parser.inc

一个用纯汇编 写的操作系统 源代码 是用 TASM 编译器写的

; flat assembler core
; Copyright (c) 1999-2006, Tomasz Grysztar.
; All rights reserved.

parser:
	mov	eax,[memory_end]
	mov	[labels_list],eax
	mov	eax,[additional_memory]
	mov	[free_additional_memory],eax
	xor	eax,eax
	mov	[current_locals_prefix],eax
	mov	[anonymous_reverse],eax
	mov	[anonymous_forward],eax
	mov	[hash_tree],eax
	mov	[blocks_stack],eax
	mov	esi,[memory_start]
	mov	edi,[source_start]
      parser_loop:
	mov	[current_line],esi
	lea	eax,[edi+100h]
	cmp	eax,[labels_list]
	jae	out_of_memory
	
	cmp	byte [esi+16],0
	je	empty_line
	
	mov	al,0Fh
	stos	byte [edi]
	mov	eax,esi
	stos	dword [edi]
	add	esi,16
      parse_line:
	cmp	byte [esi],1Ah
	jne	empty_instruction
	push	edi
	add	esi,2
	movzx	ecx,byte [esi-1]
	cmp	byte [esi+ecx],':'
	je	simple_label
	cmp	byte [esi+ecx],'='
	je	constant_label
	
	call	get_instruction
	jnc	main_instruction_identified

	cmp	byte [esi+ecx],1Ah
	jne	no_data_label
	push	esi ecx
	lea	esi,[esi+ecx+2]
	movzx	ecx,byte [esi-1]
	call	get_data_directive
	jnc	data_label
	pop	ecx esi
      no_data_label:
	call	get_data_directive
	jnc	main_instruction_identified
	pop	edi
	sub	esi,2
	xor	bx,bx
	call	parse_line_contents
	jmp	parse_next_line
      simple_label:
	pop	edi
	call	identify_label
	mov	byte [edi],2
	inc	edi
	stos	dword [edi]
	inc	esi
	xor	al,al
	stos	byte [edi]
	jmp	parse_line
      constant_label:
	pop	edi
	call	get_label_id
	mov	byte [edi],3
	inc	edi
	stos	dword [edi]
	xor	al,al
	stos	byte [edi]
	inc	esi
	xor	bx,bx
	call	parse_line_contents
	jmp	parse_next_line
      data_label:
	pop	ecx edx
	pop	edi
	push	eax ebx esi
	mov	esi,edx
	movzx	ecx,byte [esi-1]
	call	identify_label
	mov	byte [edi],2
	inc	edi
	stos	dword [edi]
	pop	esi ebx eax
	stos	byte [edi]
	push	edi
	
	
      main_instruction_identified:
	pop	edi
	mov	dl,al
	mov	al,1
	stos	byte [edi]
	mov	ax,bx
	stos	word [edi]
	mov	al,dl
	stos	byte [edi]
	
	cmp	bx,if_directive-assembler
	je	parse_block

	cmp	bx,repeat_directive-assembler
	je	parse_block

	cmp	bx,while_directive-assembler
	je	parse_block

	cmp	bx,end_directive-assembler
	je	parse_end_directive

	cmp	bx,else_directive-assembler
	je	parse_else

      common_parse:
	call	parse_line_contents
	jmp	parse_next_line

      empty_instruction:
	lods	byte [esi]
	or	al,al
	jz	parse_next_line
	cmp	al,':'
	je	invalid_name
	cmp	al,3Bh
	je	skip_preprocessed_symbol
	dec	esi
	call	parse_argument
	jmp	parse_next_line

      skip_preprocessed_symbol:
	lods	byte [esi]
	movzx	eax,al
	add	esi,eax

      skip_next:
	lods	byte [esi]
	or	al,al
	jz	parse_next_line
	cmp	al,1Ah
	je	skip_preprocessed_symbol
	cmp	al,3Bh
	je	skip_preprocessed_symbol
	cmp	al,22h
	je	skip_preprocessed_string
	jmp	skip_next

      skip_preprocessed_string:
	lods	dword [esi]
	add	esi,eax
	jmp	skip_next

      empty_line:
	add	esi,17

      parse_next_line:
	cmp	esi,[source_start]
	jb	parser_loop

      source_parsed:
	cmp	[blocks_stack],0
	je	blocks_stack_ok
	pop	eax
	pop	[current_line]
	jmp	missing_end_directive

