improved benchmark suite

YahooArchive · Aug 30, 2015 · 45dd4c0 · 45dd4c0
1 parent fa24fa2
commit 45dd4c0
Show file tree

Hide file tree

Showing 2 changed files with 71 additions and 48 deletions.
diff --git a/bin/html-purifier.js b/bin/html-purifier.js
@@ -8,65 +8,87 @@ See the accompanying LICENSE file for terms.
 /*
 This utility prints out sanitized html content
 */
-var Debug = require("debug"),
-    progname = 'HTML-Purifier';
-var Purifier = require('../src/html-purify');
-
-Debug.enable(progname);
+var Purifier = require('../src/html-purify'),
+    Benchmark = require('benchmark');
 
 (function() {
     var fs = require('fs'),
-    file,
-    lineByLine = 0,
-    noofargs = 0;
+        file,
+        lineByLine = false,
+        benchmark = false,
+        noofargs = process.argv.length;
 
     process.argv.forEach(function(val, index) {
-        ++noofargs;
         if (index === 2) {
             file = val;
-        }
-	if (index === 3) {
+        } else if (index === 3) {
             if (val === "-l") {
-	        lineByLine = 1;
-	    }
-	}
+                lineByLine = true;
+            } else if (val === "--benchmark") {
+                benchmark = true;
+            }
+        }
     });
 
-    if (noofargs >= 3) {
-        if (fs.existsSync(file)) {
-            var data = fs.readFileSync(file, 'utf-8');
-	    	var i;
-	    	var output = '';
-	    	if (lineByLine) {
-	        	// reading and processing line by line
-	        	var data2 = data.split(/\n/);
-	        	for (i = 0; i < data2.length; i++) {
-		    		if (data2[i].length !== 0) {
-	                	output = (new Purifier()).purify(data2[i]);
-		        		console.log("*****");
-	                	console.log("input  ==> " + data2[i]);
-	                	console.log("output ==> " + output);
-		    		}
-				}
 
-	    	} else {
-				var start = +new Date();
-				output = (new Purifier()).purify(data);
-	    		/*for (i = 0; i < 100; i++) {
-	        		output = (new Purifier()).purify(data);
-	    		}*/	
-				var end = +new Date();
-	    		console.log(output);
-				//console.log("html-purify runs at a speed of " + 10/((end - start)/1000) + " MB per seconds [" + (end-start)/10/1000 + " second per MB].");
-	    	}
-	    process.exit(0);
-	} else {
-		console.log("[ERROR] "+file+" not exist");
-		process.exit(1);
-	}
-    } else {
-	    console.log("Usage: html-purifier <any html file>");
-	    process.exit(1);
+    if (noofargs < 3) {
+        console.log("Usage: html-purifier <any html file>");
+        process.exit(1);
+    }
+
+    if (!fs.existsSync(file)) {
+        console.log("[ERROR] "+file+" not exist");
+        process.exit(1);
+    }
+
+    var data = fs.readFileSync(file, 'utf-8'), i, output = '';
+
+    // The following is disabled as it might generate insecure html, since linebreak can exists within attr value state, while purify() assumes data state for each call
+    // if (lineByLine) {
+    //     // reading and processing line by line
+    //     var data2 = data.split(/\n/);
+    //     for (i = 0; i < data2.length; i++) {
+    //         if (data2[i].length !== 0) {
+    //             output = (new Purifier()).purify(data2[i]);
+    //             console.log("*****");
+    //             console.log("input  ==> " + data2[i]);
+    //             console.log("output ==> " + output);
+    //         }
+    //     }
+    //     process.exit(0);
+    // }
+
+    if (!benchmark) {
+        console.log((new Purifier()).purify(data));
+    	process.exit(0);
+    } 
+    else if (benchmark) {
+
+        var suite = new Benchmark.Suite;
+        var purifier1a = new Purifier();
+        var purifier1b = new Purifier({enableTagBalancing:false});
+
+        suite.add('default', function() {
+            purifier1a.purify(data);
+        })
+        .add('disabled tag balancing', function() {
+            purifier1b.purify(data);
+        })
+        // add listeners
+        .on('cycle', function(event) {
+            console.log(String(event.target));
+        })
+        .on('complete', function() {
+            console.log('Fastest is     ', this.filter('fastest').pluck('name'));
+
+            var t = this.filter('fastest')[0].stats.mean;
+            console.log('Speed/Time is  ', data.length/1000000/t + 'MB/s', t + 's');
+        })
+        // run async
+        .run({
+        	// 'minSamples': 10,
+        	'async': true
+        });
     }
 
 }).call(this);
diff --git a/package.json b/package.json
@@ -35,6 +35,7 @@
     "css-js": "1.0.3"
   },
   "devDependencies": {
+    "benchmark": "^1.0.0",
     "expect.js": "^0.3.1",
     "grunt": "^0.4.5",
     "grunt-browserify": "^3.8.0",