Problem setting up GzipFilter in Jetty - filter

I'm trying to setup Jetty to serve compressed html content. In web.xml I setup GzipFilter and mapped it to /* but this doesn't seem to work. Here's the filter configuration:
<filter>
<filter-name>GZipFilter</filter-name>
<display-name>Jetty's GZip Filter</display-name>
<description>Filter that zips all the content on-the-fly</description>
<filter-class>org.mortbay.servlet.GzipFilter</filter-class>
<init-param>
<param-name>mimeTypes</param-name>
<param-value>text/html</param-value>
</init-param>
</filter>
<filter-mapping>
<filter-name>GZipFilter</filter-name>
<url-pattern>/*</url-pattern>
</filter-mapping>
I'm just starting to use Jetty, so the solution might be ridiculously simple. If you can link me to documentation that might help me, that would be great too.

GZIP Compression
GZIP Compression can be used to reduce the amount of data being sent "over the wire". Compression is applied as a transport encoding. This can greatly improve webapplication performance, however it can also consume more CPU and some content (eg images) cannot be well compressed.
Static Content
The Jetty Default Servlet can serve precompressed static content as a transport encoding and avoid the expense of on-the-fly compression. If the "gzip" init parameter is set to true, then Jetty will look for compressed static resources. So if a request for "foo.txt" is received and the file "foo.txt.gz" exists, then it will be served as "foo.txt" with a gzip transport encoding.
GzipFilter
The Jetty Gzip Filter is a compression filter that can be applied to almost any dynamic resource (servlet). It fixes many of the bugs in commonly available compression filters (eg handles all ways that content length may be set) and has been testing with Jetty continuations and suspending requests.
Some user-agents may be excluded from compression, so as to avoid some common browser bugs (yes this means IE!).
refer from jetty doc:
http://docs.codehaus.org/display/JETTY/GZIP+Compression
you can look Gzipfilter source code,here is a lot of useful comments :
http://download.eclipse.org/jetty/stable-7/xref/org/eclipse/jetty/servlets/GzipFilter.html

I'm gonna answer this too, since I've had a huge headake trying to make this work, and I finally did it. Also, I'm not a major expert in the fine details of HTTP so I'll give a non-professional answer.
First, here's how I checked if my GZipFilter was working or not. Started Firefox, made sure I had the Firebug addon, started the Firebug addon, went to the "Net" tab. Then I accessed the URL which should return a GZipped response. Here's what Firebug shows:
The "Size" column shows the size of the response. If you hover over the "Size" column label with your mouse, it will tell you that if the response is compressed, then it will display the compressed size of the response.
This all was done with the GZip filter enabled in Jetty. I then removed the GZip filter declaration from my web.xml, restarted Jetty and repeated the test. This time around the response had the exact same size as before, which clearly indicated that the GZip compression was not working.
After multiple trial and errors, what I did is look in Firebug at the "Request Headers" section to see the value for the "Accept" header. I've noticed that here this had values such as "application/xml" and "text/xml", but the way I had configured my GZIp filter's init param "mimeTypes" only contained "text/xml" (and was missing "application/xml"). It was configured like so:
<filter>
<filter-name>GzipFilter</filter-name>
<filter-class>org.eclipse.jetty.servlets.GzipFilter</filter-class>
<init-param>
<param-name>mimeTypes</param-name>
<param-value>text/html,text/plain,text/xml,application/xhtml+xml,text/css,application/javascript,image/svg+xml,application/json,application/xml; charset=UTF-8</param-value>
</init-param>
</filter>
<filter-mapping>
<filter-name>GzipFilter</filter-name>
<url-pattern>/*</url-pattern>
</filter-mapping>
After adding the "application/xml" value to the list like so:
<filter>
<filter-name>GzipFilter</filter-name>
<filter-class>org.eclipse.jetty.servlets.GzipFilter</filter-class>
<init-param>
<param-name>mimeTypes</param-name>
<param-value>text/html,text/plain,text/xml,application/xhtml+xml,application/xml,text/css,application/javascript,image/svg+xml,application/json,application/xml; charset=UTF-8</param-value>
</init-param>
</filter>
<filter-mapping>
<filter-name>GzipFilter</filter-name>
<url-pattern>/*</url-pattern>
</filter-mapping>
I redid my previous test, and sure enough now the reported size of the response was much smaller:
Also notice that now, the reported Response Headers contain an extra field called "Content-Encoding" with a value of "gzip".
So basically the idea is to check what kind of values you send in your Request "Accept" header and make sure that all those values are configured in the GZip filter's "mimeTypes" init param.

Sometimes using Gzipfilter has some problems, depending on how you are handling buffers and flushing. As such, using org.eclipse.jetty.servlets.IncludableGzipFilter (which actually an extends GzipFilter) may solve your problems.

On jetty 9.3:
edit jetty.conf and include the xml file "jetty-gzip.xml"
edit start.ini and add "--module=servlets"
edit jetty-gzip.xml and configure the mime-types you want.
Restart jetty and test again.

