X-Git-Url: https://git.draconx.ca/gitweb/cdecl99.git/blobdiff_plain/802e3b1cc242d149fc23cc65f1714c99c1f8520a..c733aff6d0b2b107c88e37a1491f270db0634819:/src/scan.l

diff --git a/src/scan.l b/src/scan.l
index 1688873..0a4db93 100644
--- a/src/scan.l
+++ b/src/scan.l
@@ -29,6 +29,7 @@
 #include <ctype.h>
 #include "cdecl-internal.h"
 #include "cdecl.h"
+#include "errmsg.h"
 
 #if HAVE_STRTOUMAX
 /* Best case, implementation provides strtoumax. */
@@ -44,19 +45,75 @@
 #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); \
+		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
@@ -117,10 +174,14 @@ INTEGER 0x[[:xdigit:]]+|0[0-7]+|[[:digit:]]+
 
 	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"));
+	if (errno == ERANGE) {
+		cdecl__errmsg(CDECL__ERANGE);
+		return T_LEX_ERROR;
+	}
+	if (*end) {
+		cdecl__errmsg(CDECL__EBADINT);
+		return T_LEX_ERROR;
+	}
 
 	return T_UINT;
 }
@@ -142,28 +203,9 @@ INTEGER 0x[[:xdigit:]]+|0[0-7]+|[[:digit:]]+
 
 [[:space:]]+
 . {
-	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);
-		}
-	}
+	char buf[8];
 
-	lex_error(_("syntax error, unexpected '%s'"), buf);
+	to_readable_ch(buf, yytext[0]);
+	cdecl__err(CDECL_ENOPARSE, _("syntax error, unexpected %s"), buf);
+	return T_LEX_ERROR;
 }