From 5aaf9ded3646f5312e26fc7b3502f141c2e556f5 Mon Sep 17 00:00:00 2001
From: Eugen Wissner <belka@caraus.de>
Date: Tue, 22 Apr 2025 22:51:11 +0200
Subject: [PATCH] Support global variables and constants

---
 boot/asm-boot.s    |  13 -
 boot/common-boot.s | 108 +++++++-
 boot/echo-boot.s   | 656 +++++++++++++++++++++++++++++++++++++++++++-
 boot/stage2.elna   | 666 +++++++++++++++++++++++++++++++++++++++++++--
 4 files changed, 1386 insertions(+), 57 deletions(-)

diff --git a/boot/asm-boot.s b/boot/asm-boot.s
index 80167a2..fca0d3f 100644
--- a/boot/asm-boot.s
+++ b/boot/asm-boot.s
@@ -354,19 +354,6 @@ _compile_call:
 	la a2, token_comma
 	call  _token_compare
 
-	/* DEBUG. Write the current token to stderr.
-	addi a0, zero, STDERR
-	addi a1, s1, 0
-	li a2, 4 #(sp)
-	addi a7, zero, SYS_WRITE
-	ecall
-	addi a0, zero, STDERR
-	la a1, token_open_square
-	li a2, 1
-	addi a7, zero, SYS_WRITE
-	ecall
-	DEBUG. End */
-
 	bnez a0, .Lcompile_call_paren
 
 	lw t0, 12(sp) # Argument count for a procedure call.
diff --git a/boot/common-boot.s b/boot/common-boot.s
index bfd1dbb..2228a57 100644
--- a/boot/common-boot.s
+++ b/boot/common-boot.s
@@ -1,4 +1,4 @@
-.global is_alpha, is_digit, is_alnum, write_out, read_file, exit
+.global is_alpha, is_digit, is_alnum, is_upper, is_lower, write_out, read_file, exit, memcmp, write_error
 
 .section .rodata
 
@@ -9,30 +9,112 @@
 .equ STDOUT, 1
 .equ STDERR, 2
 
+new_line: .ascii "\n"
+
 .section .text
 
+# Write the current token to stderr.
+# a0 - String pointer.
+# a1 - String length.
+.type write_error, @function
+write_error:
+	mv t0, a0
+	mv t1, a1
+
+	li a0, STDERR
+	mv a1, t0
+	mv a2, t1
+	li a7, SYS_WRITE
+	ecall
+
+	li a0, STDERR
+	la a1, new_line
+	li a2, 1
+	li a7, SYS_WRITE
+	ecall
+
+	ret
+
+# a0 - First pointer.
+# a1 - Second pointer.
+# a2 - The length to compare.
+#
+# Returns 0 in a0 if memory regions are equal.
+.type memcmp, @function
+memcmp:
+	mv t0, a0
+	li a0, 0
+
+.Lmemcmp_loop:
+	beqz a2, .Lmemcmp_end
+
+	lbu t1, (t0)
+	lbu t2, (a1)
+	sub a0, t1, t2
+
+	bnez a0, .Lmemcmp_end
+
+	addi t0, t0, 1
+	addi a1, a1, 1
+	addi a2, a2, -1
+
+	j .Lmemcmp_loop
+
+.Lmemcmp_end:
+	ret
+
+# Detects if a0 is an uppercase character. Sets a0 to 1 if so, otherwise to 0.
+is_upper:
+	li t0, 'A' - 1
+	sltu t1, t0, a0 # t1 = a0 >= 'A'
+
+	sltiu t2, a0, 'Z' + 1 # t2 = a0 <= 'Z'
+	and a0, t1, t2 # t1 = a0 >= 'A' & a0 <= 'Z'
+
+	ret
+
+# Detects if a0 is an lowercase character. Sets a0 to 1 if so, otherwise to 0.
+.type is_lower, @function
+is_lower:
+	li t0, 'a' - 1
+	sltu t2, t0, a0 # t2 = a0 >= 'a'
+
+	sltiu t3, a0, 'z' + 1 # t3 = a0 <= 'z'
+	and a0, t2, t3 # t2 = a0 >= 'a' & a0 <= 'z'
+
+	ret
+
 # Detects if the passed character is a 7-bit alpha character or an underscore.
 # The character is passed in a0.
 # Sets a0 to 1 if the character is an alpha character or underscore, sets it to 0 otherwise.
 .type is_alpha, @function
 is_alpha:
-	li t0, 'A' - 1
-	sltu t1, t0, a0 # t1 = a0 >= 'A'
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
 
