X-Git-Url: https://git.draconx.ca/gitweb/cdecl99.git/blobdiff_plain/363f15bd4b53bd90b86610b143695762037ae706..ffc2c34e83af80b4e1cbcbbbbdb7693a338f43a6:/src/scan.l

diff --git a/src/scan.l b/src/scan.l
index 75657fb..0a4db93 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
@@ -17,23 +17,120 @@
  *  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 reentrant
+%option nodefault noyywrap bison-locations reentrant never-interactive
+%option extra-type="_Bool"
+%option prefix="cdecl__yy"
 
 %{
-#define lex_error(msg) do { \
-	yyerror(yylloc, NULL, NULL, (msg)); \
-	return T_LEX_ERROR; \
+#include <ctype.h>
+#include "cdecl-internal.h"
+#include "cdecl.h"
+#include "errmsg.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 dup_token() do { \
+	yylval->strval = malloc(yyleng+1); \
+	if (!yylval->strval) { \
+		cdecl__errmsg(CDECL__ENOMEM); \
+		return T_LEX_ERROR; \
+	} \
+	strcpy(yylval->strval, yytext); \
 } while(0)
+
+static char *to_octal(char *dst, unsigned val)
+{
+	unsigned i;
+
+	for (i = 0; i < 3; i++) {
+		*dst++ = '0' + ((val >> 6) & 7u);
+		val <<= 3;
+	}
+
+	return dst;
+}
+
+/*
+ * Convert a single character to a C-style character constant, including quote
+ * characters.  At most 7 bytes are written to the buffer for the longest
+ * octal encoding, e.g., '\177'
+ */
+static void to_readable_ch(char *dst, char c)
+{
+	unsigned char uc = c;
+	unsigned i;
+	char esc;
+
+	/*
+	 * The 7 standard C control characters are contiguous in ASCII,
+	 * permitting a simple and compact lookup table; separating their
+	 * handling from backslash and quote characters hopefully allows
+	 * the compiler to recognize that.
+	 */
+	switch (c) {
+	case '\a': i = 0; break;
+	case '\b': i = 1; break;
+	case '\t': i = 2; break;
+	case '\n': i = 3; break;
+	case '\v': i = 4; break;
+	case '\f': i = 5; break;
+	case '\r': i = 6; break;
+	default:   i = 7; break;
+	}
+	esc = "abtnvfr"[i];
+
+	/* Otherwise printable characters that should still be escaped. */
+	switch (c) {
+	case '\\': case '\'': esc = c; break;
+	}
+
+	*dst++ = '\'';
+	if (esc) {
+		*dst++ = '\\';
+		*dst++ = esc;
+	} else if (isprint(uc)) {
+		*dst++ = c;
+	} else {
+		*dst++ = '\\';
+		dst = to_octal(dst, uc);
+	}
+	*dst++ = '\'';
+	*dst++ = 0;
+}
+
 %}
 
+%s ENGLISH
+
 IDENT [_[:alpha:]][_[:alnum:]]*
 INTEGER 0x[[:xdigit:]]+|0[0-7]+|[[:digit:]]+
 
 %%
 
+%{
+	if (yyextra) {
+		yyextra = 0;
+		BEGIN(ENGLISH);
+		return T_ENGLISH;
+	}
+%}
+
 "..." return T_ELLIPSIS;
 ";"   return T_SEMICOLON;
 "*"   return T_ASTERISK;
@@ -76,27 +173,39 @@ INTEGER 0x[[:xdigit:]]+|0[0-7]+|[[:digit:]]+
 	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");
+	yylval->uintval = STRTOUMAX(yytext, &end, 0);
+	if (errno == ERANGE) {
+		cdecl__errmsg(CDECL__ERANGE);
+		return T_LEX_ERROR;
+	}
+	if (*end) {
+		cdecl__errmsg(CDECL__EBADINT);
+		return T_LEX_ERROR;
+	}
 
 	return T_UINT;
 }
 
-{IDENT} {
-	yylval->strval = malloc(yyleng+1);
-	if (!yylval->strval)
-		lex_error("failed to allocate memory");
-
-	strcpy(yylval->strval, yytext);
-	return T_IDENT;
+<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[8];
+
+	to_readable_ch(buf, yytext[0]);
+	cdecl__err(CDECL_ENOPARSE, _("syntax error, unexpected %s"), buf);
+	return T_LEX_ERROR;
 }