      blocks_stack_ok:
	xor	al,al
	stos	byte [edi]
	mov	eax,[error_line]
	mov	[current_line],eax
	cmp	[anonymous_forward],0
	jne	invalid_value
	add	edi,0Fh
	and	edi,not 0Fh
	mov	[code_start],edi
	ret

      parse_block:
	mov	eax,esp
	sub	eax,100h
	jc	stack_overflow
	cmp	eax,[stack_limit]
	jb	stack_overflow
	push	[current_line]
	mov	ax,bx
	shl	eax,16
	push	eax
	inc	[blocks_stack]
	cmp	bx,if_directive-assembler
	je	parse_if
	cmp	bx,while_directive-assembler
	je	parse_while
	call	parse_line_contents
	jmp	parse_next_line

      parse_end_directive:
	cmp	byte [esi],1Ah
	jne	common_parse
	push	edi
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_instruction
	pop	edi
	jnc	parse_end_block
	sub	esi,2
	jmp	common_parse

      parse_end_block:
	mov	dl,al
	mov	al,1
	stos	byte [edi]
	mov	ax,bx
	stos	word [edi]
	mov	al,dl
	stos	byte [edi]
	lods	byte [esi]
	or	al,al
	jnz	extra_characters_on_line
	cmp	bx,if_directive-assembler
	je	close_parsing_block
	cmp	bx,repeat_directive-assembler
	je	close_parsing_block
	cmp	bx,while_directive-assembler
	je	close_parsing_block
	jmp	parse_next_line
      close_parsing_block:
	cmp	[blocks_stack],0
	je	unexpected_instruction
	cmp	bx,[esp+2]
	jne	unexpected_instruction
	dec	[blocks_stack]
	pop	eax edx
	cmp	bx,if_directive-assembler
	jne	parse_next_line
	test	al,1100b
	jz	parse_next_line
	test	al,10000b
	jnz	parse_next_line
	sub	edi,8
	jmp	parse_next_line
      parse_if:
	push	edi
	call	parse_line_contents
	xor	al,al
	stos	byte [edi]
	xchg	esi,[esp]
	mov	edi,esi
	call	preevaluate_logical_expression
	pop	esi
	cmp	al,'0'
	je	parse_false_condition_block
	cmp	al,'1'
	je	parse_true_condition_block
	or	byte [esp],10000b
	jmp	parse_next_line
      parse_while:
	push	edi
	call	parse_line_contents
	xor	al,al
	stos	byte [edi]
	xchg	esi,[esp]
	mov	edi,esi
	call	preevaluate_logical_expression
	pop	esi
	cmp	al,'0'
	je	parse_false_condition_block
	cmp	al,'1'
	jne	parse_next_line
	stos	byte [edi]
	jmp	parse_next_line
      parse_false_condition_block:
	or	byte [esp],1
	sub	edi,4
	jmp	skip_parsing
      parse_true_condition_block:
	or	byte [esp],100b
	sub	edi,4
	jmp	parse_next_line
      parse_else:
	cmp	[blocks_stack],0
	je	unexpected_instruction
	cmp	word [esp+2],if_directive-assembler
	jne	unexpected_instruction
	lods	byte [esi]
	or	al,al
	jz	parse_pure_else
	cmp	al,1Ah
	jne	extra_characters_on_line
	push	edi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_instruction
	jc	extra_characters_on_line
	pop	edi
	cmp	bx,if_directive-assembler
	jne	extra_characters_on_line
	test	byte [esp],100b
	jnz	skip_true_condition_else
	mov	dl,al
	mov	al,1
	stos	byte [edi]
	mov	ax,bx
	stos	word [edi]
	mov	al,dl
	stos	byte [edi]
	jmp	parse_if
      skip_true_condition_else:
	sub	edi,4
	or	byte [esp],1
	jmp	skip_parsing_contents
      parse_pure_else:
	bts	dword [esp],1
	jc	unexpected_instruction
	test	byte [esp],100b
	jz	parse_next_line
	sub	edi,4
	or	byte [esp],1
	jmp	skip_parsing
      skip_parsing:
	cmp	esi,[source_start]
	jae	source_parsed
	mov	[current_line],esi
	add	esi,16
      skip_parsing_line:
	cmp	byte [esi],1Ah
	jne	skip_parsing_contents
