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 HTMLFILE_DEFAULT = "index.html";
var CHECKSFILE_DEFAULT = "checks.json";
var URL_DEFAULT = "http://google.com";

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_pro_process_exit_code
    }
    return instr;
};

var getHtmlFromUrl = function(url) {
    rest.get(url).on('complete',function(result) {
    if(result instanceof Error) {
       console.log("An error getting the Url occured");
       this.retry(5000);
    } else {
	checkAndPrint(String(result),program.checks);
    }
});
};

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

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;
	out[checks[ii]] = present;
    }
    return out;
};

var clone = function(fn) {
    //workaround for commander.js issue.
    //http://stackoverflow.com/a/67726648
    return fn.bind({});
};


var checkAndPrint = function(htmlfile,checks){
    var checkJson = checkHtmlFile(htmlfile,checks);
    var outJson = JSON.stringify(checkJson,null,4);
    console.log(outJson);
};

//var mainprog = function(){
if(require.main == module) {
    program
	.option('-c, --checks <check_file>','Path to checks.json',clone(assertFileExists),CHECKSFILE_DEFAULT)
	.option('-f, --file <html_file>','Path to index.html',clone(assertFileExists),HTMLFILE_DEFAULT)
	.option('-u, --url <url>','Url to check')
	.parse(process.argv);
    if(program.url) {
	getHtmlFromUrl(program.url,program.checks);
    } else {
	checkAndPrint(fs.readFileSync(program.file),program.checks);
    }
} else {
    exports.checkHtmlFile = checkHtmlFile;
}