Retool: a 1G1R tool for Redump and No-Intro

[jaerik]

Wanted to thank you for this!  Been having way too many versions and this is just amazing to me.

 

[unexpectedpanda]

4 hours ago, jaerik said:

Wanted to thank you for this!  Been having way too many versions and this is just amazing to me.

 

You're welcome :)

[BlockABoots]

Any chance of a quick guide how to do a 1G1R dat for Final Burn Neo??. I have created a DAT (arcade only) file from FBN program itself and imported that into the latest build at the time of this post and left everything as default (as it seems to have selected the correct order Region and Languages) selected process DAT bus the command window says,

"No titles in the input DAT match your preferences. No DAT file has been created."

Any ideas?

 

[unexpectedpanda]

43 minutes ago, BlockABoots said:

Any chance of a quick guide how to do a 1G1R dat for Final Burn Neo??. I have created a DAT (arcade only) file from FBN program itself and imported that into the latest build at the time of this post and left everything as default (as it seems to have selected the correct order Region and Languages) selected process DAT bus the command window says,

"No titles in the input DAT match your preferences. No DAT file has been created."

Any ideas?

 

Retool only supports No-Intro and Redump DAT files. Here's why FinalBurn Neo and MAME aren't supported, from another feature request I had:

Quote

There's a reason there isn't a decent 1G1R solution for MAME/FBNeo DAT files -- the data structure doesn't lend itself to that sort of processing. The lack of naming standard in the DAT descriptions makes it extremely challenging to automate anything, meaning the most likely path for filtering is an inflexible, curated, massive, and high-effort list that needs frequent updating.

In short, the DAT files are built to service the emulator, not user customization. Deduping or filtering beyond parent/clone wasn't a thing they had in mind when designing the format, and even then that relationship was only established to facilitate split sets. I don't see the situation changing either -- useful filter tools like catver. ini are still separate from MAME and updated by an external party manually, decades later. I think I've seen some MAME developer resistance too in regards to letting users split arcade from software lists -- filtering just doesn't seem to be part of the ethos.

On a personal note, supporting No-Intro and Redump already takes up significant amounts of my time.

I can definitely understand the desire for this, but the combination of unsuitable source data and time expense makes it infeasible for the foreseeable future.

TOSEC is also not supported:

Quote

Support is technically possible, but largely pointless to implement without also supplying clone lists. This isn't a small effort — with over 3,000 TOSEC DAT files as of 2022, and with TOSEC's tendency to include many, many variants, there'd have to be significant community effort not only to bridge that gap, but to test the clone lists too. At this point in time, the momentum doesn't exist to justify the work.

 

Edited January 7, 2024 by unexpectedpanda

[UnluckyForSome]

On 10/1/2020 at 10:43 PM, unexpectedpanda said:

Howdy folks,

I've been beavering away at this as a project to learn Python.

Retool converts Redump and No-Intro dats to 1G1R, doing a better job than dat managers with parent/clone dats. It has both GUI and CLI versions.

You'll still need a dat manager to use the files Retool creates, such as CLRMamePro, RomVault, or Romcenter — you just won't need to use their 1G1R modes, as Retool will have already done the work for you.

Why not use CLRMAMEPro or Romcenter's 1G1R mode with a parent/clone dat?

The short version:

Dat managers don't do a great job of picking canonical 1G1R titles, and No-Intro often misses clones in their dats.

The long version:

Historically if you wanted to create a 1G1R set, you'd use a parent/clone dat in combination with a dat manager like CLRMAMEPro or Romcenter. After loading the dat into the dat manager, you'd set your desired regions and region order, and whether or not to filter by languages (assuming the dat has <release> tags properly set up — something which is vanishingly rare). You'd then trust the dat manager to choose the perfect parent title for you from your favorite region, discarding the clones from other regions.

Here's the thing. The parent/clone dat format was created for MAME, to enable one of the ways in which it organizes its ROMs. A "parent" ROM in MAME contains the base or common files for a game, while "clone" ROMs only contain files that are different from the parent. If you load a clone game in MAME, it's smart enough to load the base files from the parent, and any of the modified files it needs from the clone.

