cmailleux 2005/09/22 14:42:46 CEST
Modified files:
war/src/java/com/codeva/webapps/webclipping Rewriter.java
war/src/java/com/codeva/webapps/webclipping/servlet
WebClippingServlet.java
Log:
Add management of area tag
Add more accurate error handling
Revision Changes Path
1.12 +97 -18
webclipping/war/src/java/com/codeva/webapps/webclipping/Rewriter.java
http://jahia.mine.nu:8080/cgi-bin/cvsweb.cgi/webclipping/war/src/java/com/codeva/webapps/webclipping/Rewriter.java.diff?r1=1.11&r2=1.12&f=h
1.16 +7 -4
webclipping/war/src/java/com/codeva/webapps/webclipping/servlet/WebClippingServlet.java
http://jahia.mine.nu:8080/cgi-bin/cvsweb.cgi/webclipping/war/src/java/com/codeva/webapps/webclipping/servlet/WebClippingServlet.java.diff?r1=1.15&r2=1.16&f=h
Index: Rewriter.java
===================================================================
RCS file:
/home/cvs/repository/webclipping/war/src/java/com/codeva/webapps/webclipping/Rewriter.java,v
retrieving revision 1.11
retrieving revision 1.12
diff -u -r1.11 -r1.12
--- Rewriter.java 4 Jul 2005 09:56:32 -0000 1.11
+++ Rewriter.java 22 Sep 2005 12:42:45 -0000 1.12
@@ -148,7 +148,7 @@
* @param responseBody
* @param request
* @param response
- * @return
+ * @return outpuDocument
*/
public OutputDocument rewriteBody(String responseBody,
HttpServletRequest request,
@@ -171,6 +171,7 @@
rewriteBackgroundAttribute(source, stringBuffer, document);
rewriteFormTag(source, stringBuffer, request, response, document);
rewriteATag(source, stringBuffer, request, response, document);
+ rewriteAreaTag(source, stringBuffer, request, response, document);
rewriteImgTag(source, stringBuffer, document);
rewriteInputImageTag(source, stringBuffer, document);
rewriteLinkTag(source, stringBuffer, document);
@@ -196,11 +197,11 @@
} else {
body = (StartTag) new
Source(document.toString()).findAllStartTags(Tag.BODY).get(0);
}
- buffer.append('<' + body.getName() + ' ');
+ buffer.append('<').append(body.getName()).append(' ');
Iterator attributes = body.getAttributes().iterator();
while (attributes.hasNext()) {
Attribute attribute = (Attribute) attributes.next();
- buffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
buffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
buffer.append('>');
document.add(new StringOutputSegment(body, buffer.toString()));
@@ -242,7 +243,7 @@
* @param request
* @param response
* @param sourceUrl
- * @return
+ * @return string
*/
private String getRewritedUrl(HttpServletRequest request,
HttpServletResponse response, String sourceUrl)
throws MalformedURLException {
@@ -319,7 +320,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if
(!"href".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() +
"=" + attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
} else {
@@ -329,7 +330,7 @@
Attribute attribute = (Attribute) atList.next();
if (!"href".equalsIgnoreCase(attribute.getKey())
&&
!"target".equalsIgnoreCase(attribute.getKey())) {
- stringBuffer.append(attribute.getName() +
"=" + attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
}
@@ -345,7 +346,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if
(!"href".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() +
"=" + attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -357,7 +358,85 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if
(!"target".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() +
"=" + attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
+ }
+ }
+ stringBuffer.append('>');
+ document.add(new StringOutputSegment(startTag,
+ stringBuffer.toString()));
+ }
+ }
+ }
+ }
+ }
+
+ private void rewriteAreaTag(Source source, StringBuffer stringBuffer,
HttpServletRequest httpServletRequest,
+ HttpServletResponse response, OutputDocument
document) throws MalformedURLException {
+ List aStartTag = source.findAllStartTags(Tag.AREA);
+ for (int i = 0; i < aStartTag.size(); i++) {
+ StartTag startTag = (StartTag) aStartTag.get(i);
+ Attributes attributes = startTag.getAttributes();
+ Attribute href = attributes.get("href");
+ if (href != null && href.getValue().length() > 0) {
+ String hrefUrl = href.getValue().trim();
+ log.debug("Now we must rewrite this anchor tag for url : " +
hrefUrl);
+ stringBuffer.setLength(0);
+ if (!hrefUrl.toLowerCase().startsWith("http") &&
!hrefUrl.toLowerCase().startsWith("javascript")
+ && !hrefUrl.toLowerCase().startsWith("mailto") &&
!hrefUrl.toLowerCase().startsWith("ftp")
+ && !hrefUrl.toLowerCase().startsWith("news") &&
!hrefUrl.toLowerCase().startsWith("wais")
+ && !hrefUrl.toLowerCase().startsWith("gopher")) {
+ Matcher matcher = pattern.matcher(hrefUrl);
+ if (!matcher.find()) {
+ String rewritedUrl =
getRewritedUrl(httpServletRequest, response,
+ hrefUrl.replaceAll("\\?", "&"));
+ final Attribute target = attributes.get("target");
+ if (target != null &&
!target.getValue().equals("_self")) {
+ rewritedUrl = getAbsoluteURL(hrefUrl);
+ }
+ stringBuffer.append("<area
href=\"").append(rewritedUrl).append("\" ");
+ Iterator atList = attributes.iterator();
+ while (atList.hasNext()) {
+ Attribute attribute = (Attribute) atList.next();
+ if
(!"href".equalsIgnoreCase(attribute.getName())) {
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
+ }
+ }
+ } else {
+ stringBuffer.append("<area target=\"new\"
href=\"").append(getAbsoluteURL(hrefUrl)).append("\" ");
+ Iterator atList = attributes.iterator();
+ while (atList.hasNext()) {
