--- name: mindmap description: 生成交互式思维导图，支持从文本/摘要构建层级结构，直接输出 HTML、PNG、JPG、SVG、PDF、XMind 格式。所有图片格式仅需 pillow（pip install pillow），无需任何系统级C库。当用户要求生成思维导图、脑图、可视化结构图时调用。 user-invocable: true metadata: {"openclaw": {"requires": {"bins": ["python3"], "pip": ["pillow"]}}} --- # Mind Map Generator — 思维导图生成器 ## 触发时机 - 用户说"生成思维导图"、"画脑图"、"做一个思维导图" - 用户总结文章/内容后说"帮我做成思维导图" - 用户说"把这些内容可视化"、"帮我梳理结构" - 用户说"导出为 XMind / PNG / JPG / PDF"等格式 --- ## 输出格式与依赖 **所有平台（Windows / macOS / Linux）统一依赖，只需 `pip install pillow`：** | 格式 | 说明 | 依赖 | |------|------|------| | `html` | **默认**。交互式网页，支持拖拽/折叠/缩放/右键编辑/8种布局切换 | 无 | | `svg` | 矢量图，无限缩放不失真 | 无 | | `xmind` | XMind 8 / 2020+ 格式，可继续编辑 | 无 | | `png` | 高清位图，适合嵌入文档/PPT/分享 | pillow | | `jpg` | JPEG 位图，体积小 | pillow | | `pdf` | PDF 文档 | pillow | > **不需要安装任何系统级 C 库。** 如果 Pillow 未安装，脚本会**自动执行 `pip install pillow`**。 ### 首次安装（一行命令） ```bash pip install pillow ``` --- ## 推荐的多格式输出策略用户要求生成思维导图时，**默认同时输出 HTML + PNG + XMind 三种格式**，除非用户明确指定只要某种格式： ```bash DATA='{ "central": "核心洞见", "branches": [ {"label": "🔬 维度1", "color": "#4A90D9", "children": ["证据A", "证据B"]} ] }' # 1. HTML（交互式，浏览器查看/编辑） python3 {baseDir}/generate_mindmap.py \ --title "标题" --format html \ --output ~/.openclaw/workspace/标题.html --data "$DATA" # 2. PNG（高清图片，直接分享） python3 {baseDir}/generate_mindmap.py \ --title "标题" --format png --scale 2.0 \ --output ~/.openclaw/workspace/标题.png --data "$DATA" # 3. XMind（专业思维导图软件中编辑） python3 {baseDir}/generate_mindmap.py \ --title "标题" --format xmind \ --output ~/.openclaw/workspace/标题.xmind --data "$DATA" ``` 参数说明： - `--format html/png/jpg/svg/pdf/xmind`（默认 html） - `--output` 输出路径（可省略，默认保存到 `~/.openclaw/workspace/`） - `--scale` PNG/JPG/PDF 的像素密度（默认 2.0，建议 1.0–3.0） - `--quality` JPG 质量 1–100（默认 92） --- ## 认知科学基础：为什么这样做思维导图以下原则基于四大认知理论，指导 AI 生成高质量思维导图： ### 放射性思考（Buzan）— 结构原则 Tony Buzan 提出思维导图模拟大脑神经元的放射状结构：从中心向外发散，每条分支是一个独立的联想链。 **对生成的要求：** 中心节点是唯一的焦点，所有分支从它辐射而出，不存在"并列的两个中心"。 ### 双重编码理论（Paivio）— 记忆原则 Allan Paivio 发现：同时通过语言通道和视觉通道编码的信息，记忆率提升约 32%。具体词（可想象的）比抽象词记忆效果好得多。 **对生成的要求：** 1. **每个主分支必须带 emoji**（激活视觉通道）—— 如 `"🔬 技术演进"` 而非 `"技术演进"` 2. **优先使用具体词**而非抽象概括 —— 如 `"训练成本降至 6 万"` 而非 `"成本下降"` 3. **颜色即编码** —— 每个分支的颜色应与语义一致，同色系 = 同语义域 ### 认知负荷与米勒定律（Miller）— 容量原则工作记忆同时处理的信息块为 7±2 个。Beel & Langer 对 19,379 张思维导图的实证研究发现：典型思维导图平均 31 个节点，每节点 1-3 个词。 **对生成的硬约束：** - 主分支 **3–6 个**（不超过 7） - 每个分支的子节点 **2–5 个** - 每个节点 **4–12 字**，叶节点不超过 15 字 - **全图总节点数 ≤ 40 个**（超过则拆分为多张图或合并子节点） - 全图最多 **4 层**深度 ### 语义网络（Quillian & Collins）— 关联原则知识在大脑中以语义网络形式存储：概念是节点，关系是连线。**节点间的距离代表语义距离**——相关的概念应该在视觉上靠近。 **对生成的要求：** 语义相近的子节点应聚合在同一个分支下（聚类原则），而不是按原文出现顺序排列。 --- ## 第一步：确定中心节点（最重要）中心节点决定整张图的质量。它是**核心洞见**，不是话题标签。 **错误示范（话题标签）：** - ❌ "AI 发展趋势" - ❌ "产品发布计划" - ❌ "Python 学习" **正确示范（核心洞见）：** - ✅ "AI 正从工具变为协作者" - ✅ "产品成功 = 用户价值 × 执行速度" - ✅ "Python 的价值在于生态而非语法" **判断方法：** 1. 问自己"关于这个内容，最重要的一件事是什么？" 答案就是中心节点。 2. 如果中心节点换成另一个话题，现有分支还能成立 → 说明太泛，需要更精确。 --- ## 第二步：设计主分支（维度而非章节）主分支是"理解中心洞见的独立视角"，不是原文的章节目录。 ### 按内容类型选择维度框架 | 内容类型 | 分支维度框架 | 推荐布局 | |----------|-------------|----------| | 分析/研究类 | 是什么 · 为什么 · 怎么做 · 结果如何 | ⇆ 左右均衡 | | 问题解决类 | 根因 · 影响 · 方案 · 评估 | 🐟 鱼骨图 | | 学习知识类 | 核心概念 · 运作机制 · 适用场景 · 常见误区 | ⇆ 左右均衡 | | 产品/项目类 | 目标 · 策略 · 执行 · 风险 | → 树形 | | 比较分析类 | 相同点 · 差异点 · 各自优势 · 选择建议 | ⇆ 左右均衡 | | 叙事/报告类 | 背景 · 发现 · 影响 · 行动 | → 树形 | | 历史/阶段/流程 | 按时间或步骤顺序排列 | ⏩ 时间线 | | 概念分解/归类 | 整体 → 部分1 + 部分2 + ... | } 括弧图 | | **头脑风暴/创意发散** | **自由联想，不预设框架** | **✶ 辐射或 ⚡ 力导向** | ### 分支设计原则 **原则 1：分支之间必须相互独立**（语义网络的聚类原则）每个分支回答关于中心节点的一个不同问题。语义相近的内容聚合在同一分支下。 **原则 2：所有分支必须共同服务于中心节点** 反问："去掉这个分支，对理解中心节点有损失吗？" 无损失 → 该分支冗余，删除。 **原则 3：分支数量 3–6 个**（米勒定律）不以"原文有几节"为准。超过 7 个分支时必须合并。 **原则 4：同级分支的抽象层次必须一致** ❌ 错误：同级出现"技术原理"（抽象）和"GPT-4"（具体实例） ✅ 正确：同级都是维度（"技术路线"、"商业模式"、"社会影响"） ### 每个主分支必须带 emoji（双重编码） emoji 的作用是激活视觉编码通道，使分支在记忆中形成"图像锚点"。选择规则： - emoji 应直观反映该维度的语义核心 - 同一张图内 emoji 不重复 - 置于 label 最前面：`"🔬 技术演进"` `"💰 商业模式"` `"⚠️ 风险挑战"` | 语义域 | 推荐 emoji | |--------|-----------| | 技术/科学 | 🔬 🧪 ⚙️ 🔧 💻 | | 商业/金融 | 💰 📈 🏦 💼 🎯 | | 人物/用户 | 👤 👥 🧑‍💻 🎓 | | 风险/问题 | ⚠️ 🚧 ❗ 🔴 | | 趋势/时间 | 📅 🔮 📊 🕐 | | 成果/价值 | ✅ 🏆 💡 ⭐ | | 流程/步骤 | 🔄 📋 🗺️ 🛤️ | | 背景/历史 | 📖 🏛️ 🌍 🗂️ | --- ## 第三步：填充子节点（证据而非子话题）子节点不是"这个分支下还有哪些话题"，而是"**让这个分支成立的具体依据**"。 ### 优先使用具体词（双重编码）具体的、可想象的词比抽象词记忆效果好 1.5-2 倍（Paivio 实验数据）。 | 子节点类型 | 示例 | 作用 | |------------|------|------| | 具体机制 | "注意力机制使模型聚焦关键词" | 解释"为什么" | | 量化数据 | "GPT-4 训练成本超 1 亿美元" | 使声明可信 | | 典型案例 | "Copilot 使代码效率提升 55%" | 使抽象具体 | | 对比参照 | "vs 传统搜索：主动生成 vs 被动检索" | 揭示差异 | | 行动指南 | "先用小数据集验证再扩规模" | 指导实践 | | 关键限制 | "幻觉率随任务复杂度非线性上升" | 防止误用 | ### 层级深度规则 ``` 中心节点（1 个） └── 主分支（3–6 个） ← 理解维度，带 emoji，抽象词组 └── 子节点（2–5 个） ← 支撑证据，具体词组或短句 └── 叶节点（可选，1–3 个） ← 最具体的事实/数据/步骤 ``` - 全图最多 4 层，**总节点 ≤ 40 个** - 每个主分支至少 2 个子节点（1 个说明未充分挖掘） - 叶节点不超过 3 个（太多说明需要再拆一层） --- ## 第四步：节点文字规范 | 规则 | 正确 | 错误 | |------|------|------| | 使用原文关键词，不过度意译 | "联邦学习" | "一种保护数据的分布式方法" | | 动词短语比名词堆砌更有力 | "降低推理延迟 40%" | "推理延迟优化相关内容" | | 长度 4–12 字 | "本地训练不传原始数据" | 过长的完整句子 | | 并列内容拆成独立节点 | 三个节点："医疗" / "教育" / "金融" | 一个节点："医疗教育金融" | | 越深层越具体（认知负荷递减） | 叶节点写具体数字/步骤/案例 | 叶节点写抽象标签 | --- ## 第五步：质量自查（生成后必检）生成 JSON 后，逐条检查，不合格立即修改： **容量检查（米勒定律）：** - [ ] 总节点数是否 ≤ 40？（超过则合并或拆分） - [ ] 主分支是否 3–6 个？ - [ ] 每个节点是否 ≤ 12 字？ **必要性检查（认知负荷）：** - [ ] 删去任意一个主分支，对理解中心节点的损失是否明显？（不明显 → 删除该分支） - [ ] 每个子节点是否使其父分支更可信/更具体？（否 → 删除或替换） - [ ] 是否存在"正确的废话"节点？（如"需要进一步研究" → 删除） **双重编码检查：** - [ ] 每个主分支是否带了 emoji？ - [ ] 叶节点是否用了具体的、可想象的词？（抽象标签 → 替换为具体事实） - [ ] 颜色是否与语义一致？（同类维度同色系） **结构检查（放射性思考）：** - [ ] 中心节点是核心洞见还是话题标签？ - [ ] 同级分支的抽象层次是否一致？ - [ ] 是否有只含 1 个子节点的分支？（合并或补充） --- ## 第六步：构造 JSON 并执行 > **⚠️ 强制规则：每个主分支的 label 必须以 emoji 开头。** > 格式：`"label": "🔬 技术演进"` —— emoji + 空格 + 文字。 > 这是基于双重编码理论的硬性要求，不可省略。 ```json { "central": "核心洞见（不超过 15 字）", "branches": [ { "label": "🔬 维度1（emoji 必须有）", "color": "#4A90D9", "children": [ "具体证据A（可想象的词）", "量化数据B", { "label": "需要展开的证据C", "children": ["最具体的事实1", "最具体的事实2"] } ] } ] } ``` **颜色分配建议（按分支语义选色）：** | 颜色 | 适用维度 | 搭配 emoji | |------|----------|-----------| | `#4A90D9` 蓝 | 机制/原理/方法论 | 🔬 ⚙️ 🧪 | | `#27AE60` 绿 | 成果/价值/应用 | ✅ 💡 🏆 | | `#E86C3A` 橙 | 问题/挑战/风险 | ⚠️ 🚧 ❗ | | `#9B59B6` 紫 | 背景/趋势/上下文 | 📖 🔮 🌍 | | `#F39C12` 黄 | 资源/工具/要素 | 🔧 📦 💼 | | `#1ABC9C` 青 | 流程/步骤/路径 | 🔄 🛤️ 📋 | | `#E74C3C` 红 | 警告/限制/禁忌 | 🔴 ⛔ 🚫 | 脚本路径：`{baseDir}/generate_mindmap.py` --- ## 完整示例 ```bash DATA='{ "central": "技术壁垒正被成本竞争取代", "branches": [ { "label": "🔬 技术门槛变化", "color": "#4A90D9", "children": [ {"label": "训练成本下降", "children": ["GPT-3 成本 460万刀", "同能力模型降至 6 万"]}, "开源模型追平闭源", "微调替代全量训练" ] }, { "label": "💰 主流盈利路径", "color": "#27AE60", "children": [ "API 按 token 计费", {"label": "垂直行业定制", "children": ["医疗合规要求高溢价", "法律文档审核替代人工"]}, "订阅制 Pro 用户" ] }, { "label": "⚔️ 竞争驱动力", "color": "#E86C3A", "children": ["价格战压缩利润空间", "算力即竞争壁垒", "数据飞轮强者愈强"] }, { "label": "📜 监管不确定性", "color": "#9B59B6", "children": [ {"label": "欧美监管分歧", "children": ["EU AI Act 强制备案", "美国行业自律为主"]}, "中国实名制与内容审核", "合规成本影响中小玩家" ] } ] }' python3 {baseDir}/generate_mindmap.py --title "大模型商业化" --format html --output ~/.openclaw/workspace/大模型商业化.html --data "$DATA" python3 {baseDir}/generate_mindmap.py --title "大模型商业化" --format png --output ~/.openclaw/workspace/大模型商业化.png --data "$DATA" python3 {baseDir}/generate_mindmap.py --title "大模型商业化" --format xmind --output ~/.openclaw/workspace/大模型商业化.xmind --data "$DATA" ``` **执行成功后告知用户（必须包含完整的绝对路径，方便用户查找文件）：** > 思维导图已生成三种格式，保存在以下位置： > - **HTML**：`<绝对路径>/大模型商业化.html` —— 用浏览器打开，支持交互查看、编辑节点、切换 8 种布局 > - **PNG**：`<绝对路径>/大模型商业化.png` —— 高清图片，可直接嵌入文档或分享 > - **XMind**：`<绝对路径>/大模型商业化.xmind` —— 可在 XMind 软件中打开继续编辑 > > 其中 `<绝对路径>` 替换为脚本实际输出的路径（从脚本的 stdout 中获取，格式为 `[mindmap] ✅ HTML → /完整/路径/文件名.html`）。 --- ## 内置示例文件 `{baseDir}/examples/` 目录中包含可直接打开的示例： | 文件 | 说明 | |------|------| | `examples/ai_trends.html` | AI 发展趋势（5 主分支，3 层嵌套） | | `examples/product_launch.html` | 产品发布计划 | | `examples/python_learning.html` | Python 学习路径 | FILE:generate_mindmap.py #!/usr/bin/env python3 """ generate_mindmap.py — OpenClaw Mind Map Generator v5 Layout : Balanced left-right tree (XMind style) - branches split evenly left / right - each side laid out as a vertical tree - elbow connectors (horizontal → vertical) - smooth Bezier from root to first-level branches Features: - Click to collapse / expand - Drag node to reposition; right/bottom edge handles to resize - Pan, zoom toward cursor, reset view - In-browser export: SVG / PNG / JPG / PDF / XMind - Python-side export: --format html|svg|png|jpg|pdf|xmind Usage: python3 generate_mindmap.py --title "Topic" --output /tmp/map.html \\ --data '{"central":"...","branches":[...]}' [--format html] """ import argparse, json, math, sys, zipfile, io, uuid, os, platform, subprocess from datetime import datetime from pathlib import Path def _ensure_pillow(): """Try to import Pillow; if missing, auto-install via pip and retry.""" try: from PIL import Image # noqa: F401 return True except ImportError: pass print("[mindmap] Pillow not found, installing automatically …", file=sys.stderr) try: subprocess.check_call( [sys.executable, "-m", "pip", "install", "pillow", "--quiet", "--disable-pip-version-check", "--break-system-packages"], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, ) except (subprocess.CalledProcessError, FileNotFoundError): # Retry without --break-system-packages (older pip) try: subprocess.check_call( [sys.executable, "-m", "pip", "install", "pillow", "--quiet", "--disable-pip-version-check"], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, ) except Exception: print("[mindmap] ⚠ Auto-install failed. Please run manually:", file=sys.stderr) print(" pip install pillow", file=sys.stderr) return False try: from PIL import Image # noqa: F401 print("[mindmap] ✅ Pillow installed successfully.", file=sys.stderr) return True except ImportError: print("[mindmap] ⚠ Install succeeded but import still fails.", file=sys.stderr) return False # ───────────────────────────────────────────────────────────────────────────── # Cross-platform path helpers # ───────────────────────────────────────────────────────────────────────────── def resolve_output(raw_path: str, fmt: str) -> str: """Expand ~, $HOME, %USERPROFILE%, create parent dirs, fix extension. If the path is just a filename without directory (e.g. 'mindmap.png'), it is placed under ~/.openclaw/workspace/ instead of the current dir. """ p = Path(os.path.expandvars(os.path.expanduser(raw_path))) ext_map = {"html":".html","svg":".svg","png":".png", "jpg":".jpg","pdf":".pdf","xmind":".xmind"} expected = ext_map.get(fmt, "") if expected and p.suffix.lower() != expected: p = p.with_suffix(expected) # If only a bare filename was given (no directory), use default workspace if p.parent == Path(".") or str(p.parent) == ".": workspace = Path.home() / ".openclaw" / "workspace" workspace.mkdir(parents=True, exist_ok=True) p = workspace / p.name p.parent.mkdir(parents=True, exist_ok=True) return str(p) def default_output(fmt: str) -> str: """Return a sensible default output path for the current OS. Default: ~/.openclaw/workspace/ (cross-platform, auto-adapts to username) - Windows: C:\\Users\\\\.openclaw\\workspace\\ - macOS: /Users//.openclaw/workspace/ - Linux: /home//.openclaw/workspace/ """ ext_map = {"html":".html","svg":".svg","png":".png", "jpg":".jpg","pdf":".pdf","xmind":".xmind"} ts = datetime.now().strftime("%Y%m%d_%H%M%S") ext = ext_map.get(fmt, ".html") name = f"mindmap_{ts}{ext}" workspace = Path.home() / ".openclaw" / "workspace" workspace.mkdir(parents=True, exist_ok=True) return str(workspace / name) DEFAULT_COLORS = [ "#4A90D9","#E86C3A","#27AE60","#9B59B6", "#E74C3C","#F39C12","#1ABC9C","#E91E63", "#00BCD4","#8BC34A", ] # ── Visual config per depth ──────────────────────────────────────────────────── # depth 0 = central, 1 = branch, 2 = leaf, 3 = sub-leaf CFG = [ dict(h=48, fs=16, fw="bold", rx=12, px=32, min_w=180), dict(h=38, fs=13, fw="bold", rx= 8, px=24, min_w=110), dict(h=30, fs=12, fw="normal", rx= 6, px=18, min_w= 80), dict(h=26, fs=11, fw="normal", rx= 5, px=16, min_w= 72), ] # Horizontal gap between node right-edge and children column H_GAP = [0, 64, 48, 40] # Vertical gap between sibling nodes V_GAP = [0, 20, 12, 8] FONT_STACK = "PingFang SC,Hiragino Sans GB,Microsoft YaHei,Segoe UI,Arial,sans-serif" # ───────────────────────────────────────────────────────────────────────────── # CLI # ───────────────────────────────────────────────────────────────────────────── def parse_args(): p = argparse.ArgumentParser( description="OpenClaw Mind Map Generator — cross-platform (macOS / Linux / Windows)" ) p.add_argument("--title", required=True, help="Mind map title") p.add_argument("--output", default=None, help="Output file path (default: ~/.openclaw/workspace/mindmap_.)") p.add_argument("--data", required=True, help="JSON string describing the mind map structure") p.add_argument("--format", default="html", choices=["html","svg","png","jpg","pdf","xmind"], help="Output format (default: html)") p.add_argument("--scale", type=float, default=2.0, help="Pixel density for PNG/JPG/PDF (default: 2.0)") p.add_argument("--quality", type=int, default=92, help="JPEG quality 1-100 (default: 92)") return p.parse_args() # ───────────────────────────────────────────────────────────────────────────── # Tree helpers # ───────────────────────────────────────────────────────────────────────────── def normalize_node(node): if isinstance(node, str): return {"label": node, "children": []} if isinstance(node, dict): node.setdefault("children", []) node["children"] = [normalize_node(c) for c in node["children"]] return node return {"label": str(node), "children": []} def build_tree(data): if "central" not in data: raise ValueError("JSON must contain a 'central' key.") branches = [] for i, b in enumerate(data.get("branches", [])): nb = normalize_node(b) nb.setdefault("color", DEFAULT_COLORS[i % len(DEFAULT_COLORS)]) # ── Auto-inject emoji if branch label doesn't start with one ── _auto_inject_emoji(nb, i) branches.append(nb) return {"central": data["central"], "branches": branches} # ── Emoji auto-injection (Dual-Coding Theory) ──────────────────────────── # If the AI didn't add emoji to branch labels, the script selects an emoji # by matching the label text against a semantic keyword dictionary. # This ensures the emoji visually represents the *meaning* of the branch, # not just its color. # Semantic keyword → emoji mapping (Chinese + English, ordered by specificity) # More specific keywords must come before general ones. _SEMANTIC_EMOJI = [ # ── 人物/角色 ── (["人物", "角色", "团队", "成员", "员工", "用户", "画像", "character", "people", "team", "user", "member", "staff", "player", "persona"], "👥"), (["作者", "作家", "创始人", "author", "writer", "founder", "creator"], "✍️"), (["领导", "管理层", "CEO", "leader", "management", "executive"], "👔"), (["客户", "顾客", "消费者", "受众", "customer", "client", "consumer", "audience"], "🧑‍💼"), (["合作", "伙伴", "联盟", "协作", "partner", "collaborat", "alliance", "cooperat"], "🤝"), # ── 时间/历史 ── (["历史", "背景", "起源", "演变", "发展史", "history", "background", "origin", "evolution", "heritage"], "🏛️"), (["时间", "阶段", "时期", "年代", "时间线", "timeline", "period", "phase", "era", "date", "schedule"], "📅"), (["未来", "趋势", "展望", "预测", "前景", "future", "trend", "forecast", "outlook", "vision"], "🔮"), # ── 技术/科学 ── (["技术", "科技", "研发", "算法", "tech", "technology", "algorithm", "engineering", "R&D"], "🔬"), (["AI", "人工智能", "机器学习", "深度学习", "模型", "artificial intelligence", "machine learning", "neural", "GPT", "LLM"], "🤖"), (["代码", "编程", "开发", "软件", "code", "programming", "software", "develop", "API", "debug"], "💻"), (["数据", "数据库", "统计", "指标", "data", "database", "analytics", "statistics", "metric", "KPI"], "📊"), (["网络", "互联网", "通信", "社交媒体", "社交", "social media", "network", "internet", "communication", "web", "online"], "🌐"), (["安全", "加密", "隐私", "防护", "security", "encryption", "privacy", "cyber", "protect"], "🔒"), # ── 商业/金融 ── (["商业", "盈利", "收入", "营收", "利润", "变现", "business", "revenue", "profit", "income", "monetiz"], "💰"), (["市场", "营销", "推广", "品牌", "口碑", "传播", "market", "marketing", "brand", "promotion", "advertis", "PR"], "📈"), (["销售", "卖点", "转化", "获客", "增长", "拉新", "sale", "conver", "acquisition", "growth", "retain", "funnel"], "📊"), (["竞争", "对手", "竞品", "壁垒", "competi", "rival", "barrier"], "⚔️"), (["成本", "费用", "预算", "投资", "融资", "cost", "budget", "invest", "expense", "funding", "capital"], "💵"), (["价格", "定价", "收费", "报价", "price", "pricing", "fee", "charge", "quota"], "🏷️"), (["战略", "策略", "规划", "布局", "strategy", "strategic", "planning", "roadmap"], "🎯"), (["产品", "功能", "特性", "需求", "product", "feature", "function", "requirement", "spec"], "📦"), (["渠道", "分发", "分销", "平台", "channel", "distribut", "platform"], "🔗"), (["供应链", "供应", "采购", "库存", "procurement", "supply", "inventory", "sourcing", "vendor"], "🏭"), (["运营", "运维", "运作", "operation", "ops", "maintain"], "⚙️"), # ── 客户服务 ── (["服务", "客服", "售后", "保障", "支持", "service", "support", "after-sale", "warranty", "helpdesk"], "🎧"), (["体验", "满意度", "反馈", "口碑", "评价", "experience", "satisfaction", "feedback", "review", "NPS", "UX"], "⭐"), (["留存", "复购", "忠诚", "黏性", "retention", "loyalty", "repeat", "churn", "stickiness"], "🔁"), # ── 教育/学习 ── (["学习", "教育", "课程", "培训", "教学", "learn", "education", "course", "training", "teach", "study"], "🎓"), (["知识", "概念", "理论", "原理", "knowledge", "concept", "theory", "principle"], "📚"), (["考试", "测试", "评估", "评价", "评分", "exam", "test", "assessment", "evaluation", "grading"], "📝"), (["启蒙", "早教", "入门", "基础", "beginner", "fundamental", "basic", "introduct", "primer"], "🌟"), # ── 职业/求职 ── (["职业", "求职", "面试", "简历", "career", "job", "interview", "resume", "CV", "hiring", "recruit"], "💼"), (["薪资", "工资", "薪酬", "待遇", "salary", "wage", "compensation", "pay", "bonus"], "💳"), (["晋升", "升职", "成长", "发展", "职级", "promotion", "advancement", "career path", "growth"], "📶"), (["技能", "能力", "技巧", "skill", "ability", "competenc", "expertise", "proficienc"], "🎯"), # ── 文学/艺术 ── (["作品", "文学", "小说", "著作", "文章", "writing", "literature", "novel", "book", "article"], "📖"), (["艺术", "风格", "美学", "art", "style", "aesthetic"], "🎨"), (["音乐", "歌曲", "music", "song", "melody", "album"], "🎵"), (["电影", "视频", "影视", "film", "movie", "video", "cinema"], "🎬"), (["文化", "传统", "文明", "民俗", "culture", "tradition", "civilization", "folk"], "🌍"), (["摄影", "拍摄", "相机", "镜头", "photo", "camera", "lens", "shoot", "image"], "📷"), (["设计", "UI", "UX", "排版", "视觉", "design", "layout", "visual", "graphic", "typograph"], "🎨"), # ── 结构/组织 ── (["结构", "组织", "架构", "框架", "structure", "organization", "framework", "architect"], "🏗️"), (["流程", "步骤", "过程", "方法", "工艺", "process", "step", "procedure", "method", "workflow", "SOP"], "🔄"), (["分类", "类型", "类别", "种类", "category", "type", "classification", "kind", "taxonomy"], "🗂️"), (["情节", "故事", "叙事", "剧情", "plot", "story", "narrative", "chapter"], "📜"), (["标准", "规范", "验收", "准则", "质检", "standard", "specification", "criteria", "QA", "QC", "inspect", "quality"], "✅"), # ── 问题/风险 ── (["问题", "挑战", "困难", "障碍", "痛点", "problem", "challenge", "difficult", "obstacle", "issue", "pain point"], "⚠️"), (["风险", "危险", "威胁", "risk", "danger", "threat", "hazard"], "🚨"), (["限制", "缺点", "不足", "局限", "短板", "limitation", "disadvantage", "weakness", "restrict", "drawback"], "🚧"), (["监管", "法规", "合规", "政策", "法律", "regulat", "compliance", "policy", "law", "legal", "govern"], "⚖️"), # ── 成果/价值 ── (["成果", "成就", "成功", "价值", "优势", "亮点", "achievement", "success", "value", "advantage", "benefit", "highlight"], "🏆"), (["目标", "愿景", "使命", "goal", "objective", "mission", "OKR", "KR"], "🎯"), (["创新", "突破", "改进", "优化", "innovati", "breakthrough", "improve", "optimiz"], "💡"), (["影响", "效果", "作用", "成效", "impact", "effect", "influence", "outcome", "result"], "💫"), (["应用", "实践", "案例", "场景", "用途", "application", "practice", "case", "scenario", "use case", "usage"], "🛠️"), # ── 资源/工具 ── (["资源", "工具", "设备", "器材", "装备", "resource", "tool", "equipment", "instrument", "gear", "device"], "🔧"), (["材料", "原料", "物料", "素材", "material", "ingredient", "raw material", "substance", "fabric"], "📦"), (["环境", "生态", "自然", "绿色", "environment", "ecology", "nature", "climate", "green", "sustain"], "🌱"), # ── 健康/医疗 ── (["健康", "医疗", "疾病", "治疗", "症状", "health", "medical", "disease", "treatment", "clinical", "symptom", "diagnosis"], "🏥"), (["营养", "饮食", "膳食", "维生素", "nutrition", "diet", "vitamin", "supplement", "calorie"], "🥗"), (["疫苗", "免疫", "防疫", "vaccine", "immuniz", "prevention"], "💉"), (["健身", "运动", "锻炼", "体能", "fitness", "workout", "exercise", "gym", "training"], "💪"), (["减肥", "瘦身", "体重", "weight loss", "slim", "body fat"], "⚖️"), (["睡眠", "休息", "作息", "sleep", "rest", "insomnia", "nap"], "😴"), (["压力", "焦虑", "减压", "放松", "stress", "anxiety", "relax", "mindful", "meditation"], "🧘"), # ── 