(no commit message)

[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 dafbc832b5f12583a96f119f6e86953496661c7d..2f2e5f5bf95cdf9cc866c182551da757de820769 100644 (file)
--- a/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java
+++ b/crawler/basic/src/org/wamblee/crawler/impl/ConfigurationParser.java
@@ -36,24 +36,33 @@ import org.wamblee.crawler.PostPageRequest;
  * Parsing of the configuration from an XML file.
  */
 public class ConfigurationParser {
-    
-    private static final String ELEM_URL = "url"; 
+
+    private static final String ELEM_URL = "url";
+
     private static final String ELEM_TYPE = "type";
-    private static final String ELEM_PATTERN = "pattern"; 
-    private static final String ELEM_METHOD= "method"; 
-    private static final String ELEM_XSLT = "xslt"; 
-    private static final String ELEM_PARAM = "param"; 
+
+    private static final String ELEM_PATTERN = "pattern";
+
+    private static final String ELEM_METHOD = "method";
+
+    private static final String ELEM_XSLT = "xslt";
+
+    private static final String ELEM_PARAM = "param";
+
     private static final String AT_NAME = "name";
+
     private static final String AT_VALUE = "value";
-    
+
     private static final String METHOD_POST = "post";
+
     private static final String METHOD_GET = "get";
-    
-    private static final int MAX_TRIES = 3; 
+
+    private static final int MAX_TRIES = 3;
+
     private static final int MAX_DELAY = 5000;
-    
-    private PrintStream _os; 
-    
+
+    private PrintStream _os;
+
     public ConfigurationParser(PrintStream aOs) {
         _os = aOs;
     }
@@ -62,10 +71,10 @@ public class ConfigurationParser {
         try {
             SAXReader reader = new SAXReader();
             Document document = reader.read(aStream);
-            
-            Element root = document.getRootElement(); 
+
+            Element root = document.getRootElement();
             List<UrlConfig> urlConfigs = parseUrlConfigs(root);
-            List<PageTypeConfig> pageTypeConfigs = parsePageTypeConfigs(root); 
+            List<PageTypeConfig> pageTypeConfigs = parsePageTypeConfigs(root);
             return new ConfigurationImpl(urlConfigs, pageTypeConfigs);
         } catch (DocumentException e) {
             throw new RuntimeException("Problem parsing config file", e);
@@ -73,36 +82,36 @@ public class ConfigurationParser {
     }
 
     /**
-     * @param root
+     * @param aRoot
      * @return
      */
-    private List<UrlConfig> parseUrlConfigs(Element root) {
+    private List<UrlConfig> parseUrlConfigs(Element aRoot) {
         List<UrlConfig> configs = new ArrayList<UrlConfig>();
-        for (Iterator i = root.elementIterator(ELEM_URL); i.hasNext(); ) { 
-            Element url = (Element)i.next();
+        for (Iterator i = aRoot.elementIterator(ELEM_URL); i.hasNext();) {
+            Element url = (Element) i.next();
             UrlConfig config = parseUrlConfig(url);
             configs.add(config);
         }
         return configs;
     }
-    
-    private List<PageTypeConfig> parsePageTypeConfigs(Element root) {
+
+    private List<PageTypeConfig> parsePageTypeConfigs(Element aRoot) {
         List<PageTypeConfig> configs = new ArrayList<PageTypeConfig>();
-        for (Iterator i = root.elementIterator(ELEM_TYPE); i.hasNext(); ) { 
-            Element url = (Element)i.next();
+        for (Iterator i = aRoot.elementIterator(ELEM_TYPE); i.hasNext();) {
+            Element url = (Element) i.next();
             PageTypeConfig config = parsePageTypeConfig(url);
             configs.add(config);
         }
         return configs;
     }
-    
-    private UrlConfig parseUrlConfig(Element aUrlElem) { 
+
+    private UrlConfig parseUrlConfig(Element aUrlElem) {
         String pattern = aUrlElem.elementText(ELEM_PATTERN);
         PageRequest request = parseRequestConfig(aUrlElem);
         return new UrlConfig(pattern, request);
     }
-    
-    private PageTypeConfig parsePageTypeConfig(Element aTypeElem) { 
+
+    private PageTypeConfig parsePageTypeConfig(Element aTypeElem) {
         String pattern = aTypeElem.elementText(ELEM_PATTERN);
         PageRequest request = parseRequestConfig(aTypeElem);
         return new PageTypeConfig(pattern, request);
@@ -113,30 +122,32 @@ public class ConfigurationParser {
      * @return
      */
     private PageRequest parseRequestConfig(Element aUrlElem) {
-        String method = aUrlElem.elementText(ELEM_METHOD); 
+        String method = aUrlElem.elementText(ELEM_METHOD);
         String xslt = aUrlElem.elementText(ELEM_XSLT);
         List<NameValuePair> params = new ArrayList<NameValuePair>();
-        for (Iterator i = aUrlElem.elementIterator(ELEM_PARAM); i.hasNext(); ) { 
-            Element paramElem = (Element)i.next();
+        for (Iterator i = aUrlElem.elementIterator(ELEM_PARAM); i.hasNext();) {
+            Element paramElem = (Element) i.next();
             NameValuePair param = parseParameter(paramElem);
             params.add(param);
         }
-    
+
         NameValuePair[] paramsArray = params.toArray(new NameValuePair[0]);
-        PageRequest request; 
-        if ( METHOD_POST.equals(method)) { 
-            request = new PostPageRequest(MAX_TRIES, MAX_DELAY, paramsArray, xslt, _os);
-        }
-        else if ( METHOD_GET.equals(method) || method == null ){
-            request = new GetPageRequest(MAX_TRIES, MAX_DELAY, paramsArray, xslt, _os);
-        } else { 
-            throw new RuntimeException("Unknown request method '" + method + "'. Only " + 
-                    METHOD_GET + " and " + METHOD_POST + " are supported");
+        PageRequest request;
+        if (METHOD_POST.equals(method)) {
+            request = new PostPageRequest(MAX_TRIES, MAX_DELAY, paramsArray,
+                    xslt, _os);
+        } else if (METHOD_GET.equals(method) || method == null) {
+            request = new GetPageRequest(MAX_TRIES, MAX_DELAY, paramsArray,
+                    xslt, _os);
+        } else {
+            throw new RuntimeException("Unknown request method '" + method
+                    + "'. Only " + METHOD_GET + " and " + METHOD_POST
+                    + " are supported");
         }
         return request;
     }
-    
-    private NameValuePair parseParameter(Element aParam) { 
+
+    private NameValuePair parseParameter(Element aParam) {
         String name = aParam.attributeValue(AT_NAME);
         String value = aParam.attributeValue(AT_VALUE);
         return new NameValuePair(name, value);