start translation phase 4

05/constants.b

@@ -106,3 +106,27 @@
 :str_hash_hash
 	string ##
 	byte 0
+:str_error
+	string error
+	byte 0
+:str_define
+	string define
+	byte 0
+:str_include
+	string include
+	byte 0
+:str_ifdef
+	string ifdef
+	byte 0
+:str_ifndef
+	string ifndef
+	byte 0
+:str_if
+	string if
+	byte 0
+:str_elif
+	string elif
+	byte 0
+:str_else
+	string else
+	byte 0

05/main.b

@@ -8,6 +8,13 @@ byte 0
 byte 0
 goto main
 
+
+global object_macros_size
+global function_macros_size
+; these are allocated in main()
+global object_macros
+global function_macros
+
 function compile_error
 	argument file
 	argument line
@@ -15,12 +22,12 @@ function compile_error
 	fputs(2, file)
 	fputc(2, ':)
 	fputn(2, line)
-	fputs(2, .str_error)
+	fputs(2, .str_error_prefix)
 	fputs(2, message)
 	fputc(2, 10)
 	exit(1)
 	
-:str_error
+:str_error_prefix
 	string : Error:
 	byte 32
 	byte 0
@@ -38,6 +45,9 @@ function main
 	local output_filename
 	local pptokens
 	
+	object_macros = malloc(4000000)
+	function_macros = malloc(4000000)
+	
 	input_filename = .str_default_input_filename
 	output_filename = .str_default_output_filename
 	if argc == 1 goto have_filenames
@@ -47,6 +57,10 @@ function main
 	:have_filenames
 	pptokens = split_into_preprocessing_tokens(input_filename)
 	print_pptokens(pptokens)
+	print_separator()
+	pptokens = translation_phase_4(input_filename, pptokens)
+	;print_pptokens(pptokens)
+	print_object_macros()
 	exit(0)
 
 :usage_error

05/main.c

@@ -1,10 +1,13 @@
-#include <stdio.h>
+-#include <stdio.h>
 
 int test(int, double, ...);\
 /* here is a nice
 comment it is
 here
 */
+#
+#define A 33
+#define		 _Hello_there5 -76 + sqrt(A)
 int main(void) {
 	printf("\"Hello, world!%c\n\"", '\'');
 	_X55 = Y4_C_;

