Versatile C++ game scraper: Skyscraper

screech

@muldjord Just checked for the Screenscraper DB ;) "Amiga CD32", "Amiga AGA", "Amiga CDTV",... are child of "Amiga" ;) so if you scrape a CD32 rom on "Amiga" system, it normally return the good game referenced on "Amiga CD32" ;)

It's the same for "MAME" for exemple, if you scrape a CPS1 game in the MAME System, it will return the good game (not needed to specify CPS1 ;) ).

muldjord

@screech Yes, this is true, but switching the platform (in Skyscraper) is necessary for the match to occur. If you scrape the 'amiga' platform with Skyscraper, it will not accept an entry for the 'Amiga CD32' platform because some scraping modules are name search based. So it needs to change the platform in order to accept the returned entry. :)

With Screenscraper this is less important since it's checksum based. But for the others, it's a necessity since many games are multiplatform, and Skyscraper needs to find the right one. :)

tacodog

I have a couple question about the usage of Skyscraper that I was hoping someone could help with.

My first issue is that I use m3u files for my psx games because it handles multi disc games organically giving me just one entry in emulationstation. My m3u file points to ../psx-discs/...(disc 1).bin ../psx-discs/...(disc 2).bin etc. If I scrape by m3u filename, nothing is found. If I check by disc, I get duplicates for each disk, and the gamelist file doesn't point to the m3u. Is there a way to search by filename rather than checksums?

Secondly, I have scraped all of my systems using screenscraper, and I am missing a handful of files.

Local database cache stats:
'screenscraper' module
  Titles       : 428
  Platforms    : 428
  Descriptions : 426
  Publishers   : 418
  Developers   : 412
  Ages         : 171
  Tags         : 414
  Ratings      : 279
  ReleaseDates : 316
  Covers       : 428
  Screenshots  : 428
  Wheels       : 423
  Marquees     : 423
  Videos       : 0

I'd like to fill in these small gaps (just 3 wheels) from a second source, but don't necessary need to grab everything all over again. Is there a way to tell Skyscraper to only get the missing files? Or can I see which files are missing so I can try to grab them manually?

Thanks

muldjord

@tacodog You can scrape single files simply by providing the full or partial path to the rom on command line. You can also use the '--startat' and '--endat' command line options to only scrape a span of roms. As for seeing exactly which roms are missing the data, that is currently not possible.

For the m3u issue, the reason it isn't working for those is that the source you are using (ScreenScraper) doesn't have those exact files in their database. You'd need to overwrite the checksum manually, which can be done with '--query'. Try checking out the release notes for 2.7.5 here: https://github.com/muldjord/skyscraper/releases/tag/2.7.5

It's described in there. You can also try scraping it with another source, such as "mobygames" or "thegamesdb". They search by name instead of checksum, so if your filenames are good, they should provide good data. Not for wheels though, as those sources don't support wheels.

hermit

would it be possible to scrape Mess roms?
example for Amstrad GX4000 :
barb2.zip = Barbarian 2
brubber.zip = Burnin' Rubber

muldjord

@hermit Skyscraper supports many of the MESS platforms simply by scraping them with their own names such as:

$ Skyscraper -p nes -s screenscraper

But GX4000 in particular isn't supported.

annomatik

Hi! Skyscraper scraps many games for me, but some it does not scrap are quite common which I find odd. E.g. Chaos Engine and Lemmings.

