more robustness, now a detailed report is always sent, also if crawling

[utils] / crawler / basic / src / org / wamblee / crawler / AbstractPageRequest.java

diff --git a/crawler/basic/src/org/wamblee/crawler/AbstractPageRequest.java b/crawler/basic/src/org/wamblee/crawler/AbstractPageRequest.java
index 66627ca523faa4c56a112811e7ccdf9381107825..432ebb4ac4f72010f30bd5371a15cce2d48e4f42 100644 (file)
--- a/crawler/basic/src/org/wamblee/crawler/AbstractPageRequest.java
+++ b/crawler/basic/src/org/wamblee/crawler/AbstractPageRequest.java
@@ -39,9 +39,8 @@ import org.apache.xml.serialize.OutputFormat;
 import org.apache.xml.serialize.XMLSerializer;
 import org.w3c.dom.Document;
 import org.w3c.tidy.Tidy;
-import org.wamblee.xml.ClasspathUriResolver;
-import org.wamblee.xml.DOMUtility;
-import org.wamblee.xml.XSLT;
+import org.wamblee.xml.DomUtils;
+import org.wamblee.xml.XslTransformer;
 
 /**
  * General support claas for all kinds of requests.
@@ -59,6 +58,8 @@ public abstract class AbstractPageRequest implements PageRequest {
     private NameValuePair[] _params;
 
     private String _xslt;
+    
+    private XslTransformer _transformer; 
 
     /**
      * Constructs the request.
@@ -73,7 +74,7 @@ public abstract class AbstractPageRequest implements PageRequest {
      *            XSLT used to convert the response.
      */
     protected AbstractPageRequest(int aMaxTries, int aMaxDelay,
-            NameValuePair[] aParams, String aXslt) {
+            NameValuePair[] aParams, String aXslt, XslTransformer aTransformer) {
         if (aParams == null) {
             throw new IllegalArgumentException("aParams is null");
         }
@@ -84,6 +85,7 @@ public abstract class AbstractPageRequest implements PageRequest {
         _maxDelay = aMaxDelay;
         _params = aParams;
         _xslt = aXslt;
+        _transformer = aTransformer;
     }
 
     /*
@@ -154,9 +156,8 @@ public abstract class AbstractPageRequest implements PageRequest {
             aMethod = executeWithRedirects(aClient, aMethod);
             byte[] xhtmlData = getXhtml(aMethod);
 
-            XSLT xsltProcessor = new XSLT(new ClasspathUriResolver());
-            Document transformed = xsltProcessor.transform(xhtmlData,
-                    xsltProcessor.resolve(_xslt));
+            Document transformed = _transformer.transform(xhtmlData,
+                    _transformer.resolve(_xslt));
             ByteArrayOutputStream os = new ByteArrayOutputStream(); 
             Transformer transformer = TransformerFactory.newInstance()
                     .newTransformer();
@@ -196,7 +197,7 @@ public abstract class AbstractPageRequest implements PageRequest {
         // in a system wide way.
         ByteArrayOutputStream os = new ByteArrayOutputStream(); 
         Document w3cDoc = tidy.parseDOM(aMethod.getResponseBodyAsStream(), os);
-        DOMUtility.removeDuplicateAttributes(w3cDoc);
+        DomUtils.removeDuplicateAttributes(w3cDoc);
         LOG.debug("Content of response is \n" + os.toString()); 
 
         ByteArrayOutputStream xhtml = new ByteArrayOutputStream();
@@ -251,7 +252,7 @@ public abstract class AbstractPageRequest implements PageRequest {
             // recursion.
         }
         default: {
-            throw new RuntimeException("Method failed: "
+            throw new IOException("Method failed: "
                     + aMethod.getStatusLine());
         }
         }