Endpoint Lasso

Endpoint Lasso is a public Supernote plugin framework for sending the current lasso selection, or highlighted DOC text, to your own HTTP endpoint.

It is intentionally generic. Instead of baking in one personal backend, it gives you a reusable transport layer that you can point at your own service for OCR, indexing, search, note capture, flashcard generation, or custom LLM pipelines.

What It Does

• works from the lasso toolbar in NOTE and DOC
• also adds a DOC text-selection toolbar button for highlighted text
• sends structured metadata for the current file, page, lasso rectangle, and selected elements
• can optionally attach a generated full-page NOTE PNG
• supports either multipart or json request formats
• is configured from environment variables before build, so no source edits are required for normal setup

Quick Start

1. Copy .env.example to .env.
2. Set SN_ENDPOINT_URL.
3. Optionally set auth headers, timeout, button label, PNG behavior, and request format.
4. Install dependencies:

npm install

5. Build the plugin:

npm run build:plugin

The packaged plugin is written to:

build/outputs/supernote-endpoint-lasso.snplg

Configuration

The build step reads .env plus the current shell environment and generates src/runtimeConfig.js.

Required:

• SN_ENDPOINT_URL: destination endpoint URL

Common optional variables:

• SN_REQUEST_FORMAT: multipart or json (default: multipart)
• SN_AUTH_HEADER_NAME: auth header name, for example Authorization
• SN_AUTH_HEADER_VALUE: auth header value, for example Bearer ...
• SN_EXTRA_HEADERS_JSON: JSON object of extra headers
• SN_BUTTON_NAME: toolbar label (default: Send Lasso)
• SN_DOC_SELECTION_BUTTON_NAME: DOC text-selection label (default: Send Selection)
• SN_TIMEOUT_MS: request timeout in milliseconds (default: 20000)
• SN_INCLUDE_PAGE_PNG: true or false (default: true)
• SN_EXPORT_DIR: device path for temporary PNG export
• SN_PNG_SCALE: PNG scale factor (default: 1)
• SN_PNG_BACKGROUND_TYPE: 0 transparent or 1 white (default: 1)
• SN_PAYLOAD_FIELD_NAME: multipart field name for the JSON payload (default: payload)
• SN_IMAGE_FIELD_NAME: multipart field name for the page PNG (default: page_png)
• SN_SUCCESS_DIALOG_TITLE: success dialog title shown on-device

Payload Shape

For lasso selections, Endpoint Lasso builds a payload shaped like this:

{
  "framework": {
    "name": "Endpoint Lasso",
    "version": "0.1.3"
  },
  "generated_at": "2026-04-14T00:00:00.000Z",
  "source": {
    "file_path": "...",
    "file_kind": "note",
    "page_num": 3,
    "page_size": {"width": 1404, "height": 1872},
    "lasso_rect": {"left": 100, "top": 200, "right": 600, "bottom": 900}
  },
  "elements": [
    {
      "type": 0,
      "uuid": "...",
      "page_num": 3,
      "layer_num": 1,
      "recognize_result": null,
      "text": null,
      "text_rect": null,
      "contours": [[{"x": 1, "y": 2}]]
    }
  ],
  "attachments": {
    "page_png_included": true,
    "page_png_mode": "note-export"
  }
}

If SN_REQUEST_FORMAT=multipart, the payload is sent as a JSON field plus an optional PNG file field.

If SN_REQUEST_FORMAT=json, the payload is sent as JSON only.

For highlighted DOC text, the payload uses the same transport but sets source.selection_kind to doc_text and includes both a structured selection object and a convenience top-level selected_text field:

{
  "framework": {
    "name": "Endpoint Lasso",
    "version": "0.1.3"
  },
  "generated_at": "2026-04-22T00:00:00.000Z",
  "source": {
    "file_path": "...",
    "file_kind": "document",
    "page_num": 3,
    "page_size": null,
    "selection_kind": "doc_text",
    "selection_api": "PluginDocAPI.getLastSelectedText"
  },
  "selection": {
    "kind": "doc_text",
    "text": "highlighted text from the document"
  },
  "selected_text": "highlighted text from the document",
  "elements": [],
  "attachments": {
    "page_png_included": false,
    "page_png_mode": "not-supported-for-doc-selection"
  }
}

Notes

• NOTE and DOC lasso metadata are both supported.
• DOC highlighted text is read with PluginDocAPI.getLastSelectedText, falling back to getSelectedText on older SDKs.
• Full-page PNG export is only attempted for NOTE files.
• The toolbar action is headless (showType=0) and uses native dialogs for success/failure feedback.
• .env and generated runtime config are intentionally ignored by git.

Intended Uses

• OCR pipelines
• document or note clipping
• search and indexing
• personal knowledge system ingestion
• Anki or flashcard generation
• self-hosted or remote LLM workflows

License

MIT