Here's the list of some of the currently not scraped games I'm experiencing with Skyscraper:

Amiga

AlienBreed1_v2.3a_0998.lha (2 + 3D scrap fine)
"Bad Ninjas (Homebrew 2018).adf"
Cadaver&CadaverThePayoff_v2.2_0900.lha
ChaosEngine1_v1.2_AGA_1324.lha
ChaosEngine2_v2.1_AGA_0173.lha
CrazyCars1_v1.2.lha
CrazyCars2_v1.1_0311.lha
CrazyCars3_v2.1.lha
DefenderOfTheCrown_v3.1_0317.lha
DeluxePacMan_v1.0_AGA.lha
EyeOfTheBeholder1_v1.5_0116.lha
EyeOfTheBeholder2_v1.1_0834.lha
GhostsNGoblins_v1.2.lha
GhoulsNGhosts_v1.0_2252.lha
Gobliiins1_v1.3_0065.lha
Gobliins2_v1.4_0185.lha (Goblins3 scraps fine)
GreatGianaSisters_v1.6_2945.lha
"Inviyya (Homebrew Demo 2018).adf"
ItCameFromTheDesert1_v2.0_0014.lha
ItCameFromTheDesert2_v2.0.lha
JamesPond1_v2.1.lha
JamesPond2_v2.0_AGA_1354.lha
JamesPond3_v1.0_AGA_0688.lha
"Lemmings 3 AGA (1 of 4).adf"
Lemmings1_v1.5_Files_2089.lha (Lemmings 2 scraps fine)
Llamatron_v1.1_1MB.lha
Lotus1_v1.1_0774.lha
Lotus2_v1.11_0497.lha (Lotus 3 scraps fine)
MegaLoMania_v1.7_0272.lha
MontyPythonsFlyingCircus_v1.3_0273.lha
Nebulus1_v1.3_0361.lha (Nebulus 2 scraps fine)
"Operation Lemming v1.2.adf"
Populous1&DataDisks_v1.1_0069&1217.lha
Populous2&ChallengeGames_v1.3_0079.lha
ProjectX1_v1.3_0886.lha
ProjectX2_v1.3_0289.lha
ProjectXSE_v1.5_0927.lha
RType1_v1.4.1_0940.lha (RType 2 scraps fine)
SensibleTrainSpotting_v1.0.lha
ShadowOfTheBeast1_v2.2_1357.lha
ShadowOfTheBeast2_v1.3a_1359.lha
ShadowOfTheBeast3_v1.6_0016.lha
Speedball1_v2.0_0581.lha (Speedball 2 scraps fine)
SuperMethaneBros_v1.3.lha
TitusTheFox_v2.0_0226.lha
Turrican1_v2.0_0092.lha ( Turrican 2 + 3 scrap fine)
Xenon1_v1.2_0399.lha (Xenon 2 scraps fine)
"Zerosphere (Homebrew 2015).adf"
NES

"Balloon Fight (USA) 2P.nes"
"Banana Prince (GER).nes"
"Castlevania 2 (USA).nes" (Castlevania 1 scraps fine)
"Donkey Kong Jr. (World).nes" (Donkey Kong scraps fine)
"Galaga (USA).nes"
"Ninja Gaiden 1 (USA).nes"
"SCAT (USA) 2P.nes"
"StarTropics 1 (USA).nes"
"Tetris (Nintendo).nes" (Tetris (Tengen) scraps fine)

muldjord

@annomatik Most of the LHA's are not yet in the screenscraper database. It will work for some of them with the openretro module, but overall they are poorly supported at this time (for all scrapers, not just Skyscraper). We are in the process of making them work with both openretro and screenscraper.

As for the nes games, try looking them up on the screenscraper site and check if your checksums for your roms match those in their database. If they do please provide some documented examples of matches that don't return a result and I can look into this further. Thank you. :)

annomatik

thanks, will do. Which checksum tool should I use (either Windows or Raspbian would be fine)?

mitu

@annomatik On Linux you can use md5sum <filename> or sha1sum <filename>, on Windows I think you can use Powershell to compute them - https://docs.microsoft.com/en-us/powershell/module/microsoft.powershell.utility/get-filehash?view=powershell-6.

muldjord

Hi guys, please check here: https://retropie.org.uk/forum/topic/19588/skyscraper-now-officially-part-of-retropie-please-test

:)

annomatik

ok, here are the check sums for the NES games... which database should I compare them to?

