341 lines
16 KiB
Python
341 lines
16 KiB
Python
#!/usr/bin/env python3
|
||
# -*- coding: utf-8 -*-
|
||
|
||
"""
|
||
文字内容服务层
|
||
封装现有功能，提供API调用
|
||
"""
|
||
|
||
import logging
|
||
import uuid
|
||
from typing import List, Dict, Any, Optional, Tuple
|
||
from datetime import datetime
|
||
|
||
from core.config import ConfigManager, GenerateTopicConfig, GenerateContentConfig
|
||
from core.ai import AIAgent
|
||
from utils.file_io import OutputManager
|
||
from tweet.topic_generator import TopicGenerator
|
||
from tweet.content_generator import ContentGenerator
|
||
from tweet.content_judger import ContentJudger
|
||
from api.services.prompt_builder import PromptBuilderService
|
||
from api.services.prompt_service import PromptService
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
|
||
class TweetService:
|
||
"""文字内容服务类"""
|
||
|
||
def __init__(self, ai_agent: AIAgent, config_manager: ConfigManager, output_manager: OutputManager):
|
||
"""
|
||
初始化文字内容服务
|
||
|
||
Args:
|
||
ai_agent: AI代理
|
||
config_manager: 配置管理器
|
||
output_manager: 输出管理器
|
||
"""
|
||
self.ai_agent = ai_agent
|
||
self.config_manager = config_manager
|
||
self.output_manager = output_manager
|
||
|
||
# 初始化各个组件
|
||
self.topic_generator = TopicGenerator(ai_agent, config_manager, output_manager)
|
||
self.content_generator = ContentGenerator(ai_agent, config_manager, output_manager)
|
||
self.content_judger = ContentJudger(ai_agent, config_manager, output_manager)
|
||
|
||
# 初始化提示词服务和构建器
|
||
self.prompt_service = PromptService(config_manager)
|
||
self.prompt_builder = PromptBuilderService(config_manager, self.prompt_service)
|
||
|
||
async def generate_topics(self, dates: Optional[str] = None, numTopics: int = 5,
|
||
styles: Optional[List[str]] = None,
|
||
audiences: Optional[List[str]] = None,
|
||
scenic_spots: Optional[List[str]] = None,
|
||
products: Optional[List[str]] = None,
|
||
style_objects: Optional[List[Dict[str, Any]]] = None,
|
||
audience_objects: Optional[List[Dict[str, Any]]] = None,
|
||
scenic_spot_objects: Optional[List[Dict[str, Any]]] = None,
|
||
product_objects: Optional[List[Dict[str, Any]]] = None) -> Tuple[str, List[Dict[str, Any]]]:
|
||
"""
|
||
生成选题
|
||
|
||
Args:
|
||
dates: 日期字符串，可能为单个日期、多个日期用逗号分隔或范围
|
||
numTopics: 要生成的选题数量
|
||
styles: 风格列表
|
||
audiences: 受众列表
|
||
scenic_spots: 景区列表
|
||
products: 产品列表
|
||
style_objects: 风格对象列表
|
||
audience_objects: 受众对象列表
|
||
scenic_spot_objects: 景区对象列表
|
||
product_objects: 产品对象列表
|
||
|
||
Returns:
|
||
请求ID和生成的选题列表
|
||
"""
|
||
logger.info(f"开始生成选题，日期: {dates}, 数量: {numTopics}")
|
||
|
||
# 获取并更新配置
|
||
topic_config = self.config_manager.get_config('topic_gen', GenerateTopicConfig)
|
||
if dates:
|
||
topic_config.topic.date = dates
|
||
topic_config.topic.num = numTopics
|
||
|
||
# 使用PromptBuilderService构建提示词
|
||
system_prompt, user_prompt = self.prompt_builder.build_topic_prompt(
|
||
products=products,
|
||
scenic_spots=scenic_spots,
|
||
styles=styles,
|
||
audiences=audiences,
|
||
dates=dates,
|
||
numTopics=numTopics,
|
||
