slim down

CHANGELOG.md

@@ -1,5 +1,9 @@
 ## Version 2.0
 
+- **v2.4.3**
+
+  - Slimmed down the size/complexity of the built regexps.js file.  The regexps are now stored in a normal array.
+
 - **v2.4.0**
 
   - Forked and renamed to useragent-ng

benchmark/package.json

@@ -6,6 +6,7 @@
         "microtime": "",
         "benchmark": "",
         "useragent": "2.3.0",
+        "useragent-ng": "2.4.1",
         "ua-parser": "0.3.1",
         "useragent_parser": "1.0.0",
         "useragent-parser": "0.1.1"

benchmark/regression.js

@@ -13,7 +13,8 @@ var microtime = require('microtime')
 /**
  * Useragent parsers.
  */
-var useragent2 = require('../')
+var useragentNg2 = require('../')
+  , useragentNg = require('useragent-ng')
   , useragent = require('useragent');
 
 /**
@@ -37,12 +38,17 @@ var useragentlist = path.join(__dirname, '..', 'test', 'fixtures', file+'.yaml')
 var froomfroom = new benchmark.Suite;
 
 froomfroom
-.add('useragent latest', function () {
+.add('useragent-ng latest', function () {
   for (var i = 0; i < length; i++ ) {
-    useragent2.parse(testcases[i]);
+    useragentNg2.parse(testcases[i]);
   }
 })
-.add('useragent1', function () {
+.add('useragent-ng npm', function () {
+  for (var i = 0; i < length; i++ ) {
+    useragentNg.parse(testcases[i]);
+  }
+})
+.add('useragent classic', function () {
   for (var i = 0; i < length; i++ ) {
     useragent.parse(testcases[i]);
   }

lib/build.js

@@ -23,15 +23,19 @@ function getKnown() {
     return null;
   }
 
-  var old = require('./regexps');
   var known = new Set();
-  for (var type in old) {
-    if (!old.hasOwnProperty(type)) continue;
-    var group = old[type];
-    for (var i = 0; i < group.length; i++) {
-      var regex = group[i][0].source;
-      known.add(regex);
+  try {
+    var old = require('./regexps');
+    for (var type in old) {
+      if (!old.hasOwnProperty(type)) continue;
+      var group = old[type];
+      for (var i = 0; i < group.length; i++) {
+        var regex = group[i][0].source;
+        known.add(regex);
+      }
     }
+  } catch (e) {
+    console.warn("Error parsing old regexps.js file.  Ignoring and continuing...", e)
   }
   return known;
 }
@@ -150,76 +154,77 @@ exports.parse = function parse(sources, callback) {
     results[details.resource] = results[details.resource] || [];
 
     var resources = data[details.resource]
-      , name = details.resource.replace('_parsers', '')
       , resource
       , regex
-      , source
-      , parser;
+      , parser
+      , compiled;
 
     for (var i = 0, l = resources.length; i < l; i++) {
       resource = resources[i];
       regex = resource.regex;
 
-      source = new RegExp(regex).source;
-      if (known && !known.has(source)) {
+
+      compiled = new RegExp(regex, resource.regex_flag);
+      if (known && !known.has(compiled.source)) {
         // A quick check, regexes not matching those are clearly safe
         // This check excludes about 35% of all regexps we have
-        if (!isSafe(source)) {
-          unsafe.push(source);
+        if (!isSafe(compiled.source)) {
+          unsafe.push(compiled.source);
         }
-        known.add(source);
+        known.add(compiled.source);
       }
 
       // We need to JSON stringify the data to properly add slashes escape other
       // kinds of crap in the RegularExpression. If we don't do thing we get
-      // some illegal token warnings.
-      parser = 'parser = Object.create(null);\n';
-      parser += 'parser[0] = new RegExp('+ JSON.stringify(regex)
-      if (resource.regex_flag) {
-        parser += ',"' + resource.regex_flag + '"';
-      }
-      parser += ');\n';
+      // some illegal token warnings./(Linux)(?:[ /](\d+)\.(\d+)(?:\.(\d+)|)|)/
+
+      // parser = 'parser = Object.create(null);\n';
+      parser = '  [' + compiled.toString();
 
