TravelContentCreator/document/content_transformer.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

"""
内容转换器模块
使用LLM将解析的文档内容转换为标准化的景区和产品资料格式
"""

import logging
from typing import Dict, Any, Optional, List
from dataclasses import dataclass
from datetime import datetime
import uuid

from .content_integrator import IntegratedContent
from core.ai.ai_agent import AIAgent
from core.config.manager import ConfigManager
from utils.file_io import OutputManager

logger = logging.getLogger(__name__)

@dataclass
class TransformedContent:
    """转换后的内容"""
    original_content: IntegratedContent
    transformed_text: str
    format_type: str
    transformation_metadata: Dict[str, Any]
    transformed_at: datetime

class ContentTransformer:
    """内容转换器 - 将整合的内容转换为指定格式"""
    
    def __init__(self, config: Optional[Dict[str, Any]] = None):
        self.config = config or {}
        self.supported_formats = {
            'attraction_standard': self._transform_to_attraction_standard,
            'product_sales': self._transform_to_product_sales,
            'travel_guide': self._transform_to_travel_guide,
            'blog_post': self._transform_to_blog_post,
            'summary': self._transform_to_summary
        }
    
    def transform_content(self, 
                         integrated_content: IntegratedContent,
                         format_type: str = 'summary',
                         custom_prompt: Optional[str] = None) -> TransformedContent:
        """转换内容
        
        Args:
            integrated_content: 整合后的内容
            format_type: 转换格式类型
            custom_prompt: 自定义提示词
            
        Returns:
            TransformedContent: 转换后的内容
        """
        if format_type not in self.supported_formats:
            raise ValueError(f"不支持的格式类型: {format_type}")
        
        logger.info(f"开始转换内容，格式: {format_type}")
        
        # 执行转换
        transform_func = self.supported_formats[format_type]
        transformed_text = transform_func(integrated_content, custom_prompt)
        
        # 生成转换元数据
        transformation_metadata = {
            'format_type': format_type,
            'source_document_count': integrated_content.document_count,
            'source_content_length': integrated_content.total_content_length,
            'transformed_content_length': len(transformed_text),
            'key_topics_used': integrated_content.key_topics,
            'custom_prompt_used': custom_prompt is not None
        }
        
        return TransformedContent(
            original_content=integrated_content,
            transformed_text=transformed_text,
            format_type=format_type,
            transformation_metadata=transformation_metadata,
            transformed_at=datetime.now()
        )
    
    def _transform_to_attraction_standard(self, content: IntegratedContent, custom_prompt: Optional[str] = None) -> str:
        """转换为景点标准格式"""
        template = """
# 景点信息整理

## 基本信息
- 文档来源: {document_count}个文档
- 主要主题: {key_topics}

## 详细内容
{combined_content}

## 内容摘要
{content_summary}

---
*基于提供的文档整理，如需更多信息请参考原始文档*
"""
        
        return template.format(
            document_count=content.document_count,
            key_topics=", ".join(content.key_topics[:5]),
            combined_content=content.combined_content,
            content_summary=content.content_summary
        )
    
    def _transform_to_product_sales(self, content: IntegratedContent, custom_prompt: Optional[str] = None) -> str:
        """转换为产品销售格式"""
        template = """
# 产品销售资料

## 产品特色
基于{document_count}个文档的信息整理：

{content_summary}

## 详细介绍
{combined_content}

## 关键卖点
{key_topics}

---
*内容整理自提供的文档资料*
"""
        
        key_points = "\n".join([f"• {topic}" for topic in content.key_topics[:8]])
        
        return template.format(
            document_count=content.document_count,
            content_summary=content.content_summary,
            combined_content=content.combined_content,
            key_topics=key_points
        )
    
    def _transform_to_travel_guide(self, content: IntegratedContent, custom_prompt: Optional[str] = None) -> str:
        """转换为旅游指南格式"""
        template = """
# 旅游指南

## 概述
{content_summary}

## 详细信息
{combined_content}

## 重要提示
- 信息来源: {document_count}个文档
- 关键主题: {key_topics}

---
*本指南基于提供的文档整理，出行前请核实最新信息*
"""
        
        return template.format(
            content_summary=content.content_summary,
            combined_content=content.combined_content,
            document_count=content.document_count,
            key_topics=", ".join(content.key_topics[:5])
        )
    
    def _transform_to_blog_post(self, content: IntegratedContent, custom_prompt: Optional[str] = None) -> str:
        """转换为博客文章格式"""
        template = """
# 博客文章

## 前言
本文基于{document_count}个文档资料整理而成。

## 主要内容

{combined_content}

## 总结
{content_summary}

## 相关主题
{key_topics}

---
*本文内容整理自多个文档资料*
"""
        
        topics_list = "\n".join([f"- {topic}" for topic in content.key_topics[:10]])
        
        return template.format(
            document_count=content.document_count,
            combined_content=content.combined_content,
            content_summary=content.content_summary,
            key_topics=topics_list
        )
    
    def _transform_to_summary(self, content: IntegratedContent, custom_prompt: Optional[str] = None) -> str:
        """转换为摘要格式"""
        template = """
# 文档内容摘要

## 文档统计
- 文档数量: {document_count}
- 文档类型: {document_types}
- 内容长度: {content_length}字符

## 内容摘要
{content_summary}

## 关键主题
{key_topics}

## 完整内容
{combined_content}
"""
        
        doc_types = ", ".join([f"{k}({v}个)" for k, v in content.document_types.items()])
        topics_list = "\n".join([f"• {topic}" for topic in content.key_topics])
        
        return template.format(
            document_count=content.document_count,
            document_types=doc_types,
            content_length=content.total_content_length,
            content_summary=content.content_summary,
            key_topics=topics_list,
            combined_content=content.combined_content
        )
    
    def get_supported_formats(self) -> List[str]:
        """获取支持的格式列表"""
        return list(self.supported_formats.keys())
    
    def add_custom_format(self, format_name: str, transform_func):
        """添加自定义格式"""
        self.supported_formats[format_name] = transform_func
        logger.info(f"添加自定义格式: {format_name}")