feat: support header_type UInt32

[henry2004y]

This PR addresses two main issues related to parsing VTK files, improving compatibility with files generated by third-party tools like ParaView/PyVista (#78):

1. Support for UInt32 Header Types: Previously, the dataset header type was hardcoded to UInt64. This PR updates the VTKFile parser to dynamically read the header_type XML attribute and properly support both UInt32 and UInt64 headers across the package.

2. Support for Multi-Block Zlib Compressed Data: Fixes a ZlibError: incorrect header check (code: -3) encountered when reading appended or binary data that has been compressed into multiple blocks (a common behavior in ParaView/PyVista output). The previous implementation assumed a hardcoded compression header size of 4 integers, which is only valid for single-block compression (such as the default used by WriteVTK.jl). The parsing logic in get_data has been updated to reliably read num_blocks from the start of the header, dynamically calculate the correct header length (3 + num_blocks integers), and accurately define the byte range for the Zlib decompression step.

Changes Made

• src/ReadVTK.jl:
  • Added a header_type::DataType field to the VTKFile struct.
  • Updated the VTKFile constructor to parse the header_type (validating it as "UInt32" or "UInt64").
  • Updated get_data for FormatBinary and FormatAppended arrays to calculate offsets dynamically using header_len = 3 + num_blocks instead of a hardcoded length of 4.
  • Verified that ZLib decompressor gracefully handles the multi-block payload natively once the byte range is correctly identified.
• Documentation: Updated README.md and docs/src/index.md to remove the caveat stating that header_type must be UInt64.

Note that I have not added new test for this feature since I don't know how to add a proper minimal test. Feel free to come up with a suitable test in the dataset.

[ranocha]

Thanks a lot for your contribution!

Note that I have not added new test for this feature since I don't know how to add a proper minimal test. Feel free to come up with a suitable test in the dataset.

We use the repository https://github.com/JuliaVTK/ReadVTK_examples to store example files. Please

• create some example files locally that require the new features you added and test whether ReadVTK.jl can read them correctly with your changes
• create a PR to https://github.com/JuliaVTK/ReadVTK_examples with your example files
• update the commit hash in

  ReadVTK.jl/test/runtests.jl

  Line 9 in e4f0366

  TEST_EXAMPLES_COMMIT = "1f961a7ad38e76a8d552b8392d8cd661ab9f79b4"

  in this PR
• add tests in this PR where you read the example files you added

[henry2004y]

After merging JuliaVTK/ReadVTK_examples#3, I will push the new test here.

[ranocha]

I merged your PR adding the new example file.

[ranocha]

Thanks a lot! I will give @sloede a chance to review this PR as well.

[sloede]

Very nice PR, thanks a lot for this feature! I have only two minor remarks, after which this PR is ready to merge.

[sloede]

Do you have a reference where this information can be found that you can put here?

[henry2004y]

Not really. It is found via experiments and I do not really guarantee it is universal.

[sloede]

Does this work even if the data is compressed in multiple blocks? That is, it is not necessary to decompress block by block?

[henry2004y]

Yes. I have tried it on one sample data with 19 blocks, and it also works. The sample test file I added has only 1 block though.

[ranocha]

Thanks a lot! I will merge this now and release the new version. It would be great if you could create a second PR (to the examples repository and afterward to this repository) with another file having multiple blocks to ensure that this feature continues to work in the future.