地理/旅游 ── (["地理", "地点", "位置", "区域", "国家", "geography", "location", "region", "country", "city"], "🗺️"), (["旅游", "旅行", "攻略", "出行", "度假", "travel", "trip", "tour", "vacation", "journey", "itinerary"], "✈️"), (["住宿", "酒店", "民宿", "hotel", "accommodation", "hostel", "lodging", "Airbnb"], "🏨"), (["景点", "景区", "名胜", "观光", "sight", "attraction", "landmark", "scenic", "destination"], "🏞️"), (["交通", "物流", "运输", "出行", "transport", "logistics", "shipping", "commut", "transit"], "🚚"), # ── 食物/烹饪 ── (["食物", "美食", "菜肴", "菜谱", "烹饪", "做菜", "food", "cuisine", "cooking", "recipe", "meal", "dish", "chef"], "🍽️"), # ── 家庭/育儿 ── (["育儿", "孩子", "儿童", "宝宝", "亲子", "parent", "child", "baby", "kid", "toddler", "nurtur"], "👶"), (["家庭", "家居", "家装", "居家", "family", "home", "household", "domestic"], "🏠"), (["装修", "装饰", "翻新", "改造", "renovati", "decorat", "interior", "remodel", "furnish"], "🏠"), # ── 宠物 ── (["宠物", "猫", "狗", "喂养", "兽医", "pet", "cat", "dog", "vet", "animal", "breed", "groom"], "🐾"), # ── 体育/运动 ── (["体育", "赛事", "联赛", "比赛", "竞技", "sport", "league", "competition", "athletic", "tournament", "championship"], "⚽"), # ── 法律/制度 ── (["制度", "规则", "条例", "法案", "条款", "system", "rule", "regulation", "act", "institution", "clause"], "📜"), # ── 情感/心理 ── (["情感", "心理", "性格", "认知", "行为", "personality", "emotion", "psychology", "mental", "character", "cognitive", "behavior"], "🧠"), # ── 建筑/空间 ── (["建筑", "空间", "场所", "场地", "building", "space", "place", "architecture", "venue", "facility"], "🏗️"), # ── 财务/理财 ── (["理财", "投资", "基金", "股票", "债券", "保险", "finance", "invest", "fund", "stock", "bond", "insurance", "portfolio"], "💹"), (["税", "税务", "纳税", "报税", "tax", "taxation", "fiscal"], "🧾"), (["退休", "养老", "pension", "retire", "annuity"], "🏖️"), (["资产", "财富", "净值", "asset", "wealth", "net worth", "estate", "property"], "🏦"), # ── 制造/生产 ── (["制造", "生产", "工厂", "产线", "manufactur", "production", "factory", "assembly", "fabricat"], "🏭"), (["包装", "封装", "标签", "packaging", "labeling", "wrapping", "container"], "📦"), (["检测", "测量", "检验", "校准", "detect", "measure", "inspect", "calibrat", "monitor"], "🔎"), (["配方", "研发", "recipe", "formul", "R&D", "develop"], "🧪"), # ── 概述/介绍/总结 ── (["概述", "简介", "总结", "综述", "overview", "introduction", "summary", "abstract", "brief", "recap"], "📋"), # ── 军事/战争 ── (["战役", "战争", "军事", "战斗", "作战", "攻防", "battle", "war", "military", "combat", "warfare", "campaign"], "⚔️"), # ── 思想/哲学 ── (["思想", "哲学", "主题", "观点", "理念", "意识形态", "thought", "philosophy", "theme", "ideology", "idea", "viewpoint"], "💭"), # ── 研究/学术 ── (["研究", "学术", "流派", "论文", "学科", "实验", "research", "academic", "paper", "discipline", "experiment", "scholar"], "🔍"), # ── 版本/变更 ── (["版本", "变更", "更新", "迭代", "修订", "version", "update", "revision", "iteration", "changelog"], "📄"), # ── 护理/日常 ── (["护理", "保养", "维护", "保健", "日常", "care", "maintenance", "routine", "daily", "upkeep", "hygiene"], "🧴"), # ── 沟通/表达 ── (["沟通", "表达", "演讲", "谈判", "对话", "communicat", "express", "speech", "negotiat", "dialog", "present"], "🗣️"), # ── 选择/推荐/对比 ── (["选择", "推荐", "对比", "评测", "排名", "选型", "choose", "recommend", "compare", "review", "rank", "select", "pick", "best"], "🔖"), # ── 指南/攻略/技巧 ── (["指南", "攻略", "技巧", "诀窍", "要点", "秘诀", "guide", "tip", "trick", "hack", "how-to", "tutorial", "cheat sheet"], "📌"), # ── 构图/光线 (摄影/视觉) ── (["构图", "光线", "色彩", "曝光", "composit", "lighting", "exposure", "color", "tone", "contrast"], "🖼️"), # ── 后期/编辑 ── (["后期", "编辑", "修图", "剪辑", "渲染", "edit", "post-process", "retouch", "render", "montage"], "✂️"), ] _FALLBACK_EMOJIS = ["📌", "📎", "🔹", "🔸", "▪️", "🔻", "🔺", "💠", "🔘", "📍"] def _has_emoji_prefix(text): """Check if text starts with an emoji (Unicode emoji range).""" if not text: return False cp = ord(text[0]) if cp >= 0x1F300: return True if 0x2600 <= cp <= 0x27BF: return True if 0x2300 <= cp <= 0x23FF: return True if 0xFE00 <= cp <= 0xFEFF: return True if 0x200D <= cp <= 0x200D: return True if len(text) > 1: cp2 = ord(text[1]) if cp2 >= 0x1F300 or cp2 == 0xFE0F or cp2 == 0x20E3: return True return False def _match_emoji_by_semantic(label): """Match an emoji by scanning the label against the keyword dictionary.""" text = label.lower() for keywords, emoji in _SEMANTIC_EMOJI: for kw in keywords: if kw.lower() in text: return emoji return None def _auto_inject_emoji(branch, index): """Add semantically matched emoji prefix to branch label if missing.""" label = branch.get("label", "") if _has_emoji_prefix(label): return # already has emoji # 1. Try semantic keyword matching emoji = _match_emoji_by_semantic(label) # 2. Fallback: use index-based neutral emoji if not emoji: emoji = _FALLBACK_EMOJIS[index % len(_FALLBACK_EMOJIS)] branch["label"] = f"{emoji} {label}" def measure_w(text, depth): c = CFG[min(depth, len(CFG)-1)] w = sum(c["fs"] * (0.92 if ord(ch) > 127 else 0.58) for ch in str(text)) return max(c["min_w"], w + c["px"] * 2) _nid = [0] def annotate(node, depth): node["_id"] = "n" + str(_nid[0]); _nid[0] += 1 node["_depth"] = depth node["_w"] = measure_w(node.get("label") or node.get("central", ""), depth) node["_h"] = CFG[min(depth, len(CFG)-1)]["h"] for ch in node.get("children", []): annotate(ch, depth + 1) for b in node.get("branches", []): annotate(b, 1) def flatten_tree(tree): result, q = [], [tree] while q: n = q.pop(0); result.append(n) for c in n.get("children", []) + n.get("branches", []): q.append(c) return result # ───────────────────────────────────────────────────────────────────────────── # LAYOUT — Balanced left-right tree # # Right side: branch node → children stacked vertically to its right # Left side: branch node → children stacked vertically to its left # (node x = -(branch_x + branch_w/2 + gap + child_w/2)) # # Positions are stored as node centres. # ───────────────────────────────────────────────────────────────────────────── def subtree_height(node): """Total vertical space this subtree needs (including V_GAP between siblings).""" d = node["_depth"] kids = node.get("children", []) vg = V_GAP[min(d, len(V_GAP)-1)] if not kids: return node["_h"] ch_total = sum(subtree_height(k) for k in kids) + vg * (len(kids) - 1) return max(node["_h"], ch_total) def layout_subtree(node, cx, cy, side, positions): """ Place `node` centred at (cx, cy). Then place its children on `side` (+1 = right, -1 = left). """ positions[node["_id"]] = {"x": cx, "y": cy, "parent_id": positions.get(node["_id"], {}).get("parent_id")} kids = node.get("children", []) if not kids: return d = node["_depth"] vg = V_GAP[min(d+1, len(V_GAP)-1)] hg = H_GAP[min(d+1, len(H_GAP)-1)] # x-centre of children column child_w = max(k["_w"] for k in kids) child_cx = cx + side * (node["_w"] / 2 + hg + child_w / 2) # total height block of children heights = [subtree_height(k) for k in kids] total_h = sum(heights) + vg * (len(kids) - 1) # start