What was the error? Are you getting classpath problems or something else? If classpath, you need to make sure the gzipfilter class is available to the jetty runtime or it will die.

Are you sending the request with the "Content-Encoding: gzip" request header?

Related

HttpHeaderSecurityFilter is being weirdly ignored - it always sets max-age to 0 and ignores all parameters

This is the most bizarre tomcat configuration problem I've ever seen. I have the following entry in web.xml:
<filter>
<filter-name>httpHeaderSecurity</filter-name>
<filter-class>org.apache.catalina.filters.HttpHeaderSecurityFilter</filter-class>
<async-supported>true</async-supported>
<hstsEnabled>true</hstsEnabled>
<hstsMaxAgeSeconds>631</hstsMaxAgeSeconds>
<hstsIncludeSubDomains>true</hstsIncludeSubDomains>
<hstsPreload>true</hstsPreload>
</filter>
I intentionally put in a weird value for max-age so that it would be easy to see if the configuration is active or not.
And it's activated:
<filter-mapping>
<filter-name>httpHeaderSecurity</filter-name>
<url-pattern>/*</url-pattern>
<dispatcher>REQUEST</dispatcher>
</filter-mapping>
But it doesn't do what it should do. The resulting headers are always:
strict-transport-security: max-age=0
which is almost exactly what I don't want.
Any ideas on this? I can't even understand how this could be happening and it certainly goes against all the documentation. This is all with Tomcat 9.0.29, JDK 13. It seems like it should "just work" and seems like a very simple configuration.
To be clear, this is in the Tomcat web.xml (in $CATALINA/conf/web.xml), not in an application-specific web.xml. HSTS preload is for the entire domain (including subdomains which might be on other servers), not for any set of URIs within the domain, so it only makes sense to set preload in the server-wide web.xml.
EDIT: This is with a Spring Boot application. It seems that Spring Boot might be controlling this header and overriding whatever is being set in web.xml. This is unfortunate because Spring Boot is for a specific application in a specific context, whereas this header really applies to the entire domain, or even to sub-domains, and so is beyond any specific web application in a specific context. I think I need to figure out how to get Spring Boot to do nothing to HSTS headers and let Tomcat handle it all.

Java project unknown enum in web.xml

I am learning about filters. So I have the following code in my web.xml:
<filter>
<filter-name>asyncFilter</filter-name>
<filter-class>el.test.AnyRequestFilter</filter-class>
<async-supported>true</async-supported>
</filter>
<filter-mapping>
<filter-name>asyncFilter</filter-name>
<url-pattern>/*</url-pattern>
<dispatcher>ASYNC</dispatcher>
</filter-mapping>
When I write is is autocompletes me ASYNC, but after that it becomes in red colour and IntelliJ show me that the error is that is is unknown enum.
Why is that?
Thanks for your attention.
P.S. I have problems with following code too:
<error-page>
<location>/WEB-INF/jsp/view/badRequest.jsp</location>
<error-code>404</error-code>
</error-page>
on <error-code> it is written no child element expected at this point, but the code is working(I can run it).
from https://docs.oracle.com/cd/E19879-01/819-3669/bnagf/index.html
4.In the Add Filter Mapping dialog, select one of the following dispatcher types:
REQUEST: Only when the request comes directly from the client
FORWARD: Only when the request has been forwarded to a component (see Transferring Control to Another Web Component)
INCLUDE: Only when the request is being processed by a component that has been included (see Including Other Resources in the Response)
ERROR: Only when the request is being processed with the error page mechanism (see Handling Servlet Errors)
So ASYNC is not a valid value.
Edit : it seems it was added in javaee 6 : http://docs.oracle.com/javaee/6/tutorial/doc/bnagb.html
So maybe you have some configuration problem of JRE/JDK version with your project and IDE.

WebSphere 7.0 won't run filter for root URL

I have a WAR file that defines a filter to run on all URLs:
<!DOCTYPE web-app PUBLIC
"-//Sun Microsystems, Inc.//DTD Web Application 2.3//EN"
"http://java.sun.com/dtd/web-app_2_3.dtd">
...
<filter>
<filter-name>OurRedirectServletFilter</filter-name>
<filter-class>com.mycompany.RedirectServletFilter</filter-class>
</filter>
...
<filter-mapping>
<filter-name>OurRedirectServletFilter</filter-name>
<url-pattern>/*</url-pattern>
</filter-mapping>
The filter is designed to perform some redirects from 'convenience' URLs to a corresponding 'actual' URL, but I don't think that's really relevant to the problem.
On WebSphere 7.0, this filter doesn't run for requests to the root URL, e.g. /ctxroot or /ctxroot/; instead I just get a 404 response. It does run for /ctxroot/blah, whether blah is a valid or invalid path.
I've tried adding additional filter mappings for URL patterns <url-pattern>/</url-pattern> and <url-pattern></url-pattern>, but I get the same behavior.
I've tested on base WAS 7.0.0.0, and with the latest fix pack applied, i.e. WAS 7.0.0.27.
The filter works as expected on WAS 8.5 and I'm pretty sure on WAS 8.0, as well as on every version of WebLogic, JBoss, and Tomcat that I've tried. This then seems to be a bug with WAS 7.0, but I'd still like to find a workaround. Anybody know of one?
I eventually looked at the body of the 404 error response and saw error code SRVE0190E, which led me to this helpful page. The issue is that filters aren't called by default for URLs that correspond to resources that don't exist (though I swear I tested that for a URL other than the context root, and my filter was called).
It's possible to configure WebSphere to call filters in this situation by setting a custom property as further described in the linked page:
com.ibm.ws.webcontainer.invokefilterscompatibility=true
I also found that for the case of the context root URL, setting a welcome-file entry in web.xml that maps to an existing resource causes the filter to be called:
<welcome-file-list>
<welcome-file>fakehome.html</welcome-file>
</welcome-file-list>

