Added a webView flag in Browser. Supported for IE10+ for now.

Also a few more bot detections.

Author: Pierre

src/main/java/net/pieroxy/ua/detection/BotsHelper.java

@@ -34,10 +34,9 @@ public static String[] getGroups(String regexp, String ua, int ... groups) {
     static private Map<String, Bot> genericBotsLiteral;
     static private Bot genericBotBase = new Bot(Brand.OTHER, BotFamily.ROBOT, "", "");
     static private GenericBot[]genericBotsPatterns = new GenericBot[] {
-        new GenericBot("Mozilla/5\\.0 \\(compatible; ?([^\\);/]+)/([0-9\\.]+); ?(MirrorDetector; )?(\\+? ?https?://[^\\)]+)\\)", new int[]{1,2,4}, true),
+        new GenericBot("Mozilla/5\\.0 \\(compatible; ?([^\\);/]+)/([0-9\\.]+[ab]?); ?(MirrorDetector; )?(\\+? ?https?://[^\\)]+)\\)", new int[]{1,2,4}, true),
         new GenericBot("Mozilla/5\\.0 \\(compatible; ([^\\);/]+)\\-([0-9\\.]+); (\\+? ?https?://[^\\)]+)\\)", new int[]{1,2,3}, true),
-        new GenericBot("Mozilla/5\\.0 \\(compatible; ([^\\);/]+); (\\+? ?https?://[^\\)]+)\\)", new int[]{1,0,2}, true),
-        new GenericBot("Mozilla/5\\.0 \\(compatible; ([^\\);/ ]+) (\\+? ?https?://[^\\)]+)\\)", new int[]{1,0,2}, true),
+        new GenericBot("Mozilla/5\\.0 \\(compatible; ([^\\);/]+);? (\\+? ?https?://[^\\)]+)\\)", new int[]{1,0,2}, true),
         new GenericBot("([^\\(\\);/]+)/([0-9RC\\.]+) \\((\\+?https?://[^\\);]+)\\)( .*)?", new int[]{1,2,3}, true),
         new GenericBot("([^\\(\\);]+) \\((\\+?https?://[^\\);]+)\\)( .*)?", new int[]{1,0,2}, true),
         new GenericBot("([^\\(\\);/]+)/([0-9RC\\.]+) \\(([A-Za-z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,})\\)( .*)?", new int[]{1,2,0}, true),
@@ -274,7 +273,7 @@ else if ((ver=context.getcVersionAfterPattern("Diffbot/", MatchingType.BEGINS,Ma
         } else if ((ver = context.getcVersionAfterPattern("Mail.RU_Bot/",  MatchingType.BEGINS, MatchingRegion.PARENTHESIS))!=null) {
             return new Bot(Brand.MAILRU,BotFamily.CRAWLER,"Mail.ru crawler", ver, consumeUrlAndMozilla(context, "http://go.mail.ru"));
         } else if ((ver = context.getcVersionAfterPattern("MJ12bot/",  MatchingType.BEGINS, MatchingRegion.PARENTHESIS))!=null) {
-            return new Bot(Brand.MAJESTIC12,BotFamily.CRAWLER,"Majestic 12", ver, consumeUrlAndMozilla(context, "http://www.majestic12"));
+            return new Bot(Brand.MAJESTIC12,BotFamily.CRAWLER,"Majestic 12", ver, consumeUrlAndMozilla(context, "http://"));
         } else if ((ver = context.getcVersionAfterPattern("GigablastOpenSource/",  MatchingType.BEGINS, MatchingRegion.REGULAR))!=null) {
             return new Bot(Brand.OTHER,BotFamily.CRAWLER,"GigaBlast Crawler", ver);
         } else if (context.getUA().equals("NetLyzer FastProbe")) {

src/main/java/net/pieroxy/ua/detection/Browser.java

@@ -9,6 +9,7 @@ public class Browser extends VersionedObject {
     BrowserFamily family;
     String description;
     RenderingEngine renderingEngine;
+    boolean inWebView;
 
     /**
      * This is the most detailed constructor of the Browser object, where everything can be specified.
@@ -131,5 +132,13 @@ public String getDescription() {
     public RenderingEngine getRenderingEngine() {
         return renderingEngine;
     }
+    /** @return true if the browser is a webview, false if not or unknown */
+    public boolean isInWebView() {
+        return inWebView;
+    }
+    /** @param value true if the browser is a webview, false if not or unknown */
+    public void setInWebView(boolean value) {
+        inWebView = value;
+    }
 