style_objects=style_objects,
|
||
audience_objects=audience_objects,
|
||
scenic_spot_objects=scenic_spot_objects,
|
||
product_objects=product_objects
|
||
)
|
||
|
||
# 使用预构建的提示词生成选题
|
||
topics = await self.topic_generator.generate_topics_with_prompt(system_prompt, user_prompt)
|
||
if not topics:
|
||
logger.error("未能生成任何选题")
|
||
return str(uuid.uuid4()), []
|
||
|
||
# 生成请求ID
|
||
requestId = f"topic-{datetime.now().strftime('%Y%m%d-%H%M%S')}-{str(uuid.uuid4())[:8]}"
|
||
|
||
logger.info(f"选题生成完成，请求ID: {requestId}, 数量: {len(topics)}")
|
||
return requestId, topics
|
||
|
||
async def generate_content(self, topic: Optional[Dict[str, Any]] = None,
|
||
style_objects: Optional[List[Dict[str, Any]]] = None,
|
||
audience_objects: Optional[List[Dict[str, Any]]] = None,
|
||
scenic_spot_objects: Optional[List[Dict[str, Any]]] = None,
|
||
product_objects: Optional[List[Dict[str, Any]]] = None,
|
||
autoJudge: bool = False) -> Tuple[str, str, Dict[str, Any]]:
|
||
"""
|
||
为选题生成内容
|
||
|
||
Args:
|
||
topic: 选题信息
|
||
styles: 风格列表
|
||
audiences: 受众列表
|
||
scenic_spots: 景区列表
|
||
products: 产品列表
|
||
autoJudge: 是否自动进行内容审核
|
||
|
||
Returns:
|
||
请求ID、选题索引和生成的内容（包含judgeSuccess状态）
|
||
"""
|
||
if not topic:
|
||
topic = {"index": "1", "date": "2024-07-01"}
|
||
|
||
topicIndex = topic.get('index', 'N/A')
|
||
logger.info(f"开始为选题 {topicIndex} 生成内容{'（含审核）' if autoJudge else ''}")
|
||
|
||
# 核心修改：创建一个增强版的topic，将所有需要的信息预先填充好
|
||
enhanced_topic = topic.copy()
|
||
if style_objects:
|
||
enhanced_topic['style_object'] = style_objects[0]
|
||
enhanced_topic['style'] = style_objects[0].get('styleName')
|
||
if audience_objects:
|
||
enhanced_topic['audience_object'] = audience_objects[0]
|
||
enhanced_topic['targetAudience'] = audience_objects[0].get('audienceName')
|
||
if scenic_spot_objects:
|
||
enhanced_topic['scenic_spot_object'] = scenic_spot_objects[0]
|
||
enhanced_topic['object'] = scenic_spot_objects[0].get('name')
|
||
if product_objects:
|
||
enhanced_topic['product_object'] = product_objects[0]
|
||
enhanced_topic['product'] = product_objects[0].get('productName')
|
||
|
||
# 使用PromptBuilderService构建提示词，现在它只需要enhanced_topic
|
||
system_prompt, user_prompt = self.prompt_builder.build_content_prompt(enhanced_topic, "content")
|
||
|
||
# 使用预构建的提示词生成内容
|
||
content = await self.content_generator.generate_content_with_prompt(enhanced_topic, system_prompt, user_prompt)
|
||
if not content:
|
||
logger.error(f"未能为选题 {topicIndex} 生成内容")
|
||
return str(uuid.uuid4()), topicIndex, {}
|
||
|
||
# 如果启用自动审核，进行内嵌审核
|
||
if autoJudge:
|
||
try:
|
||
logger.info(f"开始对选题 {topicIndex} 的内容进行内嵌审核")
|
||
|
||
# 使用PromptBuilderService构建审核提示词
|
||
judge_system_prompt, judge_user_prompt = self.prompt_builder.build_judge_prompt(enhanced_topic, content)
|
||
|
||
# 进行内容审核
|
||
judged_content = await self.content_judger.judge_content_with_prompt(content, enhanced_topic, judge_system_prompt, judge_user_prompt)
|
||
|
||
# 统一输出格式：始终包含judgeSuccess状态
|
||
# content_judger返回的是judge_success字段（下划线命名）
|
||
judge_success = judged_content.get('judge_success', False)
