idnconv.c

bind-3.2.

	if (local_ace_hack) {
		actions1 = IDN_IDNCONV;
		if (flags & FLAG_ROUNDTRIPCHECK)
			actions1 |= IDN_RTCHECK;
	} else {
		actions1 = IDN_LOCALCONV;
	}

	actions2 = IDN_IDNCONV;
	if (flags & FLAG_DELIMMAP)
		actions2 |= IDN_DELIMMAP;
	if (flags & FLAG_LOCALMAP)
		actions2 |= IDN_LOCALMAP;
	if (flags & FLAG_MAP)
		actions2 |= IDN_MAP;
	if (flags & FLAG_NORMALIZE)
		actions2 |= IDN_NORMALIZE;
	if (flags & FLAG_PROHIBITCHECK)
		actions2 |= IDN_PROHCHECK;
	if (flags & FLAG_UNASSIGNCHECK)
		actions2 |= IDN_UNASCHECK;
	if (flags & FLAG_BIDICHECK)
		actions2 |= IDN_BIDICHECK;
	if (flags & FLAG_ASCIICHECK)
		actions2 |= IDN_ASCCHECK;
	if (flags & FLAG_LENGTHCHECK)
		actions2 |= IDN_LENCHECK;

	strbuf_init(&buf1);
	strbuf_init(&buf2);
	line_number = 1;
	while (strbuf_getline(&buf1, fp) != NULL) {
		/*
		 * Trim newline at the end.  This is needed for
		 * those ascii-comatible encodings such as UTF-5 or RACE
		 * not to try converting newlines, which will result
		 * in `invalid encoding' error.
		 */
		nl_trimmed = trim_newline(&buf1);

		/*
		 * Convert input line to UTF-8.
		 */
		if (local_ace_hack)
			r = convert_line(&buf1, &buf2, conf2, actions1,
					 FLAG_REVERSE|FLAG_SELECTIVE);
		else
			r = convert_line(&buf1, &buf2, conf1, actions1,
					 0);
				 
		if (r != idn_success) {
			errormsg("conversion failed at line %d: %s\n",
				 line_number,
				 idn_result_tostring(r));
			goto error;
		}
		if (!idn_utf8_isvalidstring(strbuf_get(&buf2))) {
			errormsg("conversion to utf-8 failed at line %d\n",
				 line_number);
			goto error;
		}

		/*
		 * Perform local mapping and NAMEPREP, and convert to
		 * the output codeset.
		 */
		r = convert_line(&buf2, &buf1, conf1, actions2,
				 flags & FLAG_SELECTIVE);
				 
		if (r != idn_success) {
			errormsg("error in nameprep or output conversion "
				 "at line %d: %s\n",
				 line_number, idn_result_tostring(r));
			goto error;
		}

		fputs(strbuf_get(&buf1), stdout);
		if (nl_trimmed)
			putc('\n', stdout);

		if (flush_every_line)
			fflush(stdout);

		line_number++;
	}

	strbuf_reset(&buf1);
	strbuf_reset(&buf2);
	return (0);

 error:
	strbuf_reset(&buf1);
	strbuf_reset(&buf2);
	return (1);
}

static int
decode_file(idn_resconf_t conf1, idn_resconf_t conf2, FILE *fp, int flags) {
	idn_result_t r;
	idnconv_strbuf_t buf1, buf2;
	idn_action_t actions1, actions2;
	int nl_trimmed;
	int local_ace_hack, idn_ace_hack;
	idn_converter_t conv;

	/*
	 * See if the input codeset is an ACE.
	 */
	conv = idn_resconf_getidnconverter(conf1);
	if (conv != NULL && idn_converter_isasciicompatible(conv) &&
	    (flags & FLAG_SELECTIVE))
		idn_ace_hack = 1;
	else
		idn_ace_hack = 0;
	if (conv != NULL)
		idn_converter_destroy(conv);

	conv = idn_resconf_getlocalconverter(conf1);
	if (conv != NULL && idn_converter_isasciicompatible(conv) &&
	    (flags & FLAG_SELECTIVE))
		local_ace_hack = 1;
	else
		local_ace_hack = 0;
	if (conv != NULL)
		idn_converter_destroy(conv);

	actions1 = IDN_IDNCONV;

