This is a variation on run length encoding.
Note that find(diff(A)) returns the index of the end of each phase (where a phase is a run of zeros or ones); also note that the last element in A will always be the end of a phase, as well as the element '0' (i.e. the element before A(1)).
Taking the diff of these indices tells you the length of each phase. Putting it all together:
rl = diff([0, find(diff(A)), numel(A)]); % run lengths
B = rl(1:2:end);
C = rl(2:2:end);
Note that whether B corresponds to 1 or 0 depends on A(1); you could either write an if statement to assign B and C accordingly, or do something more obfuscated to assign them automagically.
Please accept this answer if it helps, or let me know in the comments if I've missed something.
