• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (c) 2014 SGI.
3  * Copyright (c) 2018 Collabora Ltd.
4  * All rights reserved.
5  *
6  * This program is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU General Public License as
8  * published by the Free Software Foundation.
9  *
10  * This program is distributed in the hope that it would be useful,
11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13  * GNU General Public License for more details.
14  *
15  */
16 
17 /*
18  * This code is adapted from the Linux Kernel.  We have a
19  * userspace version here such that the hashes will match that
20  * implementation.
21  */
22 
23 #include "config.h"
24 #include <stdint.h>
25 #include <unistd.h>
26 #include <string.h>
27 #include <limits.h>
28 #include <errno.h>
29 
30 #include "ext2_fs.h"
31 #include "ext2fs.h"
32 #include "ext2fsP.h"
33 
34 /* Encoding a unicode version number as a single unsigned int. */
35 #define UNICODE_MAJ_SHIFT		(16)
36 #define UNICODE_MIN_SHIFT		(8)
37 
38 #define UNICODE_AGE(MAJ, MIN, REV)			\
39 	(((unsigned int)(MAJ) << UNICODE_MAJ_SHIFT) |	\
40 	 ((unsigned int)(MIN) << UNICODE_MIN_SHIFT) |	\
41 	 ((unsigned int)(REV)))
42 
43 /* Needed in struct utf8cursor below. */
44 #define UTF8HANGULLEAF	(12)
45 
46 /*
47  * Cursor structure used by the normalizer.
48  */
49 struct utf8cursor {
50 	const struct utf8data	*data;
51 	const char	*s;
52 	const char	*p;
53 	const char	*ss;
54 	const char	*sp;
55 	unsigned int	len;
56 	unsigned int	slen;
57 	short int	ccc;
58 	short int	nccc;
59 	unsigned char	hangul[UTF8HANGULLEAF];
60 };
61 
62 /*
63  * Initialize a utf8cursor to normalize a string.
64  * Returns 0 on success.
65  * Returns -1 on failure.
66  */
67 // extern int utf8cursor(struct utf8cursor *u8c, const struct utf8data *data,
68 //		      const char *s);
69 // extern int utf8ncursor(struct utf8cursor *u8c, const struct utf8data *data,
70 //		       const char *s, size_t len);
71 
72 /*
73  * Get the next byte in the normalization.
74  * Returns a value > 0 && < 256 on success.
75  * Returns 0 when the end of the normalization is reached.
76  * Returns -1 if the string being normalized is not valid UTF-8.
77  */
78 // extern int utf8byte(struct utf8cursor *u8c);
79 
80 
81 struct utf8data {
82 	unsigned int maxage;
83 	unsigned int offset;
84 };
85 
86 #define __INCLUDED_FROM_UTF8NORM_C__
87 #include "utf8data.h"
88 #undef __INCLUDED_FROM_UTF8NORM_C__
89 
90 #define ARRAY_SIZE(array)			\
91         (sizeof(array) / sizeof(array[0]))
92 
93 #if 0
94 /* Highest unicode version supported by the data tables. */
95 static int utf8version_is_supported(uint8_t maj, uint8_t min, uint8_t rev)
96 {
97 	int i = ARRAY_SIZE(utf8agetab) - 1;
98 	unsigned int sb_utf8version = UNICODE_AGE(maj, min, rev);
99 
100 	while (i >= 0 && utf8agetab[i] != 0) {
101 		if (sb_utf8version == utf8agetab[i])
102 			return 1;
103 		i--;
104 	}
105 	return 0;
106 }
107 #endif
108 
109 #if 0
110 static int utf8version_latest(void)
111 {
112 	return utf8vers;
113 }
114 #endif
115 
116 /*
117  * UTF-8 valid ranges.
118  *
119  * The UTF-8 encoding spreads the bits of a 32bit word over several
120  * bytes. This table gives the ranges that can be held and how they'd
121  * be represented.
122  *
123  * 0x00000000 0x0000007F: 0xxxxxxx
124  * 0x00000000 0x000007FF: 110xxxxx 10xxxxxx
125  * 0x00000000 0x0000FFFF: 1110xxxx 10xxxxxx 10xxxxxx
126  * 0x00000000 0x001FFFFF: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
127  * 0x00000000 0x03FFFFFF: 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
128  * 0x00000000 0x7FFFFFFF: 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
129  *
130  * There is an additional requirement on UTF-8, in that only the
131  * shortest representation of a 32bit value is to be used.  A decoder
132  * must not decode sequences that do not satisfy this requirement.
133  * Thus the allowed ranges have a lower bound.
