[Bug 748316] hlsdemux: The end offset (range_end) of a segment is not calculated properly in the m3u8 parser

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=748316

--- Comment #21 from Stavros <stavrosv at digisoft.tv> ---
Let me put some scenarios and how I think it myself. The scenarios are the
following

 - segment.stop = -1, non-inclusive range [0,-1). If there is no `if`
statement, the stop = -2, which is an invalid value, so for this reason I leave
it as a -1. So, stop = -1.
 - segment.stop = 0, non-inclusive range [0,0). As it is now, the stop = -1, is
an invalid value. I believe that we need to leave as it is, because if we
consider that segment.stop non-inclusive, then the [0,0) is invalid. So, stop =
-1.
 - segment.stop = 1, non-inclusive range [0,1). The stop = 0 and we have valid
values. So, stop = 0.

(In reply to Thiago Sousa Santos from comment #20)
> 
> Perhaps we want to keep segment_stop as non-inclusive and just subtract one
> when adding to the request headers?

If I understand correctly, the gist of what you are saying is the following

diff --git a/ext/soup/gstsouphttpsrc.c b/ext/soup/gstsouphttpsrc.c
index ad87223..ced1117 100644
--- a/ext/soup/gstsouphttpsrc.c
+++ b/ext/soup/gstsouphttpsrc.c
@@ -742,11 +742,13 @@ gst_soup_http_src_add_range_header (GstSoupHTTPSrc * src,
guint64 offset,

   gint rc;

+  guint64 stop = stop_offset - 1 < 0 ? -1 : stop_offset - 1;
+
   soup_message_headers_remove (src->msg->request_headers, "Range");
-  if (offset || stop_offset != -1) {
-    if (stop_offset != -1) {
+  if (offset || stop != -1) {
+    if (stop != -1) {
       rc = g_snprintf (buf, sizeof (buf), "bytes=%" G_GUINT64_FORMAT "-%"
-          G_GUINT64_FORMAT, offset, stop_offset);
+          G_GUINT64_FORMAT, offset, stop);
     } else {
       rc = g_snprintf (buf, sizeof (buf), "bytes=%" G_GUINT64_FORMAT "-",
           offset);

If that's the case then I believe that the problem just moves from one place to
another. If the stop_offset = -1, again the stop = -2, so we need to change
this and for this reason I have used the condition that if it's negative, than
the value will be -1.

I am not really sure what is the best/clean approach. Personally speaking I
would consider the first approach :). Maybe it would be beneficial to add a
comment there with the above edge cases.

Anyone else to comment about this?

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.