1 /* -*- mode: C; c-file-style: "gnu"; indent-tabs-mode: nil; -*- */
3 /* GLIB - Library of useful routines for C programming
4 * Copyright (C) 2008 Red Hat, Inc.
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General
17 * Public License along with this library; if not, write to the
18 * Free Software Foundation, Inc., 59 Temple Place, Suite 330,
19 * Boston, MA 02111-1307, USA.
33 * @short_description: Internet hostname utilities
36 * Functions for manipulating internet hostnames; in particular, for
37 * converting between Unicode and ASCII-encoded forms of
38 * Internationalized Domain Names (IDNs).
41 * url="http://www.ietf.org/rfc/rfc3490.txt">Internationalized Domain
42 * Names for Applications (IDNA)</ulink> standards allow for the use
43 * of Unicode domain names in applications, while providing
44 * backward-compatibility with the old ASCII-only DNS, by defining an
45 * ASCII-Compatible Encoding of any given Unicode name, which can be
46 * used with non-IDN-aware applications and protocols. (For example,
47 * "Παν語.org" maps to "xn--4wa8awb4637h.org".)
50 #define IDNA_ACE_PREFIX "xn--"
51 #define IDNA_ACE_PREFIX_LEN 4
53 /* Punycode constants, from RFC 3492. */
55 #define PUNYCODE_BASE 36
56 #define PUNYCODE_TMIN 1
57 #define PUNYCODE_TMAX 26
58 #define PUNYCODE_SKEW 38
59 #define PUNYCODE_DAMP 700
60 #define PUNYCODE_INITIAL_BIAS 72
61 #define PUNYCODE_INITIAL_N 0x80
63 #define PUNYCODE_IS_BASIC(cp) ((guint)(cp) < 0x80)
65 /* Encode/decode a single base-36 digit */
67 encode_digit (guint dig)
72 return dig - 26 + '0';
76 decode_digit (gchar dig)
78 if (dig >= 'A' && dig <= 'Z')
80 else if (dig >= 'a' && dig <= 'z')
82 else if (dig >= '0' && dig <= '9')
83 return dig - '0' + 26;
88 /* Punycode bias adaptation algorithm, RFC 3492 section 6.1 */
96 delta = firsttime ? delta / PUNYCODE_DAMP : delta / 2;
97 delta += delta / numpoints;
100 while (delta > ((PUNYCODE_BASE - PUNYCODE_TMIN) * PUNYCODE_TMAX) / 2)
102 delta /= PUNYCODE_BASE - PUNYCODE_TMIN;
106 return k + ((PUNYCODE_BASE - PUNYCODE_TMIN + 1) * delta /
107 (delta + PUNYCODE_SKEW));
110 /* Punycode encoder, RFC 3492 section 6.3. The algorithm is
111 * sufficiently bizarre that it's not really worth trying to explain
115 punycode_encode (const gchar *input_utf8,
116 gsize input_utf8_length,
119 guint delta, handled_chars, num_basic_chars, bias, j, q, k, t, digit;
120 gunichar n, m, *input;
122 gboolean success = FALSE;
124 /* Convert from UTF-8 to Unicode code points */
125 input = g_utf8_to_ucs4 (input_utf8, input_utf8_length, NULL,
126 &input_length, NULL);
130 /* Copy basic chars */
131 for (j = num_basic_chars = 0; j < input_length; j++)
133 if (PUNYCODE_IS_BASIC (input[j]))
135 g_string_append_c (output, g_ascii_tolower (input[j]));
140 g_string_append_c (output, '-');
142 handled_chars = num_basic_chars;
144 /* Encode non-basic chars */
146 bias = PUNYCODE_INITIAL_BIAS;
147 n = PUNYCODE_INITIAL_N;
148 while (handled_chars < input_length)
150 /* let m = the minimum {non-basic} code point >= n in the input */
151 for (m = G_MAXUINT, j = 0; j < input_length; j++)
153 if (input[j] >= n && input[j] < m)
157 if (m - n > (G_MAXUINT - delta) / (handled_chars + 1))
159 delta += (m - n) * (handled_chars + 1);
162 for (j = 0; j < input_length; j++)
169 else if (input[j] == n)
172 for (k = PUNYCODE_BASE; ; k += PUNYCODE_BASE)
176 else if (k >= bias + PUNYCODE_TMAX)
182 digit = t + (q - t) % (PUNYCODE_BASE - t);
183 g_string_append_c (output, encode_digit (digit));
184 q = (q - t) / (PUNYCODE_BASE - t);
187 g_string_append_c (output, encode_digit (q));
188 bias = adapt (delta, handled_chars + 1, handled_chars == num_basic_chars);
205 /* From RFC 3454, Table B.1 */
206 #define idna_is_junk(ch) ((ch) == 0x00AD || (ch) == 0x1806 || (ch) == 0x200B || (ch) == 0x2060 || (ch) == 0xFEFF || (ch) == 0x034F || (ch) == 0x180B || (ch) == 0x180C || (ch) == 0x180D || (ch) == 0x200C || (ch) == 0x200D || ((ch) >= 0xFE00 && (ch) <= 0xFE0F))
208 /* Scan @str for "junk" and return a cleaned-up string if any junk
209 * is found. Else return %NULL.