-	sltiu t2, a0, 'Z' + 1 # t2 = a0 <= 'Z'
-	and t1, t1, t2 # t1 = a0 >= 'A' & a0 <= 'Z'
+	sw a0, 4(sp)
 
-	li t0, 'a' - 1
-	sltu t2, t0, a0 # t2 = a0 >= 'a'
+	call is_upper
+	sw a0, 0(sp)
 
-	sltiu t3, a0, 'z' + 1 # t3 = a0 <= 'z'
-	and t2, t2, t3 # t2 = a0 >= 'a' & a0 <= 'z'
+	lw a0, 4(sp)
+	call is_lower
 
-	xori t3, a0, '_'
-	seqz t3, t3
+	lw t0, 4(sp)
+	xori t1, t0, '_'
+	seqz t1, t1
 
-	or a0, t1, t2
-	or a0, a0, t3
+	lw t0, 0(sp)
+	or a0, a0, t0
+	or a0, a0, t1
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
 	ret
 
 .type is_digit, @function
diff --git a/boot/echo-boot.s b/boot/echo-boot.s
index b6e65f7..a8faecb 100644
--- a/boot/echo-boot.s
+++ b/boot/echo-boot.s
@@ -1,35 +1,665 @@
-.global _start, source_code
+.global _start
+# s1 - Contains the current position in the source text.
 
 .section .rodata
-.type SOURCE_BUFFER_SIZE, @object
-.size SOURCE_BUFFER_SIZE, 4
-SOURCE_BUFFER_SIZE: .long 4096
+SOURCE_BUFFER_SIZE: .long 20480
 
 .section .bss
 .type source_code, @object
-.size source_code, 4096
-source_code: .zero 4096
+.size source_code, 20480
+source_code: .zero 20480
 
 .section .text
 
-_compile:
+# Reads a token and returns its length in a0.
+# _read_token doesn't change s1, it finds the length of the token s1 is pointing to.
+.type _read_token, @function
+_read_token:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	lbu t0, (s1) # t0 = Current character.
+	sw zero, 4(sp)
+
+	li t1, '.'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ','
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ':'
+	beq t0, t1, .Ltoken_character_colon
+
+	li t1, ';'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, '('
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ')'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, '['
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ']'
+	beq t0, t1, .Ltoken_character_single
+
+.Ltoken_character_loop_do: # Expect an identifier or a number.
+	lw t6, 4(sp)
+	add t1, s1, t6
+	lbu a0, (t1) # a0 = Current character.
+
+	call is_alnum
+
+	beqz a0, .Ltoken_character_end
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+	j .Ltoken_character_loop_do
+
+.Ltoken_character_single:
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+	j .Ltoken_character_end
+
+.Ltoken_character_colon:
+	lbu t0, 1(s1) # t0 = The character after the colon.
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+
+	li t1, '='
+	beq t0, t1, .Ltoken_character_single
+	j .Ltoken_character_end
+
+.Ltoken_character_end:
+	lw a0, 4(sp)
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
 	ret
 
