Add semicolons separating the statements

Rakefile

@@ -66,36 +66,3 @@ end
 file 'build/stage2b' => ['build/stage2b.s', 'boot/common-boot.s'] do |t|
   sh CROSS_GCC, '-nostdlib', '-o', t.name, *t.prerequisites
 end
-
-desc 'Print remaining lines to rewrite'
-task :statistics do
-  def is_false_positive(word)
-    word.start_with?('(*') ||
-      word.start_with?('*)') ||
-      ('A'..'Z').include?(word[0]) ||
-      /^[[:alpha:]][[:digit:]]$/.match(word) ||
-      ['end', 'if'].include?(word)
-  end
-
-  lines = File.read('boot/stage2.elna')
-    .split("\n")
-    .select { |line| line.start_with? "\t" }
-    .map { |line| line.delete_prefix("\t").split(' ') }
-    .reject { |words| is_false_positive(words.first) }
-    .group_by do |words|
-      if words.first.length < 5
-        case words.first
-        when 'goto'
-          'Statements'
-        else
-          words.first
-        end
-      else
-        'Statements'
-      end
-    end
-
-  lines.each do |key, value|
-    puts "#{key}: #{value.count}"
-  end
-end

boot/common-boot.s

@@ -1,6 +1,6 @@
 .global _is_alpha, _is_digit, _is_alnum, _is_upper, _is_lower
 .global _write_out, _read_file, _write_error, _put_char, _printi
-.global _get, _memcmp
+.global _get, _memcmp, _memchr, _memmem, _memcpy
 .global _divide_by_zero_error, _exit
 
 .section .rodata
@@ -16,7 +16,8 @@ new_line: .ascii "\n"
 
 .section .text
 
-# Write the current token to stderr.
+# Write the current token to stderr. Ends the output with a newline.
+#
 # a0 - String pointer.
 # a1 - String length.
 .type _write_error, @function
@@ -306,3 +307,120 @@ _put_char:
 _get:
 	lw a0, (a0)
 	ret
+
+# Searches for the occurences of a character in the given memory block.
+#
+# Parameters:
+# a0 - Memory block.
+# a1 - Needle.
+# a2 - Memory size.
+#
+# Sets a0 to the pointer to the found character or to null if the character
+# doesn't occur in the memory block.
+.type _memchr, @function
+_memchr:
+.Lmemchr_loop:
+	beqz a2, .Lmemchr_nil # Exit if the length is 0.
+
+	lbu t0, (a0) # Load the character from the memory block.
+	beq t0, a1, .Lmemchr_end # Exit if the character was found.
+
+	# Otherwise, continue with the next character.
+	addi a0, a0, 1
+	addi a2, a2, -1
+
+	j .Lmemchr_loop
+
+.Lmemchr_nil:
+	li a0, 0
+
+.Lmemchr_end:
+	ret
+
+# Locates a substring.
+#
+# Parameters:
+# a0 - Haystack.
+# a1 - Haystack size.
+# a2 - Needle.
+# a3 - Needle size.
+#
+# Sets a0 to the pointer to the beginning of the substring in memory or to 0
+# if the substring doesn't occur in the block.
+.type _memmem, @function
+_memmem:
+	# Prologue.
+	addi sp, sp, -24
+	sw ra, 20(sp)
+	sw s0, 16(sp)
+	addi s0, sp, 24
+
+	# Save preserved registers. They are used to keep arguments.
+	sw s1, 12(sp)
+	sw s2, 8(sp)
+	sw s3, 4(sp)
+	sw s4, 0(sp)
+
+	mv s1, a0
+	mv s2, a1
+	mv s3, a2
+	mv s4, a3
+
+.Lmemmem_loop:
+	blt s2, s3, .Lmemmem_nil # Exit if the needle length is greater than memory.
+
+	mv a0, s1
+	mv a1, s3
+	mv a2, s4
+	call _memcmp
+
+	mv t0, a0 # memcmp result.
+	mv a0, s1 # Memory pointer for the case the substring was found.
+	beqz t0, .Lmemmem_end
+
+	addi s1, s1, 1
+	add s2, s2, -1
+
+	j .Lmemmem_loop
+
+.Lmemmem_nil:
+	li a0, 0
+
+.Lmemmem_end:
+
+	# Restore the preserved registers.
+	lw s1, 12(sp)
+	lw s2, 8(sp)
+	lw s3, 4(sp)
+	lw s4, 0(sp)
+
+	# Epilogue.
+	lw ra, 20(sp)
+	lw s0, 16(sp)
+	add sp, sp, 24
+	ret
+
+# Copies memory.
+#
+# Parameters:
+# a0 - Destination.
+# a1 - Source.
+# a2 - Size.
+#
+# Preserves a0.
+.type _memcpy, @function
+_memcpy:
+	mv t0, a0
+
+.Lmemcpy_loop:
+	beqz a2, .Lmemcpy_end
+
+	lbu t1, (a1)
+	sb t1, (a0)
+
+	addi a0, a0, 1
+	addi a1, a1, 1
+
+.Lmemcpy_end:
+	mv a0, t0
+	ret

