Reuse startsIdentifier and continuesIdentifier functions (#1695)

include/util.hpp

@@ -3,6 +3,9 @@
 #ifndef RGBDS_UTIL_HPP
 #define RGBDS_UTIL_HPP
 
+bool startsIdentifier(int c);
+bool continuesIdentifier(int c);
+
 char const *printChar(int c);
 
 #endif // RGBDS_UTIL_HPP

src/asm/lexer.cpp

@@ -136,9 +136,8 @@ struct CaseInsensitive {
 	}
 };
 
-// This map lists all RGBASM keywords which `yylex_NORMAL` lexes as identifiers
+// This map lists all RGBASM keywords which `yylex_NORMAL` lexes as identifiers.
-// (see `startsIdentifier` and `continuesIdentifier` below). All non-identifier
+// All non-identifier tokens are lexed separately.
-// tokens are lexed separately.
 static std::unordered_map<std::string, int, CaseInsensitive, CaseInsensitive> keywordDict = {
     {"ADC",           T_(SM83_ADC)         },
     {"ADD",           T_(SM83_ADD)         },
@@ -612,8 +611,6 @@ static bool isMacroChar(char c) {
 static int peek();
 static void shiftChar();
 static uint32_t readNumber(int radix, uint32_t baseValue);
-static bool startsIdentifier(int c);
-static bool continuesIdentifier(int c);
 
 static uint32_t readBracketedMacroArgNum() {
 	bool disableMacroArgs = lexerState->disableMacroArgs;
@@ -1215,15 +1212,6 @@ static uint32_t readGfxConstant() {
 
 // Functions to read identifiers and keywords
 
-static bool startsIdentifier(int c) {
-	// Anonymous labels internally start with '!'
-	return (c <= 'Z' && c >= 'A') || (c <= 'z' && c >= 'a') || c == '.' || c == '_';
-}
-
-static bool continuesIdentifier(int c) {
-	return startsIdentifier(c) || (c <= '9' && c >= '0') || c == '#' || c == '$' || c == '@';
-}
-
 static Token readIdentifier(char firstChar, bool raw) {
 	std::string identifier(1, firstChar);
 	int tokenType = firstChar == '.' ? T_(LOCAL) : T_(SYMBOL);

src/asm/symbol.cpp

@@ -2,6 +2,7 @@
 
 #include "asm/symbol.hpp"
 
+#include <algorithm>
 #include <inttypes.h>
 #include <stdio.h>
 #include <unordered_map>
@@ -9,6 +10,7 @@
 
 #include "error.hpp"
 #include "helpers.hpp" // assume
+#include "util.hpp"
 #include "version.hpp"
 
 #include "asm/fstack.hpp"
@@ -130,6 +132,11 @@ static void updateSymbolFilename(Symbol &sym) {
 	}
 }
 
+static bool isValidIdentifier(std::string const &s) {
+	return !s.empty() && startsIdentifier(s[0])
+	       && std::all_of(s.begin() + 1, s.end(), [](char c) { return continuesIdentifier(c); });
+}
+
 static void alreadyDefinedError(Symbol const &sym, char const *asType) {
 	if (sym.isBuiltin && !sym_FindScopedValidSymbol(sym.name)) {
 		// `DEF()` would return false, so we should not claim the symbol is already defined
@@ -142,11 +149,13 @@ static void alreadyDefinedError(Symbol const &sym, char const *asType) {
 		fputs(" at ", stderr);
 		dumpFilename(sym);
 		if (sym.type == SYM_EQUS) {
-			fprintf(
+			if (std::string const &contents = *sym.getEqus(); isValidIdentifier(contents)) {
-			    stderr,
+				fprintf(
-			    "    (should it be {interpolated} to define its contents \"%s\"?)\n",
+				    stderr,
-			    sym.getEqus()->c_str()
+				    "    (should it be {interpolated} to define its contents \"%s\"?)\n",
-			);
+				    contents.c_str()
+				);
+			}
 		}
 	}
 }

src/link/output.cpp

@@ -15,6 +15,7 @@
 #include "helpers.hpp"
 #include "linkdefs.hpp"
 #include "platform.hpp"
+#include "util.hpp"
 
