The R Project SVN R

Rev

Rev 63219 | Rev 64516 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 63219 Rev 63223
Line 778... Line 778...
778 
   R_alloc stack */
 778 
   R_alloc stack */
779 
const char *translateChar(SEXP x)
 779 
const char *translateChar(SEXP x)
780 
{
 780 
{
781 
    void * obj;
 781 
    void * obj;
782 
    const char *inbuf, *ans = CHAR(x);
 782 
    const char *inbuf, *ans = CHAR(x);
783 
    char *outbuf, *p;
 783 
    char *outbuf;
784 
    size_t inb, outb, res;
 784 
    size_t inb, outb, res;
785 
    cetype_t ienc = getCharCE(x);
 785 
    cetype_t ienc = getCharCE(x);
786 
    R_StringBuffer cbuff = {NULL, 0, MAXELTSIZE};
 786 
    R_StringBuffer cbuff = {NULL, 0, MAXELTSIZE};
787 
 
 787 
 
788 
    if(TYPEOF(x) != CHARSXP)
 788 
    if(TYPEOF(x) != CHARSXP)
Line 874... Line 874...
874 
	}
 874 
	}
875 
	goto next_char;
 875 
	goto next_char;
876 
    }
 876 
    }
877 
    *outbuf = '\0';
 877 
    *outbuf = '\0';
878 
    res = strlen(cbuff.data) + 1;
 878 
    res = strlen(cbuff.data) + 1;
879 
    p = R_alloc(res, 1);
 879 
    char *p = R_alloc(res, 1);
880 
    memcpy(p, cbuff.data, res);
 880 
    memcpy(p, cbuff.data, res);
881 
    R_FreeStringBuffer(&cbuff);
 881 
    R_FreeStringBuffer(&cbuff);
882 
    return p;
 882 
    return p;
883 
}
 883 
}
884 
 
