There is no Celtic scripture and no Celtic mythographer. Every route to this material passes through someone else's hands, and each route distorts in its own direction.
The first is Greek and Roman ethnography, above all Caesar, who describes the gods of Gaul by giving them Roman names and arranging them by function. That passage has shaped every later account, and it tells us at least as much about Roman habits of classification as about Gaulish belief.
The second is epigraphy: dedications carved in the Roman provinces, many of them pairing a native name with a Roman one. These are contemporary and they are evidence of real cult. They are also the product of a period in which Celtic religion was already being reorganised under Roman rule, and they preserve names far more reliably than meanings.
The third is the medieval Irish and Welsh manuscripts, written by Christian scholars centuries after conversion. The Irish material in particular is vast and internally structured, and it is also frankly literary, with the old gods recast as ancient peoples in a pseudo-historical scheme designed to fit Ireland into biblical chronology.
The druids, who might have resolved much of this, taught orally and by policy did not write their doctrine down. Their absence from the record is not an accident of preservation but a decision they made, and it means the tradition's own account of itself was never committed to a surface that could survive.