+# Skips the spaces till the next non space character.
+.type _skip_spaces, @function
+_skip_spaces:
+.Lspace_loop_do:
+	lbu t0, (s1) # t0 = Current character.
+
+	li t1, ' '
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\t'
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\n'
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\r'
+	beq t0, t1, .Lspace_loop_repeat
+
+	j .Lspace_loop_end
+.Lspace_loop_repeat:
+	addi s1, s1, 1
+	j .Lspace_loop_do
+
+.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
+_skip_comment:
+	add s1, s1, a0
+	addi s1, s1, 1 # Skip the new line.
+	ret
+
+# Parameters:
+# a0 - Line length.
+.type _compile_assembly, @function
+_compile_assembly:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	sw a0, 4(sp) # a0 - Line length.
+
+	# Write the source to the standard output.
+	mv a0, s1
+	lw a1, 4(sp)
+	call write_out
+
+	lw t0, 4(sp)
+	add s1, s1, t0
+
+	li t0, '\n'
+	sb t0, 0(sp)
+	addi a0, sp, 0
+	li a1, 1
+	call write_out
+
+	addi s1, s1, 1 # Skip the new line.
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+.type _compile_program, @function
+_compile_program:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	# .global _start
+	li t0, 0x0a7472 # rt\n
+	sw t0, 20(sp)
+	li t0, 0x6174735f # _sta
+	sw t0, 16(sp)
+	li t0, 0x206c6162 # bal_
+	sw t0, 12(sp)
+	li t0, 0x6f6c672e # .glo
+	sw t0, 8(sp)
+
+	addi a0, sp, 8
+	li a1, 16
+	call write_out
+
+	addi s1, s1, 8 # program\n.
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _compile_constant_section, @function
+_compile_constant_section:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	# .section .rodata
+	li t0, 0x0a # \n
+	sw t0, 20(sp)
+	li t0, 0x61746164 # data
+	sw t0, 16(sp)
+	li t0, 0x6f722e20 # _.ro
+	sw t0, 12(sp)
+	li t0, 0x6e6f6974 # tion
+	sw t0, 8(sp)
+	li t0, 0x6365732e # .sec
+	sw t0, 4(sp)
+
+	addi a0, sp, 4
+	li a1, 17
+	call write_out
+
+	addi s1, s1, 6 # const\n.
+
+.Lcompile_constant_section_item:
+	call _skip_spaces
+	lbu a0, (s1)
+	call is_upper
+	beqz a0, .Lcompile_constant_section_end
+
+	call _compile_constant
+	j .Lcompile_constant_section_item
+
+.Lcompile_constant_section_end:
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _compile_constant, @function
+_compile_constant:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	call _read_token
+
+	mv a1, a0 # The identifier length from _read_token should be in a1.
+	mv a0, s1 # Save the identifier pointer before advancing it.
+	add s1, s1, a1
+	call write_out
+
+	call _skip_spaces
+	call _read_token
+	addi s1, s1, 2 # Skip the assignment sign.
+
+	# : .long
+	li t0, 0x20676e6f # ong_
+	sw t0, 4(sp)
+	li t0, 0x6c2e203a # : .l
+	sw t0, 0(sp)
+	mv a0, sp
+	li a1, 8
+	call write_out
+
+	call _skip_spaces
+	call _read_token
+
+	mv a1, a0 # The literal length from _read_token should be in a1.
+	mv a0, s1 # Save the literal pointer before advancing it.
+	add s1, s1, a1
+	call write_out
+
+	li t0, '\n'
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 1
+	call write_out
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+.type _compile_variable_section, @function
+_compile_variable_section:
