1 /*
2  *  linux/fs/isofs/joliet.c
3  *
4  *  (C) 1996 Gordon Chaffee
5  *
6  *  Joliet: Microsoft's Unicode extensions to iso9660
7  */
8 
9 #include <linux/string.h>
10 #include <linux/nls.h>
11 #include <linux/slab.h>
12 #include <linux/iso_fs.h>
13 #include <asm/unaligned.h>
14 
15 /*
16  * Convert Unicode 16 to UTF8 or ASCII.
17  */
18 static int
uni16_to_x8(unsigned char * ascii,u16 * uni,int len,struct nls_table * nls)19 uni16_to_x8(unsigned char *ascii, u16 *uni, int len, struct nls_table *nls)
20 {
21 	wchar_t *ip, ch;
22 	unsigned char *op;
23 
24 	ip = uni;
25 	op = ascii;
26 
27 	while ((ch = get_unaligned(ip)) && len) {
28 		int llen;
29 		ch = be16_to_cpu(ch);
30 		if ((llen = nls->uni2char(ch, op, NLS_MAX_CHARSET_SIZE)) > 0)
31 			op += llen;
32 		else
33 			*op++ = '?';
34 		ip++;
35 
36 		len--;
37 	}
38 	*op = 0;
39 	return (op - ascii);
40 }
41 
42 /* Convert big endian wide character string to utf8 */
43 static int
wcsntombs_be(__u8 * s,const __u8 * pwcs,int inlen,int maxlen)44 wcsntombs_be(__u8 *s, const __u8 *pwcs, int inlen, int maxlen)
45 {
46 	const __u8 *ip;
47 	__u8 *op;
48 	int size;
49 	__u16 c;
50 
51 	op = s;
52 	ip = pwcs;
53 	while ((*ip || ip[1]) && (maxlen > 0) && (inlen > 0)) {
54 		c = (*ip << 8) | ip[1];
55 		if (c > 0x7f) {
56 			size = utf8_wctomb(op, c, maxlen);
57 			if (size == -1) {
58 				/* Ignore character and move on */
59 				maxlen--;
60 			} else {
61 				op += size;
62 				maxlen -= size;
63 			}
64 		} else {
65 			*op++ = (__u8) c;
66 		}
67 		ip += 2;
68 		inlen--;
69 	}
70 	return (op - s);
71 }
72 
73 int
get_joliet_filename(struct iso_directory_record * de,unsigned char * outname,struct inode * inode)74 get_joliet_filename(struct iso_directory_record * de, unsigned char *outname, struct inode * inode)
75 {
76 	unsigned char utf8;
77 	struct nls_table *nls;
78 	unsigned char len = 0;
79 
80 	utf8 = inode->i_sb->u.isofs_sb.s_utf8;
81 	nls = inode->i_sb->u.isofs_sb.s_nls_iocharset;
82 
83 	if (utf8) {
84 		len = wcsntombs_be(outname, de->name,
85 				   de->name_len[0] >> 1, PAGE_SIZE);
86 	} else {
87 		len = uni16_to_x8(outname, (u16 *) de->name,
88 				  de->name_len[0] >> 1, nls);
89 	}
90 	if ((len > 2) && (outname[len-2] == ';') && (outname[len-1] == '1')) {
91 		len -= 2;
92 	}
93 
94 	/*
95 	 * Windows doesn't like periods at the end of a name,
96 	 * so neither do we
97 	 */
98 	while (len >= 2 && (outname[len-1] == '.')) {
99 		len--;
100 	}
101 
102 	return len;
103 }
104