212 remove_junk (const gchar *str,
215 GString *cleaned = NULL;
219 for (p = str; len == -1 ? *p : p < str + len; p = g_utf8_next_char (p))
221 ch = g_utf8_get_char (p);
222 if (idna_is_junk (ch))
226 cleaned = g_string_new (NULL);
227 g_string_append_len (cleaned, str, p - str);
231 g_string_append_unichar (cleaned, ch);
235 return g_string_free (cleaned, FALSE);
240 static inline gboolean
241 contains_uppercase_letters (const gchar *str,
246 for (p = str; len == -1 ? *p : p < str + len; p = g_utf8_next_char (p))
248 if (g_unichar_isupper (g_utf8_get_char (p)))
254 static inline gboolean
255 contains_non_ascii (const gchar *str,
260 for (p = str; len == -1 ? *p : p < str + len; p++)
262 if ((guchar)*p > 0x80)
268 /* RFC 3454, Appendix C. ish. */
269 static inline gboolean
270 idna_is_prohibited (gunichar ch)
272 switch (g_unichar_type (ch))
274 case G_UNICODE_CONTROL:
275 case G_UNICODE_FORMAT:
276 case G_UNICODE_UNASSIGNED:
277 case G_UNICODE_PRIVATE_USE:
278 case G_UNICODE_SURROGATE:
279 case G_UNICODE_LINE_SEPARATOR:
280 case G_UNICODE_PARAGRAPH_SEPARATOR:
281 case G_UNICODE_SPACE_SEPARATOR:
284 case G_UNICODE_OTHER_SYMBOL:
285 if (ch == 0xFFFC || ch == 0xFFFD ||
286 (ch >= 0x2FF0 && ch <= 0x2FFB))
290 case G_UNICODE_NON_SPACING_MARK:
291 if (ch == 0x0340 || ch == 0x0341)
300 /* RFC 3491 IDN cleanup algorithm. */
302 nameprep (const gchar *hostname,
305 gchar *name, *tmp = NULL, *p;
307 /* It would be nice if we could do this without repeatedly
308 * allocating strings and converting back and forth between
309 * gunichars and UTF-8... The code does at least avoid doing most of
310 * the sub-operations when they would just be equivalent to a
314 /* Remove presentation-only characters */
315 name = remove_junk (hostname, len);
322 name = (gchar *)hostname;
324 /* Convert to lowercase */
325 if (contains_uppercase_letters (name, len))
327 name = g_utf8_strdown (name, len);
333 /* If there are no UTF8 characters, we're done. */
334 if (!contains_non_ascii (name, len))
336 if (name == (gchar *)hostname)
337 return len == -1 ? g_strdup (hostname) : g_strndup (hostname, len);
343 name = g_utf8_normalize (name, len, G_NORMALIZE_NFKC);
347 /* KC normalization may have created more capital letters (eg,
348 * angstrom -> capital A with ring). So we have to lowercasify a
349 * second time. (This is more-or-less how the nameprep algorithm
350 * does it. If tolower(nfkc(tolower(X))) is guaranteed to be the
351 * same as tolower(nfkc(X)), then we could skip the first tolower,
352 * but I'm not sure it is.)
