    async def _generate_summary(self, destinations: List[DestinationRecommendation]) -> str:
+        """生成报告摘要"""
+        if self.ranker.client:
+            try:
+                dest_data = [
+                    {
+                        "name": d.name,
+                        "score": d.score,
+                        "reason": d.reason,
+                    }
+                    for d in destinations[:10]
+                ]
+
+                prompt = REPORT_SUMMARY_PROMPT.format(
+                    destinations_data=json.dumps(dest_data, ensure_ascii=False)
+                )
+
+                response = self.ranker.client.messages.create(
+                    model="claude-sonnet-4-5",
+                    max_tokens=1024,
+                    messages=[{"role": "user", "content": prompt}]
+                )
+
+                return response.content[0].text
+
+            except Exception as e:
+                print(f"摘要生成失败：{e}")
+
+        # 本地生成简化摘要
+        top3 = destinations[:3]
+        summary = f"本期推荐 {len(destinations)} 个热门旅行目的地。"
+        summary += f"排名前三的目的地分别是："
+        for i, dest in enumerate(top3):
+            summary += f"{i + 1}. {dest.name}（得分：{dest.score}）"
+            if i < len(top3) - 1:
+                summary += "，"
+        summary += "。这些目的地在社交媒体上热度较高，且当前季节适宜出行。"
+        return summary
+
+    def save(
+        self,
+        report: Report,
+        output_path: Optional[str] = None,
+    ) -> Path:
+        """保存报告到文件"""
+        if output_path:
+            path = Path(output_path)
+        else:
+            # 默认路径
+            filename = f"report_{report.generated_at.strftime('%Y%m%d_%H%M%S')}.md"
+            path = self.settings.output_path / filename
+
+        path.parent.mkdir(parents=True, exist_ok=True)
+
+        with open(path, "w", encoding="utf-8") as f:
+            f.write(report.markdown_content)
+
+        print(f"报告已保存：{path}")
+        return path