dsh-paddle-ocr DSH plugin

dsh-paddle-ocr is a DSH plugin that converts PDFs and images into Markdown with embedded images, leveraging Baidu PaddleOCR-VL. It offers three host tools for agents, a configuration card in the browser settings, and a task panel for drag-and-drop processing. The plugin handles page limit splitting, queue congestion with exponential backoff, and stores tokens securely in DSH credentials vault.

vision ui-productivity

How to install the dsh-paddle-ocr DSH plugin

dsh plugin --profile web add link:/path/to/dsh-paddle-ocr

Copying does not run this command. Review the repository and version before installing the dsh-paddle-ocr DSH plugin.

dsh-paddle-ocr DSH plugin data source

dsh-paddle-ocr DSH plugin snapshot date: Aug 16, 2026

discovered

What the dsh-paddle-ocr DSH plugin can do

Parse PDF/Image pages into Markdown with image files saved locally
Automatic splitting of oversized PDFs (1000 pages / 50MB for async, 100 pages for sync)
Exponential backoff retry (up to 5 times) when queue is full (errorCode 10010)
Async mode (default) and sync mode with 300s timeout
Browser UI: settings card for API token, test connection, and default processing options
Task panel: drag-and-drop files, progress bar, result preview tab, and save to workspace
Agent-accessible tools: `paddle_ocr_layout`, `paddle_ocr_test`, `paddle_split_pdf`

Where the dsh-paddle-ocr DSH plugin fits

Convert scanned PDF documents into editable Markdown for further processing
Batch parse images with layout analysis, preserving structure and embedded images
Integrate OCR capabilities into DSH agents for automated document workflows
Quickly test API token connectivity with a built-in test image
Manually split large PDFs into smaller chunks before processing

Who the dsh-paddle-ocr DSH plugin is for

DSH users who need to extract text and layout from PDFs or images
Developers building document processing pipelines inside DSH agents
Users who prefer a visual interface for OCR tasks within DSH

dsh-paddle-ocr DSH plugin limitations

Requires an external API token from Baidu AI Studio (PaddleOCR service)
Queue saturation is common; submissions may be delayed or require retry
Task panel upload limit is 30MB (larger files must use agent tools)
Only supports PaddleOCR-VL model; no other OCR backends
Async mode has a 1000-page limit per job; sync mode has 100-page limit

dsh-paddle-ocr DSH plugin: from the repository README

Quoted from the omdsh-dev/dsh-paddle-ocr README, the upstream source of the dsh-paddle-ocr DSH plugin. Copyright remains with the original authors.

DSH 插件：百度 **PaddleOCR-VL** 文档布局解析。把 PDF / 图片逐页解析为 Markdown （文字 + 图片落盘），提供三个宿主工具、一个浏览器配置卡和一个任务面板。 - 宿主工具（agent 可用）： - `paddle_ocr_layout(file, {mode, fileType, orientation, unwarping, chart, visualize, outputDir})` —— 逐页解析落盘（`doc_N.md` + 图片），默认异步模式；超限 PDF 自动拆分；队列满（10010）自动指数退避重试（最多 5 次），最终仍满则给出友好排队提示。 - `paddle_ocr_test()` —— 用内置小测试图验证 token 与接口连通性（✅/❌ + 详情）。 - `paddle_split_pdf(file, {maxPages, outputDir})` —— 手动拆分超限 PDF。 - 浏览器 UI： - 设置 → 插件 → **PaddleOCR** 卡片：API token（只进 DSH 凭据保险箱，永不回传、输入框常空、只显示“已配置”徽章）、[测试连接] 按钮、处理默认值（异步/同步、方向矫正、展平、图表识别、可视化标注图）。 - 右下角**任务面板**：拖入/选择文件 → 进度条（排队中 / 解析第 x 页 / 下载图片）→ 结果页签（Markdown 预览 + 图片缩略图）→ [落到工作区]。队列满（10010）渲染为 “当前排队中（稍后重试）”按钮，而不是错误字符串。 ## 安装 ```bash # 在 DSH 部署机上，把本包加入 web profile（同时打进 host 组合与浏览器 bundle） dsh plugin --profile web add link:/path/to/dsh-paddle-ocr ``` 插件自带 `dsh.bundle.patch`（组合插入行）与 `dsh.client` 声明；`dsh plugin` 会自动把包名加入 `dsh.profile.bundles`。重启 `dsh web` 后生效。 ## 获取 API token 1. 打开星河社区 <https://aistudio.baidu.com/paddleocr/task>； 2. 登录后进入 **个人中心 → 访问令牌（token）**； 3. 复制 token，粘贴到设置 → 插件 → PaddleOCR 卡片，点保存，再点 [测试连接]。也可以设置环境变量 `PADDLE_OCR_TOKEN`（环境变量优先、只读，卡片会显示只读徽章）。 > ⚠️ **安全提醒：请轮换你此前在 sk

Read the full READMERepository license: BSD-3-Clause

dsh-paddle-ocr DSH plugin questions

How do I install the dsh-paddle-ocr plugin?

Run the command `dsh plugin --profile web add link:/path/to/dsh-paddle-ocr` on the DSH deployment machine. The plugin will automatically add its bundle to the web profile. After that, restart the DSH web service for the changes to take effect.

Where do I get the API token for PaddleOCR?

Go to the Baidu AI Studio platform (aistudio.baidu.com/paddleocr/task), log in, and navigate to Personal Center → Access Token. Copy the token and paste it into the DSH settings → Plugins → PaddleOCR card. You can also set the environment variable `PADDLE_OCR_TOKEN` which takes priority and is read-only.

What happens if the OCR queue is full?

When the queue is full (errorCode 10010), the plugin will automatically retry with exponential backoff (2s, 4s, 8s, ... up to 5 times). If retries are exhausted, a friendly 'queued, please retry later' message is shown. In the task panel, a button allows you to retry later.

Can I process files larger than 30MB?

The task panel has a 30MB upload limit due to loopback RPC capacity. For larger files, use the agent tool `paddle_ocr_layout(file, ...)` directly, which can handle larger files without the upload limit. The plugin also automatically splits oversized PDFs (1000 pages async, 100 pages sync).

Is my API token safe?

Yes. The token is stored only in the DSH credentials vault, never written to code, config, or logs. The settings card always shows an empty input field and a 'configured' badge when a token is set. Environment variables are read-only and not exposed. It is recommended to rotate any token previously used in skill repositories.