+	# Prologue.
+	addi sp, sp, -24
+	sw ra, 20(sp)
+	sw s0, 16(sp)
+	addi s0, sp, 24
+
+	# .section .bss
+	li t0, 0x0a73 # s\n
+	sw t0, 12(sp)
+	li t0, 0x73622e20 # _.bs
+	sw t0, 8(sp)
+	li t0, 0x6e6f6974 # tion
+	sw t0, 4(sp)
+	li t0, 0x6365732e # .sec
+	sw t0, 0(sp)
+
+	addi a0, sp, 0
+	li a1, 14
+	call write_out
+
+	addi s1, s1, 4 # var\n.
+
+.Lcompile_variable_section_item:
+	call _skip_spaces
+	lbu a0, (s1)
+	call is_lower
+	beqz a0, .Lcompile_variable_section_end
+
+	call _compile_variable
+	j .Lcompile_variable_section_item
+
+.Lcompile_variable_section_end:
+	# Epilogue.
+	lw ra, 20(sp)
+	lw s0, 16(sp)
+	addi sp, sp, 24
+	ret
+
+.type _compile_variable, @function
+_compile_variable:
+	# Prologue.
+	addi sp, sp, -40
+	sw ra, 36(sp)
+	sw s0, 32(sp)
+	addi s0, sp, 40
+
+	call _read_token
+
+	# Save the identifier on the stack since it should emitted multiple times.
+	sw s1, 28(sp)
+	sw a0, 24(sp)
+	add s1, s1, a0
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the colon in front of the type.
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the opening bracket.
+
+	call _read_token
+
+	# Save the array size on the stack since it has to be emitted multiple times.
+	sw s1, 20(sp)
+	sw a0, 16(sp)
+	add s1, s1, a0
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the closing bracket.
+
+	call _skip_spaces
+	call _read_token
+	add s1, s1, a0 # Skip the type.
+
+	# .type identifier, @object
+	li t0, 0x2065 # e_
+	sw t0, 12(sp)
+	li t0, 0x7079742e # .typ
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 6
+	call write_out
+
+	lw a0, 28(sp)
+	lw a1, 24(sp)
+	call write_out
+
+	li t0, 0x0a74 # t\n
+	sw t0, 12(sp)
+	li t0, 0x63656a62 # bjec
+	sw t0, 8(sp)
+	li t0, 0x6f40202c # , @o
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 10
+	call write_out
+
+	# .size identifier, size
+	li t0, 0x2065 # e_
+	sw t0, 12(sp)
+	li t0, 0x7a69732e # .siz
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 6
+	call write_out
+
+	lw a0, 28(sp)
+	lw a1, 24(sp)
+	call write_out
+
+	li t0, 0x202c # ,
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 2
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a # \n
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 1
+	call write_out
+
+	# identifier: .zero size
+	lw a0, 28(sp)
+	lw a1, 24(sp)
+	call write_out
+
+	li t0, 0x206f7265 # ero_
+	sw t0, 12(sp)
+	li t0, 0x7a2e203a # : .z
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 8
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a # \n
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 1
+	call write_out
+
+	# Epilogue.
+	lw ra, 36(sp)
+	lw s0, 32(sp)
+	addi sp, sp, 40
+	ret
+
+.type _compile_procedure, @function
+_compile_procedure:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	addi s1, s1, 5 # Skip proc_
+	call _read_token
+	sw s1, 20(sp)
+	sw a0, 16(sp)
+	add s1, s1, a0
+
+	# .type identifier, @function
+	li t0, 0x2065 # e_
+	sw t0, 12(sp)
+	li t0, 0x7079742e # .typ
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 6
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a6e6f69 # ion\n
+	sw t0, 12(sp)
+	li t0, 0x74636e75 # unct
+	sw t0, 8(sp)
