Cannot parse replays after Last Chieftains DLC update

[emissions-dev]

Hi, since the February 17 2026 update (Americas/Chieftains DLC), mgz fails to parse replays with an assertion error in de_string().

Error:

File "/venv/lib/python3.12/site-packages/mgz/fast/header.py", line 46, in de_string
    assert data.read(2) == b'\x60\x0a'
AssertionError
Full traceback:
Traceback (most recent call last):
  File "mgz/fast/header.py", line 744, in parse
    de = parse_de(header, version, save)
  File "mgz/fast/header.py", line 455, in parse_de
    de_string(data)
  File "mgz/fast/header.py", line 46, in de_string
    assert data.read(2) == b'\x60\x0a'
AssertionError

RuntimeError: could not parse:

Environment:

• mgz version: 1.8.51 (latest from GitHub as of Feb 20)
• Python: 3.12
• Game version: Age of Empires II: DE post-Americas DLC (Feb 17, 2026)

Notes:

The \x60\x0a pattern still exists in the replay file but at a different offset (found at position 27823)
Replays from before the Feb.. 17 update parsed correctly?

[denniske]

@happyleavesaoc Hi, I have the same issue. Any updates on this?

[Kjir]

There's 2 PRs that address the issues with the fast parser, #139 and #142 (they do essentially the same thing)

[manuel-arias-torcal]

Hi,
Not sure if you have already seen it, but in setup.py file, in the install_requires list, there is the reference to aocref>=2.0.35, but its current version is 2.0.37. The dataset from AoE2 DE has been recently updated.
I applied localy changes proposed in #142 and got an error on file
File "c:\Users\mat_m\OneDrive\Documentos\AOE2\aoc-mgz-master\mgz\model_init_.py", line 174, in parse_match
dataset['civilizations'][str(player['civilization_id'])]['name'],
~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
KeyError: '58'

This civilization_id 58 corresponds with Mapuche.

[Kjir]

Hi, Not sure if you have already seen it, but in setup.py file, in the install_requires list, there is the reference to aocref>=2.0.35, but its current version is 2.0.37. The dataset from AoE2 DE has been recently updated. I applied localy changes proposed in #142 and got an error on file File "c:\Users\mat_m\OneDrive\Documentos\AOE2\aoc-mgz-master\mgz\model__init__.py", line 174, in parse_match dataset['civilizations'][str(player['civilization_id'])]['name'], ~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ KeyError: '58'

This civilization_id 58 corresponds with Mapuche.

Try the changes in #139 ,which also deal with missing civ IDs.
Or update aocref, 2.0.37 does match >=2.0.25

[manuel-arias-torcal]

Thanks a lot, @Kjir. With suggested changes, I have successfully parsed one game played yesterday using parse_match from mgz.model.

[happyleavesaoc]

Hi folks, I'd like to merge this but traditionally we also update the full parser (but this is a lot more work). We have a few options here:

• Merge this as-is and update the compatibility table in the README
• Wait for full parser support

We could also simply drop full parser support forever. I think this will make it harder to maintain the fast parser. Thoughts?

[denniske]

I vote for "Merge as is" 👍