uni/inf2270/hw3/oppgave.s

	.extern	fread, fwrite

	.text

#########################
# int readbyte(FILE *f) #
#########################
	.globl	readbyte
readbyte:
	pushl	%ebp		#
	movl	%esp,%ebp	#

	subl	$4,%esp
	movl	$0,-4(%ebp)	# initialize with 0
	leal	-4(%ebp),%eax	# put address in eax
	pushl	8(%ebp)		# FILE *stream
	pushl	$1		# int nmemb
	pushl	$1		# int size
	pushl	%eax		# void *ptr
	call	fread

	cmp	$0,%eax		# if return is 0
	je	rb_e		# jump to error
	movl	-4(%ebp),%eax	# return eax
	jmp	rb_x		# exit

rb_e:	movl	$-1,%eax	# return -1
	jmp	rb_x		# exit

rb_x:	movl	%ebp,%esp	#
	popl	%ebp		#
	ret			#

##############################
# long readutf8char(FILE *f) #
##############################
	.globl	readutf8char
readutf8char:
	pushl	%ebp		#
	movl	%esp,%ebp	#

# Read the first byte into eax
	pushl	8(%ebp)		# FILE *f
	call	readbyte

# If the character is less than 0xC0,
# meaning it's just one byte, just return it.
# This also covers the case where it's -1.

	cmp	$0xC0,%eax	# if >= 0xC0,
	jge	ru8_main	# do the whole thing
	jmp	ru8_x		# else return the one byte

ru8_main:
	subl	$16,%esp	# allocate 4 ints
				# -4(%ebp): unicode character
				# -8(%ebp): current byte
				# -12(%ebp): mask
				# -16($ebp): number of bits left

	movl	$0,-4(%ebp)	# zero out the unicode character
	movl	%eax,-8(%ebp)	# store the current byte

	cmp	$0xF0,%eax	# first byte is >= 0xF0 means 4 bytes
	jge	ru8_4bytes
	cmp	$0xE0,%eax	# >= 0xE0 means 3 bytes
	jge	ru8_3bytes
	jmp	ru8_2bytes	# else, 2 bytes (as 1 byte is already handled)

ru8_4bytes:
	movl	$18,-16(%ebp)		# 18 bits to read for 4 bytes
	movl	$0b00000111,-12(%ebp)	# only read the lowest 3 bits
	jmp	ru8_loop

ru8_3bytes:
	movl	$12,-16(%ebp)		# 12 bits to read for 3 bytes
	movl	$0b00001111,-12(%ebp)	# only read the lowest 4 bits
	jmp	ru8_loop

ru8_2bytes:
	movl	$6,-16(%ebp)		# 6 bits to read for 2 bytes
	movl	$0b00011111,-12(%ebp)	# only read the lowest 5 bits
	jmp	ru8_loop

ru8_loop:
	movl	-8(%ebp),%eax	# using eax as a temporary arithmetic register
	andl	-12(%ebp),%eax	# AND with the mask
	movl	-16(%ebp),%ecx	# move number of bits left to ecx for shll
	shll	%cl,%eax	# shift ecx bits to the left
	orl	%eax,-4(%ebp)	# OR the unicode character with eax

	cmp	$0,-16(%ebp)	# end if -8(%ebp) is 0, as that means there's
	je	ru8_done	# no more bytes to read

	movl	$63,-12(%ebp)	# subsequent iterations will use 00111111 as mask
	subl	$6,-16(%ebp)	# we read 6 bits at a time

