Generally halo atom on safrol is a bad bad leaving group, however, wouldn't it be more conveniant to simply "oxymercurate" safrol to corresponding sec. alcohol, followed by mesylation, that would convert OH (bad leaving group) to one of best leaving groups, this can then be attacked by a nucleophile such as methylamine in sn2 fashion... the only problem is if not 2 different mechanisms (sn2/E2) will take place one (sn2) leading to MDMA and (E2) leading to isosafrol + H2O
, this of corse can be controlled (kinetic/thermal) controll.... i dunno, just a thought..
Never underestimate the power of retrosynthesis.