y so the block is centred on cy cur_y = cy - total_h / 2 for kid, h in zip(kids, heights): kid_cy = cur_y + h / 2 positions[kid["_id"]] = {"x": child_cx, "y": kid_cy, "parent_id": node["_id"]} layout_subtree(kid, child_cx, kid_cy, side, positions) cur_y += h + vg def compute_layout(tree): positions = {} positions[tree["_id"]] = {"x": 0, "y": 0, "parent_id": None} branches = tree.get("branches", []) if not branches: return positions n = len(branches) n_right = math.ceil(n / 2) # right side gets ceiling n_left = n - n_right right_branches = branches[:n_right] left_branches = branches[n_right:] root_h = CFG[0]["h"] root_w = tree["_w"] def place_side(side_branches, side): """Stack branches vertically, centred on root y=0.""" vg = V_GAP[1] heights = [subtree_height(b) for b in side_branches] total_h = sum(heights) + vg * (len(heights) - 1) hg = H_GAP[1] # branch column x-centre branch_cx = side * (root_w / 2 + hg + (max(b["_w"] for b in side_branches) / 2 if side_branches else 0)) cur_y = -total_h / 2 for branch, h in zip(side_branches, heights): bcy = cur_y + h / 2 positions[branch["_id"]] = {"x": branch_cx, "y": bcy, "parent_id": tree["_id"]} layout_subtree(branch, branch_cx, bcy, side, positions) cur_y += h + vg place_side(right_branches, 1) place_side(left_branches, -1) return positions def bounding_box(positions, nodes, pad=60): xs, ys = [], [] for n in nodes: p = positions.get(n["_id"]) if not p: continue xs += [p["x"] - n["_w"]/2, p["x"] + n["_w"]/2] ys += [p["y"] - n["_h"]/2, p["y"] + n["_h"]/2] if not xs: return -pad, -pad, pad, pad return min(xs)-pad, min(ys)-pad, max(xs)+pad, max(ys)+pad def build_color_map(tree): cmap = {tree["_id"]: "#7c8cf8"} def fill(node, color): cmap[node["_id"]] = color for ch in node.get("children", []): fill(ch, color) for b in tree.get("branches", []): fill(b, b.get("color", "#888")) return cmap # ───────────────────────────────────────────────────────────────────────────── # Edge path helpers # ───────────────────────────────────────────────────────────────────────────── def edge_path(px, py, pw, cx, cy, cw, depth): """ All depths: smooth cubic Bezier from parent side-edge to child side-edge. Matches the HTML interactive version (edgePath0) — elegant S-curves instead of hard elbow connectors. """ if cx >= px: # right side x1, x2 = px + pw/2, cx - cw/2 else: # left side x1, x2 = px - pw/2, cx + cw/2 # Nearly horizontal alignment → straight line if abs(cy - py) < 3: return f"M{x1:.1f},{py:.1f} L{x2:.1f},{cy:.1f}" # Tension: depth-1 uses 0.5 (standard S-curve), deeper levels tighten slightly t = 0.5 if depth == 1 else 0.45 cpx = x1 + (x2 - x1) * t return f"M{x1:.1f},{py:.1f} C{cpx:.1f},{py:.1f} {cpx:.1f},{cy:.1f} {x2:.1f},{cy:.1f}" # ───────────────────────────────────────────────────────────────────────────── # Static SVG (Python-side export) # ───────────────────────────────────────────────────────────────────────────── def _xml(s): return str(s).replace("&","&").replace("<","<").replace(">",">").replace('"',""") def render_svg_static(tree, positions, include_xml_header=True): nodes = flatten_tree(tree) minx, miny, maxx, maxy = bounding_box(positions, nodes) W, H = maxx - minx, maxy - miny cmap = build_color_map(tree) L = [] if include_xml_header: L.append('') L.append(f'') return "\n".join(L) # ───────────────────────────────────────────────────────────────────────────── # XMind export # ───────────────────────────────────────────────────────────────────────────── def build_xmind(tree, title): """ Generate an .xmind file compatible with both XMind 8 and XMind 2020+. Includes content.xml (XMind 8 format) AND content.json (XMind 2020 format). """ def uid(): return uuid.uuid4().hex[:26] # ── content.json (XMind 2020+) ───────────────────────────────────────── def xnode_json(node): children = node.get("branches", []) + node.get("children", []) obj = {"id": uid(), "class": "topic", "title": node.get("label") or node.get("central", "")} if children: obj["children"] = {"attached": [xnode_json(c) for c in children]} if node.get("color"): obj["style"] = {"id": uid(), "properties": { "line-color": node["color"], "line-width": "2pt", "background-color": node["color"] + "33", "border-line-color": node["color"], "shape-class": "org.xmind.topicShape.roundedRect", }} return obj root_json = xnode_json(tree) root_json["structureClass"] = "org.xmind.ui.map.unbalanced" sheet_id = uid() content_json = [{ "id": sheet_id, "class": "sheet", "title": title, "rootTopic": root_json, "theme": {}, "extensions": [], }] # ── content.xml (XMind 8) ────────────────────────────────────────────── def _xe(s): return (str(s).replace("&", "&").replace("<", "<") .replace(">", ">").replace('"', """)) def xnode_xml(node, depth=0): children = node.get("branches", []) + node.get("children", []) label = node.get("label") or node.get("central", "") color = node.get("color", "") ind = " " * depth lines = [f'{ind}{_xe(label)}') if color: lines.append(f'{ind} ') if children: lines.append(f'{ind} ') lines.append(f'{ind} ') for child in children: lines.extend(xnode_xml(child, depth + 3).split("\n")) lines.append(f'{ind} ') lines.append(f'{ind} ') lines.append(f'{ind}') return "\n".join(lines) def build_styles(node, styles=None): if styles is None: styles = [] color = node.get("color", "") if color: styles.append( f' ' ) for c in node.get("branches", []) + node.get("children", []): build_styles(c, styles) return styles xml_sheet_id = uid() xml_root = xnode_xml(tree, 0) styles = build_styles(tree) content_xml = f""" {xml_root} {_xe(title)} """ styles_xml = ('\n' '\n' + "\n".join(styles) + "\n") # ── metadata ─────────────────────────────────────────────────────────── metadata = { "modifier": "", "created": datetime.now().strftime("%Y-%m-%dT%H:%M:%S.000+0000"), "creator": {"name": "OpenClaw MindMap", "version": "5.0", "platform": ""}, } manifest = {"file-entries": { "content.json": {"media-type": "application/json"}, "content.xml": {"media-type": "text/xml"}, "styles.xml": {"media-type": "text/xml"}, "metadata.json": {"media-type": "application/json"}, "manifest.json": {"media-type": "application/json"}, }} buf = io.BytesIO() with zipfile.ZipFile(buf, "w", zipfile.ZIP_DEFLATED) as zf: zf.writestr("manifest.json", json.dumps(manifest, ensure_ascii=False)) zf.writestr("content.json", json.dumps(content_json, ensure_ascii=False, indent=2)) zf.writestr("content.xml", content_xml.encode("utf-8")) zf.writestr("styles.xml", styles_xml.encode("utf-8")) zf.writestr("metadata.json", json.dumps(metadata, ensure_ascii=False, indent=2)) return buf.getvalue() # ───────────────────────────────────────────────────────────────────────────── # HTML template (raw string — no f-string, JS braces are literal) # ───────────────────────────────────────────────────────────────────────────── _HTML = r""" __TITLE__

