There are deficiencies in the HATCH - related rendering, and the FONT does not wrap.

[hqbHawke]

[vagran]

Some sample file needed. It may be some synthetic minimalistic sample illustrating some particular problem.

[hqbHawke]

Of course, I'm happy to provide them. Here are the drawings with bugs. https://aigc-1364883968.cos.ap-guangzhou.myqcloud.com/aigc/2025/08/05/00971c0ae2844f0dab39832f2058b4f0.dxf

[eXponenta]

Complex hatch wont stable render due to unstable tesseltaion. earcut that used for tesselation not handle a lot of cases when fill contours intersected or collinear themself, or when one of point included in more that 1 contour.

For that i use tess2 ( GNU TESS implementation)

@hqbHawke You can check my fork and if ok i going to make PR.

https://github.com/eXponenta/dxf-viewer/

Sample:

Source: complex.zip

About mtext: its to hard stable resolve mtext wrapping in Japanese/Chinese/Hundi and all Arabic languages due to hieroglyphics and ligatures. Only browser know layout rules for that languages. This is because when you decompose non-ASCII based text it can be broken ( ligatures and hieroglyphics is 2+ byte long and if you split one then you may got 2 another different characters)

[eXponenta]

@hqbHawke

[hqbHawke]

@hqbHawke

The analysis of "hatch" seems correct. For the mtext entity, there will be a font width corresponding to 40 in the parsed result. We can recognize this width to achieve automatic line wrapping. I have made adjustments based on your work, and now it can be parsed normally. You can try this approach

[hqbHawke]

@eXponenta

It can be seen that the text line breaks are displayed correctly. then,The issue with the text lies in the encoding format. Some use GBK while others use UTF-8. You can adopt certain methods to identify the file encoding, as I notice you're using UTF-8 by default for all files.

[eXponenta]

@hqbHawke Yep, encoding existing in document header. There are a way got it before parsing stage and redecode file in new code page.

This require rewrite DXFFetcher and insert to it validation pass.

Currently DXFFetcher always read files as utf-8. By spec a DXF greater than 2004 is always utf-8. Can you check that ezdxf also parse it correctly)?

https://ezdxf.readthedocs.io/en/stable/dxfinternals/fileencoding.html