boot/stage1.s

@@ -207,11 +207,66 @@ _build_binary_expression:
 	addi sp, sp, 32
 	ret
 
+# Checks whether the given identifier starts with "loca".
+# Parameters:
+# a0 - Pointer to the identifier.
+# a1 - Identifier length.
+#
+# Sets a0 to 1 if the identifier starts with "loca", otherwise to 0.
+.type _is_local_identifier, @function
+_is_local_identifier:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	li t0, 0x61636f6c # loca
+	sw t0, 4(sp)
+	# a0 is already set.
+	addi a1, sp, 4
+	li a2, 4
+	call _memcmp
+	seqz a0, a0
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+# Checks whether the given identifier is a saved register name, like s1 or s2.
+# Parameters:
+# a0 - Pointer to the identifier.
+# a1 - Identifier length.
+#
+# Sets a0 to 1 if the identifier is a preserved register, otherwise to 0.
+.type _is_register_identifier, @function
+_is_register_identifier:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+	lbu a0, (a0)
+	addi a1, a1, -2
+	seqz a1, a1
+	addi t0, a0, -'s'
+	seqz t0, t0
+	and a0, a1, t0
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
 # Parameters:
 # a0 - Identifier length.
 # a1 - Register number as character.
-.type _build_identifier_expression, @function
+.type _compile_identifier_expression, @function
-_build_identifier_expression:
+_compile_identifier_expression:
 	# Prologue.
 	addi sp, sp, -32
 	sw ra, 28(sp)
@@ -221,22 +276,15 @@ _build_identifier_expression:
 	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
+	lw a1, 20(sp)
-	li a2, 4
+	call _is_local_identifier
-	call _memcmp
+	bnez a0, .Lcompile_identifier_expression_local
-	beqz a0, .Lbuild_identifier_expression_local
 
-	lbu a0, (s1)
+	mv a0, s1
-	lw t0, 20(sp) 
+	lw a1, 20(sp)
-	addi t0, t0, -2
+	call _is_register_identifier
-	seqz t0, t0
+	bnez a0, .Lcompile_identifier_expression_saved
-	addi t1, a0, -'s'
-	seqz t1, t1
-	and t0, t0, t1
-	bnez t0, .Lbuild_identifier_expression_saved
 
 	# Global identifier.
 	lw t1, 16(sp)
@@ -258,7 +306,7 @@ _build_identifier_expression:
 
 	lbu a0, (s1)
 	call _is_upper
-	beqz a0, .Lbuild_identifier_expression_end
+	beqz a0, .Lcompile_identifier_expression_end
 
 	lw t1, 16(sp)
 	li t0, 0x0a290061 # a\0)\n
@@ -274,9 +322,9 @@ _build_identifier_expression:
 	li a1, 12
 	call _write_out
 
-	j .Lbuild_identifier_expression_end
+	j .Lcompile_identifier_expression_end
 
-.Lbuild_identifier_expression_saved:
+.Lcompile_identifier_expression_saved:
 	li t0, 0x00202c00 # \0,_
 	lw t1, 16(sp)
 	or t0, t0, t1
@@ -294,9 +342,9 @@ _build_identifier_expression:
 	li a0, '\n'
 	call _put_char
 
-	j .Lbuild_identifier_expression_end
+	j .Lcompile_identifier_expression_end
 
-.Lbuild_identifier_expression_local:
+.Lcompile_identifier_expression_local:
 	lw t1, 16(sp)
 	li t0, 0x00202c00 # \0,_
 	or t0, t0, t1
@@ -321,9 +369,9 @@ _build_identifier_expression:
 	li a0, '\n'
 	call _put_char
 
