Strict and verbose Hack assembler written in POSIX-compliant AWK
Hack is an educational computer architecture designed for the Nand2Tetris courses. The assembly language for the Hack architecture is described in chapter 6 of The Elements of Computing Systems, the book accompanying the courses.
Hawk will assemble the contents of the first file specified on the command line:
$ ./hawk.awk Prog.asm
Each invocation of Hawk can assemble only one assembly.
Hawk processes an assembly in two passes. This enables it to expand label symbols prior to their definition, as described in the book. Unfortunately, this means that if Hawk is to read assembly from standard input, the assembly must be passed twice:
$ { cat Prog.asm; cat Prog.asm } | ./hawk.awk
Note that cat (or whatever program you're trying to pipe into Hawk) must be
invoked twice (i.e. cat Prog.asm Prog.asm won't do) for Hawk to be able to
tell the two passes apart (with an EOF). Therefore, passing assembly through
standard input should be avoided. Instead, the output of the command you're
trying to pipe into Hawk should be redirected to a file that can be passed to
Hawk on the command line:
$ cat Prog.asm > Prog.tmp && ./hawk.awk Prog.tmp
(cat Prog.asm in the above examples is just a placeholder for whatever
you're trying to pipe into Hawk. cat should never be used this way.)
Due to lack of bitwise operators in POSIX AWK, Hawk outputs instructions as ASCII encoded hexadecimals, rather than the ASCII encoded binary format described by the book. The format from the book can be achieved by piping the output into bc:
$ { echo "ibase=F;obase=2"; ./hawk.awk Prog.asm } | bc | awk '{printf("%016d\n",$0)}'
The last awk command can also be modified to split the output every 4 bits for readability (but this is incompatible with the format described in the book):
$ { echo "ibase=F;obase=2"; ./hawk.awk Prog.asm } | bc | awk '{$0=sprintf("%016d",$0); gsub(".{4}","& ")}'