fitting a complex interaction in GLM model

Machine Learning and Modeling
1

Searching online, seems like everyone is fitting a simple interaction between 2 variables in a GLM, such as:

  1. x1 * x2
  2. x1 + x2 + x1 : x2

x1 and x2 could be either type.

However, what if there is a need to fit a more complicated interaction? For example:

ifelse(x1=="A" ,1,0) +
ifelse(x1=="B",1,0) +
ifelse(x1=="C",1,0) +
ifelse(x2==0,1,0) +
pmin(x2,10000)+
ifelse(x2>10000,1,0)

What would be a correct way to fit a x1 and x2 interaction in the above equation?

Many thanks