Chunking algorithm for files other than markdown

[sh-l]

If I understand https://docs.vrtmrz.net/LiveSync/hintandtrivia/Data+structure+of+LiveSync correctly, the plugin uses a chunking and diffing algorithm to reduce bandwidth consumption.

But the documentation mostly mentions .md files. I'm wondering to know whether the same algorithm is applied to files other than .md, and if so, how it splits such files into chunks.

My main use case would be .pdf files, and from Obsidian recently improving its PDF viewer and such, I guess PDF files are also a first-class citizen of Obsidian. I often add/modify/delete annotations to PDF files via third-party apps a lot, and these are very small operations that leaves the most of the file content unchanged, so it might be a good target for optimization. Also, internal data structure of PDF is nicely split into pages, and pages are splitted from annotations and so on, so I guess PDF files can be chunked by its page boundaries.

[vrtmrz]

Sorry for being late! And thank you for reading that document!
At processing binary files, Self-hosted LiveSync just splits the file for a specific length (after base64-encoded) and uses them for the chunk. Therefore, If the modification of the binary file is at the end of the document, the cost should be low.

However, In several formats, I have compared the PDF before and after editing and found that some references were often being modified. This would make comparisons difficult.

Nevertheless, I think that your use case might be common. Could I ask what PDF format I should use as a reference?