AI工作流自动生成公众号图文并发布到wemark

wemark平台(wemark.xinghegongyu.com)是专门免费为写作者打造的工作利器，它有多种markdown排版主题，而且支持图文编辑，只需要截图粘贴即可，同时也支持对话大模型润色文章，还有更多功能期待大家去发现，发过一篇介绍点击[这里](https://mp.weixin....%25E6%2598%25AF%25E4%25B8%2593%25E9%2597%25A8%25E5%2585%258D%25E8%25B4%25B9%25E4%25B8%25BA%25E5%2586%2599%25E4%25BD%259C%25E8%2580%2585%25E6%2589%2593%25E9%2580%25A0%25E7%259A%2584%25E5%25B7%25A5%25E4%25BD%259C%25E5%2588%25A9%25E5%2599%25A8%25EF%25BC%258C%25E5%25AE%2583%25E6%259C%2589%25E5%25A4%259A%25E7%25A7%258Dmarkdown%25E6%258E%2592%25E7%2589%2588%25E4%25B8%25BB%25E9%25A2%2598%25EF%25BC%258C%25E8%2580%258C%25E4%25B8%2594%25E6%2594%25AF%25E6%258C%2581%25E5%259B%25BE%25E6%2596%2587%25E7%25BC%2596%25E8%25BE%2591%25EF%25BC%258C%25E5%258F%25AA%25E9%259C%2580%25E8%25A6%2581%25E6%2588%25AA%25E5%259B%25BE%25E7%25B2%2598%25E8%25B4%25B4%25E5%258D%25B3%25E5%258F%25AF%25EF%25BC%258C%25E5%2590%258C%25E6%2597%25B6%25E4%25B9%259F%25E6%2594%25AF%25E6%258C%2581%25E5%25AF%25B9%25E8%25AF%259D%25E5%25A4%25A7%25E6%25A8%25A1%25E5%259E%258B%25E6%25B6%25A6%25E8%2589%25B2%25E6%2596%2587%25E7%25AB%25A0%25EF%25BC%258C%25E8%25BF%2598%25E6%259C%2589%25E6%259B%25B4%25E5%25A4%259A%25E5%258A%259F%25E8%2583%25BD%25E6%259C%259F%25E5%25BE%2585%25E5%25A4%25A7%25E5%25AE%25B6%25E5%258E%25BB%25E5%258F%2591%25E7%258E%25B0%25EF%25BC%258C%25E5%258F%2591%25E8%25BF%2587%25E4%25B8%2580%25E7%25AF%2587%25E4%25BB%258B%25E7%25BB%258D%25E7%2582%25B9%25E5%2587%25BB%255B%25E8%25BF%2599%25E9%2587%258C%255D(https%3A%2F%2Fmp.weixin.qq.com%2Fs%3F__biz%3DMzUzMTUxMzYyNQ%3D%3D%26mid%3D2247487477%26idx%3D1%26sn%3D9dc27b9517f57280c95c77fd7677201f%26scene%3D21%23wechat_redirect)%25E3%2580%2582 "http://wemark.xinghegongyu.com)%E6%98%AF%E4%B8%93%E9%97%A8%E5%85%8D%E8%B4%B9%E4%B8%BA%E5%86%99%E4%BD%9C%E8%80%85%E6%89%93%E9%80%A0%E7%9A%84%E5%B7%A5%E4%BD%9C%E5%88%A9%E5%99%A8%EF%BC%8C%E5%AE%83%E6%9C%89%E5%A4%9A%E7%A7%8Dmarkdown%E6%8E%92%E7%89%88%E4%B8%BB%E9%A2%98%EF%BC%8C%E8%80%8C%E4%B8%94%E6%94%AF%E6%8C%81%E5%9B%BE%E6%96%87%E7%BC%96%E8%BE%91%EF%BC%8C%E5%8F%AA%E9%9C%80%E8%A6%81%E6%88%AA%E5%9B%BE%E7%B2%98%E8%B4%B4%E5%8D%B3%E5%8F%AF%EF%BC%8C%E5%90%8C%E6%97%B6%E4%B9%9F%E6%94%AF%E6%8C%81%E5%AF%B9%E8%AF%9D%E5%A4%A7%E6%A8%A1%E5%9E%8B%E6%B6%A6%E8%89%B2%E6%96%87%E7%AB%A0%EF%BC%8C%E8%BF%98%E6%9C%89%E6%9B%B4%E5%A4%9A%E5%8A%9F%E8%83%BD%E6%9C%9F%E5%BE%85%E5%A4%A7%E5%AE%B6%E5%8E%BB%E5%8F%91%E7%8E%B0%EF%BC%8C%E5%8F%91%E8%BF%87%E4%B8%80%E7%AF%87%E4%BB%8B%E7%BB%8D%E7%82%B9%E5%87%BB%5B%E8%BF%99%E9%87%8C%5D(https://mp.weixin.qq.com/s?__biz=MzUzMTUxMzYyNQ==&mid=2247487477&idx=1&sn=9dc27b9517f57280c95c77fd7677201f&scene=21#wechat_redirect)%E3%80%82")

有了wemark平台，我们就可以搭建一个本地AI工作流，每天帮我们写一篇微信公众号文章发布到wemark平台，wemark平台支持实时预览并且修改，还可以支持一键复制内容到公众号，对于日日更的小伙伴非常方便，你只需要说出你要创作的主题，只需要一句话就可以创建你要的精美图文。

好了，接下来带大家开启这段愉快的旅程吧！

搭建AI工作流

项目目录：

bash 复制代码

wechat_article_workflow/
├── main.py              # 主程序入口
├── workflow.py          # LangGraph工作流定义
├── models.py            # 多模态模型接口
├── blog_publisher.py    # 博客发布器
├── config.py            # 配置管理
├── requirements.txt     # 依赖包
├── .env.example         # 环境变量示例
└── README.md           # 说明文档

使用示例：

ini 复制代码

from workflow import WechatArticleWorkflow

# 创建工作流实例
workflow = WechatArticleWorkflow()

# 运行工作流
result = workflow.run("人工智能在教育领域的应用")

# 获取生成的Markdown内容
markdown_content = result["final_markdown"]
print(markdown_content)

工作流核心代码：

python 复制代码

"""
LangGraph工作流实现
"""
import json
import uuid
import time
from typing import Dict, Any, List
from langgraph.graph import StateGraph, END
from langgraph.graph.message import add_messages
from typing_extensions import Annotated, TypedDict

from models import TextGenerator, ImageGenerator, ArticleSection
from blog_publisher import BlogPublisher

class WorkflowState(TypedDict):
    """工作流状态"""
    topic: str
    outline: Dict[str, Any]
    sections: List[ArticleSection]
    final_markdown: str
    images_generated: bool
    published: bool
    error: str

class WechatArticleWorkflow:
    """微信公众号文章生成工作流"""
    
    def __init__(self):
        self.text_generator = TextGenerator()
        self.image_generator = ImageGenerator()
        self.blog_publisher = BlogPublisher()
        self.graph = self._build_graph()
    
    def _build_graph(self) -> StateGraph:
        """构建LangGraph工作流"""
        workflow = StateGraph(WorkflowState)
        
        # 添加节点
        workflow.add_node("generate_outline", self._generate_outline)
        workflow.add_node("generate_content", self._generate_content)
        workflow.add_node("generate_images", self._generate_images)
        workflow.add_node("compile_markdown", self._compile_markdown)
        workflow.add_node("publish_article", self._publish_article)
        
        # 设置边
        workflow.set_entry_point("generate_outline")
        workflow.add_edge("generate_outline", "generate_content")
        workflow.add_edge("generate_content", "generate_images")
        workflow.add_edge("generate_images", "compile_markdown")
        workflow.add_edge("compile_markdown", "publish_article")
        workflow.add_edge("publish_article", END)
        
        return workflow.compile()
    
    def _generate_outline(self, state: WorkflowState) -> WorkflowState:
        """生成文章大纲"""
        print(f"🎯 开始为主题 '{state['topic']}' 生成大纲...")
        
        try:
            outline_text = self.text_generator.generate_article_outline(state["topic"])
            # 尝试解析JSON，如果失败则使用文本
            try:
                outline = json.loads(outline_text)
            except:
                # 如果不是JSON格式，创建基础结构
                outline = {
                    "title": f"关于{state['topic']}的深度思考",
                    "sections": [
                        {
                            "title": "引言",
                            "content_points": ["背景介绍", "问题提出"],
                            "image_description": f"{state['topic']}相关的概念图"
                        },
                        {
                            "title": "核心内容",
                            "content_points": ["主要观点", "详细分析"],
                            "image_description": f"{state['topic']}的实践应用场景"
                        },
                        {
                            "title": "总结展望",
                            "content_points": ["要点总结", "未来展望"],
                            "image_description": f"{state['topic']}的未来发展趋势"
                        }
                    ],
                    "conclusion": "总结全文要点，引导读者思考"
                }
            
            state["outline"] = outline
            print(f"✅ 大纲生成完成: {outline['title']}")
            
        except Exception as e:
            state["error"] = f"大纲生成失败: {str(e)}"
            print(f"❌ {state['error']}")
        
        return state
    
    def _generate_content(self, state: WorkflowState) -> WorkflowState:
        """生成文章内容"""
        print("📝 开始生成文章内容...")
        
        sections = []
        outline = state.get("outline", {})
        
        try:
            for section_data in outline.get("sections", []):
                print(f"  生成段落: {section_data['title']}")
                
                time.sleep(12);
                content = self.text_generator.generate_section_content(
                    section_data["title"],
                    section_data.get("content_points", []),
                    state["topic"]
                )
                
                section = ArticleSection(
                    title=section_data["title"],
                    content=content,
                    image_prompt=section_data.get("image_description", "")
                )
                sections.append(section)
            
            state["sections"] = sections
            print(f"✅ 内容生成完成，共{len(sections)}个段落")
            
        except Exception as e:
            state["error"] = f"内容生成失败: {str(e)}"
            print(f"❌ {state['error']}")
        
        return state
    
    def _generate_images(self, state: WorkflowState) -> WorkflowState:
        """生成配图"""
        print("🎨 开始生成配图...")
        
        try:
            sections = state.get("sections", [])
            
            for i, section in enumerate(sections):
                if section.image_prompt:
                    print(f"  生成图片 {i+1}: {section.image_prompt}")
                    
                    image_url = self.image_generator.generate_image(section.image_prompt)
                    if image_url:
                        section.image_url = image_url
                        print(f"  ✅ 图片生成成功")
                    else:
                        print(f"  ⚠️ 图片生成失败，将跳过")
            
            state["images_generated"] = True
            print("✅ 配图生成完成")
            
        except Exception as e:
            state["error"] = f"配图生成失败: {str(e)}"
            print(f"❌ {state['error']}")
            state["images_generated"] = False
        
        return state
    
    def _compile_markdown(self, state: WorkflowState) -> WorkflowState:
        """编译Markdown格式"""
        print("📄 开始编译Markdown...")
        
        try:
            outline = state.get("outline", {})
            sections = state.get("sections", [])
            
            # 构建Markdown内容
            markdown_parts = []
            
            # 标题
            title = outline.get("title", f"关于{state['topic']}的思考")
            markdown_parts.append(f"# {title}\n")
            
            # 各个段落
            for section in sections:
                markdown_parts.append(f"## {section.title}\n")
                
                # 添加图片（如果有）
                if section.image_url:
                    markdown_parts.append(f"![{section.title}]({section.image_url})\n")
                
                # 添加内容
                markdown_parts.append(f"{section.content}\n")
            
            # 结论
            if outline.get("conclusion"):
                markdown_parts.append(f"## 总结\n\n{outline['conclusion']}\n")
            
            # 互动引导
            markdown_parts.append("\n---\n\n💬 你对这个话题有什么看法？欢迎在评论区分享你的想法！\n")
            
            final_markdown = "\n".join(markdown_parts)
            # 确保最终内容不包含markdown代码块标记
            final_markdown = self._clean_final_markdown(final_markdown)
            state["final_markdown"] = final_markdown
            
            print("✅ Markdown编译完成")
            
        except Exception as e:
            state["error"] = f"Markdown编译失败: {str(e)}"
            print(f"❌ {state['error']}")
        
        return state
    
    def _publish_article(self, state: WorkflowState) -> WorkflowState:
        """发布文章到博客"""
        print("🚀 开始发布文章...")
        
        try:
            outline = state.get("outline", {})
            title = outline.get("title", f"关于{state['topic']}的思考")
            content = state.get("final_markdown", "")
            
            if content:
                success = self.blog_publisher.publish(title, content)
                state["published"] = success
                
                if success:
                    print("✅ 文章发布成功！")
                else:
                    print("❌ 文章发布失败")
            else:
                state["error"] = "没有内容可发布"
                print(f"❌ {state['error']}")
                
        except Exception as e:
            state["error"] = f"发布失败: {str(e)}"
            print(f"❌ {state['error']}")
            state["published"] = False
        
        return state
    
    def _clean_final_markdown(self, text: str) -> str:
        """智能清理最终markdown内容 - 只移除包装性的代码块标记，保留内容中的代码示例"""
        import re
        
        # 检查是否整个文本被```markdown包装
        if text.strip().startswith('```markdown') and text.strip().endswith('```'):
            # 移除外层的```markdown包装
            text = re.sub(r'^```markdown\s*\n?', '', text.strip())
            text = re.sub(r'\n?```\s*$', '', text)
        
        # 检查是否整个文本被```包装（无语言标识）
        elif text.strip().startswith('```') and text.strip().endswith('```'):
            lines = text.strip().split('\n')
            if len(lines) >= 2:
                # 检查第一行是否只是```（包装性的）
                first_line = lines[0].strip()
                if first_line == '```' or (first_line.startswith('```') and len(first_line) <= 10):
                    # 移除首尾的```包装
                    text = '\n'.join(lines[1:-1])
        
        # 清理多余的空行，但保持适当的段落间距
        text = re.sub(r'\n\s*\n\s*\n+', '\n\n', text)
        
        return text.strip()
    
    def run(self, topic: str) -> Dict[str, Any]:
        """运行工作流"""
        print(f"🚀 启动微信公众号文章生成工作流")
        print(f"📋 主题: {topic}")
        print("=" * 50)
        
        initial_state = WorkflowState(
            topic=topic,
            outline={},
            sections=[],
            final_markdown="",
            images_generated=False,
            published=False,
            error=""
        )
        
        try:
            final_state = self.graph.invoke(initial_state)
            
            print("\n" + "=" * 50)
            print("🎉 工作流执行完成！")
            
            if final_state.get("error"):
                print(f"⚠️ 执行过程中出现错误: {final_state['error']}")
            
            if final_state.get("published"):
                print("✅ 文章已成功发布到博客")
            
            # 输出最终的Markdown内容
            if final_state.get("final_markdown"):
                print("\n📄 生成的Markdown内容:")
                print("-" * 30)
                print(final_state["final_markdown"])
            
            return final_state
            
        except Exception as e:
            error_msg = f"工作流执行失败: {str(e)}"
            print(f"❌ {error_msg}")
            return {"error": error_msg}

启动程序：

python 复制代码

"""
主程序入口 - 微信公众号文章生成工作流
"""
import os
import sys
from workflow import WechatArticleWorkflow

def main():
    """主函数"""
    print("🤖 微信公众号文章AI生成工具")
    print("=" * 50)
    
    # 检查必要的环境变量（只需要文本生成API密钥）
    kimi_key = os.getenv("KIMI_API_KEY")
    
    if not kimi_key:
        print("⚠️ 缺少必要的环境变量:")
        print("  - KIMI_API_KEY: Kimi API密钥（用于文本生成）")
        print("\n请在.env文件中设置这个变量，或者通过export命令设置")
        print("示例:")
        print("export KIMI_API_KEY='your-kimi-key'")
        print("\n💡 图片生成使用免费服务，无需额外API密钥！")
        print("🔗 获取Kimi API密钥: https://platform.moonshot.cn/")
        return
    
    # 显示当前配置的图片生成服务
    image_model = os.getenv("IMAGE_MODEL", "pollinations")
    print(f"📸 当前图片生成服务: {image_model} (免费)")
    
    # 检查可选的图片生成API密钥
    optional_keys = {
        "HUGGINGFACE_API_KEY": "Hugging Face (免费1000次/月)",
        "STABILITY_API_KEY": "Stability AI (新用户免费额度)"
    }
    
    available_services = ["pollinations", "picsum", "unsplash"]  # 无需API密钥的服务
    for key, desc in optional_keys.items():
        if os.getenv(key):
            available_services.append(desc)
    
    print(f"🎨 可用的图片生成服务: {', '.join(available_services)}")
    
    # 获取用户输入的主题
    if len(sys.argv) > 1:
        topic = " ".join(sys.argv[1:])
    else:
        topic = input("请输入文章主题: ").strip()
    
    if not topic:
        print("❌ 请提供文章主题")
        return
    
    # 创建并运行工作流
    workflow = WechatArticleWorkflow()
    
    try:
        result = workflow.run(topic)
        
        if result.get("error"):
            print(f"\n❌ 执行失败: {result['error']}")
            sys.exit(1)
        else:
            print(f"\n🎉 任务完成！")
            
            # 保存Markdown到本地文件
            if result.get("final_markdown"):
                filename = f"article_{topic.replace(' ', '_')}.md"
                with open(filename, 'w', encoding='utf-8') as f:
                    f.write(result["final_markdown"])
                print(f"📁 文章已保存到: {filename}")
    
    except KeyboardInterrupt:
        print("\n\n⏹️ 用户中断执行")
    except Exception as e:
        print(f"\n❌ 程序执行出错: {e}")
        sys.exit(1)

if __name__ == "__main__":
    main()

启动项目

我们在启东项目的时候只需要说主题即可，比如我们这里输入：写一篇关于Python处理excel文件的使用方式，然后回车，看下日志：

perl 复制代码

(venv) ➜  wechat_article_workflow python main.py "写一篇关于Python处理excel文件的使用方式"
🤖 微信公众号文章AI生成工具
==================================================
📸 当前图片生成服务: pollinations (免费)
🎨 可用的图片生成服务: pollinations, picsum, unsplash
🚀 启动微信公众号文章生成工作流
📋 主题: 写一篇关于Python处理excel文件的使用方式
==================================================
🎯 开始为主题 '写一篇关于Python处理excel文件的使用方式' 生成大纲...
✅ 大纲生成完成: Python处理Excel文件：让数据分析更高效
📝 开始生成文章内容...
  生成段落: 引言
  生成段落: Python处理Excel的必要库
  生成段落: 读取Excel文件
  生成段落: 修改和写入Excel文件
  生成段落: 高级操作：合并与拆分Excel文件
✅ 内容生成完成，共5个段落
🎨 开始生成配图...
  生成图片 1: 一张展示Excel文件和Python图标的图片，背景为数据分析图表
🎨 Pollinations生成中: illustration artwork 一张展示Excel文件和Python图标的图片,背景为数据分析图表,social media illustration style,high quality,professional digital art high quality
🔗 图片URL: https://image.pollinations.ai/prompt/illustration%20artwork%20%E4%B8%80%E5%BC%A0%E5%B1%95%E7%A4%BAExcel%E6%96%87%E4%BB%B6%E5%92%8CPython%E5%9B%BE%E6%A0%87%E7%9A%84%E5%9B%BE%E7%89%87%2C%E8%83%8C%E6%99%AF%E4%B8%BA%E6%95%B0%E6%8D%AE%E5%88%86%E6%9E%90%E5%9B%BE%E8%A1%A8%2Csocial%20media%20illustration%20style%2Chigh%20quality%2Cprofessional%20digital%20art%20high%20quality?width=1024&height=1024&nologo=true&enhance=true
✅ 图片生成成功 (尝试 1/3)
  ✅ 图片生成成功
  生成图片 2: 一张展示pandas、openpyxl、xlrd和xlsxwriter库的图标和名称的拼图
🎨 Pollinations生成中: illustration artwork 一张展示pandas、openpyxl、xlrd和xlsxwriter库的图标和名称的拼图,social media illustration style,high quality,professional digital art high quality
🔗 图片URL: https://image.pollinations.ai/prompt/illustration%20artwork%20%E4%B8%80%E5%BC%A0%E5%B1%95%E7%A4%BApandas%E3%80%81openpyxl%E3%80%81xlrd%E5%92%8Cxlsxwriter%E5%BA%93%E7%9A%84%E5%9B%BE%E6%A0%87%E5%92%8C%E5%90%8D%E7%A7%B0%E7%9A%84%E6%8B%BC%E5%9B%BE%2Csocial%20media%20illustration%20style%2Chigh%20quality%2Cprofessional%20digital%20art%20high%20quality?width=1024&height=1024&nologo=true&enhance=true
✅ 图片生成成功 (尝试 1/3)
  ✅ 图片生成成功
  生成图片 3: 一张代码截图，展示使用pandas读取Excel文件的示例代码
🎨 Pollinations生成中: illustration artwork 一张代码截图,展示使用pandas读取Excel文件的示例代码,social media illustration style,high quality,professional digital art high quality
🔗 图片URL: https://image.pollinations.ai/prompt/illustration%20artwork%20%E4%B8%80%E5%BC%A0%E4%BB%A3%E7%A0%81%E6%88%AA%E5%9B%BE%2C%E5%B1%95%E7%A4%BA%E4%BD%BF%E7%94%A8pandas%E8%AF%BB%E5%8F%96Excel%E6%96%87%E4%BB%B6%E7%9A%84%E7%A4%BA%E4%BE%8B%E4%BB%A3%E7%A0%81%2Csocial%20media%20illustration%20style%2Chigh%20quality%2Cprofessional%20digital%20art%20high%20quality?width=1024&height=1024&nologo=true&enhance=true
⚠️ HTTP状态码: 502
🔄 等待3秒后重试...
⚠️ HTTP状态码: 502
🔄 等待3秒后重试...
✅ 图片生成成功 (尝试 3/3)
  ✅ 图片生成成功
  生成图片 4: 一张图表，展示修改前后Excel文件的数据对比
🎨 Pollinations生成中: illustration artwork 一张图表,展示修改前后Excel文件的数据对比,social media illustration style,high quality,professional digital art high quality
🔗 图片URL: https://image.pollinations.ai/prompt/illustration%20artwork%20%E4%B8%80%E5%BC%A0%E5%9B%BE%E8%A1%A8%2C%E5%B1%95%E7%A4%BA%E4%BF%AE%E6%94%B9%E5%89%8D%E5%90%8EExcel%E6%96%87%E4%BB%B6%E7%9A%84%E6%95%B0%E6%8D%AE%E5%AF%B9%E6%AF%94%2Csocial%20media%20illustration%20style%2Chigh%20quality%2Cprofessional%20digital%20art%20high%20quality?width=1024&height=1024&nologo=true&enhance=true
⏱️ 请求超时 (尝试 1/3)
🔄 等待3秒后重试...
✅ 图片生成成功 (尝试 2/3)
  ✅ 图片生成成功
  生成图片 5: 一张流程图，展示合并和拆分Excel文件的步骤
🎨 Pollinations生成中: illustration artwork 一张流程图,展示合并和拆分Excel文件的步骤,social media illustration style,high quality,professional digital art high quality
🔗 图片URL: https://image.pollinations.ai/prompt/illustration%20artwork%20%E4%B8%80%E5%BC%A0%E6%B5%81%E7%A8%8B%E5%9B%BE%2C%E5%B1%95%E7%A4%BA%E5%90%88%E5%B9%B6%E5%92%8C%E6%8B%86%E5%88%86Excel%E6%96%87%E4%BB%B6%E7%9A%84%E6%AD%A5%E9%AA%A4%2Csocial%20media%20illustration%20style%2Chigh%20quality%2Cprofessional%20digital%20art%20high%20quality?width=1024&height=1024&nologo=true&enhance=true
✅ 图片生成成功 (尝试 1/3)
  ✅ 图片生成成功
✅ 配图生成完成
📄 开始编译Markdown...
✅ Markdown编译完成
🚀 开始发布文章...
📤 正在发布文章: Python处理Excel文件：让数据分析更高效
🆔 文章ID: 207ef96c-6e6b-4339-8680-9215b6b78131
✅ 文章发布成功！

从开始理解主题->生成大纲->生成段落内容->图片->生成markdown->发布文章，整个过程都是自动完成的，我们可以看下最终发布到wemark平台中长啥样。

效果展示

以下都是来自wemark平台中此工作流自动生成上传的文章截图：

小结

惊艳到大家了吗，反正是让我如沐春风，耳目一新啊。从此解放双手，让你的热情和精力真正发挥在创造方面，其他的交给AI工作流吧！

最后想要学习的欢迎关注公众号并且加微信，需要资料的或者想要一起成长的欢迎打扰我。

历史大模型系列文章：