Implement [[ inline fragments ]]

Fixes gbdev#500

include/asm/section.h

@@ -49,6 +49,7 @@ void sect_NewSection(char const *name, uint32_t secttype, uint32_t org,
 void sect_SetLoadSection(char const *name, uint32_t secttype, uint32_t org,
 			 struct SectionSpec const *attributes, enum SectionModifier mod);
 void sect_EndLoadSection(void);
+void sect_PushInlineFragmentSection(void);
 
 struct Section *sect_GetSymbolSection(void);
 uint32_t sect_GetSymbolOffset(void);

man/rgbasm.5

@@ -851,6 +851,70 @@ first, followed by the one from
 and the one from
 .Ql bar.o
 last.
+.Ss Inline Fragments
+Inline fragments are useful for short blocks of code or data that are only referenced once.
+They are section fragments created by surrounding instructions or directives with
+.Ql [[
+double brackets
+.Ql ]] ,
+without a separate
+.Ic SECTION FRAGMENT
+declaration.
+.Pp
+The content of an inline fragment becomes a
+.Ic SECTION FRAGMENT ,
+sharing the same name and bank as its parent ROM section, but without any other constraints.
+The parent section also becomes a
+.Ic FRAGMENT
+if it was not one already, so that it can be merged with its inline fragments.
+RGBLINK merges the fragments in no particular order.
+.Pp
+An inline fragment can take the place of any 16-bit integer constant
+.Ql n16
+from the
+.Xr gbz80 7
+documentation, as well as a
+.Ic DW
+item.
+The inline fragment then evaluates to its starting address.
+For example, you can
+.Ic CALL
+or
+.Ic JP
+to an inline fragment.
+.Pp
+This code using named labels:
+.Bd -literal -offset indent
+FortyTwo:
+    call Sub1
+    jp Sub2
+Sub1:
+    ld a, [Twenty]
+    ret
+Sub2:
+    inc a
+    add a
+    ret
+Twenty: db 20
+dw FortyTwo
+.Ed
+.Pp
+is equivalent to this code using inline fragments:
+.Bd -literal -offset indent
+dw [[
+    call [[
+        ld a, [ [[db 20]] ]
+        ret
+    ]]
+    jp [[
+        inc a
+        add a
+        ret
+    ]]
+]]
+.Ed
+.Pp
+The difference is that the example using inline fragments does not declare a particular order for its pieces.
 .Sh SYMBOLS
 RGBDS supports several types of symbols:
 .Bl -hang

src/asm/lexer.c

@@ -342,6 +342,7 @@ struct LexerState {
 	uint32_t lineNo;
 	uint32_t colNo;
 	int lastToken;
+	int nextToken;
 
 	struct IfStack *ifStack;
 
@@ -365,6 +366,7 @@ static void initState(struct LexerState *state)
 	state->mode = LEXER_NORMAL;
 	state->atLineStart = true; /* yylex() will init colNo due to this */
 	state->lastToken = T_EOF;
+	state->nextToken = 0;
 
 	state->ifStack = NULL;
 
@@ -1285,6 +1287,7 @@ static uint32_t readGfxConstant(void)
 static bool startsIdentifier(int c)
 {
 	// Anonymous labels internally start with '!'
+	// Section fragment labels internally start with '$'
 	return (c <= 'Z' && c >= 'A') || (c <= 'z' && c >= 'a') || c == '.' || c == '_';
 }
 
@@ -1762,6 +1765,13 @@ static int yylex_SKIP_TO_ENDC(void); // forward declaration for yylex_NORMAL
 
 static int yylex_NORMAL(void)
 {
+	if (lexerState->nextToken) {
+		int token = lexerState->nextToken;
+
+		lexerState->nextToken = 0;
+		return token;
+	}
+
 	for (;;) {
 		int c = nextChar();
 		char secondChar;
@@ -1786,10 +1796,6 @@ static int yylex_NORMAL(void)
 			yylval.symName[1] = '\0';
 			return T_ID;
 
-		case '[':
-			return T_LBRACK;
-		case ']':
-			return T_RBRACK;
 		case '(':
 			return T_LPAREN;
 		case ')':
@@ -1799,6 +1805,26 @@ static int yylex_NORMAL(void)
 
