[WIP] Allow longer strings, not '\0'-terminated

Still to port from ISSOtm's strings branch:
- lexer.c
- parser.y
- symbol.c

Fixes gbdev#650
Fixes gbdev#505

Makefile

@@ -66,6 +66,7 @@ rgbasm_obj := \
 	src/asm/parser.o \
 	src/asm/rpn.o \
 	src/asm/section.o \
+	src/asm/string.o \
 	src/asm/symbol.o \
 	src/asm/util.o \
 	src/asm/warning.o \

include/asm/format.h

@@ -35,10 +35,12 @@ struct FormatSpec {
 	bool valid;
 };
 
+struct String;
+
 struct StrFmtArg {
 	union {
 		uint32_t number;
-		char *string;
+		struct String *string;
 	};
 	bool isNumeric;
 };
@@ -57,7 +59,7 @@ bool fmt_IsValid(struct FormatSpec const *fmt);
 bool fmt_IsFinished(struct FormatSpec const *fmt);
 void fmt_UseCharacter(struct FormatSpec *fmt, int c);
 void fmt_FinishCharacters(struct FormatSpec *fmt);
-void fmt_PrintString(char *buf, size_t bufLen, struct FormatSpec const *fmt, char const *value);
-void fmt_PrintNumber(char *buf, size_t bufLen, struct FormatSpec const *fmt, uint32_t value);
+struct String *fmt_PrintString(struct FormatSpec const *fmt, struct String const *value);
+struct String *fmt_PrintNumber(struct FormatSpec const *fmt, uint32_t value);
 
 #endif /* RGBDS_FORMAT_SPEC_H */

include/asm/fstack.h

@@ -52,6 +52,7 @@ struct FileStackNamedNode { /* NODE_FILE, NODE_MACRO */
 extern size_t maxRecursionDepth;
 
 struct MacroArgs;
+struct String;
 
 void fstk_Dump(struct FileStackNode const *node, uint32_t lineNo);
 void fstk_DumpCurrent(void);
@@ -71,10 +72,10 @@ bool fstk_FindFile(char const *path, char **fullPath, size_t *size);
 
 bool yywrap(void);
 void fstk_RunInclude(char const *path);
-void fstk_RunMacro(char const *macroName, struct MacroArgs *args);
+void fstk_RunMacro(struct String const *macroName, struct MacroArgs *args);
 void fstk_RunRept(uint32_t count, int32_t reptLineNo, char *body, size_t size);
-void fstk_RunFor(char const *symName, int32_t start, int32_t stop, int32_t step,
-		     int32_t reptLineNo, char *body, size_t size);
+void fstk_RunFor(struct String const *symName, int32_t start, int32_t stop, int32_t step,
+		 int32_t reptLineNo, char *body, size_t size);
 void fstk_StopRept(void);
 bool fstk_Break(void);
 

include/asm/lexer.h

@@ -11,8 +11,6 @@
 
 #include <stdbool.h>
 
-#define MAXSTRLEN	255
-
 struct LexerState;
 extern struct LexerState *lexerState;
 extern struct LexerState *lexerStateEOL;

include/asm/section.h

@@ -17,9 +17,10 @@
 extern uint8_t fillByte;
 
 struct Expression;
+struct String;
 
 struct Section {
-	char *name;
+	struct String *name;
 	enum SectionType type;
 	enum SectionModifier modifier;
 	struct FileStackNode *src; /* Where the section was defined */

