%top{
/*
* Scanner for C declarations.
- * Copyright © 2011 Nick Bowler
+ * Copyright © 2011, 2021, 2023-2024 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
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
- #include "parse.h"
+#include <config.h>
+#include "parse.h"
+
+/* Disable various generated code we don't use */
+#define YY_INPUT(a, b, c) do {} while (0)
+#define YY_NO_INPUT 1
+#define YY_NO_UNPUT 1
+}
+
+%option nodefault noyywrap bison-locations reentrant never-interactive
+%option extra-type="int"
+%option prefix="cdecl__yy"
+
+%{
+#include <ctype.h>
+#include "cdecl-internal.h"
+#include "cdecl.h"
+#include "errmsg.h"
+#include "intconv.h"
+
+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;
}
-%option noyywrap bison-locations
+/*
+ * 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;
+}
+
+%}
+
+IDENT [_[:alpha:]][-_[:alnum:]]*
%%
+%{
+ int intconv_base;
+ char *c;
+%}
+
+"..."|[][;*(),] {
+ unsigned char *match;
+ static const unsigned char tab[2][8] = {
+ "*[](),.;",
+ {
+ PACK_TOKEN(T_ASTERISK),
+ PACK_TOKEN(T_LBRACKET),
+ PACK_TOKEN(T_RBRACKET),
+ PACK_TOKEN(T_LPAREN),
+ PACK_TOKEN(T_RPAREN),
+ PACK_TOKEN(T_COMMA),
+ PACK_TOKEN(T_ELLIPSIS),
+ PACK_TOKEN(T_SEMICOLON)
+ }
+ };
+
+ match = memchr(&tab, yytext[0], sizeof tab[0]);
+ return UNPACK_TOKEN(match[sizeof tab[0]]);
+}
+
+0[0-7]* { intconv_base = INTCONV_OCTAL; goto int_parse; }
+[1-9][0-9]* { intconv_base = INTCONV_DECIMAL; goto int_parse; }
+0[Xx][[:xdigit:]]+ {
+ unsigned char d;
+ uintmax_t v;
+
+ yytext += 2;
+ intconv_base = INTCONV_HEXADECIMAL;
+int_parse:
+ for (v = 0; (d = *yytext++);) {
+ if (!intconv_shift(&v, intconv_base, intconv_digit(d))) {
+ cdecl__errmsg(CDECL__ERANGE);
+ return T_LEX_ERROR;
+ }
+ }
+
+ yylval->uintval = v;
+ return T_UINT;
+}
+0[Xx]|[0-9]+ {
+ cdecl__errmsg(CDECL__EBADINT);
+ return T_LEX_ERROR;
+}
+
+{IDENT} {
+ int len = yyleng, tok;
+ unsigned x;
+
+ x = cdecl__to_keyword(yytext, len, yyextra);
+ yylval->spectype = UNPACK_SPEC(x & 0xff);
+ if ((tok = (x >> 8)) == PACK_TOKEN(T_IDENT)) {
+ /*
+ * Our IDENT pattern includes hyphens so we can match
+ * "variable-length" as a keyword. In all other cases a
+ * hyphen is an error.
+ *
+ * We could use yyless to re-scan the hyphen and hit the
+ * error catch-all, but jumping straight to the error code
+ * seems to produce better results with gcc with no obvious
+ * downsides.
+ */
+#if 1
+ if ((c = memchr(yytext, '-', len)))
+ goto invalid_char;
+#else
+ yyless(strcspn(yytext, "-"));
+#endif
+ if (!(yylval->item = cdecl__alloc_item(len+1)))
+ return T_LEX_ERROR;
+ memcpy(yylval->item->s, yytext, len+1);
+ }
+ return UNPACK_TOKEN(tok);
+}
+
+[[:space:]]+
+. {
+ char buf[8];
+
+ c = yytext;
+invalid_char:
+ to_readable_ch(buf, *c);
+ cdecl__err(CDECL_ENOPARSE, _("syntax error, unexpected %s"), buf);
+ return T_LEX_ERROR;
+}