-	j .Lbuild_identifier_expression_end
+	j .Lcompile_identifier_expression_end
 
-.Lbuild_identifier_expression_end:
+.Lcompile_identifier_expression_end:
 
 	# Epilogue.
 	lw ra, 28(sp)
@@ -368,7 +416,7 @@ _build_expression:
 
 	lw a0, 20(sp)
 	lw a1, 28(sp)
-	call _build_identifier_expression
+	call _compile_identifier_expression
 
 	j .Lbuild_expression_advance
 
@@ -455,6 +503,86 @@ _build_expression:
 	addi sp, sp, 40
 	ret
 
+# Compiles an lvalue.
+#
+# Parameters:
+# a0 - Pointer to the identifier.
+# a1 - Identifier length.
+.type _compile_designator_expression, @function
+_compile_designator_expression:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	sw a0, 20(sp) # Identifier pointer.
+	sw a1, 16(sp) # Identifier length.
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call _is_local_identifier
+	bnez a0, .Lcompile_designator_expression_local
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call _is_register_identifier
+	bnez a0, .Lcompile_designator_expression_saved
+
+.Lcompile_designator_expression_local:
+	li t0, 0x202c30 # 0,_
+	sw t0, 12(sp)
+	li t0, 0x61207773 # sw a
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 7
+	call _write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(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, 12(sp)
+	li t0, 0x29707328 # (sp)
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 5
+	call _write_out
+
+	j .Lcompile_designator_expression_end
+
+.Lcompile_designator_expression_saved:
+	li t0, 0x20766d # mv_
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 3
+	call _write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call _write_out
+
+	li t0, 0x0a # \n
+	sw t0, 12(sp)
+	li t0, 0x3061202c # , a0
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 5
+	call _write_out
+
+	j .Lcompile_designator_expression_end
+
+.Lcompile_designator_expression_end:
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
 # Compiles a statement beginning with an identifier.
 #
 # Left values should be variables named "loca n", where n is the offset
@@ -508,28 +636,9 @@ _compile_identifier:
 
 .Lcompile_identifier_assign:
 	call _build_binary_expression
-
-	li t0, 0x202c30 # 0,_
-	sw t0, 12(sp)
-	li t0, 0x61207773 # sw a
-	sw t0, 8(sp)
-	addi a0, sp, 8
-	li a1, 7
-	call _write_out
-
 	lw a0, 20(sp)
 	lw a1, 16(sp)
-	addi a0, a0, 4 # Skip the "loca" variable prefix.
+	call _compile_designator_expression
-	addi a1, a1, -4 # Skip the "loca" variable prefix.
-	call _write_out
-
-	li t0, '\n'
-	sw t0, 12(sp)
-	li t0, 0x29707328 # (sp)
-	sw t0, 8(sp)
-	addi a0, sp, 8
-	li a1, 5
-	call _write_out
 
 	j .Lcompile_identifier_end
 
@@ -1310,11 +1419,7 @@ _compile_label:
 	addi s0, sp, 16
 
 	sw a0, 0(sp) # Save the line length.
-
+	mv a1, a0 # Argument for _write_out later.
-	# Write the whole line as is.
-	mv a0, s1
-	lw a1, 0(sp)
-	call _write_out
 
 	lw t0, 0(sp) # Line length.
 	mv t1, s1 # Line start.
@@ -1323,8 +1428,15 @@ _compile_label:
 	addi t1, t1, -1 # Last character on the line.
 
 	lbu t1, (t1)
-	li t2, ':'
+	li t2, ';'
-	beq t1, t2, .Lcompile_label_colon
+	bne t1, t2, .Lcompile_label_colon
+
+	addi a1, a1, -1
+
+.Lcompile_label_colon:
+	# Write the whole line as is.
+	mv a0, s1
+	call _write_out
 
 	li t0, 0x3a # :
 	sw t0, 4(sp)
@@ -1332,7 +1444,6 @@ _compile_label:
 	li a1, 1
 	call _write_out
 
-.Lcompile_label_colon:
 	li t0, '\n'
 	sw t0, 4(sp)
 	addi a0, sp, 4
@@ -1411,12 +1522,6 @@ _compile_if:
 	call _put_char
 