 		/* Handle ambiguous 1- or 2-char tokens */
 
+		case '[': /* Either [ or [[ */
+			if (peek() == '[') {
+				shiftChar();
+				return T_2LBRACK;
+			}
+			return T_LBRACK;
+
+		case ']': /* Either ] or ]] */
+			if (peek() == ']') {
+				shiftChar();
+				/*
+				 * [[ Inline fragments ]] inject a T_EOL token to
+				 * end their contents even without a newline.
+				 * Retroactively lex the ]] after it.
+				 */
+				lexerState->nextToken = T_2RBRACK;
+				return T_EOL;
+			}
+			return T_RBRACK;
+
 		case '+': /* Either += or ADD */
 			if (peek() == '=') {
 				shiftChar();

src/asm/parser.y

@@ -37,6 +37,7 @@
 #include "platform.h" // strncasecmp, strdup
 
 static struct CaptureBody captureBody; /* Captures a REPT/FOR or MACRO */
+static uint32_t inlineFragmentID = 0; /* Incrementing unique ID for inline fragment labels */
 
 static void upperstring(char *dest, char const *src)
 {
@@ -524,6 +525,7 @@ enum {
 %token	T_COMMA ","
 %token	T_COLON ":"
 %token	T_LBRACK "[" T_RBRACK "]"
+%token	T_2LBRACK "[[" T_2RBRACK "]]"
 %token	T_LPAREN "(" T_RPAREN ")"
 %token	T_NEWLINE "newline"
 
@@ -586,6 +588,7 @@ enum {
 %type	<symName> redef_id
 %type	<symName> scoped_id
 %type	<symName> scoped_anon_id
+%type	<symName> inline_fragment
 %token	T_POP_EQU "EQU"
 %token	T_POP_SET "SET"
 %token	T_POP_EQUAL "="
@@ -682,6 +685,7 @@ enum {
 %type	<expr>		op_mem_ind
 %type	<assertType>	assert_type
 
+%token T_EOL "end of line"
 %token T_EOB "end of buffer"
 %token T_EOF 0 "end of file"
 %start asmfile
@@ -695,7 +699,7 @@ lines		: %empty
 		| lines line
 ;
 
-endofline	: T_NEWLINE | T_EOB
+endofline	: T_NEWLINE | T_EOL | T_EOB
 ;
 
 plain_directive	: label
@@ -1365,14 +1369,25 @@ reloc_16bit	: relocexpr {
 			rpn_CheckNBit(&$1, 16);
 			$$ = $1;
 		}
+		| inline_fragment { rpn_Symbol(&$$, $1); }
 ;
 
 reloc_16bit_no_str : relocexpr_no_str {
 			rpn_CheckNBit(&$1, 16);
 			$$ = $1;
 		}
+		| inline_fragment { rpn_Symbol(&$$, $1); }
 ;
 
+inline_fragment	: T_2LBRACK {
+			sect_PushInlineFragmentSection();
+			sprintf($<symName>$, "$%" PRIu32, inlineFragmentID++);
+			sym_AddLabel($<symName>$);
+		} asmfile T_2RBRACK {
+			sect_PopSection();
+			strcpy($$, $<symName>2);
+		}
+;
 
 relocexpr	: relocexpr_no_str
 		| string {

src/asm/section.c

@@ -473,6 +473,40 @@ void sect_EndLoadSection(void)
 	currentLoadSection = NULL;
 }
 
+void sect_PushInlineFragmentSection(void)
+{
+	if (!checkcodesection())
+		return;
+
+	if (currentLoadSection)
+		fatalerror("`LOAD` blocks cannot contain inline fragments\n");
+
+	struct Section *sect = currentSection;
+
+	// SECTION UNION (RAM-only) is incompatible with SECTION FRAGMENT (ROM-only)
+	if (sect->modifier == SECTION_UNION)
+		fatalerror("`SECTION UNION` cannot contain inline fragments\n");
+
+	// A section containing an inline fragment has to become a fragment too
+	sect->modifier = SECTION_FRAGMENT;
+
+	sect_PushSection();
+
+	// `SECTION "...", ROM0, BANK[0]` is not allowed
+	uint32_t bank = sect->bank == 0 ? -1 : sect->bank;
+
+	struct Section *newSect = createSection(sect->name, sect->type, -1, bank, 0, 0,
+						SECTION_FRAGMENT);
+
+	// Add the new section fragment to the list (after the section containing it)
+	newSect->next = sect->next;
+	sect->next = newSect;
+
+	changeSection();
+	curOffset = newSect->size;
+	currentSection = newSect;
+}
+
 struct Section *sect_GetSymbolSection(void)
 {
 	return currentLoadSection ? currentLoadSection : currentSection;