+ Attribute attribute = (Attribute) atList.next();
+ if (!"href".equalsIgnoreCase(attribute.getKey())
&&
+
!"target".equalsIgnoreCase(attribute.getKey())) {
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
+ }
+ }
+ }
+ stringBuffer.append('>');
+ document.add(new StringOutputSegment(startTag,
+ stringBuffer.toString()));
+ } else if (hrefUrl.toLowerCase().startsWith("http")) {
+ Matcher matcher = pattern.matcher(hrefUrl);
+ URL tmp = new URL(hrefUrl);
+ if (!matcher.find() &&
tmp.getHost().equalsIgnoreCase(urlProperties.getHost())) {
+ stringBuffer.append("<area
href=\"").append(getRewritedUrl(httpServletRequest, response,
hrefUrl)).append("\" ");
+ Iterator atList = attributes.iterator();
+ while (atList.hasNext()) {
+ Attribute attribute = (Attribute) atList.next();
+ if
(!"href".equalsIgnoreCase(attribute.getName())) {
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
+ }
+ }
+ stringBuffer.append('>');
+ document.add(new StringOutputSegment(startTag,
+ stringBuffer.toString()));
+ } else {
+ stringBuffer.append("<area target=\"new\" ");
+ Iterator atList = attributes.iterator();
+ while (atList.hasNext()) {
+ Attribute attribute = (Attribute) atList.next();
+ if
(!"target".equalsIgnoreCase(attribute.getName())) {
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -381,12 +460,12 @@
String hrefUrl = href.getValue().trim();
stringBuffer.setLength(0);
if (!hrefUrl.startsWith("http")) {
- stringBuffer.append("<" + startTag.getName() + "
background=\"").append(getAbsoluteURL(hrefUrl)).append("\" ");
+
stringBuffer.append("<").append(startTag.getName()).append("
background=\"").append(getAbsoluteURL(hrefUrl)).append("\" ");
Iterator atList = attributes.iterator();
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if
(!"background".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() +
"=" + attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -416,7 +495,7 @@
Attribute attribute = (Attribute) atList.next();
final String name = attribute.getName();
if (!"action".equalsIgnoreCase(name) &&
!"method".equalsIgnoreCase(name)) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
Attribute method = attributes.get("method");
@@ -443,7 +522,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if (!"src".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -470,7 +549,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if (!"src".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -497,7 +576,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if (!"src".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -524,7 +603,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if (!"src".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -550,7 +629,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if (!"href".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -577,7 +656,7 @@
while (atList.hasNext()) {
Attribute attribute = (Attribute) atList.next();
if (!"src".equalsIgnoreCase(attribute.getName())) {
- stringBuffer.append(attribute.getName() + "=" +
attribute.getQuoteChar() + attribute.getValue() +
attribute.getQuoteChar()).append(' ');
+
stringBuffer.append(attribute.getName()).append("=").append(attribute.getQuoteChar()).append(attribute.getValue()).append(attribute.getQuoteChar()).append('
');
}
}
stringBuffer.append('>');
@@ -603,7 +682,7 @@
*
* @param document
* @param request
- * @return
+ * @return ouputDocument
*/
private OutputDocument trunkDocument(OutputDocument document,
HttpServletRequest request) {
// Get the context id
Index: WebClippingServlet.java
===================================================================
RCS file:
/home/cvs/repository/webclipping/war/src/java/com/codeva/webapps/webclipping/servlet/WebClippingServlet.java,v
retrieving revision 1.15
retrieving revision 1.16
diff -u -r1.15 -r1.16
--- WebClippingServlet.java 13 Jul 2005 14:56:36 -0000 1.15
+++ WebClippingServlet.java 22 Sep 2005 12:42:45 -0000 1.16
@@ -80,7 +80,7 @@
//
} catch (Throwable e) {
perfLog.error("Error doGET", e);
- throw new ServletException("error during execution of get
method");
+ throw new ServletException("error during execution of get method
: "+ e.getLocalizedMessage());
}
}
@@ -110,7 +110,7 @@
//
} catch (Throwable e) {
perfLog.error("Error doPOST", e);
- throw new ServletException("error during execution of get
method");
+ throw new ServletException("error during execution of post
method : "+ e.getLocalizedMessage());
}
}
@@ -504,6 +504,9 @@
// Usually when the user performs an action in the same portlet
//
String siteUrl = request.getParameter(Rewriter.URL_PATH_PARAM_NAME);
+ String headerReferer = request.getHeader("referer");
+ if(headerReferer==null)
+ headerReferer="";
if (siteUrl != null && !"".equals(siteUrl.trim())) {
perfLog.debug("->Request has URL " + contextId);
//
@@ -516,7 +519,7 @@
if (accessibleHostsStr != null &&
!"".equals(accessibleHostsStr.trim())) {
map.put(MAP_ACCESSIBLE_HOSTS,
Arrays.asList(accessibleHostsStr.split(",")));
}
- map.put(MAP_REFERER,request.getHeader("referer"));
+ map.put(MAP_REFERER,headerReferer);
//
return true;
}
@@ -527,7 +530,7 @@
String referer = null;
if(matcher.find())
mapReferer = matcher.group(2);
- matcher = pattern.matcher(request.getHeader("referer"));
+ matcher = pattern.matcher(headerReferer);
if(matcher.find())
referer = matcher.group(2);
if(mapReferer==null || !mapReferer.equals(referer)) {
