pre-seed from web browser emulator

[aebrer]

Hello!

I am currently struggling with something. I have a usecase where I produce a generator as a tic-80 cartridge, export it via the web export, and now wish to modify the seed for the prng from the browser.

I've explored a few different options, but what I've been trying lately is to read in the cart.tic file before it's loaded, search for the seed=[some number] line, and replace it, then saving cart.tic

It doesn't work, and I was wondering if anyone could help me figure out why? This is my javascript code "seed_fuckery.js", which I am loading prior to loading the emulator:

// Read the contents of the cart.tic file using an XHR request
const xhr = new XMLHttpRequest();
xhr.open('GET', 'cart.tic', true);
xhr.responseType = 'arraybuffer';
xhr.onload = function() {
  console.log("Loaded cart.tic");
  // Convert the data to an array of bytes
  const bytes = new Uint8Array(xhr.response);

  // Create a file-like object from the array of bytes
  const file = new Blob([bytes], { type: 'application/wasm' });

  // Read the contents of the file using the FileReader API
  const fileReader = new FileReader();
  fileReader.readAsArrayBuffer(file);

  fileReader.onload = function() {
    // Convert the data to an array of bytes
    const bytes = new Uint8Array(fileReader.result);

    // Iterate over the buffer, starting from the beginning
    let pos = 0;
    while (pos < bytes.length) {
      // Read the 4-byte header of the current chunk
      const header = bytes.slice(pos, pos + 4);
      pos += 4;

      // Extract the bank number and chunk type from the first byte of the header
      const bankNumber = header[0] >> 3;
      const chunkType = header[0] & 0x1F;

      // Extract the size of the chunk from the second and third bytes of the header
      const size = new DataView(header.buffer).getUint16(1, true);

      // Read the chunk data from the buffer
      const chunkData = bytes.slice(pos, pos + size);
      pos += size;

      // If the chunk type is 5 (code), process the chunk data
      if (chunkType === 5) {
        // Convert the chunk data from a buffer to a string
        const code = String.fromCharCode.apply(null, chunkData);

        // Find the line that reads "seed=42069" (or any number)
        const lines = code.split('\n');
        for (let i = 0; i < lines.length; i++) {
          const line = lines[i];
          console.log(line)
          if (line.startsWith('seed=')) {
            // Extract the old seed value
            const oldSeed = parseInt(line.slice(5));
            console.log(`Old seed: ${oldSeed}`);

            // Replace the old seed value with the new seed value
            const newSeed = Math.floor(fxrand()*1000000000);
            lines[i] = `seed=${newSeed}`;
            console.log(`New seed: ${newSeed}`);
          }
        }

        // Convert the modified code back to a buffer
        chunkData = Uint8Array.from(lines.join('\n').split('').map(c => c.charCodeAt(0)));
      }

      // Create a Blob from the modified chunk data and header
      const output = new Blob([header, chunkData], { type: 'application/wasm' });

      // Write the output Blob to the output.tic file using the FileSaver API
      saveAs(output, 'cart.tic');
    }
  }
};

[aebrer]

this is semi-related to #1698

Specifically, it would be a different way to solve the seeding issue I am trying to solve. @joshgoebel gave me some advice there to look into hacking the pmem, this was the best I was able to come up with.

[joshgoebel]

Use diff file names just to confirm it's using the file you think it is.

[aebrer]

my original implementation had two filenames, with the output stream saving as cart.tic; unfortunately the output just doesn't save anything... not to mention I’m not sure this implementation would even work when served over the web?  On Sat, Dec 17, 2022 at 3:35 PM, Josh Goebel ***@***.***> wrote: Use diff file names just to confirm it's using the file you think it is. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>

[joshgoebel]

Also, this is not really what I would expect to see. You're not opening a new file for output anywhere and you're not writing all the sections to it unless save as is supposed to be doing that which is very weird naming.

I also don't see you recalculating the new size of the chunk after you've made changes to it?

[aebrer]

Yes, I think it's really fair to say I have no idea what I’m doing, I’m completely out of my element so far. On Sat, Dec 17, 2022 at 3:40 PM, Josh Goebel ***@***.***> wrote: Also, this is a really what I would expect to see. You're not opening a new file for output anywhere and you're not writing all the sections to it unless save as is supposed to be doing that which is very weird naming. I also don't see you recalculating the new size of the chunk after you've made changes to it? —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>

[joshgoebel]

I forgot there isn't a PMEM chunk, but it'd be easier to use any of the other binary chunks and just modify a few bytes than all this code to try and alter strings and put them back together again. Just give up a sprite, or even add a new chunk if you aren't using all 8 banks of any given chunk.

[aebrer]

Yeah since I only need to pass in a single int I can't imagine it taking too much space. I’m not even using sprites so I'd love to try that. Could you possibly give me a little guidance in that direction?  On Sat, Dec 17, 2022 at 3:45 PM, Josh Goebel ***@***.***> wrote: I forgot there isn't a PMEM chunk, but it'd be easier to use any of the other binary chunks and just modify a few bytes than all this code to try and alter strings and put them back together again. Just give up a sprite, or even add a new chunk if you aren't using all 8 banks of any given chunk. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>

[joshgoebel]

You have the right idea, load it, modify it, save it, feed the modified cart to TIC-80, but I don't know the exact code/APIs for all of that off the top of my head... sorry.