I am not sure what you meant by "DIRECTLY" measure the frequency. One way would be to use a diffraction grating and using a collimated beam of the light. Infact you can use other interferometers for this purpose too.
If by 'directly', you mean coherent detection (instantaneous amplitude and phase), then direct measurements are possible using fast reference oscillators and heterodyne detectors for EM light up to the GHz range or so (millimeter waves). That may have been extended to the THz regime recently, but I have no reference for that.
It's possible to perform heterodyne measurements of visible light using an optical frequency standard as well, and I suppose the recent generation of an optical frequency comb could be used as an alternative, but it's not clear by your question what you mean by 'direct' measurement (i.e. counting cycles?)
Light sources never have a single frequency, they are always a mixture of different frequencies, which can be analyzed by a spectrometer for example.
Do you perhaps mean by 'direct measuring' that you take some hand-held device, point it at a light source and it spits out the light's spectrum? I don't know of any device that does this but I can't see why it shouldn't exist.