grader.js

#!/usr/bin/env node
/*
Automatically grade files for the presence of specified HTML tags/attributes.
Uses commander.js and cheerio. Teaches command line application development
and basic DOM parsing.

References:

 + cheerio
   - https://github.com/MatthewMueller/cheerio
   - http://encosia.com/cheerio-faster-windows-friendly-alternative-jsdom/
   - http://maxogden.com/scraping-with-node.html

 + commander.js
   - https://github.com/visionmedia/commander.js
   - http://tjholowaychuk.com/post/9103188408/commander-js-nodejs-command-line-interfaces-made-easy

 + JSON
   - http://en.wikipedia.org/wiki/JSON
   - https://developer.mozilla.org/en-US/docs/JSON
   - https://developer.mozilla.org/en-US/docs/JSON#JSON_in_Firefox_2
*/

var fs = require('fs');
var program = require('commander');
var cheerio = require('cheerio');
var rest = require('restler');
//var sys = require('util');
var HTMLFILE_DEFAULT = "index.html";
var CHECKSFILE_DEFAULT = "checks.json";

var assertFileExists = function(infile) {
    var instr = infile.toString();
    if(!fs.existsSync(instr)) {
        console.log("%s does not exist. Exiting.", instr);
        process.exit(1); // http://nodejs.org/api/process.html#process_process_exit_code
    }
    return instr;
};

var cheerioHtmlFile = function(htmlfile) {
    return cheerio.load(fs.readFileSync(htmlfile));
};

//var cheerioHtmlURL = function(URL) {
//    return rest.get(URL).on( 'complete',function(result){ cheerio.load(result); } ) ;
//    rest.get(URL).on('complete',function(result){
//	sys.puts(result);
//    });
//    return cheerio.load(temp);
//};

var loadChecks = function(checksfile) {
    return JSON.parse(fs.readFileSync(checksfile));
};

var checkHtmlFile = function(htmlfile, checksfile) {
    $ = cheerioHtmlFile(htmlfile);
    var checks = loadChecks(checksfile).sort();
    var out = {};
    for(var ii in checks) {
        var present = $(checks[ii]).length > 0;
  //      sys.puts(present);
	out[checks[ii]] = present;
    }
 //   sys.puts(out);
   return out;
};

var checkHtmlURL = function(url, checksfile){
    rest.get(url).on('complete',function(result){
//	sys.puts(result);
	$ = cheerio.load(result);
//	sys.puts($);
	var checks = loadChecks(checksfile).sort();
//	sys.puts(checks);
	var out ={};
	for(var ii in checks){
//	    sys.puts(ii);
//	    sys.puts($(checks[ii]));
	    var present = $(checks[ii]).length > 0;
//	    sys.puts(present);
	    out[checks[ii]] = present;
	    
	}
//	sys.puts("This is the out variable");
//	sys.puts(out);
	var outJson = JSON.stringify(out,null,4);
	console.log(outJson);
	//return out;
    });
};
//    $ = cheerioHtmlURL(url);
//    var checks = loadChecks(checksfile).sort();
//    var out = {};
//    for(var ii in checks) {
//	var present = $(checks[ii]).length > 0;
//	out[checks[ii]] = present;
//    }
//    return out;
//};

var clone = function(fn) {
    // Workaround for commander.js issue.
    // http://stackoverflow.com/a/6772648
    return fn.bind({});
};

if(require.main == module) {
    program
        .option('-f, --file <html_file>', 'Path to index.html', clone(assertFileExists), HTMLFILE_DEFAULT)
        .option('-c, --checks <check_file>', 'Path to checks.json', clone(assertFileExists), CHECKSFILE_DEFAULT)
	.option('-u, --url <url>','Path to URL')
	.parse(process.argv);
    if(program.url) {
//	sys.puts("There's a url!!");
	var checkJson = checkHtmlURL(program.url, program.checks);
    }   else {
//	sys.puts("No url here!!");
	var checkJson = checkHtmlFile(program.file, program.checks);
//	sys.puts("This is Json");
//	sys.puts(checkJson);
	var outJson = JSON.stringify(checkJson, null, 4);
	console.log(outJson);
    }
} else {
    exports.checkHtmlFile = checkHtmlFile;
}