 884 
 
- 
 
 885 
SEXP installTrChar(SEXP x)
- 
 
 886 
{
- 
 
 887 
    void * obj;
- 
 
 888 
    const char *inbuf, *ans = CHAR(x);
- 
 
 889 
    char *outbuf;
- 
 
 890 
    size_t inb, outb, res;
- 
 
 891 
    cetype_t ienc = getCharCE(x);
- 
 
 892 
    R_StringBuffer cbuff = {NULL, 0, MAXELTSIZE};
- 
 
 893 
 
- 
 
 894 
    if(TYPEOF(x) != CHARSXP)
- 
 
 895 
	error(_("'%s' must be called on a CHARSXP"), "translateChar");
- 
 
 896 
    if(x == NA_STRING || !(ENC_KNOWN(x))) return install(ans);
- 
 
 897 
    if(IS_BYTES(x))
- 
 
 898 
	error(_("translating strings with \"bytes\" encoding is not allowed"));
- 
 
 899 
    if(utf8locale && IS_UTF8(x)) return install(ans);
- 
 
 900 
    if(latin1locale && IS_LATIN1(x)) return install(ans);
- 
 
 901 
    if(IS_ASCII(x)) return install(ans);
- 
 
 902 
 
- 
 
 903 
    if(IS_LATIN1(x)) {
- 
 
 904 
	if(!latin1_obj) {
- 
 
 905 
	    obj = Riconv_open("", "latin1");
- 
 
 906 
	    /* should never happen */
- 
 
 907 
	    if(obj == (void *)(-1))
- 
 
 908 
#ifdef Win32
- 
 
 909 
		error(_("unsupported conversion from '%s' in codepage %d"),
- 
 
 910 
		      "latin1", localeCP);
- 
 
 911 
#else
- 
 
 912 
	        error(_("unsupported conversion from '%s' to '%s'"),
- 
 
 913 
		      "latin1", "");
- 
 
 914 
#endif
- 
 
 915 
	    latin1_obj = obj;
- 
 
 916 
	}
- 
 
 917 
	obj = latin1_obj;
- 
 
 918 
    } else {
- 
 
 919 
	if(!utf8_obj) {
- 
 
 920 
	    obj = Riconv_open("", "UTF-8");
- 
 
 921 
	    /* should never happen */
- 
 
 922 
	    if(obj == (void *)(-1))
- 
 
 923 
#ifdef Win32
- 
 
 924 
		error(_("unsupported conversion from '%s' in codepage %d"),
- 
 
 925 
		      "latin1", localeCP);
- 
 
 926 
#else
- 
 
 927 
	        error(_("unsupported conversion from '%s' to '%s'"),
- 
 
 928 
		      "latin1", "");
- 
 
 929 
#endif
- 
 
 930 
	    utf8_obj = obj;
- 
 
 931 
	}
- 
 
 932 
	obj = utf8_obj;
- 
 
 933 
    }
- 
 
 934 
 
- 
 
 935 
    R_AllocStringBuffer(0, &cbuff);
- 
 
 936 
top_of_loop:
- 
 
 937 
    inbuf = ans; inb = strlen(inbuf);
- 
 
 938 
    outbuf = cbuff.data; outb = cbuff.bufsize - 1;
- 
 
 939 
    /* First initialize output */
- 
 
 940 
    Riconv (obj, NULL, NULL, &outbuf, &outb);
- 
 
 941 
next_char:
- 
 
 942 
    /* Then convert input  */
- 
 
 943 
    res = Riconv(obj, &inbuf , &inb, &outbuf, &outb);
- 
 
 944 
    if(res == -1 && errno == E2BIG) {
- 
 
 945 
	R_AllocStringBuffer(2*cbuff.bufsize, &cbuff);
- 
 
 946 
	goto top_of_loop;
- 
 
 947 
    } else if(res == -1 && (errno == EILSEQ || errno == EINVAL)) {
- 
 
 948 
	if(outb < 13) {
- 
 
 949 
	    R_AllocStringBuffer(2*cbuff.bufsize, &cbuff);
- 
 
 950 
	    goto top_of_loop;
- 
 
 951 
	}
- 
 
 952 
	if (ienc == CE_UTF8) {
- 
 
 953 
	    /* if starting in UTF-8, use \uxxxx */
- 
 
 954 
	    /* This must be the first byte */
- 
 
 955 
	    size_t clen;
- 
 
 956 
	    wchar_t wc;
- 
 
 957 
	    clen = utf8toucs(&wc, inbuf);
- 
 
 958 
	    if(clen > 0 && inb >= clen) {
- 
 
 959 
		inbuf += clen; inb -= clen;
- 
 
 960 
# ifndef Win32
- 
 
 961 
		if((unsigned int) wc < 65536) {
- 
 
 962 
# endif
- 
 
 963 
		    snprintf(outbuf, 9, "<U+%04X>", (unsigned int) wc);
- 
 
 964 
		    outbuf += 8; outb -= 8;
- 
 
 965 
# ifndef Win32
- 
 
 966 
		} else {
- 
 
 967 
		    snprintf(outbuf, 13, "<U+%08X>", (unsigned int) wc);
- 
 
 968 
		    outbuf += 12; outb -= 12;
- 
 
 969 
		}
- 
 
 970 
# endif
- 
 
 971 
	    } else {
- 
 
 972 
		snprintf(outbuf, 5, "<%02x>", (unsigned char)*inbuf);
- 
 
 973 
		outbuf += 4; outb -= 4;
- 
 
 974 
		inbuf++; inb--;
- 
 
 975 
	    }
- 
 
 976 
	} else {
- 
 
 977 
	    snprintf(outbuf, 5, "<%02x>", (unsigned char)*inbuf);
- 
 
 978 
	    outbuf += 4; outb -= 4;
- 
 
 979 
	    inbuf++; inb--;
- 
 
 980 
	}
- 
 
 981 
	goto next_char;
- 
 
 982 
    }
- 
 
 983 
    *outbuf = '\0';
- 
 
 984 
    SEXP Sans = install(cbuff.data);
- 
 
 985 
    R_FreeStringBuffer(&cbuff);
- 
 
 986 
    return Sans;
- 
 
 987 
}
- 
 
 988 
 
885 
/* This may return a R_alloc-ed result, so the caller has to manage the
 989 
/* This may return a R_alloc-ed result, so the caller has to manage the
886 
   R_alloc stack */
 990 
   R_alloc stack */
887 
const char *translateChar0(SEXP x)
 991 
const char *translateChar0(SEXP x)
888 
{
 992 
{
889 
    if(TYPEOF(x) != CHARSXP)
 993 
    if(TYPEOF(x) != CHARSXP)