added support for configuring request headers.

[utils] / crawler / basic / src / org / wamblee / crawler / impl / ConfigurationParser.java

diff --git a/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java b/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java
index b45053a2ab953c32cdec4a4d954220f86e0e1974..3069104d07ec9174d7eeef969a5efcc1905006e7 100644 (file)
--- a/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java
+++ b/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java
@@ -17,7 +17,6 @@
 package org.wamblee.crawler.impl;
 
 import java.io.InputStream;
-import java.io.PrintStream;
 import java.util.ArrayList;
 import java.util.Iterator;
 import java.util.List;
@@ -31,6 +30,7 @@ import org.wamblee.crawler.Configuration;
 import org.wamblee.crawler.GetPageRequest;
 import org.wamblee.crawler.PageRequest;
 import org.wamblee.crawler.PostPageRequest;
+import org.wamblee.xml.XslTransformer;
 
 /**
  * Parsing of the configuration from an XML file.
@@ -48,6 +48,8 @@ public class ConfigurationParser {
     private static final String ELEM_XSLT = "xslt";
 
     private static final String ELEM_PARAM = "param";
+    
+    private static final String ELEM_HEADER = "header";
 
     private static final String AT_NAME = "name";
 
@@ -60,15 +62,14 @@ public class ConfigurationParser {
     private static final int MAX_TRIES = 3;
 
     private static final int MAX_DELAY = 100;
-
-    private PrintStream _os;
     
+    private XslTransformer _transformer;
+
     /**
      * Constructs the configuration parser. 
-     * @param aOs The stream for logging requests
      */
-    public ConfigurationParser(PrintStream aOs) {
-        _os = aOs;
+    public ConfigurationParser(XslTransformer aTransformer) {
+        _transformer = aTransformer; 
     }
 
     /**
@@ -150,21 +151,18 @@ public class ConfigurationParser {
     private PageRequest parseRequestConfig(Element aElem) {
         String method = aElem.elementText(ELEM_METHOD);
         String xslt = aElem.elementText(ELEM_XSLT);
-        List<NameValuePair> params = new ArrayList<NameValuePair>();
-        for (Iterator i = aElem.elementIterator(ELEM_PARAM); i.hasNext();) {
-            Element paramElem = (Element) i.next();
-            NameValuePair param = parseParameter(paramElem);
-            params.add(param);
-        }
-
+        List<NameValuePair> params = parseNameValuePairs(aElem, ELEM_PARAM); 
+        List<NameValuePair> headers = parseNameValuePairs(aElem, ELEM_HEADER);
+        
         NameValuePair[] paramsArray = params.toArray(new NameValuePair[0]);
+        NameValuePair[] headersArray = headers.toArray(new NameValuePair[0]);
         PageRequest request;
         if (METHOD_POST.equals(method)) {
-            request = new PostPageRequest(MAX_TRIES, MAX_DELAY, paramsArray,
-                    xslt);
+            request = new PostPageRequest(MAX_TRIES, MAX_DELAY, paramsArray, headersArray,
+                    xslt, _transformer);
         } else if (METHOD_GET.equals(method) || method == null) {
-            request = new GetPageRequest(MAX_TRIES, MAX_DELAY, paramsArray,
-                    xslt);
+            request = new GetPageRequest(MAX_TRIES, MAX_DELAY, paramsArray, headersArray,
+                    xslt, _transformer);
         } else {
             throw new RuntimeException("Unknown request method '" + method
                     + "'. Only " + METHOD_GET + " and " + METHOD_POST
@@ -173,6 +171,20 @@ public class ConfigurationParser {
         return request;
     }
 
+    /**
+     * @param aElem
+     * @return
+     */
+    private List<NameValuePair> parseNameValuePairs(Element aElem, String aElemName) {
+        List<NameValuePair> headers = new ArrayList<NameValuePair>();
+        for (Iterator i = aElem.elementIterator(aElemName); i.hasNext();) {
+            Element paramElem = (Element) i.next();
+            NameValuePair header = parseParameter(paramElem);
+            headers.add(header);
+        }
+        return headers;
+    }
+
     /**
      * Parses a parameter definition. 
      * @param aParam Parameter.