134  *
135  * 0x00000000 0x0000007F: 0xxxxxxx
136  * 0x00000080 0x000007FF: 110xxxxx 10xxxxxx
137  * 0x00000800 0x0000FFFF: 1110xxxx 10xxxxxx 10xxxxxx
138  * 0x00010000 0x001FFFFF: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
139  * 0x00200000 0x03FFFFFF: 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
140  * 0x04000000 0x7FFFFFFF: 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
141  *
142  * Actual unicode characters are limited to the range 0x0 - 0x10FFFF,
143  * 17 planes of 65536 values.  This limits the sequences actually seen
144  * even more, to just the following.
145  *
146  *          0 -     0x7F: 0                   - 0x7F
147  *       0x80 -    0x7FF: 0xC2 0x80           - 0xDF 0xBF
148  *      0x800 -   0xFFFF: 0xE0 0xA0 0x80      - 0xEF 0xBF 0xBF
149  *    0x10000 - 0x10FFFF: 0xF0 0x90 0x80 0x80 - 0xF4 0x8F 0xBF 0xBF
150  *
151  * Within those ranges the surrogates 0xD800 - 0xDFFF are not allowed.
152  *
153  * Note that the longest sequence seen with valid usage is 4 bytes,
154  * the same a single UTF-32 character.  This makes the UTF-8
155  * representation of Unicode strictly smaller than UTF-32.
156  *
157  * The shortest sequence requirement was introduced by:
158  *    Corrigendum #1: UTF-8 Shortest Form
159  * It can be found here:
160  *    http://www.unicode.org/versions/corrigendum1.html
161  *
162  */
163 
164 /*
165  * Return the number of bytes used by the current UTF-8 sequence.
166  * Assumes the input points to the first byte of a valid UTF-8
167  * sequence.
168  */
utf8clen(const char * s)169 static inline int utf8clen(const char *s)
170 {
171 	unsigned char c = *s;
172 
173 	return 1 + (c >= 0xC0) + (c >= 0xE0) + (c >= 0xF0);
174 }
175 
176 /*
177  * Decode a 3-byte UTF-8 sequence.
178  */
179 static unsigned int
utf8decode3(const char * str)180 utf8decode3(const char *str)
181 {
182 	unsigned int		uc;
183 
184 	uc = *str++ & 0x0F;
185 	uc <<= 6;
186 	uc |= *str++ & 0x3F;
187 	uc <<= 6;
188 	uc |= *str++ & 0x3F;
189 
190 	return uc;
191 }
192 
193 /*
194  * Encode a 3-byte UTF-8 sequence.
195  */
196 static int
utf8encode3(char * str,unsigned int val)197 utf8encode3(char *str, unsigned int val)
198 {
199 	str[2] = (val & 0x3F) | 0x80;
200 	val >>= 6;
201 	str[1] = (val & 0x3F) | 0x80;
202 	val >>= 6;
203 	str[0] = val | 0xE0;
204 
205 	return 3;
206 }
207 
208 /*
209  * utf8trie_t
210  *
211  * A compact binary tree, used to decode UTF-8 characters.
212  *
213  * Internal nodes are one byte for the node itself, and up to three
214  * bytes for an offset into the tree.  The first byte contains the
215  * following information:
216  *  NEXTBYTE  - flag        - advance to next byte if set
217  *  BITNUM    - 3 bit field - the bit number to tested
218  *  OFFLEN    - 2 bit field - number of bytes in the offset
219  * if offlen == 0 (non-branching node)
220  *  RIGHTPATH - 1 bit field - set if the following node is for the
221  *                            right-hand path (tested bit is set)
222  *  TRIENODE  - 1 bit field - set if the following node is an internal
223  *                            node, otherwise it is a leaf node
224  * if offlen != 0 (branching node)
225  *  LEFTNODE  - 1 bit field - set if the left-hand node is internal
226  *  RIGHTNODE - 1 bit field - set if the right-hand node is internal
227  *
228  * Due to the way utf8 works, there cannot be branching nodes with
229  * NEXTBYTE set, and moreover those nodes always have a righthand
230  * descendant.
231  */
232 typedef const unsigned char utf8trie_t;
233 #define BITNUM		0x07
234 #define NEXTBYTE	0x08
235 #define OFFLEN		0x30
236 #define OFFLEN_SHIFT	4
237 #define RIGHTPATH	0x40
238 #define TRIENODE	0x80
239 #define RIGHTNODE	0x40
240 #define LEFTNODE	0x80
241 
242 /*
243  * utf8leaf_t
244  *
245  * The leaves of the trie are embedded in the trie, and so the same
246  * underlying datatype: unsigned char.
247  *
248  * leaf[0]: The unicode version, stored as a generation number that is
249  *          an index into utf8agetab[].  With this we can filter code
250  *          points based on the unicode version in which they were
251  *          defined.  The CCC of a non-defined code point is 0.
252  * leaf[1]: Canonical Combining Class. During normalization, we need
253  *          to do a stable sort into ascending order of all characters
254  *          with a non-zero CCC that occur between two characters with
255  *          a CCC of 0, or at the begin or end of a string.
256  *          The unicode standard guarantees that all CCC values are
257  *          between 0 and 254 inclusive, which leaves 255 available as
258  *          a special value.
259  *          Code points with CCC 0 are known as stoppers.
260  * leaf[2]: Decomposition. If leaf[1] == 255, then leaf[2] is the
261  *          start of a NUL-terminated string that is the decomposition
262  *          of the character.
263  *          The CCC of a decomposable character is the same as the CCC
264  *          of the first character of its decomposition.
265  *          Some characters decompose as the empty string: these are
266  *          characters with the Default_Ignorable_Code_Point property.
267  *          These do affect normalization, as they all have CCC 0.
268  *
269  * The decompositions in the trie have been fully expanded, with the
270  * exception of Hangul syllables, which are decomposed algorithmically.
271  *
272  * Casefolding, if applicable, is also done using decompositions.
273  *
274  * The trie is constructed in such a way that leaves exist for all
275  * UTF-8 sequences that match the criteria from the "UTF-8 valid
276  * ranges" comment above, and only for those sequences.  Therefore a
277  * lookup in the trie can be used to validate the UTF-8 input.
278  */
279 typedef const unsigned char utf8leaf_t;
280 
281 #define LEAF_GEN(LEAF)	((LEAF)[0])
282 #define LEAF_CCC(LEAF)	((LEAF)[1])
283 #define LEAF_STR(LEAF)	((const char *)((LEAF) + 2))
284 
285 #define MINCCC		(0)
286 #define MAXCCC		(254)
287 #define STOPPER		(0)
288 #define	DECOMPOSE	(255)
289 
290 /* Marker for hangul syllable decomposition. */
291 #define HANGUL		((char)(255))
292 /* Size of the synthesized leaf used for Hangul syllable decomposition. */
293 #define UTF8HANGULLEAF	(12)
294 
295 /*
296  * Hangul decomposition (algorithm from Section 3.12 of Unicode 6.3.0)
297  *
298  * AC00;<Hangul Syllable, First>;Lo;0;L;;;;;N;;;;;
299  * D7A3;<Hangul Syllable, Last>;Lo;0;L;;;;;N;;;;;
300  *
301  * SBase = 0xAC00
302  * LBase = 0x1100
303  * VBase = 0x1161
304  * TBase = 0x11A7
305  * LCount = 19
306  * VCount = 21
307  * TCount = 28
308  * NCount = 588 (VCount * TCount)
309  * SCount = 11172 (LCount * NCount)
310  *
311  * Decomposition:
312  *   SIndex = s - SBase
313  *
314  * LV (Canonical/Full)
315  *   LIndex = SIndex / NCount
316  *   VIndex = (Sindex % NCount) / TCount
317  *   LPart = LBase + LIndex
318  *   VPart = VBase + VIndex
319  *
320  * LVT (Canonical)
321  *   LVIndex = (SIndex / TCount) * TCount
322  *   TIndex = (Sindex % TCount)
323  *   LVPart = SBase + LVIndex
324  *   TPart = TBase + TIndex
325  *
326  * LVT (Full)
327  *   LIndex = SIndex / NCount
328  *   VIndex = (Sindex % NCount) / TCount
329  *   TIndex = (Sindex % TCount)
330  *   LPart = LBase + LIndex
331  *   VPart = VBase + VIndex
332  *   if (TIndex == 0) {
333  *          d = <LPart, VPart>
334  *   } else {
335  *          TPart = TBase + TIndex
336  *          d = <LPart, TPart, VPart>
337  *   }
338  */
339 
340 /* Constants */
341 #define SB	(0xAC00)
342 #define LB	(0x1100)
343 #define VB	(0x1161)
344 #define TB	(0x11A7)
345 #define LC	(19)
346 #define VC	(21)
347 #define TC	(28)
348 #define NC	(VC * TC)
349 #define SC	(LC * NC)
350 
351 /* Algorithmic decomposition of hangul syllable. */
352 static utf8leaf_t *
utf8hangul(const char * str,unsigned char * hangul)353 utf8hangul(const char *str, unsigned char *hangul)
354 {
355 	unsigned int	si;
356 	unsigned int	li;
357 	unsigned int	vi;
358 	unsigned int	ti;
359 	unsigned char	*h;
360 
361 	/* Calculate the SI, LI, VI, and TI values. */
362 	si = utf8decode3(str) - SB;
363 	li = si / NC;
364 	vi = (si % NC) / TC;
365 	ti = si % TC;
366 
367 	/* Fill in base of leaf. */
368 	h = hangul;
369 	LEAF_GEN(h) = 2;
370 	LEAF_CCC(h) = DECOMPOSE;
371 	h += 2;
372 
373 	/* Add LPart, a 3-byte UTF-8 sequence. */
374 	h += utf8encode3((char *)h, li + LB);
375 
376 	/* Add VPart, a 3-byte UTF-8 sequence. */
377 	h += utf8encode3((char *)h, vi + VB);
378 
379 	/* Add TPart if required, also a 3-byte UTF-8 sequence. */
380 	if (ti)
381 		h += utf8encode3((char *)h, ti + TB);
382 
383 	/* Terminate string. */
384 	h[0] = '\0';
385 
386 	return hangul;
387 }
388 
389 /*
390  * Use trie to scan s, touching at most len bytes.