include/asm/string.h

@@ -0,0 +1,53 @@
+/*
+ * This file is part of RGBDS.
+ *
+ * Copyright (c) 2021, Eldred Habert and RGBDS contributors.
+ *
+ * SPDX-License-Identifier: MIT
+ */
+
+#ifndef RGBDS_STRING_H
+#define RGBDS_STRING_H
+
+#include <assert.h>
+#include <stdbool.h>
+#include <stdint.h>
+
+#include "helpers.h"
+
+struct String;
+
+#define PRI_STR ".*s"
+// WARNING: **DO NOT** pass any side-effecting parameters to the macros below!
+#define STR_FMT(str) (int)str_Len(str), str_Chars(str)
+
+#define MUTATE_STR(str, ...) do { \
+	struct String *___orig_str = str; \
+	str = __VA_ARGS__; \
+	assert(___orig_str == str); /* This shouldn't have been reallocated */ \
+} while (0)
+
+static inline bool str_IsWhitespace(int c)
+{
+	return c == ' ' || c == '\t';
+}
+
+size_t str_Len(struct String const *str) pure_;
+void str_Trunc(struct String *str, size_t len);
+char str_Index(struct String const *str, size_t i) pure_;
+bool str_Find(struct String const *str, char c) pure_;
+char const *str_Chars(struct String const *str) pure_;
+
+/**
+ * @param capacity The capacity to use, or 0 if unknown
+ */
+struct String *str_New(size_t capacity) mallocish_;
+void str_Ref(struct String *str);
+void str_Unref(struct String *str);
+
+struct String *str_Push(struct String *str, char c) warn_unused_result_;
+struct String *str_Append(struct String *lhs, struct String const *rhs) warn_unused_result_;
+struct String *str_AppendSlice(struct String *lhs, char const *rhs, size_t len) warn_unused_result_;
+void str_TrimEnd(struct String *str);
+
+#endif

include/asm/symbol.h

@@ -15,12 +15,12 @@
 #include <time.h>
 
 #include "asm/section.h"
+#include "asm/string.h"
 
 #include "platform.h" // MIN_NB_ELMS
 #include "types.h"
 
 #define HASHSIZE	(1 << 16)
-#define MAXSYMLEN	256
 
 enum SymbolType {
 	SYM_LABEL,
@@ -32,7 +32,7 @@ enum SymbolType {
 };
 
 struct Symbol {
-	char name[MAXSYMLEN + 1];
+	struct String *name;
 	enum SymbolType type;
 	bool isExported; /* Whether the symbol is to be exported */
 	bool isBuiltin;  /* Whether the symbol is a built-in */
@@ -45,13 +45,14 @@ struct Symbol {
 		/* If sym_IsNumeric */
 		int32_t value;
 		int32_t (*numCallback)(void);
-		/* For SYM_MACRO and SYM_EQUS; TODO: have separate fields */
+		/* For SYM_MACRO */
 		struct {
 			size_t macroSize;
 			char *macro;
 		};
 		/* For SYM_EQUS */
-		char const *(*strCallback)(void);
+		struct String *str;
+		struct String *(*strCallback)(void);
 	};
 
 	uint32_t ID; /* ID of the symbol in the object file (-1 if none) */
@@ -92,7 +93,7 @@ static inline bool sym_IsLabel(struct Symbol const *sym)
 
 static inline bool sym_IsLocal(struct Symbol const *sym)
 {
-	return sym_IsLabel(sym) && strchr(sym->name, '.');
+	return sym_IsLabel(sym) && str_Find(sym->name, '.');
 }
 
 static inline bool sym_IsExported(struct Symbol const *sym)
@@ -103,49 +104,49 @@ static inline bool sym_IsExported(struct Symbol const *sym)
 /*
  * Get a string equate's value
  */
-static inline char const *sym_GetStringValue(struct Symbol const *sym)
+static inline struct String *sym_GetStringValue(struct Symbol const *sym)
 {
 	if (sym->hasCallback)
 		return sym->strCallback();
-	return sym->macro;
+	return sym->str;
 }
 
 void sym_ForEach(void (*func)(struct Symbol *, void *), void *arg);
 