test/asm/code-after-endm-endr-endc.err

@@ -1,15 +1,15 @@
 error: code-after-endm-endr-endc.asm(6):
-    syntax error, unexpected PRINTLN, expecting newline or end of buffer
+    syntax error, unexpected PRINTLN, expecting newline or end of line or end of buffer
 error: code-after-endm-endr-endc.asm(7):
     Macro "mac" not defined
 error: code-after-endm-endr-endc.asm(12):
-    syntax error, unexpected PRINTLN, expecting newline or end of buffer
+    syntax error, unexpected PRINTLN, expecting newline or end of line or end of buffer
 error: code-after-endm-endr-endc.asm(17):
     syntax error, unexpected PRINTLN, expecting newline
 error: code-after-endm-endr-endc.asm(19):
-    syntax error, unexpected PRINTLN, expecting newline or end of buffer
+    syntax error, unexpected PRINTLN, expecting newline or end of line or end of buffer
 error: code-after-endm-endr-endc.asm(23):
     syntax error, unexpected PRINTLN, expecting newline
 error: code-after-endm-endr-endc.asm(25):
-    syntax error, unexpected PRINTLN, expecting newline or end of buffer
+    syntax error, unexpected PRINTLN, expecting newline or end of line or end of buffer
 error: Assembly aborted (7 errors)!

test/asm/inline-fragment-in-load.asm

@@ -0,0 +1,14 @@
+SECTION "OAMDMACode", ROM0
+OAMDMACode:
+LOAD "hOAMDMA", HRAM
+hOAMDMA::
+	ldh [$ff46], a
+	ld a, 40
+	jp [[
+:		dec a
+		jr nz, :-
+		ret
+	]]
+.end
+ENDL
+OAMDMACodeEnd:

test/asm/inline-fragment-in-load.err

@@ -0,0 +1,2 @@
+FATAL: inline-fragment-in-load.asm(7):
+    `LOAD` blocks cannot contain inline fragments

test/asm/inline-fragment-in-ram.asm

@@ -0,0 +1,9 @@
+SECTION "RAM", WRAM0
+
+wFoo:: db
+wBar:: ds 3
+	println "ok"
+wQux:: dw [[
+	ds 4
+	println "inline"
+]]

test/asm/inline-fragment-in-ram.err

@@ -0,0 +1,4 @@
+ERROR: inline-fragment-in-ram.asm(6):
+    Section 'RAM' cannot contain code or data (not ROM0 or ROMX)
+FATAL: inline-fragment-in-ram.asm(9):
+    No entries in the section stack

test/asm/inline-fragment-in-ram.out

@@ -0,0 +1,2 @@
+ok
+inline

test/asm/inline-fragment-in-union.asm

@@ -0,0 +1,5 @@
+SECTION UNION "U", ROM0
+	db $11
+	dw [[ db $22 ]]
+SECTION UNION "U", ROM0
+	db $33

test/asm/inline-fragment-in-union.err

@@ -0,0 +1,2 @@
+FATAL: inline-fragment-in-union.asm(3):
+    `SECTION UNION` cannot contain inline fragments

test/asm/inline-fragments.asm

@@ -0,0 +1,58 @@
+SECTION "1", ROM0[0]
+
+VERSION EQU $11
+GetVersion::
+	ld a, [ [[db VERSION]] ]
+	ret
+
+SECTION "2", ROM0, ALIGN[4]
+
+text: MACRO
+	db \1, 0
+ENDM
+
+text_pointer: MACRO
+	dw [[ text \1 ]]
+ENDM
+
+GetText::
+	ld hl, [[
+		dw [[ db "Alpha", 0 ]]
+		dw [[ text "Beta" ]]
+		text_pointer "Gamma"
+		dw 0
+	]]
+	ld c, a
+	ld b, 0
+	add hl, bc
+	add hl, bc
+	ld a, [hli]
+	ld h, [hl]
+	ld l, a
+	ret
+
+SECTION "C", ROM0
+
+Foo::
+	call [[ jp [[ jp [[ ret ]] ]] ]]
+	call [[
+Label::
+		call GetVersion
+MYTEXT EQU 3
+		ld a, MYTEXT
+		call GetText
+		ld b, h
+		ld c, l
+		ret
+	]]
+	jp [[
+Bar:
+		inc hl
+.loop
+		halt
+:		dec l
+		jr nz, :-
+		dec h
+		jr nz, .loop
+		ret
+	]]