|
||
if judge_success:
|
||
logger.info(f"选题 {topicIndex} 内容审核成功")
|
||
# 审核成功：使用审核后的内容，但移除judge_success，添加统一的judgeSuccess
|
||
content = {k: v for k, v in judged_content.items() if k != 'judge_success'}
|
||
content['judgeSuccess'] = True
|
||
else:
|
||
logger.warning(f"选题 {topicIndex} 内容审核失败，保持原始内容")
|
||
# 审核失败：保持原始内容，添加judgeSuccess=False标记
|
||
content['judgeSuccess'] = False
|
||
|
||
except Exception as e:
|
||
logger.error(f"选题 {topicIndex} 内嵌审核失败: {e}，保持原始内容")
|
||
# 审核异常：保持原始内容，添加judgeSuccess=False标记
|
||
content['judgeSuccess'] = False
|
||
else:
|
||
# 未启用审核：添加judgeSuccess=None标记，表示未进行审核
|
||
content['judgeSuccess'] = None
|
||
|
||
# 生成请求ID
|
||
requestId = f"content-{datetime.now().strftime('%Y%m%d-%H%M%S')}-{str(uuid.uuid4())[:8]}"
|
||
|
||
logger.info(f"内容生成完成，请求ID: {requestId}, 选题索引: {topicIndex}")
|
||
return requestId, topicIndex, content
|
||
|
||
async def generate_content_with_prompt(self, topic: Dict[str, Any], system_prompt: str, user_prompt: str) -> Tuple[str, str, Dict[str, Any]]:
|
||
"""
|
||
使用预构建的提示词为选题生成内容
|
||
|
||
Args:
|
||
topic: 选题信息
|
||
system_prompt: 系统提示词
|
||
user_prompt: 用户提示词
|
||
|
||
Returns:
|
||
请求ID、选题索引和生成的内容
|
||
"""
|
||
topicIndex = topic.get('index', 'unknown')
|
||
logger.info(f"开始使用预构建提示词为选题 {topicIndex} 生成内容")
|
||
|
||
# 直接使用预构建的提示词生成内容
|
||
content = await self.content_generator.generate_content_with_prompt(topic, system_prompt, user_prompt)
|
||
|
||
# 生成请求ID
|
||
requestId = f"content-{datetime.now().strftime('%Y%m%d-%H%M%S')}-{str(uuid.uuid4())[:8]}"
|
||
|
||
logger.info(f"内容生成完成，请求ID: {requestId}, 选题索引: {topicIndex}")
|
||
return requestId, topicIndex, content
|
||
|
||
async def judge_content(self, topic: Optional[Dict[str, Any]] = None, content: Dict[str, Any] = {},
|
||
style_objects: Optional[List[Dict[str, Any]]] = None,
|
||
audience_objects: Optional[List[Dict[str, Any]]] = None,
|
||
scenic_spot_objects: Optional[List[Dict[str, Any]]] = None,
|
||
product_objects: Optional[List[Dict[str, Any]]] = None) -> Tuple[str, str, Dict[str, Any], bool]:
|
||
"""
|
||
审核内容 (已重构)
|
||
"""
|
||
if not topic:
|
||
topic = {"index": "1", "date": "2024-07-01"}
|
||
if not content:
|
||
content = {"title": "未提供内容", "content": "未提供内容"}
|
||
|
||
topicIndex = topic.get('index', 'unknown')
|
||
logger.info(f"开始审核选题 {topicIndex} 的内容")
|
||
|
||
# 构建包含所有预取信息的enhanced_topic
|
||
enhanced_topic = topic.copy()
|
||
if style_objects:
|
||
enhanced_topic['style_object'] = style_objects[0]
|
||
if audience_objects:
|
||
enhanced_topic['audience_object'] = audience_objects[0]
|
||
if scenic_spot_objects:
|
||
enhanced_topic['scenic_spot_object'] = scenic_spot_objects[0]
|
||
if product_objects:
|
||
enhanced_topic['product_object'] = product_objects[0]
|
||
|
||
system_prompt, user_prompt = self.prompt_builder.build_judge_prompt(enhanced_topic, content)
|
||
|
||
judged_data = await self.content_judger.judge_content_with_prompt(content, enhanced_topic, system_prompt, user_prompt)
|
||
|
||
judgeSuccess = judged_data.get('judge_success', False)
|
