SEEN.TXT

Hi. I'm currently screwing around with a visual novel/eroge/bishoujo game called ALMA ずっとそばに (NOTE: While the page I linked to is work safe, links to the individual character profiles, are NOT...) I've done a click through at about somewhere between 60~80% comprehension for anything a girl says, and 10~25% for anything the protagonist thinks.

Since I'm starting over going for 100% comprehension, I'd like the text from the game so I can translate in places that...I wouldn't be able to play otherwise. (i.e. during my boring lectures that I don't pay attention in anyway.)

I've found a bunch of code here that looks helpful, especially the first one, but it segfaults when run, and the code is an absolute, non-commented mess. I'd really like the specifications to AVG32 or AVG200/RealLive to help me get through it. Even though ALMA is written in AVG2000, I'm guessing the SEEN.TXT format is mostly the same; there's probably just a few more opcodes and commands...

Does anyone have a link to the specs?

I've already STFW for "SEEN.TXT", "SEEN.TXT スペック", "SEEN.TXT 仕様書".

[Stranger]

Sorry in advance if my answer is not that helpful, espicaly that I do NOT have ALMA at the moment to test if this will work or not. But according to you, kanon, (AIR) and ALMA uses (hopefuly) the same SEEN.TXT file format (and in that case, CLANNAD as well).

If they are all using the same format for the SEEN.TXT file then, try to use Haeleth's vAVG32e to extract the game (click the "EXTRACT game" button). This should make a sub directory where the SEEN.TXT file is called scenario that will include all the SEENs as normal text file.

You might need to modify the INI of vAVGe so that it will excipt ALMA as well. The new entry in the INI file should (hopfuly) look like this:
<tt>
[GAME6]
GAME_NAME=ALMA ずっとそばに
REG_NAME=<ENTER THE REG LOCTION HERE>
EXE_NAME=<ENTER EXE NAME HERE>
</tt>
Hopefully the above will work, but as I said, I do NOT have ALMA to test if this will work or not. That shoud help in unpacking the SEEN.TXT, I have no idea how to repack it though. Sorry.

Originally Posted by Stranger

Sorry in advance if my answer is not that helpful, espicaly that I do NOT have ALMA at the moment to test if this will work or not. But according to you, kanon, (AIR) and ALMA uses (hopefuly) the same SEEN.TXT file format (and in that case, CLANNAD as well).

No problem. I'm just glad there's someone on this cruel, cold Internet kind enough to help me.

Originally Posted by Stranger

Hopefully the above will work, but as I said, I do NOT have ALMA to test if this will work or not. That shoud help in unpacking the SEEN.TXT, I have no idea how to repack it though. Sorry.

After doing that, (and a little more since for some reason when I said F:, it would look in F:\dat\), I get the very helpful dialog box: "ERROR".

As for repacking...well, I don't really have any plans to do a big translation project, mainly since I've only completed second year japanese and I'm not sure how competent I am in the language. I missed about everything that the protagonist thought, and if the game wasn't voiced, the % of the game that I understood would have dropped quite a bit. I just thought I'd try to read through after someone said just to download JWPce and start reading.

Er, it looks like I wasn't loged in a when I posted the above. ^^

I believe Haeleth was poking at it earlier... I guess we'd need him to figure anything out :)

Since I don't actually have the game, all I can say is... Scn2kdmp works for the demo :) vAVG32 is not going to work though, since they changed the way SEEN.TXT is stored. Iit is no longer a general-purpose archive; instead, it has a huge (1000 entries) table as header containing the offsets to the beginning of each block of compressed scenes, or NULL if it's not in the archive. IFAIK, each scene is the same packed data as AVG32, but XORed with some key (see scn2kdmp).

It's probably easier to look at the Waffle source code for the LZ77 decompression stuff - I've found that' to be much more understandable. Of course, that only handles AVG32, so you still need to do the masking first.