🧠 TITLE

导出

视图

历史

布局

右键菜单 · Tab 添加子节点 · Del 删除 · Ctrl+Z 撤销

""" def render_html(title, js_data): return (_HTML .replace("__TITLE__", title) .replace("__RAW_JSON__", js_data) .replace("__TITLE_JSON__", json.dumps(title, ensure_ascii=False)) .replace("__DATE__", datetime.now().strftime("%Y-%m-%d %H:%M"))) # ═══════════════════════════════════════════════════════════════════════════════ # IMAGE / PDF EXPORT — Two backends, auto-selected by availability: # # 1. Playwright (best quality — renders full interactive HTML in Chromium) # 2. Pillow (pure Python — works everywhere, only needs `pip install pillow`) # # The script tries each backend in order and uses the first one that works. # On a typical Windows machine only Pillow is available, and that's fine. # Pillow is auto-installed if missing. # ═══════════════════════════════════════════════════════════════════════════════ # ───────────────────────────────────────────────────────────────────────────── # Font discovery (cross-platform: Windows / macOS / Linux) # ───────────────────────────────────────────────────────────────────────────── _FONT_SEARCH_PATHS = { "Windows": [ # CJK fonts commonly found on Windows r"C:\Windows\Fonts\msyh.ttc", # 微软雅黑 r"C:\Windows\Fonts\msyhbd.ttc", # 微软雅黑 Bold r"C:\Windows\Fonts\simhei.ttf", # 黑体 r"C:\Windows\Fonts\simsun.ttc", # 宋体 r"C:\Windows\Fonts\simkai.ttf", # 楷体 r"C:\Windows\Fonts\STFANGSO.TTF", # 华文仿宋 r"C:\Windows\Fonts\STSONG.TTF", # 华文宋体 r"C:\Windows\Fonts\malgun.ttf", # Malgun Gothic (Korean but has CJK) r"C:\Windows\Fonts\meiryo.ttc", # Meiryo (Japanese but has CJK) r"C:\Windows\Fonts\segoeui.ttf", # Fallback: Segoe UI (no CJK) r"C:\Windows\Fonts\arial.ttf", # Fallback: Arial ], "Darwin": [ # macOS "/System/Library/Fonts/PingFang.ttc", "/System/Library/Fonts/STHeiti Medium.ttc", "/System/Library/Fonts/Hiragino Sans GB.ttc", "/Library/Fonts/Arial Unicode MS.ttf", "/System/Library/Fonts/Helvetica.ttc", ], "Linux": [ "/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc", "/usr/share/fonts/opentype/noto/NotoSansCJKsc-Regular.otf", "/usr/share/fonts/truetype/noto/NotoSansCJK-Regular.ttc", "/usr/share/fonts/truetype/noto/NotoSansCJKsc-Regular.otf", "/usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf", "/usr/share/fonts/truetype/wqy/wqy-zenhei.ttc", "/usr/share/fonts/truetype/wqy/wqy-microhei.ttc", "/usr/share/fonts/opentype/unifont/unifont.otf", "/usr/share/fonts/truetype/freefont/FreeSans.ttf", "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf", ], } _font_cache = {} # (size, bold) → ImageFont def _find_system_font(): """Return the path to the best available font for the current OS.""" system = platform.system() candidates = _FONT_SEARCH_PATHS.get(system, []) # Also merge Linux paths as generic fallback if system != "Linux": candidates = candidates + _FONT_SEARCH_PATHS["Linux"] for p in candidates: if Path(p).is_file(): return p return None def _get_font(size, bold=False): """Get a PIL ImageFont at the given size, with caching.""" key = (size, bold) if key in _font_cache: return _font_cache[key] from PIL import ImageFont font_path = _find_system_font() if font_path: try: # For .ttc files, index 0 is usually Regular, index 1 is Bold idx = 1 if bold and font_path.endswith((".ttc", ".TTC")) else 0 fnt = ImageFont.truetype(font_path, size, index=idx) _font_cache[key] = fnt return fnt except Exception: try: fnt = ImageFont.truetype(font_path, size) _font_cache[key] = fnt return fnt except Exception: pass # Ultimate fallback: Pillow default bitmap font fnt = ImageFont.load_default() _font_cache[key] = fnt return fnt # ───────────────────────────────────────────────────────────────────────────── # Backend 1: Playwright (best quality — renders full interactive HTML) # ───────────────────────────────────────────────────────────────────────────── def _has_playwright(): try: from playwright.sync_api import sync_playwright as _sp # noqa: F401 return True except ImportError: return False def _export_image_playwright(html_str: str, out: str, fmt: str, scale: float = 2.0, quality: int = 92): import tempfile, shutil pw_paths = [ os.environ.get("PLAYWRIGHT_BROWSERS_PATH", ""), "/opt/pw-browsers", str(Path.home() / ".cache" / "ms-playwright"), ] for p in pw_paths: if p and Path(p).is_dir(): os.environ["PLAYWRIGHT_BROWSERS_PATH"] = p break from playwright.sync_api import sync_playwright tmp_dir = tempfile.mkdtemp(prefix="mindmap_") tmp_html = os.path.join(tmp_dir, "mindmap.html") with open(tmp_html, "w", encoding="utf-8") as f: f.write(html_str) try: with sync_playwright() as pw: browser = pw.chromium.launch(headless=True, args=[ "--no-sandbox", "--disable-gpu", "--disable-dev-shm-usage", ]) dpr = max(1.0, min(scale, 4.0)) page = browser.new_page( viewport={"width": 2560, "height": 1440}, device_scale_factor=dpr, ) page.goto(f"file://{tmp_html}", wait_until="networkidle") page.wait_for_selector("#nodes-g .nd", timeout=8000) page.wait_for_timeout(600) bbox = page.evaluate(r"""() => { const nodesG = document.getElementById('nodes-g'); const edgesG = document.getElementById('edges-g'); if (!nodesG) return null; let minX=Infinity, minY=Infinity, maxX=-Infinity, maxY=-Infinity; for (const g of [nodesG, edgesG]) { if (!g) continue; for (const el of g.querySelectorAll('rect,text,path,circle')) { const r = el.getBoundingClientRect(); if (r.width===0 && r.height===0) continue; if (r.leftmaxX) maxX=r.right; if (r.bottom>maxY) maxY=r.bottom; } } if (minX===Infinity) return null; const pad=40; return {x:Math.max(0,minX-pad),y:Math.max(0,minY-pad), width:maxX-minX+pad*2,height:maxY-minY+pad*2}; }""") clip = bbox if (bbox and bbox["width"] > 0) else None if not clip: wrap = page.query_selector("#wrap") clip = wrap.bounding_box() if wrap else None png_bytes = page.screenshot(type="png", clip=clip) if clip else page.screenshot(type="png") browser.close() from PIL import Image as PILImage img = PILImage.open(io.BytesIO(png_bytes)) if fmt == "jpg": if img.mode in ("RGBA", "P"): bg = PILImage.new("RGB", img.size, (13, 15, 26)) if img.mode == "P": img = img.convert("RGBA") bg.paste(img, mask=img.split()[3]); img = bg elif img.mode != "RGB": img = img.convert("RGB") img.save(out, "JPEG", quality=quality, optimize=True) else: img.save(out, "PNG", optimize=True) finally: shutil.rmtree(tmp_dir, ignore_errors=True) # ───────────────────────────────────────────────────────────────────────────── # Backend 2: Pure Pillow (works everywhere — only needs `pip install pillow`) # ───────────────────────────────────────────────────────────────────────────── def _pillow_measure_text(text, font): """Measure text width and height using the font.""" bbox = font.getbbox(str(text)) return bbox[2] - bbox[0], bbox[3] - bbox[1] def _strip_emoji_for_pillow(text): """Remove emoji characters that Pillow/system fonts can't render. Keeps the text content after the emoji prefix.""" if not text: return text result = [] i = 0 while i < len(text): cp = ord(text[i]) # Skip emoji code points if (cp >= 0x1F300 or # Misc Symbols & Pictographs+ 0x2600 <= cp <= 0x27BF or # Misc Symbols, Dingbats 0x2300 <= cp <= 0x23FF or # Misc Technical cp == 0xFE0F or cp == 0xFE0E or # Variation Selectors cp == 0x200D or # Zero Width Joiner 0xE0020 <= cp <= 0xE007F or # Tags 0x1F900 <= cp <= 0x1F9FF or # Supplemental Symbols 0x1FA00 <= cp <= 0x1FA6F or # Chess Symbols 0x1FA70 <= cp <= 0x1FAFF or # Symbols Extended-A 0x2702 <= cp <= 0x27B0 or # Dingbats 0x231A <= cp <= 0x231B or # Watch, Hourglass cp == 0x2B50 or cp == 0x2B55 or # Star, Circle 0x2934 <= cp <= 0x2935 or # Arrows 0x25AA <= cp <= 0x25FE or # Geometric Shapes 0x2190 <= cp <= 0x21FF or # Arrows 0x20E3 == cp): # Keycap i += 1 continue result.append(text[i]) i += 1 return "".join(result).strip() def _hex_to_rgb(h): """Convert '#RRGGBB' or '#RRGGBBAA' to (R,G,B) or (R,G,B,A).""" h = h.lstrip("#") if len(h) == 8: return (int(h[0:2], 16), int(h[2:4], 16), int(h[4:6], 