The MD5 sums:

8bff672e12a06787abfa3372f27f2ae4 Balloon Fight (USA) 2P.nes
a1db8a18bae78e276d22e0725062b559 Banana Prince (GER).nes
73fa5a34a5c1579161a8c1e8ef1b2afc Castlevania 2 (USA).nes
e824d5b1c62bc13de3d2716d29225371 Donkey Kong Jr. (World).nes
9ab02f9f92d67a7ae91ebf981c1e2adf Galaga (USA).nes
2e4a6b4553d1f47db8f0e484aa5cfa3e Ninja Gaiden 1 (USA).nes
d8cbc5f2dbf44fae5239a170ad6a092d SCAT (USA) 2P.nes
cd7816986cbae890e15debd5a06e6414 StarTropics 1 (USA).nes
5c4ab6b03431f7298b19bcab58e858f0 Tetris (Nintendo).nes

The SHA1 sums:

3031c47d4481360a911763213fc3ee4a94779d0a Balloon Fight (USA) 2P.nes
ef8214f4bad14443b37121091690416f8375fbfc Banana Prince (GER).nes
990f6b4c12812ab11f2716d82bfda59de742d636 Castlevania 2 (USA).nes
1e6fcc6af295a2e41f8370efcf2440da4f29d017 Donkey Kong Jr. (World).nes
f0c642b45424b85231e8fe678c9b741fbf16cce9 Galaga (USA).nes
82e8d9e63a2f7cdaefc89ca3a8946b793b15db76 Ninja Gaiden 1 (USA).nes
74a15693eebc2c01c7982c0c9a3f294bc26f53a7 SCAT (USA) 2P.nes
ec6ebf666e3871a8dd6f296a6e200a3ca3d82a87 StarTropics 1 (USA).nes
865849aa2064ab38a5959116e697ebbfda9f4795 Tetris (Nintendo).nes

mitu

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

ok, here are the check sums for the NES games... which database should I compare them to?

I posted in your other topic how you can check the data on the Screenscraper.fr site, each game has it's own set of name/roms/hashes associated (in the 'ROMs/ISOs' tab of the game detail page)

0_1539342857947_986b9071-805d-4840-a8a0-a9d1eab95809-image.png

annomatik

Wow, that's a ton of games there... :-)

I just checked the first couple of games. Balloon Fight, Banana Prince, Castlevania 2, Donkey Kong Jr. : many versions are there, but not with the checksums I have.

So... I can think of at least four solutions:

Do I really have to keep looking for the right roms ? That would be probably a lot of effort.
I could try adding my own versions to the lists, but why, the meta-data is already there, the roms are working fine, there's already many different versions of the same game (why not just have a list of different checksums for the same game?)...
Could Skyscrapper just match filenames, not CRC sums? How would I enable that? and, if not
Can I maybe give Skyscrapper a pointer to the right meta-data?

In Kodi, it is possible to add a .nfo file, which can contain the link to the details-page of a movie / tv show episode. It is in the exact same solution like the video, just instead of mp4/mkv/whatever, it has the extension nfo. In its simplest form, it's just a simple link, but it can also be used for storing all meta-data, if wanted. And if combined with a link, it first takes the provided meta-data from the nfo and takes the rest via scrapping from the nfo.

Maybe, if that's not yet possible, it would be possible to provide direct links somehow?

Thanks!

tacodog

@muldjord Thanks for the replay about the psx games. I was able to write a script that would read the m3u file to get the cue file, and then grab the checksum of the bin/cue files and grab the correct images using --query. It works great.

I have another question/issue. I just scraped my nes roms and Super Mario 2 seems to grab the japanese wheel no matter what I do. I've tried --region 'us', 'eu', and 'wor', but I always get the "Super Mario USA" wheel.

Is this an issue with the way I'm running Skyscraper? an issue with the program itself? or an issue with the screenscraper database? Can you repeat this?

