modify the cat parse exception

bee-engine/pom.xml

@@ -4,7 +4,7 @@
 	<parent>
 		<groupId>com.dianping.cat</groupId>
 		<artifactId>parent</artifactId>
-		<version>0.5.0</version>
+		<version>0.5.1</version>
 	</parent>
 	<modelVersion>4.0.0</modelVersion>
 	<groupId>com.dianping.bee</groupId>

cat-consumer/pom.xml

@@ -3,7 +3,7 @@
    <parent>
       <groupId>com.dianping.cat</groupId>
       <artifactId>parent</artifactId>
-      <version>0.5.0</version>
+      <version>0.5.1</version>
    </parent>
    <modelVersion>4.0.0</modelVersion>
    <artifactId>cat-consumer</artifactId>

cat-consumer/src/main/java/com/dianping/cat/consumer/transaction/TransactionReportUrlFilter.java

@@ -3,16 +3,24 @@ package com.dianping.cat.consumer.transaction;
 import java.util.ArrayList;
 import java.util.Collections;
 import java.util.Comparator;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
+
+import org.codehaus.plexus.logging.LogEnabled;
+import org.codehaus.plexus.logging.Logger;
 
 import com.dianping.cat.consumer.transaction.model.entity.TransactionName;
 import com.dianping.cat.consumer.transaction.model.entity.TransactionReport;
 import com.dianping.cat.consumer.transaction.model.entity.TransactionType;
 
-public class TransactionReportUrlFilter extends com.dianping.cat.consumer.transaction.model.transform.DefaultXmlBuilder {
+public class TransactionReportUrlFilter extends com.dianping.cat.consumer.transaction.model.transform.DefaultXmlBuilder
+      implements LogEnabled {
+
+	private int m_maxItems = 200;
 
-	private int m_maxItems = 500;
+	private Logger m_logger;
 
 	private void mergeName(TransactionName old, TransactionName other) {
 		old.setTotalCount(old.getTotalCount() + other.getTotalCount());
@@ -52,6 +60,26 @@ public class TransactionReportUrlFilter extends com.dianping.cat.consumer.transa
 	public void visitType(TransactionType type) {
 		if ("URL".equals(type.getId())) {
 			Map<String, TransactionName> transactionNames = type.getNames();
+
+			Set<String> names = transactionNames.keySet();
+			Set<String> invalidates = new HashSet<String>();
+
+			for (String temp : names) {
+				int length = temp.length();
+
+				for (int i = 0; i < length; i++) {
+					if (temp.charAt(i) > 255 || temp.charAt(i) < 0) {
+						invalidates.add(temp);
+						continue;
+					}
+				}
+			}
+
+			for (String name : invalidates) {
+				m_logger.error("remove invalidate url " + name);
+				transactionNames.remove(name);
+			}
+
 			int size = transactionNames.size();
 
 			if (size > m_maxItems) {
@@ -80,4 +108,9 @@ public class TransactionReportUrlFilter extends com.dianping.cat.consumer.transa
 			return (int) (o2.getTotalCount() - o1.getTotalCount());
 		}
 	}
+
+	@Override
+	public void enableLogging(Logger logger) {
+		m_logger = logger;
+	}
 }
\ No newline at end of file

cat-consumer/src/test/java/com/dianping/cat/consumer/transaction/TransactionReportFilterTest.java

@@ -27,15 +27,24 @@ public class TransactionReportFilterTest {
 
 		System.out.println(report.toString().length());
 		System.out.println("Name size :" + type.getNames().size());
-		
+
 		TransactionReportUrlFilter f1 = new TransactionReportUrlFilter();
 		String filterReport = f1.buildXml(report);
 		TransactionReport newReport = parser.parse(filterReport);
 
 		System.out.println(newReport.toString().length());
+
 		int newSize = newReport.findMachine("10.1.77.193").findType("URL").getNames().size();
 
-		Assert.assertEquals(501, newSize);
+		Assert.assertEquals(201, newSize);
 
+		String url = "/topic/341739Š¬g"";
+
+		size = url.length();
+		for (int i = 0; i < size; i++) {
+			if (url.charAt(i) > 255 || url.charAt(i) < 0) {
+				System.out.println(i + " " + url.charAt(i));
+			}
+		}
 	}
 }

cat-core/pom.xml

@@ -4,7 +4,7 @@
 	<parent>
 		<groupId>com.dianping.cat</groupId>
 		<artifactId>parent</artifactId>
-		<version>0.5.0</version>
+		<version>0.5.1</version>
 	</parent>
 	<modelVersion>4.0.0</modelVersion>
 	<artifactId>cat-core</artifactId>

cat-data/pom.xml

@@ -4,7 +4,7 @@
 	<parent>
 		<groupId>com.dianping.cat</groupId>
 		<artifactId>parent</artifactId>
-		<version>0.5.0</version>
+		<version>0.5.1</version>
 	</parent>
 	<modelVersion>4.0.0</modelVersion>
 	<artifactId>cat-data</artifactId>

cat-hadoop/pom.xml

@@ -4,7 +4,7 @@
 	<parent>
 		<groupId>com.dianping.cat</groupId>
 		<artifactId>parent</artifactId>
-		<version>0.5.0</version>
+		<version>0.5.1</version>
 	</parent>
 	<modelVersion>4.0.0</modelVersion>
 	<artifactId>cat-hadoop</artifactId>

cat-home/pom.xml

@@ -3,7 +3,7 @@
    <parent>
       <groupId>com.dianping.cat</groupId>
       <artifactId>parent</artifactId>
-      <version>0.5.0</version>
+      <version>0.5.1</version>
    </parent>
    <modelVersion>4.0.0</modelVersion>
    <artifactId>cat-home</artifactId>

cat-home/src/test/java/com/dianping/cat/report/analyzer/UrlTransasctionBugTest.java

@@ -37,16 +37,15 @@ public class UrlTransasctionBugTest extends ComponentTestCase {
 	@Test
 	public void test() throws Exception {
 		//fix the transaction xml parse builder
-		String dateStr = "2012-12-13 14:00:00";
+		String dateStr = "2012-12-14 19:00:00";
 		Date date = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss").parse(dateStr);
 
 		List<Report> reports = m_reportDao.findAllByDomainNameDuration(date,
-		      new Date(date.getTime() + TimeUtil.ONE_HOUR), "ShopWeb", "transaction", ReportEntity.READSET_FULL);
+		      new Date(date.getTime() + TimeUtil.ONE_HOUR), "GroupWeb", "transaction", ReportEntity.READSET_FULL);
 		File file = new File("text.txt");
 		for (Report report : reports) {
 			try {
-				TransactionReport temp = DefaultSaxParser.parse(report.getContent());
-				System.out.println(temp);
+				DefaultSaxParser.parse(report.getContent());
 			} catch (Exception e) {
 				Files.forIO().writeTo(file, report.getContent());
 			}

cat-job/pom.xml

@@ -4,7 +4,7 @@
 	<parent>
 		<groupId>com.dianping.cat</groupId>
 		<artifactId>parent</artifactId>
-		<version>0.5.0</version>
+		<version>0.5.1</version>
 	</parent>
 	<modelVersion>4.0.0</modelVersion>
 	<artifactId>cat-job</artifactId>

pom.xml

@@ -4,7 +4,7 @@
 	<modelVersion>4.0.0</modelVersion>
 	<groupId>com.dianping.cat</groupId>
 	<artifactId>parent</artifactId>
-	<version>0.5.0</version>
+	<version>0.5.1</version>
 	<name>arch-cat</name>
 	<description>Central Application Tracking</description>
 	<packaging>pom</packaging>