Since early applications such as the style linter and formatter operate on unpreprocessed source code, preprocessing constructs were incorporated directly into the grammar with limitations.
Pseudo-preprocessing refers to any number of strategies that perform partial or approximate preprocessing, by making assumptions, such as whether conditional branches are taken or not taken, or expanding macros to the extent where local definitions are available. The outcome of these “transformations” is a modified view of the original text that could result in covering different preprocessing code paths (blanking out not-taken branches), which could result in different lexing and parsing results. Any number of strategies could be used to overcome the limitations of preprocessing support in the parser, which in turn improves the outreach of tools like the linter and formatter.
Most of these are not yet implemented, but help is wanted.
... does not exist in this codebase yet. See https://github.com/chipsalliance/verible/issues/183.
SystemVerilog preprocessing is in many ways more complicated than C/C++ preprocessing.