Websphere not using welcome-file-list after mapping *.html to a servlet

I am using IBM WebSphere (WAS) 7.0.0.19 to host a java-based web-app, and I needed to map the extension *.html to a particular servlet so that I could do some server-side scrubbing of user-supplied HTML files. (The server reads the file, augments it with some extra information, and serves up the modified content transparently to the person viewing the page.)
Unfortunately, when I did this, welcome-files stopped working. Previously, if I typed in the URL for a directory, the server would look for index.html and serve up that. Now, I'm just getting a 403 forbidden rule ("Forbidden - by rule."). The access logs don't show anything more--they simply state that directory indexing is forbidden by rule for the server, which is correct. I don't want the webserver to build a table of contents for directories with no index.html, but when there is an index.html, I want it to serve up that file.
My first thought was that it was trying to serve the index.html through my servlet, the servlet was failing to find the file (because the url lacked "index.html"), and therefore it thought there was no index.html. However, I put in some debug code and am quite confident that the servlet code is never getting run when I go simply to the directory itself.
I don't really care whether index.html is served through the servlet or not--in the case of this particular file, the servlet would just spit back the original file anyway. I just want index.html to be served by something.
Here is the relevant section of my web.xml
<servlet-mapping>
<servlet-name>PageScrubber</servlet-name>
<url-pattern>*.html</url-pattern>
</servlet-mapping>
<welcome-file-list>
<welcome-file>index.html</welcome-file>
<welcome-file>index.htm</welcome-file>
<welcome-file>index.jsp</welcome-file>
</welcome-file-list>
For what it's worth, index.htm and index.jsp were not working before the addition of the servlet mapping. Only index.html worked before. However, now none of them work.
I have used the same web.xml with two Oracle products: WebLogic (WLS) and Oracle Application Server (OAS) with no issues.
I am quite confident that it is just the addition of this scrubber servlet that has caused the problem, because removing that directive caused directory indexing to start working again.
I did find some notes about welcome-file-list not working when using an extended document root, and I tried setting com.ibm.ws.webcontainer.EnablePartialURLtoExtendedDocumentRoot to be true, but that did not seem to change anything.
I'm pretty much out of ideas. Does anyone out there have any thoughts as to why it's not finding my index.html? Thanks in advance!
Caveat: I am working out of memory here.
The Welcome files used to be served by File Serving Servlet (or something that sounds similar to that).
This information would be in the WebSphere extensions file.
I would take a step back and remove your pageScrubber and get the file serving Servlet to serve the welcome files and see that things are working before getting back to using the PageScrubber.
These are my initial thoughts.
HTH

Help with Jersey auto-generation of WADL

As described here: http://wikis.sun.com/display/Jersey/WADL
I'm using Jersey 1.4 in Tomcat 6.
I have tried every possible URI with a "/application.wadl" and all I get is a 404 (not available). I'm obviously not understanding something, but every blog I read makes it sound like this is "out of the box" functionality. I'm using Tomcat 6..should this matter?
I was able to use Pavel's example on using the WadlResource object here, but it seems I shouldn't need to do this: http://markmail.org/message/lbeaw5vyr4qergdd#query:+page:1+mid:fo4dt7tbd6rb3gvi+state:results
thanks.
I think that ChrisO most likely has the answer in that your wadl would be available at http://localhost:{port}/{warname}/application.wadl
I had the same problem, and found out that default application.wadl was only working when configured at the root of the app
My conf included several url patterns like the following
<filter-mapping>
<filter-name>jersey-spring</filter-name>
<url-pattern>/admin/rest/*</url-pattern>
</filter-mapping>
And I had to add this to make it work :
<filter-mapping>
<filter-name>jersey-spring</filter-name>
<url-pattern>/application.wadl</url-pattern>
</filter-mapping>
Hope it helps
If you are using Jersey, the WADL will be given automatically when you suffix the application.wadl to your url.
Here all the /rest/ is the <url-pattern> for the <servlet-mapping>
http: // yourmacine:8080 / RESTfulExample/rest/application.wadl
You need to specify the display-name and url-pattern of the sevlet mapping(if any) before application.wadl

Resources