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	cmp	byte [esi+ecx],':'
	je	skip_parsing_label
	push	edi
	call	get_instruction
	pop	edi
	jnc	skip_parsing_instruction
	add	esi,ecx
	jmp	skip_parsing_contents
      skip_parsing_label:
	lea	esi,[esi+ecx+1]
	jmp	skip_parsing_line
      skip_parsing_instruction:
	cmp	bx,if_directive-assembler
	je	skip_parsing_block
	cmp	bx,repeat_directive-assembler
	je	skip_parsing_block
	cmp	bx,while_directive-assembler
	je	skip_parsing_block
	cmp	bx,end_directive-assembler
	je	skip_parsing_end_directive
	cmp	bx,else_directive-assembler
	je	skip_parsing_else
      skip_parsing_contents:
	lods	byte [esi]
	or	al,al
	jz	skip_parsing
	cmp	al,1Ah
	je	skip_parsing_symbol
	cmp	al,3Bh
	je	skip_parsing_symbol
	cmp	al,22h
	je	skip_parsing_string
	jmp	skip_parsing_contents
      skip_parsing_symbol:
	lods	byte [esi]
	movzx	eax,al
	add	esi,eax
	jmp	skip_parsing_contents
      skip_parsing_string:
	lods	dword [esi]
	add	esi,eax
	jmp	skip_parsing_contents
      skip_parsing_block:
	mov	eax,esp
	sub	eax,100h
	jc	stack_overflow
	cmp	eax,[stack_limit]
	jb	stack_overflow
	push	[current_line]
	mov	ax,bx
	shl	eax,16
	push	eax
	inc	[blocks_stack]
	jmp	skip_parsing_contents
      skip_parsing_end_directive:
	cmp	byte [esi],1Ah
	jne	skip_parsing_contents
	push	edi
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_instruction
	pop	edi
	jnc	skip_parsing_end_block
	add	esi,ecx
	jmp	skip_parsing_contents
      skip_parsing_end_block:
	lods	byte [esi]
	or	al,al
	jnz	extra_characters_on_line
	cmp	bx,if_directive-assembler
	je	close_skip_parsing_block
	cmp	bx,repeat_directive-assembler
	je	close_skip_parsing_block
	cmp	bx,while_directive-assembler
	je	close_skip_parsing_block
	jmp	skip_parsing
      close_skip_parsing_block:
	cmp	[blocks_stack],0
	je	unexpected_instruction
	cmp	bx,[esp+2]
	jne	unexpected_instruction
	dec	[blocks_stack]
	pop	eax edx
	test	al,1
	jz	skip_parsing
	cmp	bx,if_directive-assembler
	jne	parse_next_line
	test	al,10000b
	jz	parse_next_line
	mov	al,0Fh
	stos	byte [edi]
	mov	eax,[current_line]
	stos	dword [edi]
	mov	eax,1 + (end_directive-assembler) shl 8
	stos	dword [edi]
	mov	eax,1 + (if_directive-assembler) shl 8
	stos	dword [edi]
	jmp	parse_next_line
      skip_parsing_else:
	cmp	[blocks_stack],0
	je	unexpected_instruction
	cmp	word [esp+2],if_directive-assembler
	jne	unexpected_instruction
	lods	byte [esi]
	or	al,al
	jz	skip_parsing_pure_else
	cmp	al,1Ah
	jne	extra_characters_on_line
	push	edi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_instruction
	jc	extra_characters_on_line
	pop	edi
	cmp	bx,if_directive-assembler
	jne	extra_characters_on_line
	mov	al,[esp]
	test	al,1
	jz	skip_parsing_contents
	test	al,100b
	jnz	skip_parsing_contents
	test	al,10000b
	jnz	parse_else_if
	xor	al,al
	mov	[esp],al
	mov	al,0Fh
	stos	byte [edi]
	mov	eax,[current_line]
	stos	dword [edi]
      parse_else_if:
	mov	eax,1 + (if_directive-assembler) shl 8
	stos	dword [edi]
	jmp	parse_if
      skip_parsing_pure_else:
	bts	dword [esp],1
	jc	unexpected_instruction
	mov	al,[esp]
	test	al,1
	jz	skip_parsing
	test	al,100b
	jnz	skip_parsing
	and	al,not 1
	or	al,1000b
	mov	[esp],al
	jmp	parse_next_line