16), int(h[6:8], 16)) if len(h) == 6: return (int(h[0:2], 16), int(h[2:4], 16), int(h[4:6], 16)) return (128, 128, 128) def _blend_alpha(fg_hex, bg_rgb=(13, 15, 26)): """Blend a hex color with alpha (e.g. '#4A90D930') onto a background.""" c = _hex_to_rgb(fg_hex) if len(c) == 4: r, g, b, a = c af = a / 255.0 return ( int(r * af + bg_rgb[0] * (1 - af)), int(g * af + bg_rgb[1] * (1 - af)), int(b * af + bg_rgb[2] * (1 - af)), ) return c[:3] def _export_image_pillow(tree, positions, out: str, fmt: str, scale: float = 2.0, quality: int = 92): """ Render the mind map to PNG/JPG/PDF using only Pillow. Zero system dependencies — works on Windows/macOS/Linux with just `pip install pillow`. """ from PIL import Image, ImageDraw BG_COLOR = (13, 15, 26) # #0d0f1a nodes = flatten_tree(tree) cmap = build_color_map(tree) minx, miny, maxx, maxy = bounding_box(positions, nodes, pad=80) W = maxx - minx H = maxy - miny # Apply scale s = scale img_w, img_h = int(W * s), int(H * s) img = Image.new("RGB", (img_w, img_h), BG_COLOR) draw = ImageDraw.Draw(img, "RGBA") # RGBA for alpha blending def tx(x): return (x - minx) * s def ty(y): return (y - miny) * s # ── Draw edges ──────────────────────────────────────────────────────── for node in nodes: p = positions.get(node["_id"]) if not p or p.get("parent_id") is None: continue pp = positions.get(p["parent_id"]) if not pp: continue pnode = next((n for n in nodes if n["_id"] == p["parent_id"]), None) if not pnode: continue color_hex = cmap.get(node["_id"], "#888888") depth = node["_depth"] # Opacity simulation: blend with background if depth == 1: alpha = 0.85 elif depth == 2: alpha = 0.6 else: alpha = 0.4 ec = _hex_to_rgb(color_hex)[:3] edge_color = tuple(int(ec[i] * alpha + BG_COLOR[i] * (1 - alpha)) for i in range(3)) sw = max(1, int((2.5 if depth == 1 else (1.8 if depth == 2 else 1.3)) * s)) # Draw edge as a series of line segments approximating the Bezier px_s, py_s = tx(pp["x"]), ty(pp["y"]) cx_s, cy_s = tx(p["x"]), ty(p["y"]) pw_s = pnode["_w"] * s / 2 cw_s = node["_w"] * s / 2 if cx_s >= px_s: x1 = px_s + pw_s x2 = cx_s - cw_s else: x1 = px_s - pw_s x2 = cx_s + cw_s # Draw cubic Bezier approximation (matching SVG/HTML edge_path) # Tension: depth-1 → 0.5, deeper → 0.45 tension = 0.5 if depth == 1 else 0.45 cpx = x1 + (x2 - x1) * tension steps = 24 pts = [] for t in range(steps + 1): t_f = t / steps # Cubic Bezier: P0=(x1,py), P1=(cpx,py), P2=(cpx,cy), P3=(x2,cy) u = 1 - t_f bx = u**3 * x1 + 3 * u**2 * t_f * cpx + 3 * u * t_f**2 * cpx + t_f**3 * x2 by = u**3 * py_s + 3 * u**2 * t_f * py_s + 3 * u * t_f**2 * cy_s + t_f**3 * cy_s pts.append((bx, by)) for i in range(len(pts) - 1): draw.line([pts[i], pts[i + 1]], fill=edge_color, width=sw) # ── Draw nodes ──────────────────────────────────────────────────────── # Root gradient colors ROOT_GRAD_START = (76, 95, 219) # #4c5fdb ROOT_GRAD_END = (124, 140, 248) # #7c8cf8 for node in nodes: p = positions.get(node["_id"]) if not p: continue depth = node["_depth"] c = CFG[min(depth, len(CFG) - 1)] w_raw, h_raw = node["_w"], node["_h"] w_s, h_s = w_raw * s, h_raw * s nx = tx(p["x"]) - w_s / 2 ny = ty(p["y"]) - h_s / 2 rx = int(c["rx"] * s) color_hex = cmap.get(node["_id"], "#888888") label = node.get("label") or node.get("central", "") if depth == 0: # Root: gradient-like fill (approximate with solid blend) root_color = tuple((ROOT_GRAD_START[i] + ROOT_GRAD_END[i]) // 2 for i in range(3)) # Glow effect: draw a larger, blurred rect behind glow_pad = 6 * s glow_color = root_color + (60,) draw.rounded_rectangle( [nx - glow_pad, ny - glow_pad, nx + w_s + glow_pad, ny + h_s + glow_pad], radius=rx + int(glow_pad), fill=glow_color, ) draw.rounded_rectangle([nx, ny, nx + w_s, ny + h_s], radius=rx, fill=root_color) text_color = (255, 255, 255) elif depth == 1: fill_color = _blend_alpha(color_hex + "30", BG_COLOR) outline_color = _hex_to_rgb(color_hex)[:3] ow = max(1, int(2 * s)) draw.rounded_rectangle([nx, ny, nx + w_s, ny + h_s], radius=rx, fill=fill_color, outline=outline_color, width=ow) text_color = (255, 255, 255) elif depth == 2: fill_color = _blend_alpha(color_hex + "18", BG_COLOR) outline_color = _blend_alpha(color_hex + "bb", BG_COLOR) ow = max(1, int(1.5 * s)) draw.rounded_rectangle([nx, ny, nx + w_s, ny + h_s], radius=rx, fill=fill_color, outline=outline_color, width=ow) text_color = (224, 228, 240) else: fill_color = _blend_alpha(color_hex + "0e", BG_COLOR) outline_color = _blend_alpha(color_hex + "77", BG_COLOR) ow = max(1, int(1 * s)) draw.rounded_rectangle([nx, ny, nx + w_s, ny + h_s], radius=rx, fill=fill_color, outline=outline_color, width=ow) text_color = (168, 176, 200) # ── Draw text ───────────────────────────────────────────────── font_size = int(c["fs"] * s) is_bold = c["fw"] in ("bold", "600", "700") font = _get_font(font_size, bold=is_bold) # Strip emoji for Pillow (system fonts can't render color emoji) render_label = _strip_emoji_for_pillow(label) tw, th = _pillow_measure_text(render_label, font) text_x = tx(p["x"]) - tw / 2 text_y = ty(p["y"]) - th / 2 draw.text((text_x, text_y), render_label, fill=text_color, font=font) # ── Save ────────────────────────────────────────────────────────────── if fmt == "jpg": img = img.convert("RGB") img.save(out, "JPEG", quality=quality, optimize=True) elif fmt == "pdf": # Embed the rendered image into a single-page PDF via Pillow # Pillow can save PDF directly from an Image object img = img.convert("RGB") img.save(out, "PDF", resolution=72.0 * s) else: img.save(out, "PNG", optimize=True) # ───────────────────────────────────────────────────────────────────────────── # MAIN # ───────────────────────────────────────────────────────────────────────────── def main(): args = parse_args() fmt = args.format.lower() if args.output is None: out = default_output(fmt) else: out = resolve_output(args.output, fmt) try: raw = json.loads(args.data) except json.JSONDecodeError as e: print(f"[mindmap] ERROR: Invalid JSON — {e}", file=sys.stderr); sys.exit(1) try: tree = build_tree(raw) except ValueError as e: print(f"[mindmap] ERROR: {e}", file=sys.stderr); sys.exit(1) # ── XMind (pure Python, no deps) ──────────────────────────────────── if fmt == "xmind": open(out, "wb").write(build_xmind(tree, args.title)) print(f"[mindmap] ✅ XMind → {out}"); return # ── SVG (static, pure Python) ─────────────────────────────────────── if fmt == "svg": _nid[0] = 0 tree["_depth"] = 0; tree["label"] = tree["central"] annotate(tree, 0) positions = compute_layout(tree) svg_str = render_svg_static(tree, positions, include_xml_header=True) open(out, "w", encoding="utf-8").write(svg_str) print(f"[mindmap] ✅ SVG → {out}"); return # ── PNG / JPG / PDF — auto-select best available backend ──────────── if fmt in ("png", "jpg", "pdf"): # Backend 1: Playwright (best quality for png/jpg) if fmt in ("png", "jpg") and _has_playwright(): html_str = render_html(args.title, json.dumps(tree, ensure_ascii=False)) _export_image_playwright(html_str, out, fmt, scale=args.scale, quality=args.quality) print(f"[mindmap] ✅ {fmt.upper()} → {out} (via Playwright)"); return # Backend 2: Pillow (pure Python, works everywhere) # Auto-install if missing if not _ensure_pillow(): print("[mindmap] ERROR: Cannot export image — Pillow is required.", file=sys.stderr) print("[mindmap] Please run: pip install pillow", file=sys.stderr) sys.exit(1) _nid[0] = 0 tree["_depth"] = 0; tree["label"] = tree["central"] annotate(tree, 0) positions = compute_layout(tree) _export_image_pillow(tree, positions, out, fmt, scale=args.scale, quality=args.quality) print(f"[mindmap] ✅ {fmt.upper()} → {out} (via Pillow)"); return # ── HTML (default) ────────────────────────────────────────────────── html = render_html(args.title, json.dumps(tree, ensure_ascii=False)) open(out, "w", encoding="utf-8").write(html) print(f"[mindmap] ✅ HTML → {out}") if __name__ == "__main__": main() FILE:examples/ai_trends.html AI 发展趋势