+	li t0, 0x6640202c # , @f
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 12
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a3a # :\n
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 2
+	call write_out
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip opening argument paren.
+	call _skip_spaces
+	addi s1, s1, 1 # Skip closing argument paren.
+	call _skip_spaces
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+# Parameters:
+# a0 - Line length.
+.type _compile_line, @function
+_compile_line:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	sw a0, 20(sp) # a0 - Line length.
+
+	beqz a0, .Lcompile_line_empty # Skip an empty line.
+
+	lbu t0, (s1)
+	li t1, '#'
+	beq t0, t1, .Lcompile_line_comment
+
+	li t0, 0x0a6d6172 # ram\n
+	sw t0, 16(sp)
+	li t0, 0x676f7270 # prog
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 8
+	call memcmp
+	beqz a0, .Lcompile_line_program
+
+	li t0, 0x0a74 # t\n
+	sw t0, 16(sp)
+	li t0, 0x736e6f63 # cons
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 6
+	call memcmp
+	beqz a0, .Lcompile_line_const
+
+	li t0, 0x0a726176 # var\n
+	sw t0, 16(sp)
+	mv a0, s1
+	addi a1, sp, 16
+	li a2, 4
+	call memcmp
+	beqz a0, .Lcompile_line_var
+
+	li t0, 0x20 # _
+	sw t0, 16(sp)
+	li t0, 0x636f7270 # proc
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 5
+	call memcmp
+	beqz a0, .Lcompile_line_procedure
+
+	j .Lcompile_line_unchanged # Else.
+
+.Lcompile_line_const:
+	call _compile_constant_section
+	j .Lcompile_line_end
+
+.Lcompile_line_procedure:
+	call _compile_procedure
+	j .Lcompile_line_end
+
+.Lcompile_line_var:
+
+	/* DEBUG
+	mv a0, s1
+	li a1, 20
+	call write_error */
+
+	call _compile_variable_section
+	j .Lcompile_line_end
+
+.Lcompile_line_program:
+	call _compile_program
+	j .Lcompile_line_end
+
+.Lcompile_line_comment:
+	lw a0, 20(sp)
+	call _skip_comment
+	j .Lcompile_line_end
+
+.Lcompile_line_empty:
+	addi s1, s1, 1
+	j .Lcompile_line_end
+
+.Lcompile_line_unchanged:
+	lw a0, 20(sp)
+	call _compile_assembly
+	j .Lcompile_line_end
+
+.Lcompile_line_end:
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+# Finds the end of the line and returns its length in a0.
+.type _read_line, @function
+_read_line:
+	mv t0, s1 # Local position in the source text.
+
+.Lread_line_do:
+	lbu t1, (t0) # t1 = Current character.
+	beqz t1, .Lread_line_end # Exit the loop on the NUL character.
+	li t2, '\n'
+	beq t1, t2, .Lread_line_end # Exit the loop on the new line.
+
+	addi t0, t0, 1
+	j .Lread_line_do
+
+.Lread_line_end:
+	sub a0, t0, s1 # Return the line length.
+	ret
+
+.type _compile, @function
+_compile:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+.Lcompile_do:
+	lbu t0, (s1) # t0 = Current character.
+	beqz t0, .Lcompile_end # Exit the loop on the NUL character.
+
+	call _skip_indentation
+	call _read_line
+	call _compile_line
+
+	j .Lcompile_do
+.Lcompile_end:
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Entry point.
+.type _start, @function
 _start:
 	# Read the source from the standard input.
 	la a0, source_code
 	la a1, SOURCE_BUFFER_SIZE # Buffer size.
 	lw a1, (a1)
 	call read_file
-	mv s1, a0
 
