handwriting_answer_pad 0.1.0

A Flutter handwriting capture pad (scribble-based) with undo/redo, clear, render-to-image, and a pluggable recognizer hook — bring your own OCR / vision model. No backend coupling.

handwriting_answer_pad

A Flutter handwriting capture pad — a drawing canvas with undo / redo / clear and a "recognize" action — backed by a pluggable recognizer so you bring your own OCR or vision model. The pad never talks to a backend itself, so it stays decoupled and testable.

Great for math answers, signatures, quick sketches, or anything you want to turn into text.

Install

dependencies:
  handwriting_answer_pad: ^0.1.0

Usage

import 'dart:typed_data';
import 'package:handwriting_answer_pad/handwriting_answer_pad.dart';

HandwritingAnswerPad(
  recognizer: (Uint8List pngBytes) async {
    // Send the rendered handwriting to your OCR / vision model and return text.
    return await myVisionApi.imageToText(pngBytes);
  },
  onResult: (text) => print('Recognized: $text'),
);

The widget renders the strokes to a PNG and hands the bytes to your recognizer; the returned text is shown below the pad. Errors are caught and shown as a friendly message (the raw exception is never leaked to the user).

Programmatic control

final controller = HandwritingPadController(strokeWidth: 4);

HandwritingAnswerPad(controller: controller, recognizer: myRecognizer);

controller.undo();
controller.clear();
final png = await controller.renderPng(); // capture the drawing yourself
controller.dispose();                     // if you created it

Tips

• Pair the output with image_white_background to give vision models an opaque canvas, and with ai_rubric_grader to grade the recognized answer.
• Run recognition (and any API key) server-side; pass only the image up and the text back.

License

MIT