Support preserved registers as identifiers

… in expressions

Rakefile

@@ -37,7 +37,7 @@ end
 directory 'build'
 
 desc 'Initial stage'
-file 'build/stage1' => ['boot/stage1.s', 'boot/common-boot.s', 'build'] do |t|
+file 'build/stage1' => ['boot/stage1.s', 'boot/common-boot.s', 'boot/tokenizer.s', 'build'] do |t|
   source = t.prerequisites.filter { |prerequisite| prerequisite.end_with? '.s' }
 
   sh CROSS_GCC, '-nostdlib', '-o', t.name, *source

boot/common-boot.s

@@ -1,5 +1,6 @@
 .global _is_alpha, _is_digit, _is_alnum, _is_upper, _is_lower
-.global _write_out, _read_file, _memcmp, _write_error, _put_char, _printi
+.global _write_out, _read_file, _write_error, _put_char, _printi
+.global _get, _memcmp
 .global _divide_by_zero_error, _exit
 
 .section .rodata
@@ -187,7 +188,9 @@ _write_out:
 # a0 - Buffer pointer.
 # a1 - Buffer size.
 #
-# Returns the result in a0.
+# Sets s1 to the buffer passed in a0.
+#
+# Returns the amount of bytes written in a0.
 .type _read_file, @function
 _read_file:
 	# Prologue.
@@ -196,9 +199,11 @@ _read_file:
 	sw s0, 0(sp)
 	addi s0, sp, 8
 
-	mv a2, a1
+	mv s1, a0
-	mv a1, a0
+
 	li a0, STDIN
+	mv a2, a1
+	mv a1, s1
 	li a7, SYS_READ
 	ecall
 
@@ -293,3 +298,11 @@ _put_char:
 	lw s0, 8(sp)
 	add sp, sp, 16
 	ret
+
+# a0 - Pointer to an array to get the first element.
+#
+# Dereferences a pointer and returns what is on the address in a0.
+.type _get, @function
+_get:
+	lw a0, (a0)
+	ret

boot/stage1.s

@@ -207,6 +207,130 @@ _build_binary_expression:
 	addi sp, sp, 32
 	ret
 
+# Parameters:
+# a0 - Identifier length.
+# a1 - Register number as character.
+.type _build_identifier_expression, @function
+_build_identifier_expression:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	sw a0, 20(sp) # Identifier length.
+	sw a1, 16(sp) # Register number as character.
+
+	li t0, 0x61636f6c # loca
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 4
+	call _memcmp
+	beqz a0, .Lbuild_identifier_expression_local
+
+	lbu a0, (s1)
+	lw t0, 20(sp) 
+	addi t0, t0, -2
+	seqz t0, t0
+	addi t1, a0, -'s'
+	seqz t1, t1
+	and t0, t0, t1
+	bnez t0, .Lbuild_identifier_expression_saved
+
+	# Global identifier.
+	lw t1, 16(sp)
+	li t0, 0x00202c00 # \0,_
+	or t0, t0, t1
+	sw t0, 12(sp)
+	li t0, 0x6120616c # la a
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 7
+	call _write_out
+
+	mv a0, s1
+	lw a1, 20(sp)
+	call _write_out
+
+	li a0, '\n'
+	call _put_char
+
+	lbu a0, (s1)
+	call _is_upper
+	beqz a0, .Lbuild_identifier_expression_end
+
+	lw t1, 16(sp)
+	li t0, 0x0a290061 # a\0)\n
+	sll t2, t1, 8
+	or t0, t0, t2
+	sw t0, 12(sp)
+	li t0, 0x28202c00 # \0, (
+	or t0, t0, t1
+	sw t0, 8(sp)
+	li t0, 0x6120776c # lw a
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 12
+	call _write_out
+
+	j .Lbuild_identifier_expression_end
+
+.Lbuild_identifier_expression_saved:
+	li t0, 0x00202c00 # \0,_
+	lw t1, 16(sp)
+	or t0, t0, t1
+	sw t0, 12(sp)
+	li t0, 0x6120766d # mv a
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 7
+	call _write_out
+
+	mv a0, s1
+	lw a1, 20(sp)
+	call _write_out
+
+	li a0, '\n'
+	call _put_char
+
+	j .Lbuild_identifier_expression_end
+
+.Lbuild_identifier_expression_local:
+	lw t1, 16(sp)
+	li t0, 0x00202c00 # \0,_
+	or t0, t0, t1
+	sw t0, 12(sp)
+	li t0, 0x6120776c # lw a
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 7
+	call _write_out
+
+	mv a0, s1
+	lw a1, 20(sp)
+	addi a0, a0, 4 # Skip the "loca" variable prefix.
+	addi a1, a1, -4 # Skip the "loca" variable prefix.
+	call _write_out
+
+	li t0, 0x29707328 # (sp)
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 4
+	call _write_out
+	li a0, '\n'
+	call _put_char
+
+	j .Lbuild_identifier_expression_end
+
+.Lbuild_identifier_expression_end:
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
 # Evalutes an expression and saves the result in a0.
 #
 # a0 - X in aX, the register number to save the result.
@@ -242,31 +366,9 @@ _build_expression:
 	li t0, '_'
 	beq a0, t0, .Lbuild_expression_call
 