+	la s1, source_code # s1 = Source code position.
 	call _compile
 
-	# Write the source to the standard output.
-	la a0, source_code
-	mv a1, s1
-	call write_out
-
 	# Call exit.
 	li a0, 0   # Use 0 return code.
 	call exit
diff --git a/boot/stage2.elna b/boot/stage2.elna
index a119f4a..4338ad8 100644
--- a/boot/stage2.elna
+++ b/boot/stage2.elna
@@ -1,31 +1,661 @@
-.global _start, source_code
+program
+# s1 - Contains the current position in the source text.
 
-.equ SYS_READ, 63
-.equ SYS_WRITE, 64
-.equ SYS_EXIT, 93
-.equ STDIN, 0
-.equ STDOUT, 1
-.equ STDERR, 2
+const
+	SOURCE_BUFFER_SIZE := 20480
 
-.equ SOURCE_BUFFER_SIZE, 2048
-
-.section .bss
-.type source_code, @object
-.size source_code, SOURCE_BUFFER_SIZE
-source_code: .zero SOURCE_BUFFER_SIZE
+var
+	source_code: [20480]Byte
 
 .section .text
 
+# Reads a token and returns its length in a0.
+# _read_token doesn't change s1, it finds the length of the token s1 is pointing to.
+proc _read_token()
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	lbu t0, (s1) # t0 = Current character.
+	sw zero, 4(sp)
+
+	li t1, '.'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ','
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ':'
+	beq t0, t1, .Ltoken_character_colon
+
+	li t1, ';'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, '('
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ')'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, '['
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ']'
+	beq t0, t1, .Ltoken_character_single
+
+.Ltoken_character_loop_do: # Expect an identifier or a number.
+	lw t6, 4(sp)
+	add t1, s1, t6
+	lbu a0, (t1) # a0 = Current character.
+
+	call is_alnum
+
+	beqz a0, .Ltoken_character_end
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+	j .Ltoken_character_loop_do
+
+.Ltoken_character_single:
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+	j .Ltoken_character_end
+
+.Ltoken_character_colon:
+	lbu t0, 1(s1) # t0 = The character after the colon.
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+
+	li t1, '='
+	beq t0, t1, .Ltoken_character_single
+	j .Ltoken_character_end
+
+.Ltoken_character_end:
+	lw a0, 4(sp)
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+# Skips the spaces till the next non space character.
+.type _skip_spaces, @function
+_skip_spaces:
+.Lspace_loop_do:
+	lbu t0, (s1) # t0 = Current character.
+
+	li t1, ' '
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\t'
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\n'
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\r'
+	beq t0, t1, .Lspace_loop_repeat
+
+	j .Lspace_loop_end
+.Lspace_loop_repeat:
+	addi s1, s1, 1
+	j .Lspace_loop_do
+
+.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
+_skip_comment:
+	add s1, s1, a0
+	addi s1, s1, 1 # Skip the new line.
+	ret
+
+# Parameters:
+# a0 - Line length.
+.type _compile_assembly, @function
+_compile_assembly:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	sw a0, 4(sp) # a0 - Line length.
+
+	# Write the source to the standard output.
+	mv a0, s1
+	lw a1, 4(sp)
+	call write_out
+
+	lw t0, 4(sp)
+	add s1, s1, t0
+
+	li t0, '\n'
+	sb t0, 0(sp)
+	addi a0, sp, 0
+	li a1, 1
+	call write_out
+
+	addi s1, s1, 1 # Skip the new line.
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+.type _compile_program, @function
+_compile_program:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	# .global _start
+	li t0, 0x0a7472 # rt\n
+	sw t0, 20(sp)
+	li t0, 0x6174735f # _sta
+	sw t0, 16(sp)
+	li t0, 0x206c6162 # bal_
+	sw t0, 12(sp)
+	li t0, 0x6f6c672e # .glo
+	sw t0, 8(sp)
+
+	addi a0, sp, 8
+	li a1, 16
+	call write_out
+
+	addi s1, s1, 8 # program\n.
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _compile_constant_section, @function
+_compile_constant_section:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	# .section .rodata
+	li t0, 0x0a # \n
+	sw t0, 20(sp)
+	li t0, 0x61746164 # data
+	sw t0, 16(sp)