354 if (contains_uppercase_letters (name, -1))
356 name = g_utf8_strdown (name, -1);
361 /* Check for prohibited characters */
362 for (p = name; *p; p = g_utf8_next_char (p))
364 if (idna_is_prohibited (g_utf8_get_char (p)))
372 /* FIXME: We're supposed to verify certain constraints on bidi
373 * characters, but glib does not appear to have that information.
381 * g_hostname_to_ascii:
382 * @hostname: a valid UTF-8 or ASCII hostname
384 * Converts @hostname to its canonical ASCII form; an ASCII-only
385 * string containing no uppercase letters and not ending with a
388 * Return value: an ASCII hostname, which must be freed, or %NULL if
389 * @hostname is in some way invalid.
394 g_hostname_to_ascii (const gchar *hostname)
396 gchar *name, *label, *p;
401 out = g_string_new (NULL);
402 label = name = nameprep (hostname, -1);
407 for (p = label; *p && *p != '.'; p++)
409 if ((guchar)*p > 0x80)
417 if (!strncmp (label, IDNA_ACE_PREFIX, IDNA_ACE_PREFIX_LEN))
420 g_string_append (out, IDNA_ACE_PREFIX);
421 if (!punycode_encode (label, llen, out))
425 g_string_append_len (out, label, llen);
427 if (out->len - oldlen > 63)
431 if (*label && *++label)
432 g_string_append_c (out, '.');
437 return g_string_free (out, FALSE);
441 g_string_free (out, TRUE);
446 * g_hostname_is_non_ascii:
447 * @hostname: a hostname
449 * Tests if @hostname contains Unicode characters. If this returns
450 * %TRUE, you need to encode the hostname with g_hostname_to_ascii()
451 * before using it in non-IDN-aware contexts.
453 * Note that a hostname might contain a mix of encoded and unencoded
454 * segments, and so it is possible for g_hostname_is_non_ascii() and
455 * g_hostname_is_ascii_encoded() to both return %TRUE for a name.
457 * Return value: %TRUE if @hostname contains any non-ASCII characters
462 g_hostname_is_non_ascii (const gchar *hostname)
464 return contains_non_ascii (hostname, -1);
467 /* Punycode decoder, RFC 3492 section 6.2. As with punycode_encode(),
468 * read the RFC if you want to understand what this is actually doing.
471 punycode_decode (const gchar *input,
475 GArray *output_chars;
478 guint oldi, w, k, digit, t;
481 n = PUNYCODE_INITIAL_N;
483 bias = PUNYCODE_INITIAL_BIAS;
485 split = input + input_length - 1;
486 while (split > input && *split != '-')
490 output_chars = g_array_sized_new (FALSE, FALSE, sizeof (gunichar),
492 input_length -= (split - input) + 1;
493 while (input < split)
495 gunichar ch = (gunichar)*input++;
496 if (!PUNYCODE_IS_BASIC (ch))
498 g_array_append_val (output_chars, ch);
503 output_chars = g_array_new (FALSE, FALSE, sizeof (gunichar));
509 for (k = PUNYCODE_BASE; ; k += PUNYCODE_BASE)
513 digit = decode_digit (*input++);
514 if (digit >= PUNYCODE_BASE)
516 if (digit > (G_MAXUINT - i) / w)
521 else if (k >= bias + PUNYCODE_TMAX)
527 if (w > G_MAXUINT / (PUNYCODE_BASE - t))
529 w *= (PUNYCODE_BASE - t);
532 bias = adapt (i - oldi, output_chars->len + 1, oldi == 0);
534 if (i / (output_chars->len + 1) > G_MAXUINT - n)
536 n += i / (output_chars->len + 1);
537 i %= (output_chars->len + 1);
539 g_array_insert_val (output_chars, i++, n);
542 for (i = 0; i < output_chars->len; i++)
543 g_string_append_unichar (output, g_array_index (output_chars, gunichar, i));
544 g_array_free (output_chars, TRUE);
548 g_array_free (output_chars, TRUE);
553 * g_hostname_to_unicode:
554 * @hostname: a valid UTF-8 or ASCII hostname
556 * Converts @hostname to its canonical presentation form; a UTF-8
557 * string in Unicode normalization form C, containing no uppercase
558 * letters, no forbidden characters, and no ASCII-encoded segments,
559 * and not ending with a trailing dot.
561 * Of course if @hostname is not an internationalized hostname, then
562 * the canonical presentation form will be entirely ASCII.
564 * Return value: a UTF-8 hostname, which must be freed, or %NULL if
565 * @hostname is in some way invalid.
570 g_hostname_to_unicode (const gchar *hostname)
575 out = g_string_new (NULL);
579 llen = strcspn (hostname, ".");
580 if (!g_ascii_strncasecmp (hostname, IDNA_ACE_PREFIX, IDNA_ACE_PREFIX_LEN))
582 hostname += IDNA_ACE_PREFIX_LEN;
583 llen -= IDNA_ACE_PREFIX_LEN;
584 if (!punycode_decode (hostname, llen, out))
586 g_string_free (out, TRUE);
592 gchar *canonicalized = nameprep (hostname, llen);
594 g_string_append (out, canonicalized);
595 g_free (canonicalized);
599 if (*hostname && *++hostname)
600 g_string_append_c (out, '.');
604 return g_string_free (out, FALSE);
608 * g_hostname_is_ascii_encoded:
609 * @hostname: a hostname
611 * Tests if @hostname contains segments with an ASCII-compatible
612 * encoding of an Internationalized Domain Name. If this returns
613 * %TRUE, you should decode the hostname with g_hostname_to_unicode()
614 * before displaying it to the user.
616 * Note that a hostname might contain a mix of encoded and unencoded
617 * segments, and so it is possible for g_hostname_is_non_ascii() and
618 * g_hostname_is_ascii_encoded() to both return %TRUE for a name.
620 * Return value: %TRUE if @hostname contains any ASCII-encoded
626 g_hostname_is_ascii_encoded (const gchar *hostname)
630 if (!g_ascii_strncasecmp (hostname, IDNA_ACE_PREFIX, IDNA_ACE_PREFIX_LEN))
632 hostname = strchr (hostname, '.');
639 * g_hostname_is_ip_address:
640 * @hostname: a hostname (or IP address in string form)
642 * Tests if @hostname is the string form of an IPv4 or IPv6 address.
643 * (Eg, "192.168.0.1".)
645 * Return value: %TRUE if @hostname is an IP address
650 g_hostname_is_ip_address (const gchar *hostname)
653 gint nsegments, octet;
655 /* On Linux we could implement this using inet_pton, but the Windows
656 * equivalent of that requires linking against winsock, so we just
657 * figure this out ourselves. Tested by tests/hostutils.c.
660 p = (char *)hostname;
666 /* If it contains a ':', it's an IPv6 address (assuming it's an
667 * IP address at all). This consists of eight ':'-separated
668 * segments, each containing a 1-4 digit hex number, except that
669 * optionally: (a) the last two segments can be replaced by an
670 * IPv4 address, and (b) a single span of 1 to 8 "0000" segments
671 * can be replaced with just "::".
676 while (*p && nsegments < 8)
678 /* Each segment after the first must be preceded by a ':'.
679 * (We also handle half of the "string starts with ::" case
682 if (p != (char *)hostname || (p[0] == ':' && p[1] == ':'))
689 /* If there's another ':', it means we're skipping some segments */
690 if (*p == ':' && !skipped)
695 /* Handle the "string ends with ::" case */
702 /* Read the segment, make sure it's valid. */
703 for (end = p; g_ascii_isxdigit (*end); end++)
705 if (end == p || end > p + 4)
710 if ((nsegments == 6 && !skipped) || (nsegments <= 6 && skipped))
720 return !*p && (nsegments == 8 || skipped);
725 /* Parse IPv4: N.N.N.N, where each N <= 255 and doesn't have leading 0s. */
726 for (nsegments = 0; nsegments < 4; nsegments++)
735 /* Check the segment; a little tricker than the IPv6 case since
736 * we can't allow extra leading 0s, and we can't assume that all
737 * strings of valid length are within range.
744 for (end = p; g_ascii_isdigit (*end); end++)
745 octet = 10 * octet + (*end - '0');
747 if (end == p || end > p + 3 || octet > 255)
753 /* If there's nothing left to parse, then it's ok. */
757 #define __G_HOST_UTILS_C__
758 #include "galiasdef.c"