How do i determine the correct filesystem path

I am working on an application that has to download some external resources and make them accessible through a public/static directory in Ring.

But.. I have a problem saving the resources into a static directory in my application, when developing I use the ring-jetty-adapter; the test & production servers are running Tomcat.

I added :web-content "public" to my leiningen project and added the public directory in the root of the project, then I have a download function using http-agent and duck-streams:

(defn download
  [file-name url]
  (h/http-agent url
                :handler (fn [agnt]
                           (let [fname file-name]
                             (with-open [w (d/writer fname)]
                               (d/copy (h/stream agnt) w))))))

If I am booting Jetty from the REPL and use savepath: "public/my.file", the downloaded file is placed correctly in the public directory. But when I deploy it using a .war file to Tomcat, it looks for a public directory in the Tomcat root directory, and not under the application context path.

I tried to add a middleware wrapper to determine the context path and from there build the correct save path, but I cannot find any way to access the HttpServlet or a way to 开发者_StackOverflow社区determine if the application is running in the adapter or if it is deployed under a specific context.

Here the wrapper:

(defn wrap-context-info [handler]
  (fn [req]
    (let [resp (handler req)]
      (assoc resp :servlet (:servlet req) :req (:servlet-request  req)))))

both :servlet and :req are nil.

Looking at the ring-servlet source, it appears that the ring servlet adapter associates the HttpServlet, HttpServletRequest, and HttpServletResponse objects with the ring request map under :servlet, :servlet-request, and :servlet-response keys, accordingly.

It also adds a :servlet-context entry to the request map with the value of (.getServletContext servlet) for convenience.

Inside your handler, you might want to check for the presence of these keys in the request map, and then pull further information that you need from the associated objects.