+	li t0, 0x6f722e20 # _.ro
+	sw t0, 12(sp)
+	li t0, 0x6e6f6974 # tion
+	sw t0, 8(sp)
+	li t0, 0x6365732e # .sec
+	sw t0, 4(sp)
+
+	addi a0, sp, 4
+	li a1, 17
+	call write_out
+
+	addi s1, s1, 6 # const\n.
+
+.Lcompile_constant_section_item:
+	call _skip_spaces
+	lbu a0, (s1)
+	call is_upper
+	beqz a0, .Lcompile_constant_section_end
+
+	call _compile_constant
+	j .Lcompile_constant_section_item
+
+.Lcompile_constant_section_end:
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _compile_constant, @function
+_compile_constant:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	call _read_token
+
+	mv a1, a0 # The identifier length from _read_token should be in a1.
+	mv a0, s1 # Save the identifier pointer before advancing it.
+	add s1, s1, a1
+	call write_out
+
+	call _skip_spaces
+	call _read_token
+	addi s1, s1, 2 # Skip the assignment sign.
+
+	# : .long
+	li t0, 0x20676e6f # ong_
+	sw t0, 4(sp)
+	li t0, 0x6c2e203a # : .l
+	sw t0, 0(sp)
+	mv a0, sp
+	li a1, 8
+	call write_out
+
+	call _skip_spaces
+	call _read_token
+
+	mv a1, a0 # The literal length from _read_token should be in a1.
+	mv a0, s1 # Save the literal pointer before advancing it.
+	add s1, s1, a1
+	call write_out
+
+	li t0, '\n'
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 1
+	call write_out
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+.type _compile_variable_section, @function
+_compile_variable_section:
+	# Prologue.
+	addi sp, sp, -24
+	sw ra, 20(sp)
+	sw s0, 16(sp)
+	addi s0, sp, 24
+
+	# .section .bss
+	li t0, 0x0a73 # s\n
+	sw t0, 12(sp)
+	li t0, 0x73622e20 # _.bs
+	sw t0, 8(sp)
+	li t0, 0x6e6f6974 # tion
+	sw t0, 4(sp)
+	li t0, 0x6365732e # .sec
+	sw t0, 0(sp)
+
+	addi a0, sp, 0
+	li a1, 14
+	call write_out
+
+	addi s1, s1, 4 # var\n.
+
+.Lcompile_variable_section_item:
+	call _skip_spaces
+	lbu a0, (s1)
+	call is_lower
+	beqz a0, .Lcompile_variable_section_end
+
+	call _compile_variable
+	j .Lcompile_variable_section_item
+
+.Lcompile_variable_section_end:
+	# Epilogue.
+	lw ra, 20(sp)
+	lw s0, 16(sp)
+	addi sp, sp, 24
+	ret
+
+.type _compile_variable, @function
+_compile_variable:
+	# Prologue.
+	addi sp, sp, -40
+	sw ra, 36(sp)
+	sw s0, 32(sp)
+	addi s0, sp, 40
+
+	call _read_token
+
+	# Save the identifier on the stack since it should emitted multiple times.
+	sw s1, 28(sp)
+	sw a0, 24(sp)
+	add s1, s1, a0
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the colon in front of the type.
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the opening bracket.
+
+	call _read_token
+
+	# Save the array size on the stack since it has to be emitted multiple times.
+	sw s1, 20(sp)
+	sw a0, 16(sp)
+	add s1, s1, a0
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the closing bracket.
+
+	call _skip_spaces
+	call _read_token
+	add s1, s1, a0 # Skip the type.
+
+	# .type identifier, @object
+	li t0, 0x2065 # e_
+	sw t0, 12(sp)
+	li t0, 0x7079742e # .typ
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 6
+	call write_out
+
+	lw a0, 28(sp)
+	lw a1, 24(sp)
+	call write_out
+
+	li t0, 0x0a74 # t\n
+	sw t0, 12(sp)
+	li t0, 0x63656a62 # bjec
+	sw t0, 8(sp)
+	li t0, 0x6f40202c # , @o
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 10
+	call write_out
+
+	# .size identifier, size
+	li t0, 0x2065 # e_
+	sw t0, 12(sp)
+	li t0, 0x7a69732e # .siz
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 6
+	call write_out
+
+	lw a0, 28(sp)
+	lw a1, 24(sp)
+	call write_out
+
+	li t0, 0x202c # ,
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 2
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a # \n
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 1
+	call write_out
+
+	# identifier: .zero size
+	lw a0, 28(sp)
+	lw a1, 24(sp)
+	call write_out
+
+	li t0, 0x206f7265 # ero_
+	sw t0, 12(sp)
+	li t0, 0x7a2e203a # : .z
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 8
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a # \n
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 1
