Yeah, this took me a while. Basically, the entire file represents a sort of assembly language for a function. You need to evaluate that entire function for each pixel in the output image.
Each line starts with a line number (with a leading underscore). This is followed by an instruction, and zero, one, or two arguments. The arguments can be either floating point constants or integers (with leading underscores) which represent the results of those corresponding lines earlier in the function.
The x and y coordinates of each pixel are loaded in the assembly language using the var-x and var-y instructions.
Each line starts with a line number (with a leading underscore). This is followed by an instruction, and zero, one, or two arguments. The arguments can be either floating point constants or integers (with leading underscores) which represent the results of those corresponding lines earlier in the function.
The x and y coordinates of each pixel are loaded in the assembly language using the var-x and var-y instructions.