X-Git-Url: https://git.draconx.ca/gitweb/cdecl99.git/blobdiff_plain/f5e2e659f116a49afd3b9d9f37553646e62fa4d4..9abf0eb2e74d0fb3a4f652499571cb35697d3036:/src/scan.l

diff --git a/src/scan.l b/src/scan.l
index f2e92a8..1688873 100644
--- a/src/scan.l
+++ b/src/scan.l
@@ -1,7 +1,7 @@
 %top{
 /*
  *  Scanner for C declarations.
- *  Copyright © 2011 Nick Bowler
+ *  Copyright © 2011, 2021, 2023 Nick Bowler
  *
  *  This program is free software: you can redistribute it and/or modify
  *  it under the terms of the GNU General Public License as published by
@@ -17,70 +17,153 @@
  *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
- #include "parse.h"
+#include <config.h>
+#include "parse.h"
 }
 
-%option noyywrap bison-locations
+%option nodefault noyywrap bison-locations reentrant never-interactive
+%option extra-type="_Bool"
+%option prefix="cdecl__yy"
 
 %{
-#define lex_error(msg) do { \
-	yyerror(yylloc, NULL, (msg)); \
+#include <ctype.h>
+#include "cdecl-internal.h"
+#include "cdecl.h"
+
+#if HAVE_STRTOUMAX
+/* Best case, implementation provides strtoumax. */
+#  define STRTOUMAX strtoumax
+#elif HAVE_STRTOULL
+/* Fall back to strtoull, with possibly reduced range. */
+#define STRTOUMAX strtoull
+#elif HAVE___STRTOULL
+/* HP-UX 11 has __strtoull in <inttypes.h> */
+#define STRTOUMAX __strtoull
+#else
+/* Fall back to strtoul, with possibly reduced range. */
+#define STRTOUMAX strtoul
+#endif
+
+#define lex_error(...) do { \
+	cdecl__err(CDECL_ENOPARSE, __VA_ARGS__); \
 	return T_LEX_ERROR; \
 } while(0)
+
+#define dup_token() do { \
+	yylval->strval = malloc(yyleng+1); \
+	if (!yylval->strval) { \
+		cdecl__err(CDECL_ENOMEM); \
+		return T_LEX_ERROR; \
+	} \
+	strcpy(yylval->strval, yytext); \
+} while(0)
 %}
 
+%s ENGLISH
+
 IDENT [_[:alpha:]][_[:alnum:]]*
+INTEGER 0x[[:xdigit:]]+|0[0-7]+|[[:digit:]]+
 
 %%
 
-";" return T_SEMICOLON;
-"*" return T_ASTERISK;
-"(" return T_LPAREN;
-")" return T_RPAREN;
-"[" return T_LBRACKET;
-"]" return T_RBRACKET;
-"," return T_COMMA;
-
-"typedef"  return T_TYPEDEF;
-"extern"   return T_EXTERN;
-"static"   return T_STATIC;
-"auto"     return T_AUTO;
-"register" return T_REGISTER;
-
-"restrict" return T_RESTRICT;
-"volatile" return T_VOLATILE;
-"const"    return T_CONST;
-
-"inline"   return T_INLINE;
-
-"void"     return T_VOID;
-"char"     return T_CHAR;
-"short"    return T_SHORT;
-"int"      return T_INT;
-"long"     return T_LONG;
-"float"    return T_FLOAT;
-"double"   return T_DOUBLE;
-"signed"   return T_SIGNED;
-"unsigned" return T_UNSIGNED;
-"_Bool"    return T_BOOL;
-"_Complex" return T_COMPLEX;
-
-"struct"   return T_STRUCT;
-"union"    return T_UNION;
-"enum"     return T_ENUM;
-
-{IDENT} {
-	yylval->strval = malloc(yyleng+1);
-	if (!yylval->strval)
-		lex_error("failed to allocate memory");
-
-	strcpy(yylval->strval, yytext);
-	return T_IDENT;
+%{
+	if (yyextra) {
+		yyextra = 0;
+		BEGIN(ENGLISH);
+		return T_ENGLISH;
+	}
+%}
+
+"..." return T_ELLIPSIS;
+";"   return T_SEMICOLON;
+"*"   return T_ASTERISK;
+"("   return T_LPAREN;
+")"   return T_RPAREN;
+"["   return T_LBRACKET;
+"]"   return T_RBRACKET;
+","   return T_COMMA;
+
+"typedef"    return T_TYPEDEF;
+"extern"     return T_EXTERN;
+"static"     return T_STATIC;
+"auto"       return T_AUTO;
+"register"   return T_REGISTER;
+
+"restrict"   return T_RESTRICT;
+"volatile"   return T_VOLATILE;
+"const"      return T_CONST;
+
+"inline"     return T_INLINE;
+
+"void"       return T_VOID;
+"char"       return T_CHAR;
+"short"      return T_SHORT;
+"int"        return T_INT;
+"long"       return T_LONG;
+"float"      return T_FLOAT;
+"double"     return T_DOUBLE;
+"signed"     return T_SIGNED;
+"unsigned"   return T_UNSIGNED;
+"_Bool"      return T_BOOL;
+"_Complex"   return T_COMPLEX;
+"_Imaginary" return T_IMAGINARY;
+
+"struct"     return T_STRUCT;
+"union"      return T_UNION;
+"enum"       return T_ENUM;
+
+{INTEGER} {
+	char *end;
+
+	errno = 0;
+	yylval->uintval = STRTOUMAX(yytext, &end, 0);
+	if (errno == ERANGE)
+		lex_error(_("integer constant out of range"));
+	if (*end)
+		lex_error(_("invalid integer constant"));
+
+	return T_UINT;
 }
 
+<ENGLISH>{
+	"variable-length" return T_VLA;
+	"type"            return T_TYPE;
+	"declare"         return T_DECLARE;
+	"pointer"         return T_POINTER;
+	"function"        return T_FUNCTION;
+	"returning"       return T_RETURNING;
+	"array"           return T_ARRAY;
+	"to"              return T_TO;
+	"of"              return T_OF;
+	"as"              return T_AS;
+}
+
+{IDENT} { dup_token(); return T_IDENT; }
+
 [[:space:]]+
 . {
-	char buf[] = "syntax error, unexpected #";
-	*strchr(buf, '#') = *yytext;
-	lex_error(buf);
+	char buf[5] = { yytext[0] };
+	unsigned char c = buf[0];
+
+	if (!isprint(c) || c == '\\' || c == '\'') {
+		/* Encode nonprinting characters with C-style escapes */
+		buf[0] = '\\';
+		switch (c) {
+		case '\a': buf[1] = 'a'; break;
+		case '\b': buf[1] = 'b'; break;
+		case '\f': buf[1] = 'f'; break;
+		case '\n': buf[1] = 'n'; break;
+		case '\r': buf[1] = 'r'; break;
+		case '\t': buf[1] = 't'; break;
+		case '\v': buf[1] = 'v'; break;
+		case '\\': buf[1] = '\\'; break;
+		case '\'': buf[1] = '\''; break;
+		default:
+			buf[1] = '0' + ((c >> 6) & 3);
+			buf[2] = '0' + ((c >> 3) & 7);
+			buf[3] = '0' + ((c >> 0) & 7);
+		}
+	}
+
+	lex_error(_("syntax error, unexpected '%s'"), buf);
 }