Dat manager 1G1R is effectively a hack on top of this system. In 1G1R mode a dat manager uses the parent/clone relationships in a dat to set up a group of related titles. It then selects a single title from that group based on your region and language preferences and ignores the other titles, in an effort to only include the most desired version. While on the surface this seems reasonable, if you look a little closer you start to see the cracks.

The primary issue is that dat managers and parent/clone dats don't have an expansive concept of title priority. For example, what happens when there are two copies of the same title from the same region, but they have different names? Or different version numbers? Or were published by different companies at different times? Which title does the dat manager choose then? The answer — the first one it finds, not the newest or best.

There's another issue, and that concerns languages. To determine a 1G1R title, dat managers give each title a score, combined from your region and language priority. While selected regions are treated as a filter (don't include Europe, and you won't get European titles), languages aren't. Instead, languages are treated as a "bonus" score added to the original region score, meaning you can end up with titles in languages you don't want if a region is high enough priority. Check out LogiqX's pseudo code for an idea of how this works (search for "I do this kind of thing for a living..." to find the relevant bit).

Retool handles proper 1G1R title selection, and a whole lot more. It even identifies the languages of each title by using multiple sources — the implied language spoken in the region the title is from, languages explicitly listed in the title's filename, and languages listed on Redump's website and No-Intro's database, which aren't always included in the filenames.

After you set up the GUI or user-config.yaml to your liking, Retool's output is already 1G1R, meaning you don't need to select 1G1R mode, regions, or languages in your dat manager — just load the dat and go.

Clone lists

While Retool is smart enough to automatically match certain types of parents and clones, there are certain situations that require manual assignment, like when a title has a different name in different regions. To achieve this, Retool keeps clone lists.

At the time of writing, Retool's clone lists are the most thorough that I know of. I manually combed through titles in most dats, and cross referenced them on Wikipedia, Moby Games, Retroplace, GameTDB, VDGB, VGM, YouTube, Amazon.jp, PlayAsia, Sega Retro, PSCX2 Wiki, PlayStation DataCenter, The Cutting Room Floor, and Atari Mania. I checked out the parent/clone dats for No-Intro, and occasionally I went through Redump's site for Japanese, Korean, Russian, and Chinese characters for titles, so I could do translations and find out what they were called in other languages. When all else failed, I did some good old web searching in order to turn up information. At some point I discovered FilterQuest, a similar tool, and added some missing titles from there.

Clone lists are updated independently of the program, and are formatted as JSON files. They are stored in a subfolder called clonelists, which is in the same folder as Retool.

You can update them from the GUI using the File menu, or by running updateclonelists.py.

Features

• Region ordering and language filtering.
• Can exclude the following title types:
  • Add-ons
  • Applications
  • Audio titles
  • Bad dumps
  • BIOS and other chips
  • Bonus discs
  • Coverdiscs
  • Demos and samples
  • Educational titles
  • Manuals
  • Multimedia titles
  • Pirate titles
  • Preproduction titles (alphas, betas, prototypes)
  • Promotional titles
  • Unlicensed titles
  • Video titles
• Custom exclude and include filters.
• Output a list of the final 1G1R title names, and prefix and suffix each title with whatever you like.
• The most thorough clone lists available, which can update from the internet.
• A GUI.
• A legacy parent/clone dat creation mode, and a tool called CloneRel, so you can analyze the relationships Retool creates. This is mostly useful if you intend to update clone lists yourself.

Where can I read more or download Retool?

Over on the Retool Github page, where there are installation instructions for those familiar with Git and Python, and also for Windows users who haven't had much CLI, Git, or Python experience. Read the Wiki for the full set of instructions.

What else is out there?

There are a few different projects that all have slightly different focuses, like Pixelpiper's DCTF, AndreBrait's 1G1R romset generator, and UnluckyForSome's FilterQuest.

 

 

Hey, i’m just getting back into this when it seems you may have just finished! 😄

Just wanted to say thankyou - and your efforts blow FilterQuest (my rubbish script) out of the water 😄

 

Go well 👊