-	li t0, 0x61636f6c # loca
+	lw a0, 20(sp)
-	sw t0, 8(sp)
+	lw a1, 28(sp)
-	mv a0, s1
+	call _build_identifier_expression
-	addi a1, sp, 8
-	li a2, 4
-	call _memcmp
-	beqz a0, .Lbuild_expression_identifier
-
-	# Named identifier.
-	lw t1, 28(sp)
-	li t0, 0x00202c00 # \0,_
-	or t0, t0, t1
-	sw t0, 8(sp)
-	li t0, 0x6120616c # la a
-	sw t0, 4(sp)
-	addi a0, sp, 4
-	li a1, 7
-	call _write_out
-
-	lw a0, 24(sp)
-	lw a1, 20(sp)
-	call _write_out
-
-	li a0, '\n'
-	call _put_char
 
 	j .Lbuild_expression_advance
 
@@ -313,33 +415,6 @@ _build_expression:
 
 	j .Lbuild_expression_advance
 
-.Lbuild_expression_identifier:
-	lw t1, 28(sp)
-	li t0, 0x00202c00 # \0,_
-	or t0, t0, t1
-	sw t0, 16(sp)
-	li t0, 0x6120776c # lw a
-	sw t0, 12(sp)
-	addi a0, sp, 12
-	li a1, 7
-	call _write_out
-
-	lw a0, 24(sp)
-	lw a1, 20(sp)
-	addi a0, a0, 4 # Skip the "loca" variable prefix.
-	addi a1, a1, -4 # Skip the "loca" variable prefix.
-	call _write_out
-
-	li t0, '\n'
-	sw t0, 16(sp)
-	li t0, 0x29707328 # (sp)
-	sw t0, 12(sp)
-	addi a0, sp, 12
-	li a1, 5
-	call _write_out
-
-	j .Lbuild_expression_advance
-
 .Lbuild_expression_call:
 	lw a0, 24(sp)
 	lw a1, 20(sp)
@@ -683,24 +758,6 @@ _skip_spaces:
 .Lspace_loop_end:
 	ret
 
-# Skips tabs at the line beginning.
-.type _skip_indentation, @function
-_skip_indentation:
-.Lskip_indentation_do:
-	lbu t0, (s1)
-
-	li t1, '\t'
-	beq t0, t1, .Lskip_indentation_skip
-
-	j .Lskip_indentation_end
-
-.Lskip_indentation_skip:
-	addi s1, s1, 1
-	j .Lskip_indentation_do
-
-.Lskip_indentation_end:
-	ret
-
 # Parameters:
 # a0 - Line length.
 .type _skip_comment, @function
@@ -1134,7 +1191,7 @@ _compile_procedure:
 
 	# Generate the body of the procedure.
 .Lcompile_procedure_body:
-	call _skip_indentation
+	call _skip_spaces
 	call _read_line
 	sw a0, 12(sp)
 	li t0, 0x0a646e65 # end\n
@@ -1427,23 +1484,19 @@ _compile_line:
 	li t1, '('
 	beq t0, t1, .Lcompile_line_comment
 
-	li t0, 0x0a6d6172 # ram\n
-	sw t0, 12(sp)
 	li t0, 0x676f7270 # prog
-	sw t0, 8(sp)
+	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 8
+	addi a1, sp, 12
-	li a2, 8
+	li a2, 4
 	call _memcmp
 	beqz a0, .Lcompile_line_program
 
