Versatile C++ game scraper: Skyscraper

mo418

@muldjord I was about to ask why I get « no game found » for all my roms using screenscraper.

Thanks for the heads up. Keep us posted (and keep up the good work!)

Regards :)

muldjord

The good people of Screenscraper has been very helpful in figuring out how to move forward with this and the issue is now resolved. Stay tuned for more info soon...

mo418

@muldjord I confirm. Working like a charm now

Thanks for your support :)

Edit: Lots of games not found though, but it’s the 2nd time I use your scraper so I don’t know if it’s a setting issue or else.

muldjord

@mo418 This will probably be better once the new release (3.0.2) is out. There are issues with the screenscraper module currently in addition to the previous key ban.

muldjord

Skyscraper 3.0.2 released: https://github.com/muldjord/skyscraper

Upped the rom limit from 5 to 35 for the "igdb" module
Upped the rom limit from 25 to 35 for the "mobygames" module
Added media cache config options to module section
Add Sharp X1 platform as "x1"
Now exits nicely when running low on disk space
Added 'spaceCheck=<BOOL>' to config.ini
Fixed crash when using '--startat' and '--endat' where the '--endat' file name came before the '--startat' file name. Note! What 'ls' reports in alphabetical order is not always what Skyscraper see as it it locale specific. So be aware of this. A huge thanks to 'Gemba' for taking the time to investigate this bug thoroughly.
Fixed bug in game list metadata preservation when using relativePaths and '<folder>' entries (thank you to 'HumanRob' for reporting this)
Fixed game list entries skipping for 'relativePaths' and '<folder>' instances
Skyscraper now saves the cache and exits nicely on ctrl+c (SIGINT) (thanks to 'krcroft' for pointing this out)
The 'screenscraper' module now includes 'systemeid' in the query for better results
Now skips the game list assembling when in gathering mode
Now skips cache saving when in game list generation mode
Output now says whether it was a gathering run or a game list generation run

Fixed a bunch of stuff in the game list skipping and metadata preservation code. This was sortof b0rked before when people used relative paths. Should work as expected now. Let me know if you find cases where it doesn't work.
The screenscraper module now includes 'systemeid' in the query which should give better results for all platforms.
I've also included another quite important feature which is the "exit nicely" when user presses ctrl+c to stop a scraping. Before it would simply force the process to die. Now it let's the currently running threads finish up the entries they are working on and then saves the cache for the stuff it has gone through. Before this data was lost. So this is a huge improvement.

Brunnis

@muldjord That’s awesome! I can’t test it out until tomorrow morning. Do you have reason to believe this also fixes the issue where Skyscraper missed hits on hashes that should have been found?

muldjord

@Brunnis Yes, I can do manual md5 searches now, so that should be fixed too.

mitu

Yes, just updated and tried an MD5 search (the one reported first with the Atari 2600 E.T game) and it's found now.

scocasso

@muldjord said in Versatile C++ game scraper: Skyscraper:

Currently you have the option to either scrape from scratch, or to only scrape entries that don't already exist.

I cannot find this option. I don't want to accidentally scrape all 2000+ games, I only need to scrape about 50 which are new.

I'm using Skyscraper 3.0.1 on retropie from the Retropie tools menu.

mo418

When it says [...] PASS 1 ——— Game [gametitle] not found :( ———

Does it mean it will make other passes until it will find the rom media in screenscraper?

Updated to 3.0.2 and I still get a lot of them (while UXS found them all, except very few titles) so there must be something going on.

Also, is there a way to export the games titles that were not found?

Thanks in advance!

mitu

@mo418 When running with --unattend, the list of not found titles is saved in ~/.skyscraper/skipped-<scraper>.txt. It will tell you at the end of the run

[...]
Total number of games: 2
Successfully processed games: 1
Skipped games: 1 (Filenames saved to '~/.skyscraper/skipped-thegamesdb.txt')

muldjord

@scocasso If you run it from command line you can choose to skip existing entries (it will ask you). From the retropie menu it loads them from cache, which is really fast, but it doesn't skip them. If you want to skip them entirely, you need to run it from command line.

Brunnis

@muldjord said in Versatile C++ game scraper: Skyscraper:

@Brunnis Yes, I can do manual md5 searches now, so that should be fixed too.

I have tested this morning and there's still something not working as it should. The update fixed approximately half of the "No returned matches" occurrences. The previous example (Batman on NES) now works. However, the same issue remains for other ROMs. Here's an example:

35933222cf8658f7c6679fc7de630aaa Battletoads & Double Dragon - The Ultimate Team (U) [!].nes

This one is also found by searching with the MD5 hash directly on screenscraper.fr, so it seems Skyscraper may still have an issue here. I guess it's possible that MD5 matching simply doesn't work and that Batman started working because of some other change.