05/preprocess.b

@@ -360,3 +360,156 @@ function print_pptokens
 	:print_pptokens_loop_end
 	putc(10)
 	return
+
+function pptoken_copy_and_advance
+	argument p_in
+	argument p_out
+	local in
+	local out
+	in = *8p_in
+	out = *8p_out
+	out = strcpy(out, in)
+	in = memchr(in, 0)
+	*8p_in = in + 1
+	*8p_out = out + 1
+	return
+
+function pptoken_skip
+	argument p_in
+	local in
+	in = *8p_in
+	in = memchr(in, 0)
+	*8p_in = in + 1
+	return
+
+; skip any space tokens here
+function pptoken_skip_spaces
+	argument p_in
+	local in
+	in = *8p_in
+	:pptoken_skip_spaces_loop
+		if *1in != 32 goto pptoken_skip_spaces_loop_end
+		pptoken_skip(&in)
+		goto pptoken_skip_spaces_loop
+	:pptoken_skip_spaces_loop_end
+	*8p_in = in
+	return
+
+; phase 4:
+; Preprocessing directives are executed and macro invocations are expanded.
+; A #include preprocessing directive causes the named header or source file to be processed from phase 1 through phase 4, recursively.
+function translation_phase_4
+	argument filename
+	argument input
+	local output
+	local in
+	local out
+	local p
+	local c
+	local b
+	local line_number
+		
+	output = malloc(16000000)
+	out = output
+	in = input
+	line_number = 0
+	
+	:phase4_line
+		line_number += 1
+		c = *1in
+		if c == 0 goto phase4_end
+		if c == '# goto pp_directive ; NOTE: ## cannot appear at the start of a line
+		
+		:process_pptoken
+			c = *1in
+			if c == 10 goto phase4_next_line
+			b = isdigit(c)
+			if b != 0 goto phase4_next_pptoken
+		
+		:phase4_next_pptoken
+			pptoken_copy_and_advance(&in, &out)
+			goto process_pptoken
+		:phase4_next_line
+			pptoken_copy_and_advance(&in, &out)
+			goto phase4_line
+	:pp_directive
+		pptoken_skip(&in)
+		c = *1in
+		if c == 10 goto phase4_next_line ; "null directive" C89 § 3.8.7
+		b = str_equals(in, .str_error)
+		if b != 0 goto pp_directive_error
+		b = str_equals(in, .str_define)
+		if b != 0 goto pp_directive_define
+		goto unrecognized_directive
+	:pp_directive_error
+		fputs(2, filename)
+		fputc(2, ':)
+		fputn(2, line_number)
+		fputs(2, .str_directive_error)
+		exit(1)
+	:pp_directive_define
+		local macro_name
+		pptoken_skip(&in)
+		pptoken_skip_spaces(&in)
+		macro_name = in
+		pptoken_skip(&in)
+		c = *1in
+		if c == '( goto function_macro_definition
+			; it's an object-like macro, e.g. #define X 47
+			p = object_macros + object_macros_size
+			; copy name
+			p = strcpy(p, macro_name)
+			p += 1
+			; copy contents
+			memccpy_advance(&p, &in, 10) ; copy until newline
+			*1p = 255 ; replace newline with special "macro end" character
+			p += 1
+			object_macros_size = p - object_macros
+			in += 2 ; skip newline and following null character
+			goto phase4_line
+		:function_macro_definition
+			; a function-like macro, e.g. #define JOIN(a,b) a##b
+			byte 0xcc
+	:str_directive_error
+		string : #error
+		byte 10
+		byte 0
+	:phase4_end
+		return output
+	:unrecognized_directive
+		compile_error(filename, line_number, .str_unrecognized_directive)
+	:str_unrecognized_directive
+		string Unrecognized preprocessor directive.
+		byte 0
+
+function look_up_object_macro
+	argument name
+	local p
+	p = object_macros
+	:objmacro_lookup_loop
+		@TODO
+
+function print_object_macros
+	local p
+	local c
+	p = object_macros
+	:print_objmacros_loop
+		if *1p == 0 goto return_0 ; done!
+		puts(p)
+		putc(':)
+		putc(32)
+		p = memchr(p, 0)
+		p += 1
+		:print_objreplacement_loop
+			putc('{)
+			puts(p)
+			putc('})
+			p = memchr(p, 0)
+			p += 1
+			c = *1p
+			if c == 255 goto print_objreplacement_loop_end
+			goto print_objreplacement_loop
+		:print_objreplacement_loop_end
+			p += 1
+			fputc(1, 10)
+			goto print_objmacros_loop

05/util.b

@@ -90,6 +90,28 @@ function memchr
 	:memchr_loop_end
 	return p
 
+; copy from *p_src to *p_dest until terminator is reached, setting both to point to their respective terminators
+function memccpy_advance
+	argument p_dest
+	argument p_src
+	argument terminator
+	local src
+	local dest
+	local c
+	src = *8p_src
+	dest = *8p_dest
+	:memccpy_advance_loop
+		c = *1src
+		*1dest = c
+		if c == terminator goto memccpy_advance_loop_end
+		src += 1
+		dest += 1
+		goto memccpy_advance_loop
+	:memccpy_advance_loop_end
+	*8p_src = src
+	*8p_dest = dest
+	return
+
 function strlen
 	argument s
 	local p
@@ -101,6 +123,7 @@ function strlen
 	:strlen_loop_end
 	return p - s
 
+; like C strcpy, but returns a pointer to the terminating null character in dest
 function strcpy
 	argument dest
 	argument src
@@ -119,6 +142,20 @@ function strcpy
 	:strcpy_loop_end
 	return p
 
+function str_equals
+	argument a
+	argument b
+	local c
+	local d
+	:str_equals_loop
+		c = *1a
+		d = *1b
+		if c != d goto return_0
+		if c == 0 goto return_1
+		a += 1
+		b += 1
+		goto str_equals_loop
+
 function str_startswith
 	argument s
 	argument prefix
@@ -150,6 +187,16 @@ function puts
 	fputs(1, s)
 	return
 
+function print_separator
+	fputs(1, .str_separator)
+	return
+
+:str_separator
+	byte 10
+	string ------------------------------------------------
+	byte 10
+	byte 0
+
 function fputn
 	argument fd
 	argument n