 .Lcompile_if_loop:
-	/* DEBUG
-	mv a0, s1
-	li a1, 6
-	call _write_error
-	call _divide_by_zero_error */
-
 	call _skip_spaces
 	call _read_token
 
@@ -1532,13 +1637,15 @@ _compile_line:
 	call _memcmp
 	beqz a0, .Lcompile_line_exit
 
-	li t0, 0x61636f6c # loca
-	sw t0, 12(sp)
 	mv a0, s1
-	addi a1, sp, 12
+	lw a1, 20(sp)
-	li a2, 4
+	call _is_local_identifier
-	call _memcmp
+	bnez a0, .Lcompile_line_identifier
-	beqz a0, .Lcompile_line_identifier
+
+	mv a0, s1
+	li a1, 2
+	call _is_register_identifier
+	bnez a0, .Lcompile_line_identifier
 
 	li t0, 0x6f706d69 # impo
 	sw t0, 12(sp)
@@ -1806,9 +1913,6 @@ _main:
 .type _start, @function
 _start:
 	call _tokenizer_initialize
-	li a1, 50
-	call _write_error
-
 	call _main
 	call _compile
 

boot/tokenizer.s

@@ -1,165 +1,195 @@
 .global _tokenizer_initialize
 
 .section .rodata
+#
+# Classes:
+#
+# 0x00: Invalid
+# 0x01: Digit
+# 0x02: Character
+# 0x03: Space
+.type classes, @object
+.size classes, 128
+classes:
+	.byte 0x00 # 00 NUL
+	.byte 0x00 # 01 SOH
+	.byte 0x00 # 02 STX
+	.byte 0x00 # 03 ETX
+	.byte 0x00 # 04 EOT
+	.byte 0x00 # 05 ENQ
+	.byte 0x00 # 06 ACK
+	.byte 0x00 # 07 BEL
+	.byte 0x00 # 08 BS
+	.byte 0x00 # 09 HT
+	.byte 0x00 # 0A LF
+	.byte 0x00 # 0B VT
+	.byte 0x00 # 0C FF
+	.byte 0x00 # 0D CR
+	.byte 0x00 # 0E SO
+	.byte 0x00 # 0F SI
+	.byte 0x00 # 10 DLE
+	.byte 0x00 # 11 DC1
+	.byte 0x00 # 12 DC2
+	.byte 0x00 # 13 DC3
+	.byte 0x00 # 14 DC4
+	.byte 0x00 # 15 NAK
+	.byte 0x00 # 16 SYN
+	.byte 0x00 # 17 ETB
+	.byte 0x00 # 18 CAN
+	.byte 0x00 # 19 EM
+	.byte 0x00 # 1A SUB
+	.byte 0x00 # 1B ESC
+	.byte 0x00 # 1C FS
+	.byte 0x00 # 1D GS
+	.byte 0x00 # 1E RS
+	.byte 0x00 # 1F US
+	.byte 0x03 # 20 Space
+	.byte 0x00 # 21 !
+	.byte 0x00 # 22 "
+	.byte 0x00 # 23 #
+	.byte 0x00 # 24 $
+	.byte 0x00 # 25 %
+	.byte 0x00 # 26 &
+	.byte 0x00 # 27 '
+	.byte 0x00 # 28 (
+	.byte 0x00 # 29 )
+	.byte 0x00 # 2A *
+	.byte 0x00 # 2B +
+	.byte 0x00 # 2C ,
+	.byte 0x00 # 2D -
+	.byte 0x00 # 2E .
+	.byte 0x00 # 2F /
+	.byte 0x01 # 30 0
+	.byte 0x01 # 31 1
+	.byte 0x01 # 32 2
+	.byte 0x01 # 33 3
+	.byte 0x01 # 34 4
+	.byte 0x01 # 35 5
+	.byte 0x01 # 36 6
+	.byte 0x01 # 37 7
+	.byte 0x01 # 38 8
+	.byte 0x01 # 39 9
+	.byte 0x00 # 3A :
+	.byte 0x00 # 3B ;
+	.byte 0x00 # 3C <
+	.byte 0x00 # 3D =
+	.byte 0x00 # 3E >
+	.byte 0x00 # 3F ?
+	.byte 0x00 # 40 @
+	.byte 0x02 # 41 A
+	.byte 0x02 # 42 B
+	.byte 0x02 # 43 C
+	.byte 0x02 # 44 D
+	.byte 0x02 # 45 E
+	.byte 0x02 # 46 F
+	.byte 0x02 # 47 G
+	.byte 0x02 # 48 H
+	.byte 0x02 # 49 I
+	.byte 0x02 # 4A J
+	.byte 0x02 # 4B K
+	.byte 0x02 # 4C L
+	.byte 0x02 # 4D M
+	.byte 0x02 # 4E N
+	.byte 0x02 # 4F O
+	.byte 0x02 # 50 P
+	.byte 0x02 # 51 Q
+	.byte 0x02 # 52 R
+	.byte 0x02 # 53 S
+	.byte 0x02 # 54 T
+	.byte 0x02 # 55 U
+	.byte 0x02 # 56 V
+	.byte 0x02 # 57 W
+	.byte 0x02 # 58 X
+	.byte 0x02 # 59 Y
+	.byte 0x02 # 5A Z
+	.byte 0x00 # 5B [
+	.byte 0x00 # 5C \
+	.byte 0x00 # 5D ]
+	.byte 0x00 # 5E ^
+	.byte 0x00 # 5F _
+	.byte 0x00 # 60 `
+	.byte 0x02 # 61 a
+	.byte 0x02 # 62 b
+	.byte 0x02 # 63 c
+	.byte 0x02 # 64 d
+	.byte 0x02 # 65 e
+	.byte 0x02 # 66 f
+	.byte 0x02 # 67 g
+	.byte 0x02 # 68 h
+	.byte 0x02 # 69 i
+	.byte 0x02 # 6A j
+	.byte 0x02 # 6B k
+	.byte 0x02 # 6C l
+	.byte 0x02 # 6D m
+	.byte 0x02 # 6E n
+	.byte 0x02 # 6F o
+	.byte 0x02 # 70 p
+	.byte 0x02 # 71 q
+	.byte 0x02 # 72 r
+	.byte 0x02 # 73 s
+	.byte 0x02 # 74 t
+	.byte 0x02 # 75 u
+	.byte 0x02 # 76 v
+	.byte 0x02 # 77 w
+	.byte 0x02 # 78 x
+	.byte 0x02 # 79 y
+	.byte 0x02 # 7A z
+	.byte 0x00 # 7B {
+	.byte 0x00 # 7C |
+	.byte 0x00 # 7D }
+	.byte 0x00 # 7E ~
+	.byte 0x00 # 7F DEL
 