(A large part of this information was from Haeleth, from some time ago - I like to keep the credits straight)

[Haeleth]

AVG2000/RealLive are only similar to AVG32 on the surface, and the SEEN.TXT is completely different. Different archive format, different bytecode. There is absolutely no binary compatibility. At the moment scn2kdump is the only program I know of that handles it at all, and, of course, it doesn't work on the full version of ALMA.

As Myen says, I've done a bit of hacking on this myself, and I did get some way towards producing a usable tool. I can decompress ALMA's SEEN.TXT, but that still leaves us with raw unreadable bytecode. I never got round to finishing the disassembler... ^^

Anyway, you can get the notes I made on the format of SEEN.TXT from here. Mostly it's based on scn2kdump, but in English. ;) It should be enough for you to cobble together a decompressor. There's nothing in there on the bytecode format, I'm afraid; for the purpose of dumping text it should be enough to know that any double-byte Shift_JIS characters are almost certainly part of the game's script, so just scanning the bytecode for those should be good enough.

Note that the compression is slightly different from AVG32's, even behind the XOR "encryption". VisualArt's can't seem to resist using a slightly different LZ77 variant in every single file format they come up with...

First of all, lots of thanks for the information! I've been busy with RL, and I'll be busy until Wed., so I can't test anything... Your notes look *extremely* helpful, though.

At the moment scn2kdump is the only program I know of that handles it at all, and, of course, it doesn't work on the full version of ALMA.

So all that code in scn2kdump that looks like a bytecode interpreter (class Cmd, etc) is, in fact, the bytecode format? Or a mostly complete part of the bytecode format?

Also, (and I'm fully aware that I'm asking for speculation here), Jagarl said that he tested scn2kdump against Princess Bride, but that was released several months after ALMA. So...er, I'm not sure what I'm trying to say here... ^^;; I'd think that an older game would work while a newer one would have updates to the game engine...

Finally, my current copy of ALMA is pirated. I need to make it legal if I'm going to continue with this project. Himeya doesn't have the DVDs. Does Getchu ship overseas? What about other shops in Japan? I know amazon.co.jp doesn't ship any software and I've CDs overseas (grrr. and amazon has it if for 3,000円 less then retail...)

----
P.S. Everyone go download the ED to Princess Bride. It's great. It's on the official site I linked to above.

[Haeleth]

Originally Posted by DrBlue

I've been busy with RL...

It took me several minutes to work out that you meant "real life" not "RealLive". ^^;

Originally Posted by He then

So all that code in scn2kdump that looks like a bytecode interpreter (class Cmd, etc) is, in fact, the bytecode format? Or a mostly complete part of the bytecode format?

Yup. Well... not even mostly complete actually, it's just the subset of the bytecode format required to be able to scan the script of Princess Bride and identify the text in it with 100% accuracy. Presumably the reason it crashes on ALMA is that ALMA uses commands which Princess Bride doesn't. Bear in mind that Jagarl doesn't have access to any official specs either!

Originally Posted by Finally he

Finally, my current copy of ALMA is pirated. I need to make it legal if I'm going to continue with this project. Himeya doesn't have the DVDs. Does Getchu ship overseas? What about other shops in Japan? I know amazon.co.jp doesn't ship any software and I've CDs overseas (grrr. and amazon has it if for 3,000円 less then retail...)

Yeah, damn Amazon... they're the only place I can find that's taking preorders on Clannad, too (apart from Key themselves, and I'm pretty sure they don't ship overseas). On the other hand, Amazon charge extortionate shipping, you'd probably end up paying most of that 3000 yen anyway. ;)

Rants aside, what I'd do in your situation is just buy the CD version from Himeya. I believe (correct me if I'm wrong) that the game is exactly the same once installed. Besides, you could easily make a moral argument for keeping the pirated DVD version once you've paid for the game.

By the way, if you're interested in poking around with RealLive at all you might find this helps - it's a patch to translate the debug menus in the version of RealLive.exe that ALMA uses. Could be useful if you want to try modifying the bytecode at all.

Oh, and in the unlikely event that you can read O'Caml, I'd be willing to share the source code to my incomplete RealLive disassembler. But if you don't know the language I doubt you'd find my messy ML at all preferable to Jagarl's messy C++!