WASM in grapheme_iterator causing integer underflow

[GnomedDev]

Describe the bug
The subtraction at the below line is underflowing with a super large file.

sax-wasm/src/sax/grapheme_iterator.rs

Line 267 in e4ea343

Some((unsafe { &*ptr::slice_from_raw_parts(ptr.add(start), cursor - start) }, found))

To Reproduce
Steps to reproduce the behavior:

1. Download and extract https://osm-planet-eu-central-1.s3.dualstack.eu-central-1.amazonaws.com/planet/replication/day/000/004/543.osc.gz
2. Parse the file with the OpenTag event enabled.
3. Wait for crash.

Expected behavior
Underflow does not occur, file parses correctly.

Desktop (please complete the following information):

• OS: Linux
• Browser: Chromium 133.0.6943.98

Additional context
I am testing on a branch with #108 fixed by bumping the offset to 1024, and may be able to provide a smaller reproduction later.

[justinwilaby]

👀

[justinwilaby]

Fixing the underflow exposes a u32 overflow with the character increment on very large files.

I have a few options:

1. reset to zero when u32::MAX is reached
2. use u64 (BigInt on the JS side means breaking changes)
3. Introduce a flag that does not track line and character values.

Let me know which you prefer and I'll implement your preference.

[GnomedDev]

Using u64 sounds like the best option to me.

[justinwilaby]

@GnomedDev - Please confirm that #110 fixes the issue. I suspect this also fixes #108 but I was not able to repro this.

I'm seeing a successful parse completion of your test file at around 7 seconds or so with no errors.