Resume Halted/Errored transfers

[GoogleCodeExporter]

In the event that it stops with an error on one of the files, it would be nice 
to have some way of resuming ... I can see a few reasonable ways of handling it:

-it could have a context menu in which we could restart and/or resume the file
-if the amz is fully restarted it could skip complete/correct files and 
retry/resume incomplete/corrupt files
-or ideally it could automatically try to resume and/or restart any failed 
files.

This would be of great help to me. Thanks.

Original issue reported on code.google.com by 1m.0n.f...@gmail.com on 13 Sep 2010 at 10:59

[GoogleCodeExporter]

how often do you have failed files you wish to restart?

Each of those possible alternatives you listed come with their own set of 
challenges and corner cases that would need to be handled/dealt with. 

Original comment by SCColbert@gmail.com on 13 Sep 2010 at 7:50

[GoogleCodeExporter]

Not too terribly often, it has maybe happened two or three times since I 
started using Pymazon (I think I started using Pymazon in May, and according to 
my Amazon history I've downloaded 16 things since then). My Internet connection 
is likely at fault as it's far from perfect ... of course, I'm not the only one 
with a less than perfect connection eh?

Original comment by 1m.0n.f...@gmail.com on 13 Sep 2010 at 9:16

[GoogleCodeExporter]

Hmm, and while on the topic of making sure things finsih downloading ... some 
sort of verification couldn't hurt either, even if it's something so simple as 
checking the files size. I'm not sure if this track (Pink Floyd - Pigs (Three 
Different Ones)) finished without Error or not ... either way the file is 
truncated, much to my dismay :(

Original comment by 1m.0n.f...@gmail.com on 7 Oct 2010 at 2:24

[GoogleCodeExporter]

I had original validated the download by checking the downloaded file size 
against the file size listed in the AMZ file. Unfortunately, Amazon lists many 
file sizes incorrectly and so there is no reliable way to validate the download.

Original comment by SCColbert@gmail.com on 7 Oct 2010 at 6:07

[GoogleCodeExporter]

Okay, that really sucks :( Curiosity question though, related to that. When you 
saw them with incorrect file sizes, were they way off or were they close? If 
the off ones tend to be close file size +-1% or something could work, which 
could still be better than nothing.

Original comment by 1m.0n.f...@gmail.com on 7 Oct 2010 at 9:11

[GoogleCodeExporter]

They were off by fairly large amounts sometimes; 100Kb or more... Really too 
much and too sporadic to determine any kind of heuristic without purchasing a 
shit ton of albums and inspecting all of the data. Care to start a community 
data collection project? :)

Original comment by SCColbert@gmail.com on 12 Oct 2010 at 6:50

[GoogleCodeExporter]

Not sure what your thoughts on this would be, but the best way to collect the 
data would probably be to build it into Pymazon. After downloading it could 
just submit the expected filesize, actual filesize, and metadata. Hmm, how 
often are the filesizes off? Even if isn't reliable, if it's not wrong too 
often it could still be a meaningful suggestion. 

Original comment by 1m.0n.f...@gmail.com on 13 Oct 2010 at 12:19

[GoogleCodeExporter]

Hmm, it seems I'm having MUCH more trouble lately. I just did another and about 
half of the stuff failed to finish ... In *all* of the cases Pymazon said 
complete when they finished though. Oddly, also, I've seen the same behavior 
with other programs in the last several weeks too. Very odd.

Original comment by 1m.0n.f...@gmail.com on 13 Oct 2010 at 2:36

[GoogleCodeExporter]

Pymazon will say complete whenever urllib2 believes the file transfer is 
complete. What you are seeing could be a result of a flakey internet 
connection. Does it happen again when you try redownloading the files? (Pymazon 
will allow you to download as many times as the amazon servers will allow, in 
which case the dl will fail almost instantly.)

The other possibility is that amazon is getting crafty and sending a valid 
response when it detects Pymazon instead of its proprietary downloader and 
instead of sending some other http error. But given that *some* of your 
downloads are working, and that other programs are also experiencing 
difficulty, I doubt this is the case.

Original comment by SCColbert@gmail.com on 13 Oct 2010 at 6:42

[GoogleCodeExporter]

I downloaded most of an album: 10 tracks total and track number 10 only reached 
43% and would go no further. Attempting to restart the download results in 
pymazon trying to start with track 1 all over again.

Original comment by mobiledi...@gmail.com on 3 Dec 2011 at 8:20

[GoogleCodeExporter]

I have this problem too (my Internet connection is very flaky).

I have a proposed patch that adds a simple command line option, "-k 
--skip-existing-files". This seems like a pretty low-impact workaround for the 
problem, without introducing too many corner cases. I added basic hooks into 
the UI for logging.

http://pastebin.com/XRdD0Ufw

Let me know if you're interested in incorporating it. (Don't know is Pastebin 
is still in development.)

Original comment by zac...@gmail.com on 2 Jun 2012 at 6:31