WronGGG! Any way you put it - that is wronGGG!!!
The fundamental problem is signal-to-noise ratio in spectra of large complex molecules that comprise a very tiny fraction of all species in highly complex mixtures - like, uh, plant matter or trace analytes in a human breath.
Any doctoral or beyond level (from a top 25 university dept), honest, sophisticated instrumental analytical chemist know this intractable problem. It is a fundamental limitation that cannot be worked around.
The reason we can test for alcohol is solely because it is so ABUNDANT in the breath, urine, and blood of a drinker - 0.001 or higher on a percentage by mole of the sample - AND because ethanol is a simple molecule with a characteristic bond stretch energy which is useful because it is so abundant in the sample for spectroscopy. THC is many, many orders of magnitude less abundant and is far more complex of a molecule with no easy discrimination spectroscopic peaks/patterns that can be pulled out as a signal from an incredibly messy and noisy spectral background.
It has nothing to do with pot smoking.