Versatile C++ game scraper: Skyscraper

AnalogHero

muldjord

Skyscraper 2.0.5 released: https://github.com/muldjord/skyscraper

Added support for 'scummvm' platform in scraping module 'thegamesdb' or 'openretro'.
Now only removes 'the' from searchName if longer than 10 chars.
Now always converts underscores to spaces in search- and compare names.
Added edit distance optimization ('the sequel' will now match 'some game: the sequel' 100%).

Happy scraping! :)

muldjord

A little note on what is going on behind the scenes. I've been looking into creating some sort of GUI for Skyscraper to make it easier to first-time users to get good results. I've had a look at a bunch of sfml and sdl2 based gui frameworks, but not a single one of them are suitable. Believe me, I've been scouring the web for one that fits my requirements, and it's just not gonna happen. They are either too complex, or have too many dependencies to be feasible without making everything unnecessarily complex to install.

So, what I've basically decided to do, is to include a 'script-maker' mode when running Skyscraper. It will present the user with a bunch of questions, such as "What platform do you wish to scrape?" and other relevant stuff. And when all questions have been answered, it will create a script or possibly config file (or both) and ask if you want to run it.

Comments on this approach will be appreciated.

AnalogHero

@muldjord Maybe you can go with a shell script gui like the retropie-setup. Dont know if there is any framework there?

muldjord

@analoghero Yes, I am also considering that. Although I am partial to the "questionnaire" method since it can then be implemented directly in the software as I see fit. Thanks for suggesting it.

bleuge

@muldjord I compiled this in Windows using Cygwin some months ago, but got many problems running it. Do you think it is too much effort to make it run ok in Windows, scrapping is lot of faster and easier that connectin to Pi, etc. Thanks!

About commandline, I'm happy with it, but I know many users will be too complicated, an interactive shell as you suggest it will ok, while a graphical gui is thought and implemented. If we could run this in Windows i'll do a gui myself if it's needed. Regards!

muldjord

@bleuge You can easily cross-compile it for Windows. Just download and install Qt 5 from the official site (open source version). When installing, put a checkmark in the latest mingw32 version, I don't support MSVC++ although it might work, not sure. When it's installed you'll find a Qt specific command prompt in your start menu. From there navigate to the Skyscraper source folder and run:
$ qmake
$ mingw32-make release

That will give you a compiled version in a subfolder called "release". Then you need to copy the needed dll's from the qt5 install folder into the release folder and start an ordinary command prompt there. Then you should be able to run it.

I thiiiink it will create the .skyscraper data folder inside "C:\Users[username]... somewhere". Just keep in mind that there are some minor Linux specific stuff in Skyscraper's "make install" command that will copy relevant files to the .skyscraper folder. So you'd be missing those. But I believe I auto-create empty ones...

So yeah, shouldn't be too hard, but it's not really something I am going to look into myself at this time. Happy hacking! :)

bleuge

Thanks for tips, I already compiled ok in Cygwin (after some work with qt5). Yesterday i updated it and recompiled it. Copied the needed DLLs and ...

Running Skyscraper v2.0.5a by Lars Muldjord

Please set a valid platform with '-p [platform]'
Check '--help' for a list of supported platforms, now exiting...

What i meant about windows, was the folder/linux structure, making it portable or something like that. I'll try with some roms to see if i can make it working normally.
Thanks!

muldjord

@bleuge I am not sure what you mean. It IS cross-platform compatible (and clearly runs just fine looking at your quoted text). It uses the home folder variable to specify where to save settings, which depends on you OS. So if you are running Windows it will save stuff in the Users folder as expected. Please elaborate.

As I mentioned the only thing that isn't cross-platform is the "make install" part which I've made more convenient and automated on Linux. But it shouldn't really be necessary for it to run properly on Windows.

bleuge

@muldjord thanks for your answers, i'll elaborate.
I have subfolders with roms in c:\roms

Skyscraper.exe -p atari2600 -i c:\roms\2600 -t 8 -d c:\roms\00skdb --nobrackets --verbose

I got

Folder '/home/myuser/.skyscraper/c:\roms\2600/images' doesn't exist, trying to create it... Failed! Please check path and permissions

I've looked for .skyscraper file and found that Skyscraper created folders in

c:\home\myuser.skyscraper\dbs
c:\home\myuser.skyscraper\import...various subs here

I don't have any problem Skyscraper creating files and folders whenever it needs it. But well, i can't make it work under Windows.

Maybe i can create a config.ini file in the folder of the executable?

Thanks for your time, this scraper is very very good! Your idea of the localdb is wonderful.

muldjord

@bleuge I see a potential problem. Please try writing your paths with forward slashes instead of backward slashes.

I do find it odd that it looks for /home/myuser though. I use the builtin user home from Qt, which should adjust for Windows and use the Users folder.

A Former User

This might be a tougher request but is it possible to get .7z support for Nintendo DS? DraStic supports .7z and most rom sets are already in that format for the platform.

Thanks.

bleuge

@muldjord I updated everything and recompiled it.

This time i ran it at my main pc at home. W10 this time.