       // Check if we have replacement for the parsed family name
       if (resource[details.replacement]) {
-        parser += 'parser[1] = "'+ resource[details.replacement].replace('"', '\\"') +'";\n';
+        parser += ', "'+ resource[details.replacement].replace('"', '\\"') +'"';
       } else {
-        parser += 'parser[1] = 0;\n';
+        parser += ', 0';
       }
 
       if (details.name === 'device') {
         if (resource.brand_replacement) {
-          parser += 'parser[2] = "'+ resource.brand_replacement.replace('"', '\\"') +'";\n';
+          parser += ', "'+ resource.brand_replacement.replace('"', '\\"') +'"';
         } else if (resource.model_replacement) {
-          parser += 'parser[2] = 0;\n';
+          parser += ', 0';
         }
 
         if (resource.model_replacement) {
-          parser += 'parser[3] = "'+ resource.model_replacement.replace('"', '\\"') +'";\n';
+          parser += ', "'+ resource.model_replacement.replace('"', '\\"') +'"';
         }
       } else {
         if (resource.v1_replacement) {
-          parser += 'parser[2] = "'+ resource.v1_replacement.replace('"', '\\"') +'";\n';
+          parser += ', "'+ resource.v1_replacement.replace('"', '\\"') +'"';
         } else if (resource.v2_replacement || resource.v3_replacement) {
-          parser += 'parser[2] = 0;\n';
+          parser += ', 0';
         }
 
         if (resource.v2_replacement) {
-          parser += 'parser[3] = "'+ resource.v2_replacement.replace('"', '\\"') +'";\n';
+          parser += ', "'+ resource.v2_replacement.replace('"', '\\"') +'"';
         } else if (resource.v3_replacement) {
-          parser += 'parser[3] = 0;\n';
+          parser += ', 0';
         }
 
         if (resource.v3_replacement) {
-          parser += 'parser[4] = "'+ resource.v3_replacement.replace('"', '\\"') +'";\n';
+          parser += ', "'+ resource.v3_replacement.replace('"', '\\"') +'"';
         }
       }
 
-      parser += 'exports.'+ details.name +'['+ i +'] = parser;';
+      parser += ']';
+
+      // parser += 'exports.'+ details.name +'['+ i +'] = parser;';
       results[details.resource].push(parser);
     }
   });
 
+  // TODO: consider getting rid of these checks - we either need to scan the REs for REDoS vulnerabilities
+  // ourselves OR consider uap-core to be trustworthy-enough.  (The latter seems fine)
   if (unsafe.length > 0) {
     console.log('There are new regexps! Here they are, one per line:');
     for (var i = 0; i < unsafe.length; i++) {
@@ -245,18 +250,17 @@ exports.generate = function generate(results, callback) {
   var regexps  = [
       '"use strict";'
     , exports.LEADER
-    , 'var parser;'
-    , 'exports.browser = Object.create(null);'
-    , results.user_agent_parsers.join('\n')
-    , 'exports.browser.length = '+ results.user_agent_parsers.length +';'
-
-    , 'exports.device = Object.create(null);'
-    , results.device_parsers.join('\n')
-    , 'exports.device.length = '+ results.device_parsers.length +';'
-
-    , 'exports.os = Object.create(null);'
-    , results.os_parsers.join('\n')
-    , 'exports.os.length = '+ results.os_parsers.length +';'
+    , 'exports.browser = ['
+    , results.user_agent_parsers.join(',\n')
+    , '];'
+
+    , 'exports.device = ['
+    , results.device_parsers.join(',\n')
+    , '];'
+
+    , 'exports.os = ['
+    , results.os_parsers.join(',\n')
+    , '];'
   ].join('\n\n');
 
   // Now that we have generated the structure for the RegExps export file we