🧠 AI 发展趋势

导出

视图

历史

布局

右键菜单 · Tab 添加子节点 · Del 删除 · Ctrl+Z 撤销

FILE:examples/ai_trends.json { "central": "AI 发展趋势", "branches": [ { "label": "大语言模型", "color": "#4A90D9", "children": [ {"label": "能力提升", "children": ["多模态理解", "代码生成", "逻辑推理"]}, {"label": "开源生态", "children": ["Llama 系列", "Mistral", "Qwen"]}, "模型压缩", "长上下文" ] }, { "label": "AI Agent", "color": "#E86C3A", "children": [ "自主规划", "工具调用", {"label": "记忆系统", "children": ["短期记忆", "长期记忆", "外部知识库"]}, {"label": "多智能体", "children": ["协作框架", "角色分工"]} ] }, { "label": "行业应用", "color": "#27AE60", "children": [ {"label": "医疗健康", "children": ["辅助诊断", "新药研发"]}, {"label": "教育培训", "children": ["个性化学习", "智能批改"]}, "金融风控", "代码开发", "内容创作" ] }, { "label": "安全与治理", "color": "#9B59B6", "children": [ {"label": "技术安全", "children": ["幻觉问题", "对齐研究", "可解释性"]}, {"label": "数据隐私", "children": ["联邦学习", "差分隐私"]}, "监管政策", "伦理规范" ] }, { "label": "基础设施", "color": "#F39C12", "children": [ {"label": "算力", "children": ["GPU 集群", "专用芯片", "分布式训练"]}, "训练框架", "推理优化", "云服务平台" ] } ] } FILE:examples/product_launch.html 产品发布 v2.0

🧠 产品发布 v2.0

导出

视图

历史

布局

右键菜单 · Tab 添加子节点 · Del 删除 · Ctrl+Z 撤销

FILE:examples/product_launch.json { "central": "产品发布 v2.0", "branches": [ { "label": "需求分析", "color": "#4A90D9", "children": [ {"label": "用户调研", "children": ["问卷设计", "用户访谈", "数据分析"]}, {"label": "竞品分析", "children": ["功能对比", "定价策略"]}, "需求优先级排序", "PRD 文档" ] }, { "label": "设计阶段", "color": "#1ABC9C", "children": [ {"label": "UI 设计", "children": ["线框图", "视觉稿", "设计规范"]}, {"label": "UX 设计", "children": ["用户流程", "交互原型"]}, "设计评审", "可用性测试" ] }, { "label": "研发阶段", "color": "#E86C3A", "children": [ {"label": "前端开发", "children": ["组件库", "页面实现", "性能优化"]}, {"label": "后端开发", "children": ["API 设计", "数据库", "微服务"]}, {"label": "测试", "children": ["单元测试", "集成测试", "压力测试"]}, "代码评审" ] }, { "label": "上线准备", "color": "#9B59B6", "children": [ {"label": "灰度发布", "children": ["内测用户", "A/B 测试"]}, {"label": "运维保障", "children": ["监控告警", "回滚预案"]}, "文档更新", "客服培训" ] }, { "label": "推广运营", "color": "#F39C12", "children": [ {"label": "市场推广", "children": ["发布会", "媒体公关", "社交媒体"]}, {"label": "数据追踪", "children": ["DAU/MAU", "转化率", "留存率"]}, "用户反馈收集", "迭代规划" ] } ] } FILE:examples/python_learning.html Python 学习路径

🧠 Python 学习路径

导出

视图

历史

布局

右键菜单 · Tab 添加子节点 · Del 删除 · Ctrl+Z 撤销

FILE:examples/python_learning.json { "central": "Python 学习路径", "branches": [ { "label": "基础语法", "color": "#4A90D9", "children": [ {"label": "数据类型", "children": ["int/float", "str", "list/tuple", "dict/set"]}, {"label": "控制流", "children": ["if/elif/else", "for/while", "break/continue"]}, {"label": "函数", "children": ["定义与调用", "参数类型", "lambda", "装饰器"]}, "模块与包" ] }, { "label": "面向对象", "color": "#27AE60", "children": [ {"label": "类与对象", "children": ["属性", "方法", "__init__"]}, {"label": "继承", "children": ["单继承", "多继承", "super()"]}, "封装与多态", "魔术方法" ] }, { "label": "常用库", "color": "#E86C3A", "children": [ {"label": "数据处理", "children": ["NumPy", "Pandas", "Polars"]}, {"label": "可视化", "children": ["Matplotlib", "Seaborn", "Plotly"]}, {"label": "Web 框架", "children": ["FastAPI", "Django", "Flask"]}, {"label": "AI/ML", "children": ["PyTorch", "scikit-learn", "transformers"]} ] }, { "label": "工程实践", "color": "#9B59B6", "children": [ {"label": "项目管理", "children": ["虚拟环境", "依赖管理", "pyproject.toml"]}, {"label": "代码质量", "children": ["类型注解", "linting", "格式化"]}, {"label": "测试", "children": ["pytest", "unittest", "mock"]}, "CI/CD" ] }, { "label": "进阶专题", "color": "#00BCD4", "children": [ {"label": "并发编程", "children": ["多线程", "多进程", "asyncio"]}, {"label": "性能优化", "children": ["Cython", "numba", "profiling"]}, "元编程", "设计模式" ] } ] }

Generate Mindmap

Comments (0)