391  * Returns the leaf if one exists, NULL otherwise.
392  *
393  * A non-NULL return guarantees that the UTF-8 sequence starting at s
394  * is well-formed and corresponds to a known unicode code point.  The
395  * shorthand for this will be "is valid UTF-8 unicode".
396  */
utf8nlookup(const struct utf8data * data,unsigned char * hangul,const char * s,size_t len)397 static utf8leaf_t *utf8nlookup(const struct utf8data *data,
398 			       unsigned char *hangul, const char *s, size_t len)
399 {
400 	utf8trie_t	*trie;
401 	int		offlen;
402 	int		offset;
403 	int		mask;
404 	int		node;
405 
406 	if (!data)
407 		return NULL;
408 	if (len == 0)
409 		return NULL;
410 
411 	trie = utf8data + data->offset;
412 	node = 1;
413 	while (node) {
414 		offlen = (*trie & OFFLEN) >> OFFLEN_SHIFT;
415 		if (*trie & NEXTBYTE) {
416 			if (--len == 0)
417 				return NULL;
418 			s++;
419 		}
420 		mask = 1 << (*trie & BITNUM);
421 		if (*s & mask) {
422 			/* Right leg */
423 			if (offlen) {
424 				/* Right node at offset of trie */
425 				node = (*trie & RIGHTNODE);
426 				offset = trie[offlen];
427 				while (--offlen) {
428 					offset <<= 8;
429 					offset |= trie[offlen];
430 				}
431 				trie += offset;
432 			} else if (*trie & RIGHTPATH) {
433 				/* Right node after this node */
434 				node = (*trie & TRIENODE);
435 				trie++;
436 			} else {
437 				/* No right node. */
438 				return NULL;
439 			}
440 		} else {
441 			/* Left leg */
442 			if (offlen) {
443 				/* Left node after this node. */
444 				node = (*trie & LEFTNODE);
445 				trie += offlen + 1;
446 			} else if (*trie & RIGHTPATH) {
447 				/* No left node. */
448 				return NULL;
449 			} else {
450 				/* Left node after this node */
451 				node = (*trie & TRIENODE);
452 				trie++;
453 			}
454 		}
455 	}
456 	/*
457 	 * Hangul decomposition is done algorithmically. These are the
458 	 * codepoints >= 0xAC00 and <= 0xD7A3. Their UTF-8 encoding is
459 	 * always 3 bytes long, so s has been advanced twice, and the
460 	 * start of the sequence is at s-2.
461 	 */
462 	if (LEAF_CCC(trie) == DECOMPOSE && LEAF_STR(trie)[0] == HANGUL)
463 		trie = utf8hangul(s - 2, hangul);
464 	return trie;
465 }
466 
467 /*
468  * Use trie to scan s.
469  * Returns the leaf if one exists, NULL otherwise.
470  *
471  * Forwards to utf8nlookup().
472  */
utf8lookup(const struct utf8data * data,unsigned char * hangul,const char * s)473 static utf8leaf_t *utf8lookup(const struct utf8data *data,
474 			      unsigned char *hangul, const char *s)
475 {
476 	return utf8nlookup(data, hangul, s, (size_t)-1);
477 }
478 
479 #if 0
480 /*
481  * Maximum age of any character in s.
482  * Return -1 if s is not valid UTF-8 unicode.
483  * Return 0 if only non-assigned code points are used.
