include/linux/unicode.h

0001 /* SPDX-License-Identifier: GPL-2.0 */
0002 #ifndef _LINUX_UNICODE_H
0003 #define _LINUX_UNICODE_H
0004
0005 #include <linux/init.h>
0006 #include <linux/dcache.h>
0007
0008 struct utf8data;
0009 struct utf8data_table;
0010
0011 #define UNICODE_MAJ_SHIFT       16
0012 #define UNICODE_MIN_SHIFT       8
0013
0014 #define UNICODE_AGE(MAJ, MIN, REV)          \
0015     (((unsigned int)(MAJ) << UNICODE_MAJ_SHIFT) |   \
0016      ((unsigned int)(MIN) << UNICODE_MIN_SHIFT) |   \
0017      ((unsigned int)(REV)))
0018
0019 static inline u8 unicode_major(unsigned int age)
0020 {
0021     return (age >> UNICODE_MAJ_SHIFT) & 0xff;
0022 }
0023
0024 static inline u8 unicode_minor(unsigned int age)
0025 {
0026     return (age >> UNICODE_MIN_SHIFT) & 0xff;
0027 }
0028
0029 static inline u8 unicode_rev(unsigned int age)
0030 {
0031     return age & 0xff;
0032 }
0033
0034 /*
0035  * Two normalization forms are supported:
0036  * 1) NFDI
0037  *   - Apply unicode normalization form NFD.
0038  *   - Remove any Default_Ignorable_Code_Point.
0039  * 2) NFDICF
0040  *   - Apply unicode normalization form NFD.
0041  *   - Remove any Default_Ignorable_Code_Point.
0042  *   - Apply a full casefold (C + F).
0043  */
0044 enum utf8_normalization {
0045     UTF8_NFDI = 0,
0046     UTF8_NFDICF,
0047     UTF8_NMAX,
0048 };
0049
0050 struct unicode_map {
0051     unsigned int version;
0052     const struct utf8data *ntab[UTF8_NMAX];
0053     const struct utf8data_table *tables;
0054 };
0055
0056 int utf8_validate(const struct unicode_map *um, const struct qstr *str);
0057
0058 int utf8_strncmp(const struct unicode_map *um,
0059          const struct qstr *s1, const struct qstr *s2);
0060
0061 int utf8_strncasecmp(const struct unicode_map *um,
0062          const struct qstr *s1, const struct qstr *s2);
0063 int utf8_strncasecmp_folded(const struct unicode_map *um,
0064                 const struct qstr *cf,
0065                 const struct qstr *s1);
0066
0067 int utf8_normalize(const struct unicode_map *um, const struct qstr *str,
0068            unsigned char *dest, size_t dlen);
0069
0070 int utf8_casefold(const struct unicode_map *um, const struct qstr *str,
0071           unsigned char *dest, size_t dlen);
0072
0073 int utf8_casefold_hash(const struct unicode_map *um, const void *salt,
0074                struct qstr *str);
0075
0076 struct unicode_map *utf8_load(unsigned int version);
0077 void utf8_unload(struct unicode_map *um);
0078
0079 #endif /* _LINUX_UNICODE_H */