util/unicode.c

   1 /*
   2  * Dealing with Unicode
   3  *
   4  * Copyright (C) 2013 Red Hat, Inc.
   5  *
   6  * Authors:
   7  *  Markus Armbruster <armbru@redhat.com>
   8  *
   9  * This work is licensed under the terms of the GNU GPL, version 2 or
  10  * later.  See the COPYING file in the top-level directory.
  11  */
  12
  13 #include "qemu-common.h"
  14
  15 /**
  16  * mod_utf8_codepoint:
  17  * @s: string encoded in modified UTF-8
  18  * @n: maximum number of bytes to read from @s, if less than 6
  19  * @end: set to end of sequence on return
  20  *
  21  * Convert the modified UTF-8 sequence at the start of @s.  Modified
  22  * UTF-8 is exactly like UTF-8, except U+0000 is encoded as
  23  * "\xC0\x80".
  24  *
  25  * If @n is zero or @s points to a zero byte, the sequence is invalid,
  26  * and @end is set to @s.
  27  *
  28  * If @s points to an impossible byte (0xFE or 0xFF) or a continuation
  29  * byte, the sequence is invalid, and @end is set to @s + 1
  30  *
  31  * Else, the first byte determines how many continuation bytes are
  32  * expected.  If there are fewer, the sequence is invalid, and @end is
  33  * set to @s + 1 + actual number of continuation bytes.  Else, the
  34  * sequence is well-formed, and @end is set to @s + 1 + expected
  35  * number of continuation bytes.
  36  *
  37  * A well-formed sequence is valid unless it encodes a codepoint
  38  * outside the Unicode range U+0000..U+10FFFF, one of Unicode's 66
  39  * noncharacters, a surrogate codepoint, or is overlong.  Except the
  40  * overlong sequence "\xC0\x80" is valid.
  41  *
  42  * Conversion succeeds if and only if the sequence is valid.
  43  *
  44  * Returns: the Unicode codepoint on success, -1 on failure.
  45  */
  46 int mod_utf8_codepoint(const char *s, size_t n, char **end)
  47 {
  48     static int min_cp[5] = { 0x80, 0x800, 0x10000, 0x200000, 0x4000000 };
  49     const unsigned char *p;
  50     unsigned byte, mask, len, i;
  51     int cp;
  52
  53     if (n == 0 || *s == 0) {
  54         /* empty sequence */
  55         *end = (char *)s;
  56         return -1;
  57     }
  58
  59     p = (const unsigned char *)s;
  60     byte = *p++;
  61     if (byte < 0x80) {
  62         cp = byte;              /* one byte sequence */
  63     } else if (byte >= 0xFE) {
  64         cp = -1;                /* impossible bytes 0xFE, 0xFF */
  65     } else if ((byte & 0x40) == 0) {
  66         cp = -1;                /* unexpected continuation byte */
  67     } else {
  68         /* multi-byte sequence */
  69         len = 0;
  70         for (mask = 0x80; byte & mask; mask >>= 1) {
  71             len++;
  72         }
  73         assert(len > 1 && len < 7);
  74         cp = byte & (mask - 1);
  75         for (i = 1; i < len; i++) {
  76             byte = i < n ? *p : 0;
  77             if ((byte & 0xC0) != 0x80) {
  78                 cp = -1;        /* continuation byte missing */
  79                 goto out;
  80             }
  81             p++;
  82             cp <<= 6;
  83             cp |= byte & 0x3F;
  84         }
  85         if (cp > 0x10FFFF) {
  86             cp = -1;            /* beyond Unicode range */
  87         } else if ((cp >= 0xFDD0 && cp <= 0xFDEF)
  88                    || (cp & 0xFFFE) == 0xFFFE) {
  89             cp = -1;            /* noncharacter */
  90         } else if (cp >= 0xD800 && cp <= 0xDFFF) {
  91             cp = -1;            /* surrogate code point */
  92         } else if (cp < min_cp[len - 2] && !(cp == 0 && len == 2)) {
  93             cp = -1;            /* overlong, not \xC0\x80 */
  94         }
  95     }
  96
  97 out:
  98     *end = (char *)p;
  99     return cp;
 100 }