The wheel I want is located here:
https://screenscraper.fr/image.php?gameid=1248&media=wheel&hd=1&region=wor&num=&version=&maxwidth=600&maxheight=300

Skyscraper -p nes --refresh --region 'wor' --unpack -s screenscraper -i . -o . -g . --pretend /mnt/exhd0/roms/nes/Favorites/Super\ Mario\ Bros.\ 2\ \(USA\)\ \(Rev\ A\).zip

The md5 of my (unzipped) rom is d769d73689ea4684dbb0bd7cf3a24b2f.

Thanks

muldjord

@annomatik Skyscraper already matches just filenames when it looks for games at screenscraper. But it only works if the filename is unique in their database. So if they have 2 roms with the same filename, it won't work. This is not a Skyscraper thing, but a screenscraper thing, so not something I can change. You can try scraping the games with some of the other sources. They are filename search based.

You can, however, overwrite the checksums Skyscraper uses when looking for the games with the '--query' option. Read more about that here: https://github.com/muldjord/skyscraper/releases/tag/2.7.5

It works from 2.7.5 and newer. :)

muldjord

@tacodog I'll need to look into that when I get the time. It could be a bug of some sort. If the '--region us' doesn't pull it in, there might be a problem. If the wheel exists in their database, it should use it. If it didn't it will prioritize other regions until it finds one that does exist. So either there's a problem with the us wheel in their database, or there's an issue in Skyscraper. Should be easy to reproduce though.

tacodog

@muldjord One more request/suggestion. Would it make sense to divide the import directory by system? My understanding is that all images now go in import/screenshots. So if I wanted to use custom images for Mortal Kombat 2, I would have to remember to change the images in the import directory when building snes or megadrive databases. Not really a big problem, just an idea.

annomatik

Cool, will try the query option. Thanks!

annomatik

Didn't work :-( I used the following script, it said it scraped something, but when I started Emulationstation again, there still was no meta-data for the games I specified:

#!/bin/bash
./Skyscraper -p nes -s screenscraper --query sha1=F5E4CD9F287A8233EEE35AD1881B2770D9F2C453 "/home/pi/RetroPie/roms/nes/Balloon Fight (USA) 2P.nes"
./Skyscraper -p nes -s screenscraper --query sha1=74357C80B61FFD25AC2A802391068600F2646D98 "/home/pi/RetroPie/roms/nes/Banana Prince (GER).nes"
./Skyscraper -p nes -s screenscraper --query sha1=85793B0F013EC0F24F36FED0A2859B3B74131510 "/home/pi/RetroPie/roms/nes/Castlevania 2 (USA).nes"
./Skyscraper -p nes -s screenscraper --query sha1=02633E208732B598E3A8EB80B6E0E09926F25E83 "/home/pi/RetroPie/roms/nes/Donkey Kong Jr. (World).nes"
./Skyscraper -p nes -s screenscraper --query sha1=DA54C223D79FA59EB95437854B677CF69B5CAC8A "/home/pi/RetroPie/roms/nes/Galaga (USA).nes"
./Skyscraper -p nes -s screenscraper --query sha1=CA513F841D75EFEB33BB8099FB02BEEB39F6BB9C "/home/pi/RetroPie/roms/nes/Ninja Gaiden 1 (USA).nes"
./Skyscraper -p nes -s screenscraper --query sha1=74A15693EEBC2C01C7982C0C9A3F294BC26F53A7 "/home/pi/RetroPie/roms/nes/SCAT (USA) 2P.nes"
./Skyscraper -p nes -s screenscraper --query sha1=FDD3D68C7B5B2F9DC7EC50D9888B0A8EDBAAEC28 "/home/pi/RetroPie/roms/nes/StarTropics 1 (USA).nes"
./Skyscraper -p nes -s screenscraper --query sha1=77747840541BFC62A28A5957692A98C550BD6B2B "/home/pi/RetroPie/roms/nes/Tetris (Nintendo).nes"