Add a utility function to calculate the next slice to download

content/browser/download/download_item_impl.cc

Index: content/browser/download/download_item_impl.cc
diff --git a/content/browser/download/download_item_impl.cc b/content/browser/download/download_item_impl.cc
index 34571f5990cb2b72854feeeef99104b3017c7b33..99539a3a9ae527aef7f7e39135f355465a326663 100644
--- a/content/browser/download/download_item_impl.cc
+++ b/content/browser/download/download_item_impl.cc
@@ -45,6 +45,7 @@
 #include "content/browser/download/download_net_log_parameters.h"
 #include "content/browser/download/download_request_handle.h"
 #include "content/browser/download/download_stats.h"
+#include "content/browser/download/parallel_download_utils.h"
 #include "content/browser/renderer_host/render_view_host_impl.h"
 #include "content/browser/web_contents/web_contents_impl.h"
 #include "content/public/browser/browser_context.h"
@@ -1944,7 +1945,13 @@ void DownloadItemImpl::ResumeInterruptedDownload(
       new DownloadUrlParameters(GetURL(),
                                 storage_partition->GetURLRequestContext()));
   download_params->set_file_path(GetFullPath());
-  download_params->set_offset(GetReceivedBytes());
+  if (received_slices_.size() > 1) {

[xingliu, 2017/03/01 23:03:59]

Does FindNextSliceToDownload assume the slices are downloaded one by one?

[qinmin, 2017/03/01 23:25:27]

No,there is no such assumptions. For download resumption, however, all the
workers should have been paused, so we can pick up the first gap to start.

[xingliu, 2017/03/02 00:23:51]

Following up the size of received_slices, if the size is 1, and we have one
slice [10, 20], then the logic falls into set_offset(GetReceivedBytes()), which
is probably wrong.

Also for normal download, if the size of received_slices is guaranteed to be 0,
then maybe check greater than 0 is ok.

[xingliu, 2017/03/02 00:23:51]

So this function mainly serves for sending the first resumption request(not a
range request since we need the first response headers). That sounds good to me.

[qinmin, 2017/03/02 01:00:39]

The case you described(main request returns no slice, while one worker returns a
slice) will never happen. This is because the main request will create the
download file before DownloadItem::start() is called. So at least a [0, 0] slice
will be recorded in the received_slices_ when DownloadItem::start() is called.
That's why I put a slice size > 1 here.  Anyway, changed this to > 0 as it
shouldn't matter with the final result.

+    ReceivedSlice next_slice = FindNextSliceToDownload(received_slices_);
+    download_params->set_offset(next_slice.offset);
+    download_params->set_length(next_slice.received_bytes);
+  } else {
+    download_params->set_offset(GetReceivedBytes());
+  }
   download_params->set_last_modified(GetLastModifiedTime());
   download_params->set_etag(GetETag());
   download_params->set_hash_of_partial_file(hash_);