cc1: Add correct handling of universal character names - scc - simple c99 compiler
  HTML git clone git://git.simple-cc.org/scc
   DIR Log
   DIR Files
   DIR Refs
   DIR Submodules
   DIR README
   DIR LICENSE
       ---
   DIR commit f29cc7e4e4ff9756684411fadfca18f3ff6556d2
   DIR parent 3103a4ad89edc64ed22bd66de2a49b15a7e43d1b
  HTML Author: Roberto E. Vargas Caballero <k0ga@shike2.net>
       Date:   Tue,  7 Apr 2026 09:22:21 +0200
       
       cc1: Add correct handling of universal character names
       
       This commits adds the support only in character constants, because
       we think that adding the support for identifiers is worthless and
       it makes more sense to support (or accept in a careless way) utf8
       identifiers. The support for universal character constants in
       strings is coming.
       
       Diffstat:
         M src/cmd/scc-cc/cc1/lex.c            |      56 +++++++++++++++++++++++--------
         A tests/cc/execute/0259-multichar.c   |      15 +++++++++++++++
         M tests/cc/execute/scc-tests.lst      |       1 +
       
       3 files changed, 58 insertions(+), 14 deletions(-)
       ---
   DIR diff --git a/src/cmd/scc-cc/cc1/lex.c b/src/cmd/scc-cc/cc1/lex.c
       @@ -11,6 +11,8 @@
        #include <scc/scc.h>
        #include "cc1.h"
        
       +#define REPLACECHAR 0xFFFD
       +
        int yytoken;
        struct yystype yylval;
        char yytext[STRINGSIZ+3];
       @@ -516,10 +518,11 @@ number(void)
                return CONSTANT;
        }
        
       -static int
       -escape(void)
       +static Rune
       +escape(int multi)
        {
       -        int c, d, i, cnt, base;
       +        Rune c;
       +        int uni, d, i, cnt, base;
        
                switch (*++input->p) {
                case 'a':
       @@ -544,17 +547,21 @@ escape(void)
                        return '\\';
                case '\?':
                        return '\?';
       +        case 'U':
       +                cnt = 8;
       +                goto check_universal;
                case 'u':
       -                /*
       -                 * FIXME: universal constants are not correctly handled
       -                 */
       -                if (!isdigit(*++input->p))
       -                        warn("incorrect digit for numerical character constant");
       -                base = 10;
       +                cnt = 4;
       +        check_universal:
       +                if (!multi)
       +                        warn("multi-character character constant");
       +                ++input->p;
       +                uni = 1;
       +                base = 16;
                        break;
                case 'x':
       -                if (!isxdigit(*++input->p))
       -                        warn("\\x used with no following hex digits");
       +                ++input->p;
       +                uni = 0;
                        cnt = 2;
                        base = 16;
                        break;
       @@ -566,6 +573,7 @@ escape(void)
                case '5':
                case '6':
                case '7':
       +                uni = 0;
                        cnt = 3;
                        base = 8;
                        break;
       @@ -574,7 +582,7 @@ escape(void)
                        return ' ';
                }
        
       -        for (c = i = 0; i < cnt; ++i) {
       +        for (c = i = 0; i < cnt && isxdigit(*input->p); ++i) {
                        static char digits[] = "0123456789ABCDEF";
                        char *p = strchr(digits, toupper(*input->p));
        
       @@ -586,6 +594,26 @@ escape(void)
                }
                --input->p;
        
       +        if (base == 16 && i != cnt) {
       +                if (uni) {
       +                        warn("incorrect digit for universal character constant");
       +                        c = REPLACECHAR;
       +                } else {
       +                        warn("\\x used with no following hex digits");
       +                        c = ' ';
       +                }
       +        }
       +
       +        if (!uni)
       +                return c;
       +
       +        if (c < 0xa0 && (c == 0x24 || c == 0x40 || c == 0x60)
       +        || c >= 0xD800 && c <= 0xDFFF
       +        || c >= 0x110000) {
       +                warn("invalid universal character constant");
       +                c = REPLACECHAR;
       +        }
       +
                return c;
        }
        
       @@ -626,7 +654,7 @@ decode(int multi)
                Rune r;
        
                if (*input->p == '\\') {
       -                r = escape();
       +                r = escape(multi);
                        return r;
                }
        
       @@ -713,7 +741,7 @@ string(void)
                        esc = (c == '\\' && !esc && disescape);
        
                        if (c == '\\' && !esc)
       -                        c = escape();
       +                        c = escape(0);
        
                        if (bp == &yytext[STRINGSIZ+1]) {
                                /* too long, ignore everything until next quote */
   DIR diff --git a/tests/cc/execute/0259-multichar.c b/tests/cc/execute/0259-multichar.c
       @@ -0,0 +1,15 @@
       +
       +int
       +main(void)
       +{
       +        int a = L'\u00a1';
       +
       +        if (a != 0xa1)
       +                return 1;
       +        if (L'\u00a1' != 0xa1)
       +                return 2;
       +        if (L'\U000000b1' != 0xb1)
       +                return 3;
       +
       +        return 0;
       +}
   DIR diff --git a/tests/cc/execute/scc-tests.lst b/tests/cc/execute/scc-tests.lst
       @@ -249,3 +249,4 @@
        0256-ary.c
        0257-main.c
        0258-variadic.c
       +0259-multichar.c