If I ever did a Tetris clone, I would have a 2D array representing the game area, one array element per square. This array would initially be all zeroes. Each block is four squares, no matter what shape it is. When a block is placed, set the array elements of the grid corresponding to the position of the squares in the block to one. Then check the rows of the array that were modified for rows that are now all ones.

I hope that made sense.