Pass the filename obtained from content-disposition header for download

content/browser/android/download_controller_android_impl.cc

Index: content/browser/android/download_controller_android_impl.cc
diff --git a/content/browser/android/download_controller_android_impl.cc b/content/browser/android/download_controller_android_impl.cc
index f3a274a3ef4a7e043597cf53ab9cc59d81bdc4cb..16864ab4ab38455e3daed4a30eebed30cf02aac5 100644
--- a/content/browser/android/download_controller_android_impl.cc
+++ b/content/browser/android/download_controller_android_impl.cc
@@ -27,6 +27,7 @@
 #include "jni/DownloadController_jni.h"
 #include "net/cookies/cookie_options.h"
 #include "net/cookies/cookie_store.h"
+#include "net/http/http_content_disposition.h"
 #include "net/http/http_request_headers.h"
 #include "net/http/http_response_headers.h"
 #include "net/url_request/url_request.h"
@@ -221,10 +222,16 @@ void DownloadControllerAndroidImpl::StartAndroidDownload(
   ScopedJavaLocalRef<jstring> jreferer =
       ConvertUTF8ToJavaString(env, info.referer);
 
+  // Try parsing the content disposition header to get a
+  // explicitly specified filename if available.
+  net::HttpContentDisposition header(info.content_disposition, "");

[boliu, 2014/02/07 18:11:47]

Hmm...

If we use the full blown GetSuggestedFilename in net_util.cc here (ie take into
account whatever mimetype we have here as well), then there should be no need
for any filename guessing in java code.

Win?

[shashi, 2014/02/07 18:59:12]

I agree with Bo, that should be the best way and even easier to keep desktop and
Android implementations in sync. and we will be able to even reduce our
temporary/incomplete parsing implementation in Java.

*Slight tricky to make sure we do not break any existing scenarios (I hope we
have regression tests for different download bugs on Android over years).


On 2014/02/07 18:11:47, boliu wrote:

[Jinsuk Kim, 2014/02/08 02:49:28]

I'd like to upstream the filename decision logic eventually. I guess it takes
more understanding on other parts like Yaron mentioned in the bug entry. Mind if
I started it in the next CL while letting this one handle the bug fix only?

[Jinsuk Kim, 2014/02/08 02:49:28]

I was not aware of GetSuggestedFilename and it is a great suggestion. Thanks!
I'd like to do it in the follow up CL if you're okay.

[boliu, 2014/02/08 02:51:23]

sgtm, yaron already gave lgtm, which is all you need

+  ScopedJavaLocalRef<jstring> jfilename =
+      ConvertUTF8ToJavaString(env, header.filename());
+
   Java_DownloadController_newHttpGetDownload(
       env, GetJavaObject()->Controller(env).obj(), view.obj(), jurl.obj(),
       juser_agent.obj(), jcontent_disposition.obj(), jmime_type.obj(),
-      jcookie.obj(), jreferer.obj(), info.total_bytes);
+      jcookie.obj(), jreferer.obj(), jfilename.obj(), info.total_bytes);
 }
 
 void DownloadControllerAndroidImpl::OnDownloadStarted(