search

impiaaa / smstools

Scripts for decoding data files from Super Mario Sunshine, Zelda: Wind Waker, and more
21 stars 3 forks source link

BMG: Support STR1 chunk, extra INF1 entry bytes #5

Open TROguz opened 4 months ago

TROguz commented 4 months ago

Will there be support for repacking bmg files?

impiaaa commented 3 months ago

It was already able to convert cutscene BMG's to and from SRT. I changed it so now non-cutscene files export to CSV, and those can be brought back to BMG.

TROguz commented 3 months ago

It was already able to convert cutscene BMG's to and from SRT. I changed it so now non-cutscene files export to CSV, and those can be brought back to BMG.

I've tried and doesn't work repack csv to bmg. Here. Some example file: message_en.zip

impiaaa commented 3 months ago

Ok, this file has 2 things: first, a STR1 section, which I don't parse yet, and second, extra data in the INF1 entry pad bytes. The issue with fixing the first issue is just trying to find a good file format to convert this to. I could just do JSON but that's not very friendly for translations. The second issue is more fundamental: INF1 entries can be in a number of formats, entirely depending on which part of the game code reads it. Right now the script makes a guess depending on the size of the entries, and then the number of entries when converting back. Supporting the extra 3 pad bytes would make that format ambiguous with another. So I guess I could go with JSON and explicitly specify the entry size, or maybe a sort of "header row" in the CSV. I'll reopen this and think about it.

TROguz commented 3 months ago

Ok, this file has 2 things: first, a STR1 section, which I don't parse yet, and second, extra data in the INF1 entry pad bytes. The issue with fixing the first issue is just trying to find a good file format to convert this to. I could just do JSON but that's not very friendly for translations. The second issue is more fundamental: INF1 entries can be in a number of formats, entirely depending on which part of the game code reads it. Right now the script makes a guess depending on the size of the entries, and then the number of entries when converting back. Supporting the extra 3 pad bytes would make that format ambiguous with another. So I guess I could go with JSON and explicitly specify the entry size, or maybe a sort of "header row" in the CSV. I'll reopen this and think about it.

There are 186 bmg files in total. I can send them all privately if you want. This was just an example. We can talk on Discord.