algorithms - How CDMA receiver extract it's corresponding data from the receiving modulated & superposition-ed signal?

I have difficulty understanding CDMA.

I have read all those PN/walsh code things, and understand orthogonality very well.

But none of the resources I found demonstrate how a receiver end extract data from modulated & superposition-ed signal.

Most of them just stop at the digital encoding & decoding part and skip the modulation part.

But in real implementation signals of different walsh code are modulated then superposition-ed.

I cannot figure out how it is done.

======================================================

P.S. I am confused that does the sender add the signals before or after modulation?

If before, take BPSK as an example, there is a chance there would be >2 different values in the added result vector, which can no longer be encoded in BPSK.

If after, then the signals are simply super-positioned, the receiver would often see both 0&1, with different amplitudes, at the same time, i.e. Acos(wt)+Bcos(wt+pi), which seems a bit hard to decode.