-raw_classes:
+.section .data
-	.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
+.type _initialize_classes, @function
-_tokenizer_classes:
+_initialize_classes:
 	# Prologue.
-	addi sp, sp, -8
+	addi sp, sp, -24
-	sw ra, 4(sp)
+	sw ra, 20(sp)
-	sw s0, 0(sp)
+	sw s0, 16(sp)
-	addi s0, sp, 8
+	addi s0, sp, 24
+
+	sw s1, 12(sp) # Preserve the s1 register used for the character counter.
+	li s1, 128 # 128 ASCII characters.
+
+.Linitialize_classes_loop:
+	addi s1, s1, -1
+
+	la t0, classes
+	add t0, t0, s1
+	lbu t0, (t0)
+	li t1, 0x01
+
+	bne t0, t1, .Linitialize_classes_step
+
+	/* DEBUG */
+	li a0, 0x69676964
+	sw a0, 8(sp) # Preserve the memory address.
+	addi a0, sp, 8
+	li a1, 4
+	call _write_error
+
+.Linitialize_classes_step:
+	bnez s1, .Linitialize_classes_loop
+
+	lw s1, 12(sp) # Restore the saved register.
 
 	# Epilogue.
-	lw ra, 4(sp)
+	lw ra, 20(sp)
-	lw s0, 0(sp)
+	lw s0, 16(sp)
-	addi sp, sp, 8
+	addi sp, sp, 24
 	ret
 
 # Initializes the lookup tables.
@@ -171,8 +201,7 @@ _tokenizer_initialize:
 	sw s0, 0(sp)
 	addi s0, sp, 8
 
-	la a0, raw_classes
+	call _initialize_classes
-	call _tokenizer_classes
 
 	# Epilogue.
 	lw ra, 4(sp)