Request: Need to detect chuncks whith 0

[ivan386]

Some users share incomlete files. Shareaza must detect zero chuncks in file by ed2k or tiger set. And show warning to user that file is incomplete. If user chuse to download it than not need to download that chuncks (We alredy know what in it).

[raspopov]

Zero chunk can be normal part of file. So declined.

[old_death]

[raspopov]

Anyway its impossible to detect zero chunks with 100%.
Use zip to compress real zero chunks.

[ivan386]

Green line (hash checked) on white part (not downloaded) will be warning for advanced user.

Whith ed2k hash set we can only detect parts 9728000 bytes. [PATCH] "empty" ed2k block detector.
Whith tiger tree hash set we can detect parts from 1024 bytes to 8388608 bytes.

If user see that file incomlete but have some usefull parts he can search complete copy and join already downloaded data whith this new download.

Stadart whay to detect empty parts is "Assume file is 100% complete and reverify". But it may lose incomplete downloaded parts and it to slow for that easy task.


No need to stay in queue to get "empty" part.

[raspopov]

"No need to stay in queue to get "empty" part" is different from "Some users share incomlete files" please don't change topic of feature request!
BTW how many zero chunks you saw on eDonkey?

[ivan386]

> 1000 chuncks

some files already marked as corrupt/fake some not.

Some files have same name and size but diffrent hash.

It's greate problem for big files.

[raspopov]

Its Gnutella 2 and its just icomplete downloads.

[ivan386]

At the end of first file big "empty" part.

magnet:?xt=urn:bitprint:HYWC5ILQPVVOZOSUHRKOMKZ5YWTJP2S5.F7YN3X4LKE4CECOVBNCTXBKWRUL5NN2XTZWB3SI&xt=urn:ed2khash:b4c8aeccd780261e7015e4bbb1eaed98&xl=1523968732&dn=CBT%20NUGGETS%20LINUX%20SERIES%20(VIDEO%20TUTORIAL).7z

when you get hash set use "Assume file is 100% complete and reverify".

[raspopov]

I don't thinks so. And we speaking about eDonkey and completed downloads.

[ivan386]

just chek it.

[brov]

I checked that file and I can confirm ~180MB verified chunk without any downloading

[mojo85]

Just checked it as well. Ivan is correct that this does pose a problem if a person did believe it to be a complete file, when in fact it was just an incomplete file. The hash and file go with each other, but I can see this being of great frustration to an end user and I see this as a potential for a spam exploit. We should definately have a warning system for this scenario, the end user has to be smart enoug to decide whether to continue downloading or to cancel.

[old_death]

[mojo85]

[ailurophobe]

That is the code for detecting files that have no complete sources, which is useful and should be added to Shareaza (if it isn't already there). But what ivan was talking about was some people sharing files as complete when they are actually not and contain large empty parts.

But I am glad you brought this up, because adding a filter check-box that hides files without complete sources would be a good idea. Not sure if all networks report this data though.

[mojo85]

Yeah that is what I'm saying, not sure how eMule handles this during the download, but this is how it handles things during the search. I haven't used the eD2k network in a while, especially while using eMule (maybe 5 years ago?). But I kind of remember that servers keep a history of the last time they saw such a file complete, I can't find the lugdunum website, but these servers should have extensions that should help us identify some files. But yes this is definately worth adding to Shareaza's routine search filters if it only enhances it a bit.

Well the simplest but most brute force method is to check upon download import. The instant any file is set to download, it waits to get a hashset and assumes complete to rehash and continue downloading ... but the progress bar would show a different color for regions/ chunks/ parts that are a continuous zero. Maybe worth adding as an advanced settings feature as it would just tax cpu.

Can anyone think of a clever way of doing this? Just for the record this is a problem as it leaves room for spam and fake corrupt files to easily proliferate. Also a small file that is like this is ok, the problem is when it is 100mb and over. Than you are just wasting time and bandwidth to download something that could have been prevented.

[old_death]