484  */
485 static int utf8agemax(const struct utf8data *data, const char *s)
486 {
487 	utf8leaf_t	*leaf;
488 	int		age = 0;
489 	int		leaf_age;
490 	unsigned char	hangul[UTF8HANGULLEAF];
491 
492 	if (!data)
493 		return -1;
494 
495 	while (*s) {
496 		leaf = utf8lookup(data, hangul, s);
497 		if (!leaf)
498 			return -1;
499 
500 		leaf_age = utf8agetab[LEAF_GEN(leaf)];
501 		if (leaf_age <= data->maxage && leaf_age > age)
502 			age = leaf_age;
503 		s += utf8clen(s);
504 	}
505 	return age;
506 }
507 #endif
508 
509 #if 0
510 /*
511  * Minimum age of any character in s.
512  * Return -1 if s is not valid UTF-8 unicode.
513  * Return 0 if non-assigned code points are used.
514  */
515 static int utf8agemin(const struct utf8data *data, const char *s)
516 {
517 	utf8leaf_t	*leaf;
518 	int		age;
519 	int		leaf_age;
520 	unsigned char	hangul[UTF8HANGULLEAF];
521 
522 	if (!data)
523 		return -1;
524 	age = data->maxage;
525 	while (*s) {
526 		leaf = utf8lookup(data, hangul, s);
527 		if (!leaf)
528 			return -1;
529 		leaf_age = utf8agetab[LEAF_GEN(leaf)];
530 		if (leaf_age <= data->maxage && leaf_age < age)
531 			age = leaf_age;
532 		s += utf8clen(s);
533 	}
534 	return age;
535 }
536 #endif
537 
538 #if 0
539 /*
540  * Maximum age of any character in s, touch at most len bytes.
541  * Return -1 if s is not valid UTF-8 unicode.
542  */
543 static int utf8nagemax(const struct utf8data *data, const char *s, size_t len)
544 {
545 	utf8leaf_t	*leaf;
546 	int		age = 0;
547 	int		leaf_age;
548 	unsigned char	hangul[UTF8HANGULLEAF];
549 
550 	if (!data)
551 		return -1;
552 
553 	while (len && *s) {
554 		leaf = utf8nlookup(data, hangul, s, len);
555 		if (!leaf)
556 			return -1;
557 		leaf_age = utf8agetab[LEAF_GEN(leaf)];
558 		if (leaf_age <= data->maxage && leaf_age > age)
559 			age = leaf_age;
560 		len -= utf8clen(s);
561 		s += utf8clen(s);
562 	}
563 	return age;
564 }
565 #endif
566 
567 #if 0
568 /*
569  * Maximum age of any character in s, touch at most len bytes.
570  * Return -1 if s is not valid UTF-8 unicode.
571  */
572 static int utf8nagemin(const struct utf8data *data, const char *s, size_t len)
573 {
574 	utf8leaf_t	*leaf;
575 	int		leaf_age;
576 	int		age;
577 	unsigned char	hangul[UTF8HANGULLEAF];
578 
579 	if (!data)
580 		return -1;
581 	age = data->maxage;
582 	while (len && *s) {
583 		leaf = utf8nlookup(data, hangul, s, len);
584 		if (!leaf)
585 			return -1;
586 		leaf_age = utf8agetab[LEAF_GEN(leaf)];
587 		if (leaf_age <= data->maxage && leaf_age < age)
588 			age = leaf_age;
589 		len -= utf8clen(s);
590 		s += utf8clen(s);
591 	}
592 	return age;
593 }
594 #endif
595 
596 #if 0
597 /*
598  * Length of the normalization of s.
599  * Return -1 if s is not valid UTF-8 unicode.
600  *
601  * A string of Default_Ignorable_Code_Point has length 0.
602  */
603 static ssize_t utf8len(const struct utf8data *data, const char *s)
604 {
605 	utf8leaf_t	*leaf;
606 	size_t		ret = 0;
607 	unsigned char	hangul[UTF8HANGULLEAF];
608 
609 	if (!data)
610 		return -1;
611 	while (*s) {
612 		leaf = utf8lookup(data, hangul, s);
613 		if (!leaf)
614 			return -1;
615 		if (utf8agetab[LEAF_GEN(leaf)] > data->maxage)
616 			ret += utf8clen(s);
617 		else if (LEAF_CCC(leaf) == DECOMPOSE)
618 			ret += strlen(LEAF_STR(leaf));
619 		else
620 			ret += utf8clen(s);
621 		s += utf8clen(s);
622 	}
623 	return ret;
624 }
625 #endif
626 
627 #if 0
628 /*
629  * Length of the normalization of s, touch at most len bytes.
630  * Return -1 if s is not valid UTF-8 unicode.