I have the skyscrapper executable (and required dlls) in d:\ES\ (emulationstation folder)
Roms in c:\roms\

Skyscraper.exe -p atari2600 -i c:/roms/2600 -t 2 -d c:/roms/00skdb --nobrackets --verbose
Skyscraper.exe -p atari2600 -i c:\roms\2600 -t 2 -d c:\roms\00skdb --nobrackets --verbose
(both work) I changed -t 8 to -t 2, it seems it was hanging for no reason, maybe my connection is too slow. With -t 8 it processed some games and stay there, no more output, with -t 2 it keeps going until end.

Running Skyscraper v2.0.5a by Lars Muldjord
Platform : 'atari2600'
Scraper module : 'thegamesdb'
Input folder : 'c:/roms/2600'
Gamelist folder : 'c:/roms/2600'
Images folder : 'c:/roms/2600/images'
Local db folder : 'c:/roms/00skdb'

Folder '/home/bleuge/.skyscraper/c:/roms/2600/images' doesn't exist, trying to create it... Success!

I've used backslashes or forward, the result is the same. And... where are the files created? In d:\ (where skyscraper exe is and is run) Look the folder structure created by skyscraper:

D:\home\bleuge
D:\home\bleuge.skyscraper
D:\home\bleuge.skyscraper\c?
D:\home\bleuge.skyscraper\dbs
D:\home\bleuge.skyscraper\import
D:\home\bleuge.skyscraper\skipped-thegamesdb.txt
D:\home\bleuge.skyscraper\c?\roms
D:\home\bleuge.skyscraper\c?\roms\00skdb
D:\home\bleuge.skyscraper\c?\roms\2600
D:\home\bleuge.skyscraper\c?\roms\00skdb\covers
D:\home\bleuge.skyscraper\c?\roms\00skdb\db.xml
D:\home\bleuge.skyscraper\c?\roms\00skdb\screenshots
D:\home\bleuge.skyscraper\c?\roms\00skdb\videos
D:\home\bleuge.skyscraper\c?\roms\00skdb\covers\thegamesdb
D:\home\bleuge.skyscraper\c?\roms\00skdb\covers\thegamesdb\47775e7b9ff344126bc7f5ffe33e05c63bea5764.png
D:\home\bleuge.skyscraper\c?\roms\00skdb\screenshots\thegamesdb
D:\home\bleuge.skyscraper\c?\roms\00skdb\videos\thegamesdb
D:\home\bleuge.skyscraper\c?\roms\2600\gamelist.xml
D:\home\bleuge.skyscraper\c?\roms\2600\images
D:\home\bleuge.skyscraper\c?\roms\2600\images\Adventure (USA).png
....

I know this is not designed to work in Windows, but something is going very wrong with folders and file paths, etc... I am very bad at C++, i'll try to find something in the source code.

Look that funny c? ... it's exactly 0x3f char. Also i looked the skipped files list (skipped-thegamesdb.txt ):

--- The following is a list of skipped games ---
'Montezuma's Revenge - Featuring Panama Joe (USA)', Closest match was 'Montezuma's Revenge'
'Basketball (USA)', Closest match was 'BMX Air Master'
'Battlezone (USA)', Closest match was 'Basketball'
'Beamrider (USA)', Closest match was 'Battlezone'

This seems also wrong. Maybe something is going very bad in this windows compile?

Thanks for your time !!!

muldjord

@bleuge I've been thinking about this some more, and I think your problem comes from running it through cygwin. Cygwin is a POSIX env for Windows. If you want to run it on Windows, I suggest you use the Qt provided command line - both for compiling and running it.

Basically cygwin messes with A LOT of the environment stuff. For instance, the home folder for Cygwin is not the Windows home folder. This is a big problem in itself. And 'C:' is '/cygdrive/c' in Cygwin. So when you type in C: on your Skyscraper command line it is not going to work for that reason alone.

So there's a pointer for your further investigation. Simply DON'T use Cygwin and use the official tools provided by Qt instead.

muldjord

@livefastcyyoung I will add .7z support for all platforms, no problem. :) I will probably do a quick release with just this fix soon. Look out for it on the github page, and here.

muldjord

Skyscraper 2.0.5b released: https://github.com/muldjord/skyscraper

Added support for '*.7z' files for all platforms

Happy scraping! :)

bleuge

@muldjord ok, understood, thanks! I'll try to compile Windows native, thanks!

muldjord

@bleuge Good luck with it! :)

A Former User

Has anyone made a config file that would format images the same way sselph’s scraper does?

Every way I’ve tried changing this config to match, it’s still not quite right.

jura

@muldjord , your tool works amazingly great!
This good example when you are good at something and your are doing something for the community! :)

I have 700 nes and 700 snes and 1000 megadrive roms and scraping those took 30mins.
Which was amazing!

It found 80% of the covers for the roms but for c64 it found only 700 when i have 17 000 (17 thousand) roms. :)
Dunno is it possible to set the match percentage as a parameter to allow lower findings, or is it even necessary.

Any idea how I could increase the result persentage at least for C64?