 #include "link/main.hpp"
 #include "link/section.hpp"
@@ -260,29 +261,13 @@ static void writeROM() {
 	}
 }
 
-// Checks whether a symbol is legal for a sym file or map file.
-// Eliminates anonymous labels, which start with a '!'.
-static bool isLegalSymbol(Symbol const &sym) {
-	if (sym.name.empty()) {
-		return false;
-	}
-	char c = sym.name[0];
-	return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || c == '_';
-}
-
-// Checks whether this character is legal in a symbol's name in a sym file
-static bool isLegalForSymName(char c) {
-	return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '_'
-	       || c == '@' || c == '#' || c == '$' || c == '.';
-}
-
 // Prints a symbol's name to a file, assuming that the first character is legal.
 // Illegal characters are UTF-8-decoded (errors are replaced by U+FFFD) and emitted as '\u'/'\U'.
 static void printSymName(std::string const &name, FILE *file) {
 	for (char const *ptr = name.c_str(); *ptr != '\0';) {
 		char c = *ptr;
 
-		if (isLegalForSymName(c)) {
+		if (continuesIdentifier(c)) {
 			// Output legal ASCII characters as-is
 			putc(c, file);
 			++ptr;
@@ -369,7 +354,7 @@ static void writeSymBank(SortedSections const &bankSections, SectionType type, u
 	forEachSortedSection(sect, {
 		for (Symbol const *sym : sect->symbols) {
 			// Don't output symbols that begin with an illegal character
-			if (isLegalSymbol(*sym)) {
+			if (!sym->name.empty() && startsIdentifier(sym->name[0])) {
 				uint16_t addr = static_cast<uint16_t>(sym->label().offset + sect->org);
 				uint16_t parentAddr = addr;
 				if (auto pos = sym->name.find('.'); pos != std::string::npos) {
@@ -482,7 +467,7 @@ static void writeMapBank(SortedSections const &sectList, SectionType type, uint3
 			for (uint16_t org = sect->org; sect; sect = sect->nextu.get()) {
 				for (Symbol *sym : sect->symbols) {
 					// Don't output symbols that begin with an illegal character
-					if (isLegalSymbol(*sym)) {
+					if (!sym->name.empty() && startsIdentifier(sym->name[0])) {
 						// Space matches "\tSECTION: $xxxx ..."
 						fprintf(mapFile, "\t         $%04" PRIx32 " = ", sym->label().offset + org);
 						printSymName(sym->name, mapFile);

src/util.cpp

@@ -6,6 +6,15 @@
 #include <stdint.h>
 #include <stdio.h>
 
+bool startsIdentifier(int c) {
+	// This returns false for anonymous labels, which internally start with a '!'
+	return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || c == '.' || c == '_';
+}
+
+bool continuesIdentifier(int c) {
+	return startsIdentifier(c) || (c >= '0' && c <= '9') || c == '#' || c == '$' || c == '@';
+}
+
 char const *printChar(int c) {
 	// "'A'" + '\0': 4 bytes
 	// "'\\n'" + '\0': 5 bytes

test/asm/builtin-overwrite.err

@@ -32,34 +32,24 @@ error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(6):
     Built-in symbol '__ISO_8601_UTC__' cannot be purged
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(9):
     '__ISO_8601_UTC__' already defined at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(10):
     '__ISO_8601_UTC__' already defined at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(13):
     '__ISO_8601_UTC__' already defined as constant at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(14):
     '__ISO_8601_UTC__' already defined as constant at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(17):
     '__ISO_8601_UTC__' already defined at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(18):
     '__ISO_8601_UTC__' already defined at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(21):
     '__ISO_8601_UTC__' already defined as constant at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(22):
     '__ISO_8601_UTC__' already defined as constant at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(25):
     '__ISO_8601_UTC__' already defined as non-EQU at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(26):
     '__ISO_8601_UTC__' already defined as non-EQU at <builtin>
-    (should it be {interpolated} to define its contents ""1989-04-21T12:34:56Z""?)
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(29):
     Built-in symbol '__ISO_8601_UTC__' cannot be redefined
 error: builtin-overwrite.asm(37) -> builtin-overwrite.asm::tickle(30):