631  */
632 static ssize_t utf8nlen(const struct utf8data *data, const char *s, size_t len)
633 {
634 	utf8leaf_t	*leaf;
635 	size_t		ret = 0;
636 	unsigned char	hangul[UTF8HANGULLEAF];
637 
638 	if (!data)
639 		return -1;
640 	while (len && *s) {
641 		leaf = utf8nlookup(data, hangul, s, len);
642 		if (!leaf)
643 			return -1;
644 		if (utf8agetab[LEAF_GEN(leaf)] > data->maxage)
645 			ret += utf8clen(s);
646 		else if (LEAF_CCC(leaf) == DECOMPOSE)
647 			ret += strlen(LEAF_STR(leaf));
648 		else
649 			ret += utf8clen(s);
650 		len -= utf8clen(s);
651 		s += utf8clen(s);
652 	}
653 	return ret;
654 }
655 #endif
656 
657 /*
658  * Set up an utf8cursor for use by utf8byte().
659  *
660  *   u8c    : pointer to cursor.
661  *   data   : const struct utf8data to use for normalization.
662  *   s      : string.
663  *   len    : length of s.
664  *
665  * Returns -1 on error, 0 on success.
666  */
utf8ncursor(struct utf8cursor * u8c,const struct utf8data * data,const char * s,size_t len)667 static int utf8ncursor(struct utf8cursor *u8c, const struct utf8data *data,
668 		const char *s, size_t len)
669 {
670 	if (!data)
671 		return -1;
672 	if (!s)
673 		return -1;
674 	u8c->data = data;
675 	u8c->s = s;
676 	u8c->p = NULL;
677 	u8c->ss = NULL;
678 	u8c->sp = NULL;
679 	u8c->len = len;
680 	u8c->slen = 0;
681 	u8c->ccc = STOPPER;
682 	u8c->nccc = STOPPER;
683 	/* Check we didn't clobber the maximum length. */
684 	if (u8c->len != len)
685 		return -1;
686 	/* The first byte of s may not be an utf8 continuation. */
687 	if (len > 0 && (*s & 0xC0) == 0x80)
688 		return -1;
689 	return 0;
690 }
691 
692 #if 0
693 /*
694  * Set up an utf8cursor for use by utf8byte().
695  *
696  *   u8c    : pointer to cursor.
697  *   data   : const struct utf8data to use for normalization.
698  *   s      : NUL-terminated string.
699  *
700  * Returns -1 on error, 0 on success.
701  */
702 static int utf8cursor(struct utf8cursor *u8c, const struct utf8data *data,
703 	       const char *s)
704 {
705 	return utf8ncursor(u8c, data, s, (unsigned int)-1);
706 }
707 #endif
708 
709 /*
710  * Get one byte from the normalized form of the string described by u8c.
711  *
712  * Returns the byte cast to an unsigned char on succes, and -1 on failure.
713  *
714  * The cursor keeps track of the location in the string in u8c->s.
715  * When a character is decomposed, the current location is stored in
716  * u8c->p, and u8c->s is set to the start of the decomposition. Note
717  * that bytes from a decomposition do not count against u8c->len.
718  *
719  * Characters are emitted if they match the current CCC in u8c->ccc.
720  * Hitting end-of-string while u8c->ccc == STOPPER means we're done,
721  * and the function returns 0 in that case.
722  *
723  * Sorting by CCC is done by repeatedly scanning the string.  The
724  * values of u8c->s and u8c->p are stored in u8c->ss and u8c->sp at
725  * the start of the scan.  The first pass finds the lowest CCC to be
726  * emitted and stores it in u8c->nccc, the second pass emits the
727  * characters with this CCC and finds the next lowest CCC. This limits
728  * the number of passes to 1 + the number of different CCCs in the
729  * sequence being scanned.
730  *
731  * Therefore:
732  *  u8c->p  != NULL -> a decomposition is being scanned.
733  *  u8c->ss != NULL -> this is a repeating scan.
734  *  u8c->ccc == -1   -> this is the first scan of a repeating scan.
735  */
utf8byte(struct utf8cursor * u8c)736 static int utf8byte(struct utf8cursor *u8c)
737 {
738 	utf8leaf_t *leaf;
739 	int ccc;
740 
741 	for (;;) {
742 		/* Check for the end of a decomposed character. */
743 		if (u8c->p && *u8c->s == '\0') {
744 			u8c->s = u8c->p;
745 			u8c->p = NULL;
746 		}
747 
748 		/* Check for end-of-string. */
749 		if (!u8c->p && (u8c->len == 0 || *u8c->s == '\0')) {
750 			/* There is no next byte. */
751 			if (u8c->ccc == STOPPER)
752 				return 0;
753 			/* End-of-string during a scan counts as a stopper. */
754 			ccc = STOPPER;
755 			goto ccc_mismatch;
756 		} else if ((*u8c->s & 0xC0) == 0x80) {
757 			/* This is a continuation of the current character. */
758 			if (!u8c->p)
759 				u8c->len--;
760 			return (unsigned char)*u8c->s++;
761 		}
762 
763 		/* Look up the data for the current character. */
764 		if (u8c->p) {
765 			leaf = utf8lookup(u8c->data, u8c->hangul, u8c->s);
766 		} else {
767 			leaf = utf8nlookup(u8c->data, u8c->hangul,
768 					   u8c->s, u8c->len);
769 		}
770 
771 		/* No leaf found implies that the input is a binary blob. */
772 		if (!leaf)
773 			return -1;
774 
775 		ccc = LEAF_CCC(leaf);
776 		/* Characters that are too new have CCC 0. */
777 		if (utf8agetab[LEAF_GEN(leaf)] > u8c->data->maxage) {
778 			ccc = STOPPER;
779 		} else if (ccc == DECOMPOSE) {
780 			u8c->len -= utf8clen(u8c->s);
781 			u8c->p = u8c->s + utf8clen(u8c->s);
782 			u8c->s = LEAF_STR(leaf);
783 			/* Empty decomposition implies CCC 0. */
784 			if (*u8c->s == '\0') {
785 				if (u8c->ccc == STOPPER)
786 					continue;
787 				ccc = STOPPER;
788 				goto ccc_mismatch;
789 			}
790 
791 			leaf = utf8lookup(u8c->data, u8c->hangul, u8c->s);
792 			if (!leaf)
793 				return -1;
794 			ccc = LEAF_CCC(leaf);
795 		}
796 
797 		/*
798 		 * If this is not a stopper, then see if it updates
799 		 * the next canonical class to be emitted.
800 		 */
801 		if (ccc != STOPPER && u8c->ccc < ccc && ccc < u8c->nccc)
802 			u8c->nccc = ccc;
803 
804 		/*
805 		 * Return the current byte if this is the current
806 		 * combining class.
807 		 */
808 		if (ccc == u8c->ccc) {
809 			if (!u8c->p)
810 				u8c->len--;
811 			return (unsigned char)*u8c->s++;
812 		}
813 
814 		/* Current combining class mismatch. */
815 ccc_mismatch:
816 		if (u8c->nccc == STOPPER) {
817 			/*
818 			 * Scan forward for the first canonical class
819 			 * to be emitted.  Save the position from
820 			 * which to restart.
821 			 */
822 			u8c->ccc = MINCCC - 1;
823 			u8c->nccc = ccc;
824 			u8c->sp = u8c->p;
825 			u8c->ss = u8c->s;
826 			u8c->slen = u8c->len;
827 			if (!u8c->p)
828 				u8c->len -= utf8clen(u8c->s);
829 			u8c->s += utf8clen(u8c->s);
830 		} else if (ccc != STOPPER) {
831 			/* Not a stopper, and not the ccc we're emitting. */
832 			if (!u8c->p)
833 				u8c->len -= utf8clen(u8c->s);
834 			u8c->s += utf8clen(u8c->s);
835 		} else if (u8c->nccc != MAXCCC + 1) {
836 			/* At a stopper, restart for next ccc. */
837 			u8c->ccc = u8c->nccc;
838 			u8c->nccc = MAXCCC + 1;
839 			u8c->s = u8c->ss;
840 			u8c->p = u8c->sp;
841 			u8c->len = u8c->slen;
842 		} else {
843 			/* All done, proceed from here. */
844 			u8c->ccc = STOPPER;
845 			u8c->nccc = STOPPER;
846 			u8c->sp = NULL;
847 			u8c->ss = NULL;
848 			u8c->slen = 0;
849 		}
850 	}
851 }
852 
853 #if 0
854 /*
855  * Look for the correct const struct utf8data for a unicode version.
856  * Returns NULL if the version requested is too new.
857  *
858  * Two normalization forms are supported: nfdi and nfdicf.
859  *
860  * nfdi:
861  *  - Apply unicode normalization form NFD.
862  *  - Remove any Default_Ignorable_Code_Point.
863  *
864  * nfdicf:
865  *  - Apply unicode normalization form NFD.
866  *  - Remove any Default_Ignorable_Code_Point.
867  *  - Apply a full casefold (C + F).
868  */
869 static const struct utf8data *utf8nfdi(unsigned int maxage)
870 {
871 	int i = ARRAY_SIZE(utf8nfdidata) - 1;
872 
873 	while (maxage < utf8nfdidata[i].maxage)
874 		i--;
875 	if (maxage > utf8nfdidata[i].maxage)
876 		return NULL;
877 	return &utf8nfdidata[i];
878 }
879 #endif
880 
utf8nfdicf(unsigned int maxage)881 static const struct utf8data *utf8nfdicf(unsigned int maxage)
882 {
883 	int i = ARRAY_SIZE(utf8nfdicfdata) - 1;
884 
885 	while (maxage < utf8nfdicfdata[i].maxage)
886 		i--;
887 	if (maxage > utf8nfdicfdata[i].maxage)
888 		return NULL;
889 	return &utf8nfdicfdata[i];
890 }
891 
utf8_casefold(const struct ext2fs_nls_table * table,const unsigned char * str,size_t len,unsigned char * dest,size_t dlen)892 static int utf8_casefold(const struct ext2fs_nls_table *table,
893 			  const unsigned char *str, size_t len,
894 			  unsigned char *dest, size_t dlen)
895 {
896 	const struct utf8data *data = utf8nfdicf(table->version);
897 	struct utf8cursor cur;
898 	size_t nlen = 0;
899 
900 	if (utf8ncursor(&cur, data, (const char *) str, len) < 0)
901 		goto invalid_seq;
902 
903 	for (nlen = 0; nlen < dlen; nlen++) {
904 		int c = utf8byte(&cur);
905 
906 		dest[nlen] = c;
907 		if (!c)
908 			return nlen;
909 		if (c == -1)
910 			break;
911 	}
912 
913 	return -ENAMETOOLONG;
914 
915 invalid_seq:
916 	if (dlen < len)
917 		return -ENAMETOOLONG;
918 
919 	/* Signal invalid sequence */
920 	return -EINVAL;
921 }
922 
utf8_validate(const struct ext2fs_nls_table * table,char * s,size_t len,char ** pos)923 static int utf8_validate(const struct ext2fs_nls_table *table,
924 			 char *s, size_t len, char **pos)
925 {
926 	const struct utf8data *data = utf8nfdicf(table->version);
927 	utf8leaf_t	*leaf;
928 	unsigned char	hangul[UTF8HANGULLEAF];
929 
930 	if (!data)
931 		return -1;
932 	while (len && *s) {
933 		leaf = utf8nlookup(data, hangul, s, len);
934 		if (!leaf) {
935 			*pos = s;
936 			return 1;
937 		}
938 		len -= utf8clen(s);
939 		s += utf8clen(s);
940 	}
941 	return 0;
942 }
943 
utf8_casefold_cmp(const struct ext2fs_nls_table * table,const unsigned char * str1,size_t len1,const unsigned char * str2,size_t len2)944 static int utf8_casefold_cmp(const struct ext2fs_nls_table *table,
945 			     const unsigned char *str1, size_t len1,
946 			     const unsigned char *str2, size_t len2)
947 {
948 	const struct utf8data *data = utf8nfdicf(table->version);
949 	int c1, c2;
950 	struct utf8cursor cur1, cur2;
951 
952 	if (utf8ncursor(&cur1, data, (const char *) str1, len1) < 0)
953 		return -1;
954 	if (utf8ncursor(&cur2, data, (const char *) str2, len2) < 0)
955 		return -1;
956 
957 	do {
958 		c1 = utf8byte(&cur1);
959 		c2 = utf8byte(&cur2);
960 
961 		if (c1 < 0 || c2 < 0)
962 			return -1;
963 		if (c1 != c2)
964 			return c1 - c2;
965 	} while (c1);
966 
967 	return 0;
968 }
969 
970 static const struct ext2fs_nls_ops utf8_ops = {
971 	.casefold = utf8_casefold,
972 	.validate = utf8_validate,
973 	.casefold_cmp = utf8_casefold_cmp,
974 };
975 
976 static const struct ext2fs_nls_table nls_utf8 = {
977 	.ops = &utf8_ops,
978 	.version = UNICODE_AGE(12, 1, 0),
979 };
980 
ext2fs_load_nls_table(int encoding)981 const struct ext2fs_nls_table *ext2fs_load_nls_table(int encoding)
982 {
983 	if (encoding == EXT4_ENC_UTF8_12_1)
984 		return &nls_utf8;
985 
986 	return NULL;
987 }
988 
ext2fs_check_encoded_name(const struct ext2fs_nls_table * table,char * name,size_t len,char ** pos)989 int ext2fs_check_encoded_name(const struct ext2fs_nls_table *table,
990 			      char *name, size_t len, char **pos)
991 {
992 	return table->ops->validate(table, name, len, pos);
993 }
994 
ext2fs_casefold_cmp(const struct ext2fs_nls_table * table,const unsigned char * str1,size_t len1,const unsigned char * str2,size_t len2)995 int ext2fs_casefold_cmp(const struct ext2fs_nls_table *table,
996 			const unsigned char *str1, size_t len1,
997 			const unsigned char *str2, size_t len2)
998 {
999 	return table->ops->casefold_cmp(table, str1, len1, str2, len2);
1000 }
1001