jruberg / infersudoku

Sudoku solver using human-level inference

InferSudoku is a Coffeescript project to solve Sudoku puzzles using human-level inference. It was written as a final project for COMP360 - Topics in Artificial Intelligence at Wesleyan University in the Spring 2011 semester. The following are instructions that were provided to the instructor when the project was submitted for a grade.

---

This project is written in Coffeescript, a language which compiles to javascript.

The original source (in coffeescript) is included along with compiled javascript files which will run in a web browser; both types of files are included in the js/ directory.

The coffeescript source files are also viewable in a pretty annotated source format (via Docco) in HTML documents in the docs/ directory; these should be viewable if you navigate a web browser to these HTML files.

The CSS for the HTML page was written in SASS, a language which compiles to CSS. The compiled CSS and the source SASS are both provided in the css/ directory.

To view the app, navigate to html/sudoku.html in a web browser (preferrably Chrome, the app was tested solely in Chrome and I did not have time to test it in other browsers, so it may look funny or do funny things).

Both Coffeescript and SASS have the ability to be compiled by the browser instead of beforehand, but it involves including the appropriate compiler javascript files. If the grader would prefer this so the source code can be tinkered with without recompiling each time, then let me know and I can set up with this model instead.

The bookmarklet used to gather sudoku data was as follows: javascript:(function(){var s="";for(i=0;i<9;i++){for(j=0;j<9;j++){var v=document.getElementById('f'+i+j).value;s+=v==""?".":v}s+="\n";}var pre=document.createElement('pre');pre.innerHTML=s;var body=document.getElementsByTagName('body')[0];body.insertBefore(pre,body.children[0]);})()