 }

src/main/java/net/pieroxy/ua/detection/UserAgentDetector.java

@@ -887,6 +887,12 @@ static float tryParseVersionNumber(String s) {
         }
     }
 
+    static void setInternetExplorerWebview(UserAgentContext context, Browser b) {
+        if (context.consume("MSAppHost/",  MatchingType.BEGINS, MatchingRegion.PARENTHESIS)) {
+            b.setInWebView(true);
+        }
+    }
+
     static Browser tryGetIE(UserAgentContext context, String possibleVersions, OS os) {
 
         Browser res=null;
@@ -922,6 +928,10 @@ static Browser tryGetIE(UserAgentContext context, String possibleVersions, OS os
                 res = new Browser(Brand.MICROSOFT,BrowserFamily.IE,"IE",new RenderingEngine(Brand.MICROSOFT, RenderingEngineFamily.TRIDENT, "for IE " + iever, 2), verie);
             }
 
+            if (iever >= 10) {
+                setInternetExplorerWebview(context, res);
+            }
+
             if (possibleVersions.indexOf(String.valueOf((int)Math.floor(iever))+",")==-1) res = null;
 
             if (res != null) {
@@ -990,6 +1000,7 @@ static Browser tryGetIE(UserAgentContext context, String possibleVersions, OS os
                                                new Matcher("Gecko", MatchingType.REGEXP)
                     },
                     MatchingRegion.REGULAR);
+                    setInternetExplorerWebview(context, res);
                 }
             }
             if (res!=null) return res;

src/main/java/net/pieroxy/ua/tooling/UserAgentTester.java

@@ -136,7 +136,9 @@ private static String compare(UserAgentDetectionResult a,
     addErrorReport(result, "browser version", a.getBrowser().getVersion(),
         b.getBrowser().getVersion());
     addErrorReport(result, "browser fullVersion", a.getBrowser().getFullVersion(),
-        b.getBrowser().getFullVersion());
+            b.getBrowser().getFullVersion());
+    addErrorReport(result, "browser inWebView", a.getBrowser().isInWebView(),
+            b.getBrowser().isInWebView());
 
     addErrorReport(result, "device", a.getDevice(), b.getDevice());
     addErrorReport(result, "device architecture", a.getDevice().getArchitecture(),
@@ -197,11 +199,10 @@ private static void addErrorReport(StringBuilder errors, String name,
           .append("'\n");
     } else {
       if (expected != null && actual != null) {
-        if (expected instanceof String || expected instanceof Enum) {
+        if (expected instanceof String || expected instanceof Enum || expected instanceof Boolean) {
           errors.append("\texpected ").append(name).append(" to be '")
               .append(expected.toString()).append("' but was '")
               .append(actual.toString()).append("'\n");
-
         }
       }
     }
@@ -222,7 +223,7 @@ private static class UserAgentDetection {
         comment, ignored_tokens, unknown_tokens, device_arch, browser_vendor,
         os_vendor, bot_family, bot_vendor, bot_description, bot_version,
         bot_url, browser_version, browser_fullVersion, re_brand, re_family,
-        re_version, re_fullversion;
+        re_version, re_fullversion, browser_inwebview;
 
 
     public UserAgentDetection(String line) {
@@ -257,6 +258,7 @@ public UserAgentDetection(String line) {
       re_family = getStringOrNull(elements[30]);
       re_version  = getStringOrNull(elements[31]);
       re_fullversion = getStringOrNull(elements[32]);
+      browser_inwebview = getStringOrNull(elements[33]);
     }
 
     private String getStringOrNull(String s) {
@@ -274,7 +276,7 @@ public UserAgentDetectionResult getDetectionResult() {
             bot_description, bot_version, bot_url);
       }
 
-      return new UserAgentDetectionResult(
+      UserAgentDetectionResult res = new UserAgentDetectionResult(
           new Device(device_arch, Enum.valueOf(DeviceType.class, device_type),
               Enum.valueOf(Brand.class, device_brand), Enum.valueOf(
                   Brand.class, device_manufacturer), device),
@@ -291,7 +293,8 @@ public UserAgentDetectionResult getDetectionResult() {
               OSFamily.class, os_family), os_description, os_version),
           new Locale(lang, country), comment, ignored_tokens, unknown_tokens,
           bot);
-
+      res.getBrowser().setInWebView(browser_inwebview.equals("1"));
+      return res;
     }
   }
 }