xref: /linux/fs/unicode/utf8-selftest.c (revision 9208c05f9fdfd927ea160b97dfef3c379049fff2)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * Kernel module for testing utf-8 support.
4  *
5  * Copyright 2017 Collabora Ltd.
6  */
7 
8 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
9 
10 #include <linux/module.h>
11 #include <linux/printk.h>
12 #include <linux/unicode.h>
13 #include <linux/dcache.h>
14 
15 #include "utf8n.h"
16 
17 static unsigned int failed_tests;
18 static unsigned int total_tests;
19 
20 #define _test(cond, func, line, fmt, ...) do {				\
21 		total_tests++;						\
22 		if (!cond) {						\
23 			failed_tests++;					\
24 			pr_err("test %s:%d Failed: %s%s",		\
25 			       func, line, #cond, (fmt?":":"."));	\
26 			if (fmt)					\
27 				pr_err(fmt, ##__VA_ARGS__);		\
28 		}							\
29 	} while (0)
30 #define test_f(cond, fmt, ...) _test(cond, __func__, __LINE__, fmt, ##__VA_ARGS__)
31 #define test(cond) _test(cond, __func__, __LINE__, "")
32 
33 static const struct {
34 	/* UTF-8 strings in this vector _must_ be NULL-terminated. */
35 	unsigned char str[10];
36 	unsigned char dec[10];
37 } nfdi_test_data[] = {
38 	/* Trivial sequence */
39 	{
40 		/* "ABba" decomposes to itself */
41 		.str = "aBba",
42 		.dec = "aBba",
43 	},
44 	/* Simple equivalent sequences */
45 	{
46                /* 'VULGAR FRACTION ONE QUARTER' cannot decompose to
47                   'NUMBER 1' + 'FRACTION SLASH' + 'NUMBER 4' on
48                   canonical decomposition */
49                .str = {0xc2, 0xbc, 0x00},
50 	       .dec = {0xc2, 0xbc, 0x00},
51 	},
52 	{
53 		/* 'LATIN SMALL LETTER A WITH DIAERESIS' decomposes to
54 		   'LETTER A' + 'COMBINING DIAERESIS' */
55 		.str = {0xc3, 0xa4, 0x00},
56 		.dec = {0x61, 0xcc, 0x88, 0x00},
57 	},
58 	{
59 		/* 'LATIN SMALL LETTER LJ' can't decompose to
60 		   'LETTER L' + 'LETTER J' on canonical decomposition */
61 		.str = {0xC7, 0x89, 0x00},
62 		.dec = {0xC7, 0x89, 0x00},
63 	},
64 	{
65 		/* GREEK ANO TELEIA decomposes to MIDDLE DOT */
66 		.str = {0xCE, 0x87, 0x00},
67 		.dec = {0xC2, 0xB7, 0x00}
68 	},
69 	/* Canonical ordering */
70 	{
71 		/* A + 'COMBINING ACUTE ACCENT' + 'COMBINING OGONEK' decomposes
72 		   to A + 'COMBINING OGONEK' + 'COMBINING ACUTE ACCENT' */
73 		.str = {0x41, 0xcc, 0x81, 0xcc, 0xa8, 0x0},
74 		.dec = {0x41, 0xcc, 0xa8, 0xcc, 0x81, 0x0},
75 	},
76 	{
77 		/* 'LATIN SMALL LETTER A WITH DIAERESIS' + 'COMBINING OGONEK'
78 		   decomposes to
79 		   'LETTER A' + 'COMBINING OGONEK' + 'COMBINING DIAERESIS' */
80 		.str = {0xc3, 0xa4, 0xCC, 0xA8, 0x00},
81 
82 		.dec = {0x61, 0xCC, 0xA8, 0xcc, 0x88, 0x00},
83 	},