	if (local_ace_hack) {
		actions2 = IDN_IDNCONV;
		if (flags & FLAG_MAP)
			actions2 |= IDN_MAP;
		if (flags & FLAG_NORMALIZE)
			actions2 |= IDN_NORMALIZE;
		if (flags & FLAG_PROHIBITCHECK)
			actions2 |= IDN_PROHCHECK;
		if (flags & FLAG_UNASSIGNCHECK)
			actions2 |= IDN_UNASCHECK;
		if (flags & FLAG_BIDICHECK)
			actions2 |= IDN_BIDICHECK;
		if (flags & FLAG_ASCIICHECK)
			actions2 |= IDN_ASCCHECK;
		if (flags & FLAG_LENGTHCHECK)
			actions2 |= IDN_LENCHECK;
	} else {
		actions2 = IDN_LOCALCONV;
	}

	if (flags & FLAG_DELIMMAP)
		actions1 |= IDN_DELIMMAP;
	if (flags & FLAG_MAP)
		actions1 |= IDN_MAP;
	if (flags & FLAG_NORMALIZE)
		actions1 |= IDN_NORMALIZE;
	if (flags & FLAG_NORMALIZE)
		actions1 |= IDN_NORMALIZE;
	if (flags & FLAG_PROHIBITCHECK)
		actions1 |= IDN_PROHCHECK;
	if (flags & FLAG_UNASSIGNCHECK)
		actions1 |= IDN_UNASCHECK;
	if (flags & FLAG_BIDICHECK)
		actions1 |= IDN_BIDICHECK;
	if (flags & FLAG_ASCIICHECK)
		actions1 |= IDN_ASCCHECK;
	if (flags & FLAG_ROUNDTRIPCHECK)
		actions1 |= IDN_RTCHECK;

	strbuf_init(&buf1);
	strbuf_init(&buf2);
	line_number = 1;
	while (strbuf_getline(&buf1, fp) != NULL) {
		/*
		 * Trim newline at the end.  This is needed for
		 * those ascii-comatible encodings such as UTF-5 or RACE
		 * not to try converting newlines, which will result
		 * in `invalid encoding' error.
		 */
		nl_trimmed = trim_newline(&buf1);

		/*
		 * Treat input line as the string encoded in local
		 * encoding and convert it to UTF-8 encoded string.
		 */
		if (local_ace_hack) {
			if (strbuf_copy(&buf2, strbuf_get(&buf1)) == NULL)
				r = idn_nomemory;
			else
				r = idn_success;
		} else {
			r = convert_line(&buf1, &buf2, conf1, IDN_LOCALCONV,
					 0);
		}
		if (r != idn_success) {
			errormsg("conversion failed at line %d: %s\n",
				 line_number, idn_result_tostring(r));
			goto error;
		}

		/*
		 * Convert internationalized domain names in the line.
		 */
		if (idn_ace_hack) {
			r = convert_line(&buf2, &buf1, conf1, actions1,
					 FLAG_REVERSE|FLAG_SELECTIVE);
		} else {
			r = convert_line(&buf2, &buf1, conf1, actions1,
					 FLAG_REVERSE);
		}
		if (r != idn_success) {
			errormsg("conversion failed at line %d: %s\n",
				 line_number,
				 idn_result_tostring(r));
			goto error;
		}
		if (!idn_utf8_isvalidstring(strbuf_get(&buf1))) {
			errormsg("conversion to utf-8 failed at line %d\n",
				 line_number);
			goto error;
		}

		/*
		 * Perform round trip check and convert to the output
		 * codeset.
		 */
		if (local_ace_hack) {
			r = convert_line(&buf1, &buf2, conf2, actions2,
					 FLAG_SELECTIVE);
		} else {
			r = convert_line(&buf1, &buf2, conf1, actions2,
					 FLAG_REVERSE);
		}

		if (r != idn_success) {
			errormsg("error in nameprep or output conversion "
				 "at line %d: %s\n",
				 line_number, idn_result_tostring(r));
			goto error;
		}

		fputs(strbuf_get(&buf2), stdout);
		if (nl_trimmed)
			putc('\n', stdout);

		if (flush_every_line)
			fflush(stdout);

		line_number++;
	}
	strbuf_reset(&buf1);
	strbuf_reset(&buf2);
	return (0);

 error:
	strbuf_reset(&buf1);
	strbuf_reset(&buf2);
	return (1);
}

static int
trim_newline(idnconv_strbuf_t *buf) {
	/*
	 * If the string in BUF ends with a newline, trim it and
	 * return 1.  Otherwise, just return 0 without modifying BUF.
	 */
	char *s = strbuf_get(buf);
	size_t len = strlen(s);