 int32_t sym_GetValue(struct Symbol const *sym);
 void sym_SetExportAll(bool set);
-struct Symbol *sym_AddLocalLabel(char const *symName);
-struct Symbol *sym_AddLabel(char const *symName);
+struct Symbol *sym_AddLocalLabel(struct String *symName);
+struct Symbol *sym_AddLabel(struct String *symName);
 struct Symbol *sym_AddAnonLabel(void);
-void sym_WriteAnonLabelName(char buf[MIN_NB_ELMS(MAXSYMLEN + 1)], uint32_t ofs, bool neg);
-void sym_Export(char const *symName);
-struct Symbol *sym_AddEqu(char const *symName, int32_t value);
-struct Symbol *sym_AddSet(char const *symName, int32_t value);
+struct String *sym_WriteAnonLabelName(uint32_t ofs, bool neg);
+void sym_Export(struct String const *symName);
+struct Symbol *sym_AddEqu(struct String const *symName, int32_t value);
+struct Symbol *sym_AddSet(struct String const *symName, int32_t value);
 uint32_t sym_GetPCValue(void);
 uint32_t sym_GetConstantSymValue(struct Symbol const *sym);
-uint32_t sym_GetConstantValue(char const *symName);
+uint32_t sym_GetConstantValue(struct String const *symName);
 /*
  * Find a symbol by exact name, bypassing expansion checks
  */
-struct Symbol *sym_FindExactSymbol(char const *symName);
+struct Symbol *sym_FindExactSymbol(struct String const *symName);
 /*
  * Find a symbol by exact name; may not be scoped, produces an error if it is
  */
-struct Symbol *sym_FindUnscopedSymbol(char const *symName);
+struct Symbol *sym_FindUnscopedSymbol(struct String const *symName);
 /*
  * Find a symbol, possibly scoped, by name
  */
-struct Symbol *sym_FindScopedSymbol(char const *symName);
+struct Symbol *sym_FindScopedSymbol(struct String const *symName);
 struct Symbol const *sym_GetPC(void);
-struct Symbol *sym_AddMacro(char const *symName, int32_t defLineNo, char *body, size_t size);
-struct Symbol *sym_Ref(char const *symName);
-struct Symbol *sym_AddString(char const *symName, char const *value);
-struct Symbol *sym_RedefString(char const *symName, char const *value);
-void sym_Purge(char const *symName);
+struct Symbol *sym_AddMacro(struct String *symName, int32_t defLineNo, char *body, size_t size);
+struct Symbol *sym_Ref(struct String *symName);
+struct Symbol *sym_AddString(struct String *symName, struct String *value);
+struct Symbol *sym_RedefString(struct String *symName, struct String *value);
+void sym_Purge(struct String const *symName);
 void sym_Init(time_t now);
 
 /* Functions to save and restore the current symbol scope. */
-char const *sym_GetCurrentSymbolScope(void);
-void sym_SetCurrentSymbolScope(char const *newScope);
+struct String const *sym_GetCurrentSymbolScope(void);
+void sym_SetCurrentSymbolScope(struct String const *newScope);
 
 #endif /* RGBDS_SYMBOL_H */

include/asm/warning.h

@@ -30,7 +30,6 @@ enum WarningID {
 	WARNING_EMPTY_MACRO_ARG,      /* Empty macro argument */
 	WARNING_EMPTY_STRRPL,	      /* Empty second argument in `STRRPL` */
 	WARNING_LARGE_CONSTANT,	      /* Constants too large */
-	WARNING_LONG_STR,	      /* String too long for internal buffers */
 	WARNING_MACRO_SHIFT,	      /* Shift past available arguments in macro */
 	WARNING_NESTED_COMMENT,	      /* Comment-start delimiter in a block comment */
 	WARNING_OBSOLETE,	      /* Obsolete things */

include/helpers.h

@@ -18,6 +18,12 @@
 #ifdef __GNUC__ // GCC or compatible
 	#define format_(archetype, str_index, first_arg) \
 		__attribute__ ((format (archetype, str_index, first_arg)))
+	// Notes that a function allocates something
+	#define mallocish_ __attribute__ ((malloc))
+	// Notes that a function does not have any side effects (more or less)
+	#define pure_ __attribute__ ((pure))
+	// Notes that a function's return value should not be ignored
+	#define warn_unused_result_ __attribute__ ((warn_unused_result))
 	// In release builds, define "unreachable" as such, but trap in debug builds
 	#ifdef NDEBUG
 		#define unreachable_	__builtin_unreachable
@@ -27,6 +33,9 @@
 #else
 	// Unsupported, but no need to throw a fit
 	#define format_(archetype, str_index, first_arg)
+	#define mallocish_
+	#define pure_
+	#define warn_unused_result_
 	// This seems to generate similar code to __builtin_unreachable, despite different semantics
 	// Note that executing this is undefined behavior (declared _Noreturn, but does return)
 	static inline _Noreturn unreachable_(void) {}

src/CMakeLists.txt

@@ -54,6 +54,7 @@ set(rgbasm_src
     "asm/output.c"
     "asm/rpn.c"
     "asm/section.c"
+    "asm/string.c"
     "asm/symbol.c"
     "asm/util.c"
     "asm/warning.c"