parse_line_contents:
	mov	[parenthesis_stack],0
      parse_instruction_arguments:
	cmp	bx,prefix_instruction-assembler
	je	allow_embedded_instruction
	cmp	bx,times_directive-assembler
	je	parse_times_directive
	cmp	bx,end_directive-assembler
	je	allow_embedded_instruction
	cmp	bx,label_directive-assembler
	je	parse_label_directive
	cmp	bx,segment_directive-assembler
	je	parse_label_directive
	cmp	bx,load_directive-assembler
	je	parse_load_directive
	cmp	bx,extrn_directive-assembler
	je	parse_extrn_directive
	cmp	bx,public_directive-assembler
	je	parse_public_directive
      parse_argument:
	lea	eax,[edi+100h]
	cmp	eax,[labels_list]
	jae	out_of_memory
	lods	byte [esi]
	cmp	al,':'
	je	instruction_separator
	cmp	al,','
	je	separator
	cmp	al,'='
	je	separator
	cmp	al,'|'
	je	separator
	cmp	al,'&'
	je	separator
	cmp	al,'~'
	je	separator
	cmp	al,'>'
	je	greater
	cmp	al,'<'
	je	less
	cmp	al,')'
	je	close_parenthesis
	or	al,al
	jz	contents_parsed
	cmp	al,'['
	je	address_argument
	cmp	al,']'
	je	separator
	cmp	al,'{'
	je	unallowed_character
	cmp	al,'}'
	je	unallowed_character
	cmp	al,'#'
	je	unallowed_character
	cmp	al,'`'
	je	unallowed_character
	dec	esi
	cmp	al,1Ah
	jne	expression_argument
	push	edi
	mov	edi,directive_operators
	call	get_operator
	or	al,al
	jnz	operator_argument
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_symbol
	jnc	symbol_argument
	cmp	ecx,1
	jne	check_argument
	cmp	byte [esi],'?'
	jne	check_argument
	pop	edi
	movs	byte [edi],[esi]
	jmp	argument_parsed
      symbol_argument:
	pop	edi
	stos	word [edi]
	jmp	argument_parsed
      operator_argument:
	pop	edi
	cmp	al,85h
	je	ptr_argument
	stos	byte [edi]
	cmp	al,80h
	je	forced_expression
	cmp	al,81h
	je	forced_parenthesis
	cmp	al,82h
	je	parse_from_operator
	cmp	al,89h
	je	parse_label_operator
	jmp	argument_parsed
      allow_embedded_instruction:
	cmp	byte [esi],1Ah
	jne	parse_argument
	push	edi
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_instruction
	jnc	embedded_instruction
	call	get_data_directive
	jnc	embedded_instruction
	pop	edi
	sub	esi,2
	jmp	parse_argument
      embedded_instruction:
	pop	edi
	mov	dl,al
	mov	al,1
	stos	byte [edi]
	mov	ax,bx
	stos	word [edi]
	mov	al,dl
	stos	byte [edi]
	jmp	parse_instruction_arguments
      parse_times_directive:
	mov	al,'('
	stos	byte [edi]
	call	convert_expression
	mov	al,')'
	stos	byte [edi]
	cmp	byte [esi],':'
	jne	allow_embedded_instruction
	movs	byte [edi],[esi]
	jmp	allow_embedded_instruction
      parse_label_directive:
	cmp	byte [esi],1Ah
	jne	argument_parsed
	push	esi
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	call	identify_label
	pop	ebx
	cmp	eax,0Fh
	je	non_label_identified
	mov	byte [edi],2
	inc	edi
	stos	dword [edi]
	xor	al,al
	stos	byte [edi]
	jmp	argument_parsed
      non_label_identified:
	mov	esi,ebx
	jmp	argument_parsed
      parse_load_directive:
	cmp	byte [esi],1Ah
	jne	argument_parsed
	push	esi
	inc	esi
	movzx	ecx,byte [esi]
	inc	esi
	call	get_label_id
	pop	ebx
	cmp	eax,0Fh
	je	non_label_identified
	mov	byte [edi],2
	inc	edi
	stos	dword [edi]
	xor	al,al
	stos	byte [edi]
	jmp	argument_parsed
      parse_public_directive:
	cmp	byte [esi],1Ah
	jne	parse_argument
	inc	esi
	push	esi
	movzx	ecx,byte [esi]
	inc	esi
	mov	al,2
	stos	byte [edi]
	call	get_label_id
	stos	dword [edi]
	mov	ax,8600h
	stos	word [edi]
	pop	ebx
	push	ebx esi edi
	mov	edi,directive_operators
	call	get_operator
	pop	edi edx ebx
	cmp	al,86h
	je	argument_parsed
	mov	esi,edx
	xchg	esi,ebx
	movzx	ecx,byte [esi]
	inc	esi
	mov	ax,'('
	stos	word [edi]
	mov	eax,ecx
	stos	dword [edi]
	rep	movs byte [edi],[esi]
	xor	al,al
	stos	byte [edi]
	xchg	esi,ebx
	jmp	argument_parsed
      parse_extrn_directive: