Need to prevent character autosubstitution in URIs

[opoudjis]

http://www.example.com/...text[] is auto-substituting to a smart ellipse (which is non-ascii). Need to prevent all such substitutions in URIs.

[opoudjis]

Even worse:

https://isotc.iso.org/livelink/livelink/fetch/2000/2122/15507012/20094274/ISO_Code_of_Conduct_-_Complaints_Handling_Guidance_Process_2020.pdf?nodeid=21068087&vernum=-2

is ending up as

https://isotc.iso.org/livelink/livelink/fetch/2000/2122/15507012/20094274/ISO_Code_of_Conduct_-<em>Complaints_Handling_Guidance_Process_2020.pdf?nodeid=21068087&vernum=-2

[opoudjis]

This is an Asciidoctor issue, and has been for a decade: asciidoctor/asciidoctor#61

https://docs.asciidoctor.org/asciidoc/latest/macros/complex-urls/ are the recommended workarounds. We are not realising the workarounds are even necessary, because the text is already processed with substitutions by the time the parser reveals the link text.

I'm going to preprocess the Asciidoctor text instead, disabling substitutions on identifiable links in the markup.

[opoudjis]

I've implemented the preprocessing, which realises the Asciidoctor workaround. I now need to restrict its application so it is not done wherever macros are not expanded (passthrough, sourcecode, stem, document header, etc.)

[opoudjis]

Also need to prevent link manipulation in attributes of blocks.