diffCleanupSemantic(Lossless) can run into an infinite loop

[aforemendude]

Describe the bug

For a specifically crafted input, diffFromDelta would produce a delta that cause diffCleanupSemantic and diffCleanupSemanticLossless to hang (infinite loop).

The documentation for diffFromDelta says that it would throw on invalid input, but it's not doing that for this case. Or this could be an issue with diffCleanupSemantic and diffCleanupSemanticLossless where they are mishandling specific diffs.

This is happening due to the empty edit/equality tuples. Reproducible on the latest v2.0.1.

Reproduction:

import { diffCleanupSemantic, diffFromDelta } from 'diff-match-patch-es'

const diffs = diffFromDelta('ac', '=1\t-0\t=0\t+b\t=1')

console.log('diff:', JSON.stringify(diffs))
diffCleanupSemantic(diffs)

Potential fix (unverified):

The original Java version has some extra check:

https://github.com/google/diff-match-patch/blob/62f2e689f498f9c92dbc588c58750addec9b1654/java/src/name/fraser/neil/plaintext/diff_match_patch.java#L938-L939

        while (edit.length() != 0 && equality2.length() != 0
            && edit.charAt(0) == equality2.charAt(0)) {

But missing from the JS (and TS) version:

diff-match-patch-es/src/diff.ts

Line 880 in 4f35fb7

while (edit.charAt(0) === equality2.charAt(0)) {

Reproduction

https://github.com/aforemendude/bugs-reproduction/tree/main/diff-match-patch-es-delta-inf-loop

System Info

  System:
    OS: Linux 7.0 Ubuntu 24.04.4 LTS 24.04.4 LTS (Noble Numbat)
    CPU: (4) x64 Intel(R) N95
    Memory: 12.39 GB / 15.40 GB
    Container: Yes
    Shell: 5.2.21 - /bin/bash
  Binaries:
    Node: 24.18.0 - /usr/local/nodejs/bin/node
    npm: 11.16.0 - /usr/local/nodejs/bin/npm

Used Package Manager

npm

Validations

• Follow our Code of Conduct
• Read the Contributing Guide.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Check that this is a concrete bug. For Q&A, please open a GitHub Discussion instead.
• The provided reproduction is a minimal reproducible of the bug.

Contributions

• I am willing to submit a PR to fix this issue
• I am willing to submit a PR with failing tests (actually just go ahead and do it, thanks!)