||
|
||
if 'judge_success' in judged_data:
|
||
judged_data = {k: v for k, v in judged_data.items() if k != 'judge_success'}
|
||
judged_data['judgeSuccess'] = judgeSuccess
|
||
|
||
requestId = f"judge-{datetime.now().strftime('%Y%m%d-%H%M%S')}-{str(uuid.uuid4())[:8]}"
|
||
|
||
logger.info(f"内容审核完成，请求ID: {requestId}, 选题索引: {topicIndex}, 审核结果: {judgeSuccess}")
|
||
return requestId, topicIndex, judged_data, judgeSuccess
|
||
|
||
async def run_pipeline(self, dates: Optional[str] = None, numTopics: int = 5,
|
||
styles: Optional[List[str]] = None,
|
||
audiences: Optional[List[str]] = None,
|
||
scenic_spots: Optional[List[str]] = None,
|
||
products: Optional[List[str]] = None,
|
||
skipJudge: bool = False,
|
||
autoJudge: bool = False) -> Tuple[str, List[Dict[str, Any]], Dict[str, Dict[str, Any]], Dict[str, Dict[str, Any]]]:
|
||
"""
|
||
运行完整的内容生成流水线：生成选题 → 生成内容 → 审核内容
|
||
|
||
Args:
|
||
dates: 日期字符串，可能为单个日期、多个日期用逗号分隔或范围
|
||
numTopics: 要生成的选题数量
|
||
styles: 风格列表
|
||
audiences: 受众列表
|
||
scenic_spots: 景区列表
|
||
products: 产品列表
|
||
skipJudge: 是否跳过内容审核步骤（与autoJudge互斥）
|
||
autoJudge: 是否在内容生成时进行内嵌审核
|
||
|
||
Returns:
|
||
请求ID、选题列表、内容字典和审核后内容字典
|
||
"""
|
||
logger.info(f"开始运行完整流水线，日期: {dates}, 数量: {numTopics}, 内嵌审核: {autoJudge}")
|
||
|
||
# 生成请求ID
|
||
requestId = f"pipeline-{datetime.now().strftime('%Y%m%d-%H%M%S')}-{str(uuid.uuid4())[:8]}"
|
||
|
||
# 1. 生成选题
|
||
_, topics = await self.generate_topics(dates, numTopics, styles, audiences, scenic_spots, products)
|
||
if not topics:
|
||
logger.error("未能生成任何选题")
|
||
return requestId, [], {}, {}
|
||
|
||
# 2. 为每个选题生成内容
|
||
contents = {}
|
||
judgedContents = {}
|
||
|
||
for topic in topics:
|
||
topicIndex = topic.get('index', 'unknown')
|
||
_, _, content = await self.generate_content(topic, autoJudge=autoJudge)
|
||
|
||
if autoJudge:
|
||
# 内嵌审核模式：content已包含审核结果和judgeSuccess状态
|
||
# 创建原始内容副本（移除judgeSuccess状态，但保留其他字段）
|
||
original_content = {k: v for k, v in content.items() if k != 'judgeSuccess'}
|
||
contents[topicIndex] = original_content
|
||
judgedContents[topicIndex] = content # 包含审核结果和judgeSuccess状态
|
||
else:
|
||
# 无审核模式：直接保存内容
|
||
contents[topicIndex] = content
|
||
|
||
# 如果使用内嵌审核或跳过审核，直接返回结果
|
||
if autoJudge or skipJudge:
|
||
logger.info(f"{'使用内嵌审核' if autoJudge else '跳过内容审核步骤'}，流水线完成，请求ID: {requestId}")
|
||
if autoJudge:
|
||
return requestId, topics, contents, judgedContents
|
||
else:
|
||
return requestId, topics, contents, contents
|
||
|
||
# 3. 对每个内容进行审核
|
||
judgedContents = {}
|
||
for topicIndex, content in contents.items():
|
||
topic = next((t for t in topics if t.get('index') == topicIndex), None)
|
||
if not topic:
|
||
logger.warning(f"找不到选题 {topicIndex} 的原始数据，跳过审核")
|
||
continue
|
||
|
||
try:
|
||
_, _, judged_data, _ = await self.judge_content(topic, content)
|
||
judgedContents[topicIndex] = judged_data
|
||
except Exception as e:
|
||
logger.critical(f"为选题 {topicIndex} 处理内容审核时发生意外错误: {e}", exc_info=True)
|
||
|
||
logger.info(f"流水线完成，请求ID: {requestId}")
|
||
return requestId, topics, contents, judgedContents |