| name | XmindAnything |
| license | MIT |
| description | Converts PDF files, Markdown documents, web URLs, GitHub projects, local directories (mixed file types), or plain text into XMind mind maps (.xmind files). Use when the user asks to: (1) convert a PDF/paper/article into a mind map or 思维导图/脑图, (2) turn a Markdown document or README into a structured diagram, (3) visualize a webpage or blog post as a mind map, (4) generate architecture diagrams from GitHub repos or local project directories, (5) organize study materials (directories with mixed files) into thematic mind maps, (6) create an XMind file from any text content, (7) 将文章/论文/文档/网页/项目转为思维导图. Requires xmindmark CLI. For scanned PDFs: tesseract-ocr + poppler-utils. For xlsx/pptx: openpyxl + python-pptx (auto-install on first use).
|
XmindAnything (v2.2)
Convert any content (PDF, Markdown, URL, GitHub projects, directories, plain text, and more) into XMind mind maps using XMindMark format + xmindmark CLI.
Prerequisites
Dependency Matrix
| 功能 | 依赖 | 安装命令 | 自动安装 |
|---|
| 核心转换 | xmindmark (npm) | npm install -g xmindmark | ✅ |
| 文本 PDF 提取 | pdftotext → poppler-utils | apt install poppler-utils | ✅ |
| 扫描 PDF OCR | tesseract + poppler-utils | apt install tesseract-ocr tesseract-ocr-chi-sim poppler-utils | ✅ |
| DOCX 提取 | python-docx (pip) | pip install python-docx | ✅ |
| PPTX 提取 | python-pptx (pip) | pip install python-pptx | ✅ |
| .ppt/.doc 旧格式 | soffice → libreoffice-impress-nogui | apt install libreoffice-impress-nogui | ✅ |
| XLSX 提取 | openpyxl (pip) | pip install openpyxl | ✅ |
| 内容清洗 | python3 (≥3.8) | 系统自带 | — |
| 文件上传 | curl | 系统自带 | — |
一键安装
bash scripts/setup.sh
首次使用时会自动检测缺失依赖并安装。也可以手动运行 scripts/setup.sh 确保环境就绪。
Workflow
Input → Extract content → Structure into XMindMark → Convert to .xmind → Upload & share link
Step 1: Identify input type
| Input | Detection | Scenario |
|---|
| Directory (目录) | [ -d "$input" ] 为真 | project |
| GitHub URL | URL 含 github.com | project |
| PowerPoint (.pptx) | File extension | pptx |
| PowerPoint (.ppt) | File extension (旧格式) | pptx → LibreOffice 转 pptx |
| Word (.docx) | File extension | docx |
| Word (.doc) | File extension (旧格式) | docx → LibreOffice 转 docx |
| PDF (.pdf) | File extension or "论文/paper" | paper |
| Scanned PDF | pdftotext returns < 100 bytes | ocr-pdf |
| Markdown (.md) | File extension | markdown |
| Web URL (http/https) | Starts with http (非GitHub) | url |
| Other file types (.yaml/.json/.py/...) | File extension | format-registry |
| Plain text | None of above | Auto-structure |
Step 2: Extract and structure
After extracting raw text (from any source), ALWAYS run the cleaning pipeline:
python3 scripts/clean_content.py /tmp/raw_text.txt /tmp/clean_text.txt
This automatically:
- Removes citation markers (
[1], [2,3], [42])
- Strips page numbers, ISBNs, DOI links, watermark headers
- Truncates the References/Bibliography section
- Filters LaTeX rendering artifacts
- Merges broken sentences (pdftotext artifacts)
- Normalizes whitespace
Then proceed to structure based on input type.
输入判断逻辑（按优先级）：
- 先判是否目录
[ -d "$input" ] → project 子场景 B
- 再判是否 GitHub URL → project 子场景 A
- 再判文件扩展名 → 查 format-registry
- 其余走现有文本逻辑
Scan-detection for PDFs: First test if PDF is text-based or scanned:
pdftotext input.pdf /tmp/test.txt && wc -c /tmp/test.txt
If < 100 bytes → scanned, needs OCR. Jump to ocr-pdf.
Text-based PDFs: Use pdf tool → extract title/authors/abstract/method/results/conclusion → each section 2-4 bullet points.
Scanned PDFs (OCR): Convert to images → tesseract OCR → read result → structure. Full workflow in ocr-pdf. Prerequisite: apt-get install -y tesseract-ocr tesseract-ocr-chi-sim poppler-utils.
Markdown: Parse # = center, ## = main branch, ### = sub-branch. 1-2 sentences per node.
URLs: Use web_fetch (readability mode auto-strips nav/ads). Identify natural chapter boundaries. Read scenario-url.md for real-world example.
Plain text: Group related ideas into thematic branches. Center topic from first sentence.
其他文件格式： 查 format-registry，按对应方法提取。
- 文本类：提取结构信息（key/列名/标题），不提取全文
- 源码类：提取 class/function 签名，不提取实现细节
- 二进制类：首次使用时安装对应 Python 库（openpyxl / python-pptx）
- 跳过图片/视频/音频/压缩包/二进制库，最后标注「已跳过 N 个文件」
Step 3: Generate XMindMark
Write to .xmindmark file. Syntax reference: SYNTAX.md.
Critical rules (tested against xmindmark 0.3.2):
- Line 1 = center topic (no prefix)
- Main branches:
- prefix, no indent
- Sub-branches: 4 spaces,
* or -
- Max 4 levels deep, 3-7 children per branch
- No spaces inside
[] markers
- DO NOT use
[B]: or [S]: title syntax — crashes xmindmark 0.3.2 parser
- DO NOT use cross-branch
[1]/[^1] relationships — XMind draws ugly crossing lines
- Use
[B] boundaries and [S] summaries without titles only
Step 4: Convert
rm -f output.xmind && xmindmark -f xmind output.xmindmark
Skip SVG export (requires Chromium download, often fails on servers).
Step 5: Upload & share
RESULT=$(curl -s -F "file=@output.xmind" https://tmpfiles.org/api/v1/upload)
XMindMark Quick Syntax
Center Topic
- Main Branch 1 [B1]
* Sub-topic 1.1
- Detail A
- Detail B
* Sub-topic 1.2
- Main Branch 2 [B2]
* Point A
* Point B
- Key Takeaways [S1]
* Conclusion 1
* Conclusion 2
Remember: No [B]:/[S]: titles. No cross-branch [1]/[^1]. Keep nodes short.