	if (s[len - 1] == '\n') {
		s[len - 1] = '\0';
		return (1);
	}

	return (0);
}

static idn_result_t
convert_line(idnconv_strbuf_t *from, idnconv_strbuf_t *to,
	     idn_resconf_t conf, idn_action_t actions, int flags)
{
	idn_result_t r = idn_success;
	char *from_str = strbuf_get(from);

	for (;;) {
		char *to_str = strbuf_get(to);
		size_t to_size = strbuf_size(to);

		switch (flags & (FLAG_REVERSE|FLAG_SELECTIVE)) {
		case 0:
			r = idn_res_encodename(conf, actions, from_str,
					       to_str, to_size);
			break;
		case FLAG_REVERSE:
			r = idn_res_decodename(conf, actions, from_str,
					       to_str, to_size);
			break;
		case FLAG_SELECTIVE:
			r = selective_encode(conf, actions, from_str,
					     to_str, to_size);
			break;
		case FLAG_REVERSE|FLAG_SELECTIVE:
			r = selective_decode(conf, actions, from_str,
					     to_str, to_size);
			break;
		}
		if (r == idn_buffer_overflow) {
			/*
			 * Conversion is not successful because
			 * the size of the target buffer is not enough.
			 * Double the size and retry.
			 */
			if (strbuf_double(to) == NULL) {
				/* oops. allocation failed. */
				return (idn_nomemory);
			}
		} else {
			break;
		}
	}
	return (r);
}

static char *options[] = {
	"-in INPUT-CODESET   : specifies input codeset name.",
	"-i INPUT-CODESET    : synonym for -in",
	"-out OUTPUT-CODESET : specifies output codeset name.",
	"-o OUTPUT-CODESET   : synonym for -out",
	"-conf CONF-FILE     : specifies idnkit configuration file.",
	"-c CONF-FILE        : synonym for -conf",
	"-noconf             : do not load idnkit configuration file.",
	"-C                  : synonym for -noconf",
	"-reverse            : specifies reverse conversion.",
	"                      (i.e. IDN encoding to local encoding)",
	"-r                  : synonym for -reverse",
	"-nameprep VERSION   : specifies version name of NAMEPREP.",
	"-n VERSION          : synonym for -nameprep",
	"-nonameprep         : do not perform NAMEPREP.",
	"-N                  : synonym for -nonameprep",
	"-localmap MAPPING   : specifies local mapping.",
	"-nolocalmap         : do not perform local mapping.",
	"-L                  : synonym for -nolocalmap",
	"-nounassigncheck    : do not perform unassigned codepoint check.",
	"-U                  : synonym for -nounassigncheck",
	"-nobidicheck        : do not perform bidirectional text check.",
	"-B                  : synonym for -nobidicheck",
	"-nolengthcheck      : do not check label length.",
	"-noasciicheck       : do not check ASCII range characters.",
	"-A                  : synonym for -noasciicheck",
	"-noroundtripcheck   : do not perform round trip check.",
	"-delimiter U+XXXX   : specifies local delimiter code point.",
	"-alias alias-file   : specifies codeset alias file.",
	"-a                  : synonym for -alias",
	"-flush              : line-buffering mode.",
	"-whole              : convert the whole region instead of",
	"                      regions containing non-ascii characters.",
	"-w                  : synonym for -whole",
	"-version            : print version number, then exit.",
	"-v                  : synonym for -version",
	"",
	" The following options can be specified multiple times",
	"   -localmap, -delimiter",
	NULL,
};

static void
print_version() {
	fprintf(stderr, "idnconv (idnkit) version: %s\n"
		"library version: %s\n",
		IDNKIT_VERSION,
		idn_version_getstring());
	exit(0);
}

static void
print_usage(char *cmd) {
	int i;

	fprintf(stderr, "Usage: %s [options..] [file]\n", cmd);

	for (i = 0; options[i] != NULL; i++)
		fprintf(stderr, "\t%s\n", options[i]);

	exit(1);
}

static unsigned long
get_ucs(const char *p) {
	unsigned long v;
	char *end;

	/* Skip optional 'U+' */
	if (strncmp(p, "U+", 2) == 0)
		p += 2;

	v = strtoul(p, &end, 16);
	if (*end != '\0') {
		fprintf(stderr, "invalid UCS code point \"%s\"\n", p);
		exit(1);
	}

	return v;
}