Actually, Aristarchus, a contemporary of Eratosthenes (slightly earlier in fact) determined the ratio distance to the sun/distance to the moon to be about 20 (actual answer is about 390), and that the distance to the moon was about 20 Earth radii (although Archimedes credited him with an approximation closer to 80 Earth radii; the actual value is about 60). This would have validated any assumption about the sun being far enough away to generate essentially parallel rays.
Incidentally, the fact that the world was round was probably already known by Eratosthenes' time, since the Greek society was essentially a group of seafarers, and when you watch a ship go out to sea, in addition to appearing smaller as it gets further away, it also appears to sink downwards. They would doubtless have noticed this, and concluded that the Earth was probably round.