Important

Only works with PNG screenshots of NYT grids (currently).

Todo

• document the pre processing command
• test the Grid.String() method
• test the /read-grid endpoint
• get a "best attempt" OCR flow going again
• support Sudoku.com grids
• make it work for other file formats

summary

• api.go -> basic web-server to expose grid processing
• grid.go -> identifies the grid boundaries, splits out each cell into it's on entity, orchestrates cell processing via grid_worker.go
• grid_worker.go -> thread pool of cell processors, is orchestrated by the grid, calls processing methods on each cell
• cell.go -> in-charge of placeholder and value identification, manages pre-processing via grid_image.go
• grid_image.go -> low-level wrapper around image.Image, executes image pre-processing via ImageMagick, executes OCR via Tesseract

starting

docker

• docker build . -t grid-reader
• docker run -p 8080:8080 grid-reader
• curl --form file='@grids/3/grid.png' localhost:8080/read-grid

local

• you'll need imagemagick installed (https://github.com/gographics/imagick)
• go run main.go