	# read the next byte into -8(%ebp)
	pushl	8(%ebp)		# f
	call	readbyte
	movl	%eax,-8(%ebp)	# move byte to memory

	jmp	ru8_loop	# since we haven't exited yet, just loop

ru8_done:
	movl	-4(%ebp),%eax	# move the unicode character to eax
				# then exit

ru8_x:	movl	%ebp,%esp	#
	popl	%ebp		#
	ret			#

############################################
# void writebyte(FILE *f, unsigned char b) #
############################################
	.globl	writebyte
writebyte:
	pushl	%ebp		#
	movl	%esp,%ebp	#

	leal	12(%esp),%eax	# put address in eax
	pushl	8(%esp)		# FILE *stream
	pushl	$1		# nmemb
	pushl	$1		# size
	pushl	%eax		# void *ptr
	call	fwrite

	movl	%ebp,%esp	#
	popl	%ebp		#
	ret			#

################################################
# void writeutf8char(FILE *f, unsigned long u) #
################################################
	.globl	writeutf8char
writeutf8char:
	pushl	%ebp		#
	movl	%esp,%ebp	#

	movl	12(%ebp),%ecx	# put u in ecx

	cmp	$0x0080,%ecx	# if the number doesn't fit in one byte,
	jge	wu8_main	# skip the early return version

	pushl	%ecx		# unsigned char u
	pushl	8(%ebp)		# FILE *f
	call	writebyte
	jmp	wu8_x		# exit

# wu8_main is responsible for jumping to the correct section depending on
# the size of the unicode character.
wu8_main:
	subl	$12,%esp	# stack allocate 3 ints

	cmp	$0x10000,%ecx	# numbers >= 0x10000 need 4 bytes
	jge	wu8_4bytes
	cmp	$0x0800,%ecx	# numbers >= 0x0800 need 3 bytes
	jge	wu8_3bytes
	jmp	wu8_2bytes	# numbers >= 0x0080 need 2 bytes
				# we already dealt with those which need 1 byte

# This section uses 3 stack allocated ints.
# They're here mainly for this expression:
# %-4(%ebp) | ((u >> %-8(%ebp)) & %-12(%ebp))

wu8_4bytes:
	movl	$18,-8(%ebp)		# 4 bytes means there are 18 bits to write
	movl	$0b11110000,-4(%ebp)	# the first 4 bits should be 1
	movl	$0b00000111,-12(%ebp)	# only the last 3 bits is data
	jmp	wu8_loop

wu8_3bytes:
	movl	$12,-8(%ebp)		# 3 bytes means there are 12 bits to write
	movl	$0b11100000,-4(%ebp)	# the first 3 bits should be 1
	movl	$0b00001111,-12(%ebp)	# only the last 4 bits is data
	jmp	wu8_loop

wu8_2bytes:
	movl	$6,-8(%ebp)		# 2 bytes means there are 6 bits to write
	movl	$0b11000000,-4(%ebp)	# the first 2 bits should be 1
	movl	$0b00011111,-12(%ebp)	# only the last 3 bits is data
	jmp	wu8_loop

wu8_loop:
	movl	12(%ebp),%eax	# using eax as a temporary arithmetic register
	movl	-8(%ebp),%ecx	# move -8(%ebp) to %ecx for use with shrl
	shrl	%cl,%eax	# shift ecx (or rather cl) bits to the right
	andl	-12(%ebp),%eax	# AND with -4(%ebp) to remove the upper two bits
	orl	-4(%ebp),%eax	# OR with ecx to set the leftmost bit

	pushl	%eax		# unsigned char u
	pushl	8(%ebp)		# FILE *f
	call	writebyte	# call
	addl	$8,%esp		# restore esp

	cmp	$0,-8(%ebp)	# end if -8(%ebp) is 0, as that means there's
	je	wu8_x		# no bytes left to write

	movl	$63,-12(%ebp)	# subsequent iterations will AND with 00111111
	movl	$128,-4(%ebp)	# subsequent iterations will OR with 10000000

	subl	$6,-8(%ebp)	# remove 6 from -8(%ebp), as we write 6 bits at a time

	jmp	wu8_loop	# since we haven't exited yet, just loop

wu8_x:	movl	%ebp,%esp	#
	popl	%ebp		#
	ret			#