+	call write_out
+
+	# Epilogue.
+	lw ra, 36(sp)
+	lw s0, 32(sp)
+	addi sp, sp, 40
+	ret
+
+.type _compile_procedure, @function
+_compile_procedure:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	addi s1, s1, 5 # Skip proc_
+	call _read_token
+	sw s1, 20(sp)
+	sw a0, 16(sp)
+	add s1, s1, a0
+
+	# .type identifier, @function
+	li t0, 0x2065 # e_
+	sw t0, 12(sp)
+	li t0, 0x7079742e # .typ
+	sw t0, 8(sp)
+	addi a0, sp, 8
+	li a1, 6
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a6e6f69 # ion\n
+	sw t0, 12(sp)
+	li t0, 0x74636e75 # unct
+	sw t0, 8(sp)
+	li t0, 0x6640202c # , @f
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 12
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	li t0, 0x0a3a # :\n
+	sw t0, 12(sp)
+	addi a0, sp, 12
+	li a1, 2
+	call write_out
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip opening argument paren.
+	call _skip_spaces
+	addi s1, s1, 1 # Skip closing argument paren.
+	call _skip_spaces
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+# Parameters:
+# a0 - Line length.
+.type _compile_line, @function
+_compile_line:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	sw a0, 20(sp) # a0 - Line length.
+
+	beqz a0, .Lcompile_line_empty # Skip an empty line.
+
+	lbu t0, (s1)
+	li t1, '#'
+	beq t0, t1, .Lcompile_line_comment
+
+	li t0, 0x0a6d6172 # ram\n
+	sw t0, 16(sp)
+	li t0, 0x676f7270 # prog
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 8
+	call memcmp
+	beqz a0, .Lcompile_line_program
+
+	li t0, 0x0a74 # t\n
+	sw t0, 16(sp)
+	li t0, 0x736e6f63 # cons
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 6
+	call memcmp
+	beqz a0, .Lcompile_line_const
+
+	li t0, 0x0a726176 # var\n
+	sw t0, 16(sp)
+	mv a0, s1
+	addi a1, sp, 16
+	li a2, 4
+	call memcmp
+	beqz a0, .Lcompile_line_var
+
+	li t0, 0x20 # _
+	sw t0, 16(sp)
+	li t0, 0x636f7270 # proc
+	sw t0, 12(sp)
+	mv a0, s1
+	addi a1, sp, 12
+	li a2, 5
+	call memcmp
+	beqz a0, .Lcompile_line_procedure
+
+	j .Lcompile_line_unchanged # Else.
+
+.Lcompile_line_const:
+	call _compile_constant_section
+	j .Lcompile_line_end
+
+.Lcompile_line_procedure:
+	call _compile_procedure
+	j .Lcompile_line_end
+
+.Lcompile_line_var:
+
+	/* DEBUG
+	mv a0, s1
+	li a1, 20
+	call write_error */
+
+	call _compile_variable_section
+	j .Lcompile_line_end
+
+.Lcompile_line_program:
+	call _compile_program
+	j .Lcompile_line_end
+
+.Lcompile_line_comment:
+	lw a0, 20(sp)
+	call _skip_comment
+	j .Lcompile_line_end
+
+.Lcompile_line_empty:
+	addi s1, s1, 1
+	j .Lcompile_line_end
+
+.Lcompile_line_unchanged:
+	lw a0, 20(sp)
+	call _compile_assembly
+	j .Lcompile_line_end
+
+.Lcompile_line_end:
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+# Finds the end of the line and returns its length in a0.
+.type _read_line, @function
+_read_line:
+	mv t0, s1 # Local position in the source text.
+
+.Lread_line_do:
+	lbu t1, (t0) # t1 = Current character.
+	beqz t1, .Lread_line_end # Exit the loop on the NUL character.
+	li t2, '\n'
+	beq t1, t2, .Lread_line_end # Exit the loop on the new line.
+
+	addi t0, t0, 1
+	j .Lread_line_do
+
+.Lread_line_end:
+	sub a0, t0, s1 # Return the line length.
+	ret
+
+.type _compile, @function
+_compile:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+.Lcompile_do:
+	lbu t0, (s1) # t0 = Current character.
+	beqz t0, .Lcompile_end # Exit the loop on the NUL character.
+
+	call _skip_indentation
+	call _read_line
+	call _compile_line
+
+	j .Lcompile_do
+.Lcompile_end:
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Entry point.
+.type _start, @function
 _start:
 	# Read the source from the standard input.
 	la a0, source_code
-	li a1, SOURCE_BUFFER_SIZE # Buffer size.
+	la a1, SOURCE_BUFFER_SIZE # Buffer size.
+	lw a1, (a1)
 	call read_file
 
-	# Write the source to the standard output.
-	mv a1, a0
-	la a0, source_code
-	call write_out
+	la s1, source_code # s1 = Source code position.
+	call _compile
 
 	# Call exit.
 	li a0, 0   # Use 0 return code.