Make getCharset to support null parameter.

Author: sutra

webmagic-core/src/main/java/us/codecraft/webmagic/utils/UrlUtils.java

@@ -116,6 +116,10 @@ public static List<String> convertToUrls(Collection<Request> requests) {
     private static final Pattern patternForCharset = Pattern.compile("charset\\s*=\\s*['\"]*([^\\s;'\"]*)", Pattern.CASE_INSENSITIVE);
 
     public static String getCharset(String contentType) {
+        if (contentType == null) {
+            return null;
+        }
+
         Matcher matcher = patternForCharset.matcher(contentType);
         if (matcher.find()) {
             String charset = matcher.group(1);

webmagic-core/src/test/java/us/codecraft/webmagic/utils/UrlUtilsTest.java

@@ -1,5 +1,7 @@
 package us.codecraft.webmagic.utils;
 
+import static org.junit.Assert.assertNull;
+
 import org.junit.Assert;
 import org.junit.Test;
 
@@ -43,5 +45,9 @@ public void testGetDomain(){
         Assert.assertEquals("www.dianping.com",UrlUtils.getDomain(url));
     }
 
+    @Test
+    public void testGetCharset() {
+        assertNull(UrlUtils.getCharset(null));
+    }
 
 }