math

npm install

node server.js

npm run start

For Mobile AR

HTTPS=true npm start

For Node.js

When getting the node-canvas error for M1 Mac

brew install pkg-config cairo pango libpng jpeg giflib librsvg

For Python

brew install swig

pip install trace_skeleton
pip install svgwrite

For CnSTD

pyenv install 3.10
pyenv global 3.10
pip install cnstd cnocr
cnstd analyze -m mfd -i sample-1.jpg -o output-1.jpg

For Google OCR

Get key.json

node lib/google-ocr.js

Preparation Pipeline

You can test and convert your textbook data for free with the following:

Step 1. Get a sample PDF page (e.g., /public/sample/sample-01.pdf)
Step 2. Convert PDF to JPG with Ezgif and save it (e.g., /public/sample/sample-01.jpg)
Step 3. Perform OCR with Google Cloud Vision API and save response as a JSON file (e.g., /public/sample/ocr-01.json)
Step 4. Perform math OCR with MathPix Web Snip Tool and save it as a markdown file (e.g., /public/sample/mathpix-01.md)
Step 5. Perform CnST by running $ python lib/math-extract.py
Step 6. Get figure contour with $ python lib/figure-extract.py
Step 7. Get figure line trace with $ node lib/figure-extract.js

if you are willing to pay for the API usage, please use lib/google-ocr.js and lib/mathpix.js

TexToSVG MathML

mi: [x, y], mo: [+, =, ()], mn: [1, 2, 3], msup: [^2]
1D466: y, 1D465: x, ...
30: 0, 31: 1, 32: 2, ...
- x^2 = msup-mi-1D465, msup-mn-32
- 10  = mn-31-30
- \sqrt{x} = msqrt-mo-221A, msqrt-mi-1D465