Also, after the update to version 3.0.2, Skyscraper will no longer create the gamelist.xml file after scraping. I upgraded from version 3.0.1 (via the RetroPie-Setup script), where it worked fine (even after manually deleting gamelist.xml before scraping). Any ideas?

EDIT: I'm initiating scraping from RetroPie-Setup. Should I try the command line for one of the troublesome ROMs? Any suggested command to execute that could help diagnose the issue?

muldjord

@Brunnis said in Versatile C++ game scraper: Skyscraper:

Also, after the update to version 3.0.2, Skyscraper will no longer create the gamelist.xml file after scraping. I upgraded from version 3.0.1 (via the RetroPie-Setup script), where it worked fine (even after manually deleting gamelist.xml before scraping). Any ideas?

You need to run the game list generation pass after gathering the resources, please read the docs, it's explained there. This was also the case with 3.0.1.

I'll check that one example you mention and see if I can find out why it won't work.

EDIT: Checked the Double Dragon file you mentioned and that particular file does not exist in their database for the NES platform (at leas I can't find it, so please provide a direct link to the NES version of this exact file if you did find it). So that is the reason. Whatever UXS does I can't tell you why it recognizes it. From the screenscraper APIv2, I can't get a hit on that file when it doesn't exist. I'm guessing this is the case with all of the remaining files you have trouble with.

EDIT2: I just manually looked over all the Battletoads roms for the NES platform in their database. I can't find a file with that file name.

Brunnis

@muldjord said in Versatile C++ game scraper: Skyscraper:

You need to run the game list generation pass after gathering the resources, please read the docs, it's explained there. This was also the case with 3.0.1.

Okay, that's weird. I haven't even seen "gamelist generation" mentioned in the RetroPie-Setup script menu from where I initiate the scraping. I though I'd be handled automatically? With Skyscraper 3.0.1 I could delete gamelist.xml, enter RetroPie-Setup, perform the scraping and the gamelist.xml would be generated.

But I'll read the docs and see if I can get a handle on it.

I'll check that one example you mention and see if I can find out why it won't work.

EDIT: Checked the Double Dragon file you mentioned and that particular file does not exist in their database for the NES platform (at leas I can't find it, so please provide a direct link to the NES version of this exact file if you did find it). So that is the reason. Whatever UXS does I can't tell you why it recognizes it. From the screenscraper APIv2, I can't get a hit on that file when it doesn't exist. I'm guessing this is the case with all of the remaining files you have trouble with.

No idea why you can't get a hit on that file. Just search from the screenscraper.fr front page with the MD5 hash and it'll list a single result for NES: https://www.screenscraper.fr/rominfos.php?romid=265259

muldjord

@Brunnis said in Versatile C++ game scraper: Skyscraper:

With Skyscraper 3.0.1 I could delete gamelist.xml, enter RetroPie-Setup, perform the scraping and the gamelist.xml would be generated

If you chose the "generate game list(s)" option yes. Not while scraping.

I see the file now. It is odd that it won't recognize it. I even tried renaming it to match the same name, but still no luck. I have no explanation for this currently, but I'll dig a bit deeper. Thanks.

Brunnis

@muldjord said in Versatile C++ game scraper: Skyscraper:

@Brunnis said in Versatile C++ game scraper: Skyscraper:

With Skyscraper 3.0.1 I could delete gamelist.xml, enter RetroPie-Setup, perform the scraping and the gamelist.xml would be generated

If you chose the "generate game list(s)" option yes. Not while scraping.

I had apparently not updated the RetroPie-Setup script in a few weeks. I see now that the whole menu has been redone. Problem solved.

I see the file now. It is odd that it won't recognize it. I even tried renaming it to match the same name, but still no luck. I have no explanation for this currently, but I'll dig a bit deeper. Thanks.

Okay, thanks.

muldjord

@muldjord I figured out why now. Stay tuned...

EDIT: Yes, this is because of un-escaped amperands in the returned xml. The file actually returns a result, but my parser quits because it doesn't see valid xml. I've reported this to the Screenscraper team, hopefully they will be interested in fixing it. If they do it will work without updating Skyscraper.

Brunnis

@muldjord said in Versatile C++ game scraper: Skyscraper:

@muldjord I figured out why now. Stay tuned...

EDIT: Yes, this is because of un-escaped amperands in the returned xml. The file actually returns a result, but my parser quits because it doesn't see valid xml. I've reported this to the Screenscraper team, hopefully they will be interested in fixing it. If they do it will work.

Nice that you found it! That's a pretty major issue... Yep, let's hope they'll fix it.

muldjord

@Brunnis I've worked around the invalid xml by simply fixing it internally before I parse it. I've just released Skyscraper 3.0.3 where this is fixed. Please try it out with the roms you are having trouble with and let me know how it goes. I've already tested it with the Battletoads file and that worked brilliantly.