-	li t0, 0x0a74 # t\n
-	sw t0, 12(sp)
 	li t0, 0x736e6f63 # cons
-	sw t0, 8(sp)
+	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 8
+	addi a1, sp, 12
-	li a2, 6
+	li a2, 4
 	call _memcmp
 	beqz a0, .Lcompile_line_const
 
@@ -1455,23 +1508,19 @@ _compile_line:
 	call _memcmp
 	beqz a0, .Lcompile_line_var
 
-	li t0, 0x20 # _
-	sw t0, 12(sp)
 	li t0, 0x636f7270 # proc
-	sw t0, 8(sp)
+	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 8
+	addi a1, sp, 12
-	li a2, 5
+	li a2, 4
 	call _memcmp
 	beqz a0, .Lcompile_line_procedure
 
-	li t0, 0x0a6e # n\n
-	sw t0, 12(sp)
 	li t0, 0x69676562 # begi
-	sw t0, 8(sp)
+	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 8
+	addi a1, sp, 12
-	li a2, 6
+	li a2, 4
 	call _memcmp
 	beqz a0, .Lcompile_line_begin
 
@@ -1491,13 +1540,11 @@ _compile_line:
 	call _memcmp
 	beqz a0, .Lcompile_line_identifier
 
-	li t0, 0x7472 # rt
-	sw t0, 12(sp)
 	li t0, 0x6f706d69 # impo
-	sw t0, 8(sp)
+	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 8
+	addi a1, sp, 12
-	li a2, 6
+	li a2, 4
 	call _memcmp
 	beqz a0, .Lcompile_line_import
 
@@ -1509,13 +1556,11 @@ _compile_line:
 	call _memcmp
 	beqz a0, .Lcompile_line_goto
 
-	li t0, 0x6e72 # rn
-	sw t0, 12(sp)
 	li t0, 0x75746572 # retu
-	sw t0, 8(sp)
+	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 8
+	addi a1, sp, 12
-	li a2, 6
+	li a2, 4
 	call _memcmp
 	beqz a0, .Lcompile_line_return
 
@@ -1716,7 +1761,7 @@ _compile:
 	lbu t0, (s1) # t0 = Current character.
 	beqz t0, .Lcompile_end # Exit the loop on the NUL character.
 
-	call _skip_indentation
+	call _skip_spaces
 	call _read_line
 	lw a1, 4(sp)
 	call _compile_line
@@ -1736,16 +1781,35 @@ _compile:
 	addi sp, sp, 16
 	ret
 
-# Entry point.
+.type _main, @function
-.type _start, @function
+_main:
-_start:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
 	# Read the source from the standard input.
 	la a0, source_code
 	li a1, SOURCE_BUFFER_SIZE # Buffer size.
 	call _read_file
 
 	li s2, 1
-	la s1, source_code # s1 = Source code position.
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Entry point.
+.type _start, @function
+_start:
+	call _tokenizer_initialize
+	li a1, 50
+	call _write_error
+
+	call _main
 	call _compile
 
 	# Call exit.

boot/states.txt

@@ -0,0 +1,20 @@
+- start
+digit: integer
+upper: identifier
+lower: identifier
+space: start
+invalid: error
+
+- identifier
+digit: identifier
+upper: identifier
+lower: identifier
+space: end
+invalid: end
+
+- integer:
+digit: integer
+upper: end
+lower: end
+space: end
+invalid: end

boot/tokenizer.s

@@ -0,0 +1,181 @@
+.global _tokenizer_initialize
+
+.section .rodata
+
+raw_classes:
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "space\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "digit\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "upper\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "lower\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+	.ascii "invalid\n"
+
+.section .bss
+.type class_names, @object
+.size class_names, 1024
+class_names: .zero 1024
+
+.section .data
+.type classes, @object
+.size classes, 512 # 128 characters * 4 byte.
+classes:
+
+.section .text
+
+# Initializes the classification table.
+#
+# Paramaters:
+# a0 - Raw input for the classification table.
+.type _tokenizer_classes, @function
+_tokenizer_classes:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Initializes the lookup tables.
+.type _tokenizer_initialize, @function
+_tokenizer_initialize:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+	la a0, raw_classes
+	call _tokenizer_classes
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret