# 2026 研发效能平台

## 1. 代码托管
### 1.1 GitLab 自建
### 1.2 镜像同步
## 2. CI/CD
### 2.1 流水线引擎
- Argo Workflows
- Tekton
### 2.2 制品库
- Harbor
- Nexus
## 3. 可观测性
### 3.1 Metrics
- Prometheus
- VictoriaMetrics
### 3.2 Tracing
- Jaeger
- Tempo
## 4. AI 辅助
### 4.1 代码补全
### 4.2 图表生成

openapi: 3.1.0
info:
  title: Order Service
  version: 1.0.0
paths:
  /orders:
    post:
      x-upstream: [auth-service, user-service, stock-service]
      x-downstream: [kafka, order-db]

你是 Mermaid 语法专家。请将以下描述转换为 Mermaid flowchart。

约束：
1. 只输出 ```mermaid 代码块，不要任何解释
2. 节点 ID 使用英文驼峰，标签使用中文
3. 条件分支必须使用 {} 菱形节点
4. 异常路径用虚线 -.-> 表示
5. 并行流程使用 subgraph 包裹
6. 每个节点标签不超过 12 个汉字
7. 必须包含 START 和 END 节点（如果存在明确起止）

描述：
{用户输入}

Mermaid 语法自检清单：
- [ ] 所有 subgraph 有 end
- [ ] 特殊字符（括号、冒号）用引号包裹
- [ ] 没有使用保留字（end, graph, subgraph 作为节点 ID）

repo/
├── docs/
│   └── architecture/
│       ├── system-overview.mmd
│       ├── order-flow.mmd
│       ├── data-model.mmd
│       └── generated/          # CI 生成，加入 .gitignore
│           ├── system-overview.svg
│           └── system-overview.png
├── .github/
│   └── workflows/
│       └── render-diagrams.yml
└── scripts/
    └── render_all.py

name: Render Diagrams

on:
  pull_request:
    paths:
      - 'docs/architecture/**.mmd'
  push:
    branches: [main]
    paths:
      - 'docs/architecture/**.mmd'

jobs:
  render:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Setup Node
        uses: actions/setup-node@v4
        with:
          node-version: '20'

      - name: Install mermaid-cli
        run: npm install -g @mermaid-js/mermaid-cli@11

      - name: Render all .mmd to SVG
        run: |
          mkdir -p docs/architecture/generated
          for f in docs/architecture/*.mmd; do
            name=$(basename "$f" .mmd)
            mmdc -i "$f" -o "docs/architecture/generated/${name}.svg" -t neutral -b transparent
            mmdc -i "$f" -o "docs/architecture/generated/${name}.png" -t neutral -s 3
          done

      - name: Upload artifacts
        uses: actions/upload-artifact@v4
        with:
          name: diagrams
          path: docs/architecture/generated/

      - name: Comment PR with preview
        if: github.event_name == 'pull_request'
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            const files = fs.readdirSync('docs/architecture/generated')
              .filter(f => f.endsWith('.png'));
            let body = '## 📊 图表预览\n\n';
            for (const f of files) {
              body += `### ${f}\n\n`;
            }
            github.rest.issues.createComment({
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
              body
            });

import { parseMermaidToExcalidraw } from "@excalidraw/mermaid-to-excalidraw";
import { convertToExcalidrawElements } from "@excalidraw/excalidraw";

const mermaidSource = `
flowchart TD
    A[用户] --> B[网关]
    B --> C[订单服务]
    C --> D[(数据库)]
`;

const { elements } = await parseMermaidToExcalidraw(mermaidSource);
const excalidrawElements = convertToExcalidrawElements(elements);
// 保存为 .excalidraw 文件

#!/bin/bash
set -euo pipefail

INPUT_DIR="${1:-./docs/architecture}"
OUTPUT_DIR="${2:-./docs/architecture/generated}"
FORMATS=("svg" "png" "pdf")

mkdir -p "$OUTPUT_DIR"

for mmd in "$INPUT_DIR"/*.mmd; do
  name=$(basename "$mmd" .mmd)
  echo "Processing: $name"

  for fmt in "${FORMATS[@]}"; do
    case "$fmt" in
      svg)
        mmdc -i "$mmd" -o "$OUTPUT_DIR/$name.svg" -t neutral -b transparent
        ;;
      png)
        mmdc -i "$mmd" -o "$OUTPUT_DIR/$name.png" -t neutral -s 3 -b white
        ;;
      pdf)
        mmdc -i "$mmd" -o "$OUTPUT_DIR/$name.pdf" -t neutral -b white
        ;;
    esac
  done
done

echo "Done. Output in $OUTPUT_DIR"

#!/usr/bin/env python3
"""
AI Diagram Pipeline
读取 Markdown 大纲 → LLM 生成 Mermaid → 校验 → 渲染 → 输出报告
Author: 2026 Architecture Team
"""

import os
import re
import json
import subprocess
import hashlib
from pathlib import Path
from typing import Literal
from dataclasses import dataclass, field

import httpx
from pydantic import BaseModel, Field
from tenacity import retry, stop_after_attempt, wait_exponential
from rich.console import Console
from rich.table import Table
from rich.progress import track

console = Console()

# ============ 配置 ============
LLM_BASE_URL = os.getenv("LLM_BASE_URL", "https://api.openai.com/v1")
LLM_API_KEY = os.getenv("LLM_API_KEY", "")
LLM_MODEL = os.getenv("LLM_MODEL", "gpt-4o-2026")

INPUT_DIR = Path("./docs/architecture")
OUTPUT_DIR = Path("./docs/architecture/generated")
CACHE_DIR = Path("./.diagram-cache")

OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
CACHE_DIR.mkdir(parents=True, exist_ok=True)


# ============ 数据模型 ============
class DiagramSpec(BaseModel):
    """LLM 返回的结构化图表规格"""
    diagram_type: Literal["flowchart", "sequenceDiagram", "mindmap", "classDiagram", "erDiagram", "gantt"]
    mermaid_code: str = Field(..., description="完整的 Mermaid 源码")
    title: str = Field(..., description="图表标题")
    reasoning: str = Field("", description="生成理由")


@dataclass
class RenderResult:
    source_file: str
    diagram_type: str
    title: str
    mermaid_code: str
    svg_path: str = ""
    png_path: str = ""
    success: bool = False
    error: str = ""
    cached: bool = False


# ============ LLM 调用 ============
SYSTEM_PROMPT = """你是 Mermaid 语法专家。根据用户提供的 Markdown 大纲，生成最合适的 Mermaid 图表。

规则：
1. 如果大纲是层级结构（# ## ###），使用 mindmap
2. 如果大纲描述流程步骤，使用 flowchart TD
3. 如果大纲描述系统交互，使用 sequenceDiagram
4. 节点 ID 用英文驼峰，标签用中文
5. 条件分支用 {} 菱形
6. 异常路径用 -.-> 虚线
7. 只返回 JSON，格式：{"diagram_type": "...", "mermaid_code": "...", "title": "...", "reasoning": "..."}
8. mermaid_code 不要包含 ```mermaid 包裹

Mermaid 语法自检：
- subgraph 必须有 end
- 特殊字符用引号包裹
- 节点 ID 不能用保留字（end, graph, subgraph, class）
"""


@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=2, max=10))
def call_llm(markdown_content: str) -> DiagramSpec:
    """调用 LLM 生成图表规格"""
    with httpx.Client(timeout=120) as client:
        resp = client.post(
            f"{LLM_BASE_URL}/chat/completions",
            headers={
                "Authorization": f"Bearer {LLM_API_KEY}",
                "Content-Type": "application/json",
            },
            json={
                "model": LLM_MODEL,
                "messages": [
                    {"role": "system", "content": SYSTEM_PROMPT},
                    {"role": "user", "content": markdown_content},
                ],
                "response_format": {"type": "json_object"},
                "temperature": 0.2,
            },
        )
        resp.raise_for_status()
        content = resp.json()["choices"][0]["message"]["content"]
        return DiagramSpec.model_validate_json(content)


# ============ 语法校验 ============
def validate_mermaid(code: str) -> tuple[bool, str]:
    """用 mmdc 校验 Mermaid 语法"""
    tmp = CACHE_DIR / "validate.mmd"
    tmp.write_text(code, encoding="utf-8")
    try:
        result = subprocess.run(
            ["mmdc", "-i", str(tmp), "-o", str(CACHE_DIR / "validate.svg")],
            capture_output=True,
            text=True,
            timeout=30,
        )
        if result.returncode == 0:
            return True, ""
        return False, result.stderr[:500]
    except subprocess.TimeoutExpired:
        return False, "Validation timeout"
    except FileNotFoundError:
        return False, "mmdc not found. Install: npm i -g @mermaid-js/mermaid-cli"


# ============ 渲染 ============
def render_mermaid(code: str, name: str) -> tuple[str, str, str]:
    """渲染 Mermaid 为 SVG 和 PNG"""
    mmd_file = OUTPUT_DIR / f"{name}.mmd"
    svg_file = OUTPUT_DIR / f"{name}.svg"
    png_file = OUTPUT_DIR / f"{name}.png"

    mmd_file.write_text(code, encoding="utf-8")

    errors = []
    for fmt, out in [("svg", svg_file), ("png", png_file)]:
        cmd = ["mmdc", "-i", str(mmd_file), "-o", str(out), "-t", "neutral"]
        if fmt == "png":
            cmd += ["-s", "3", "-b", "white"]
        else:
            cmd += ["-b", "transparent"]

        result = subprocess.run(cmd, capture_output=True, text=True, timeout=60)
        if result.returncode != 0:
            errors.append(f"{fmt}: {result.stderr[:200]}")

    return (
        str(svg_file) if svg_file.exists() else "",
        str(png_file) if png_file.exists() else "",
        "; ".join(errors),
    )


# ============ 缓存 ============
def content_hash(content: str) -> str:
    return hashlib.sha256(content.encode()).hexdigest()[:16]


def get_cached(content_hash_val: str) -> DiagramSpec | None: