I don't have the exact lines anymore, but it wasn't too fancy (it mainly consisted of grep and sed). What I had were two xml-files with two language descriptions. My lines transformed them into yacc&lex-files building a compiler from the one language to the other (if you must know: the input language was the output of an interactive medical experimentation program, the output was the input for the compiler of a laboratory robot. Sounds stupid? It was; but the robot manufacturer didn't give us the oh-so-precious hardware specifications and didn't manage to build something like a java machine on it; instead they provided a crappy compiler for a even crappier input language. Having seen how they work and whom they equip, I'm somewhat surprised that there actually is progress in medical research...).
Privately about 14,000 lines of Turbo Pascal / Assembler which unfortunately I lost when multiple floppy disks died simultaneously during a file transfer due to a bad floppy disk drive :(
Professionally I work on a project that easily has 1,000,000 lines of C/C++ code, of which I own probably 75,000.