84 
85 };
86 
87 static const struct {
88 	/* UTF-8 strings in this vector _must_ be NULL-terminated. */
89 	unsigned char str[30];
90 	unsigned char ncf[30];
91 } nfdicf_test_data[] = {
92 	/* Trivial sequences */
93 	{
94 		/* "ABba" folds to lowercase */
95 		.str = {0x41, 0x42, 0x62, 0x61, 0x00},
96 		.ncf = {0x61, 0x62, 0x62, 0x61, 0x00},
97 	},
98 	{
99 		/* All ASCII folds to lower-case */
100 		.str = "ABCDEFGHIJKLMNOPQRSTUVWXYZ0.1",
101 		.ncf = "abcdefghijklmnopqrstuvwxyz0.1",
102 	},
103 	{
104 		/* LATIN SMALL LETTER SHARP S folds to
105 		   LATIN SMALL LETTER S + LATIN SMALL LETTER S */
106 		.str = {0xc3, 0x9f, 0x00},
107 		.ncf = {0x73, 0x73, 0x00},
108 	},
109 	{
110 		/* LATIN CAPITAL LETTER A WITH RING ABOVE folds to
111 		   LATIN SMALL LETTER A + COMBINING RING ABOVE */
112 		.str = {0xC3, 0x85, 0x00},
113 		.ncf = {0x61, 0xcc, 0x8a, 0x00},
114 	},
115 	/* Introduced by UTF-8.0.0. */
116 	/* Cherokee letters are interesting test-cases because they fold
117 	   to upper-case.  Before 8.0.0, Cherokee lowercase were
118 	   undefined, thus, the folding from LC is not stable between
119 	   7.0.0 -> 8.0.0, but it is from UC. */
120 	{
121 		/* CHEROKEE SMALL LETTER A folds to CHEROKEE LETTER A */
122 		.str = {0xea, 0xad, 0xb0, 0x00},
123 		.ncf = {0xe1, 0x8e, 0xa0, 0x00},
124 	},
125 	{
126 		/* CHEROKEE SMALL LETTER YE folds to CHEROKEE LETTER YE */
127 		.str = {0xe1, 0x8f, 0xb8, 0x00},
128 		.ncf = {0xe1, 0x8f, 0xb0, 0x00},
129 	},
130 	{
131 		/* OLD HUNGARIAN CAPITAL LETTER AMB folds to
132 		   OLD HUNGARIAN SMALL LETTER AMB */
133 		.str = {0xf0, 0x90, 0xb2, 0x83, 0x00},
134 		.ncf = {0xf0, 0x90, 0xb3, 0x83, 0x00},
135 	},
136 	/* Introduced by UTF-9.0.0. */
137 	{
138 		/* OSAGE CAPITAL LETTER CHA folds to
139 		   OSAGE SMALL LETTER CHA */
140 		.str = {0xf0, 0x90, 0x92, 0xb5, 0x00},
141 		.ncf = {0xf0, 0x90, 0x93, 0x9d, 0x00},
142 	},
143 	{
144 		/* LATIN CAPITAL LETTER SMALL CAPITAL I folds to
145 		   LATIN LETTER SMALL CAPITAL I */
146 		.str = {0xea, 0x9e, 0xae, 0x00},
147 		.ncf = {0xc9, 0xaa, 0x00},
148 	},
149 	/* Introduced by UTF-11.0.0. */
150 	{
151 		/* GEORGIAN SMALL LETTER AN folds to GEORGIAN MTAVRULI
152 		   CAPITAL LETTER AN */
153 		.str = {0xe1, 0xb2, 0x90, 0x00},
154 		.ncf = {0xe1, 0x83, 0x90, 0x00},
155 	}
156 };
157 
158 static ssize_t utf8len(const struct unicode_map *um, enum utf8_normalization n,
159 		const char *s)
160 {
161 	return utf8nlen(um, n, s, (size_t)-1);
162 }
163 
164 static int utf8cursor(struct utf8cursor *u8c, const struct unicode_map *um,
165 		enum utf8_normalization n, const char *s)
166 {
167 	return utf8ncursor(u8c, um, n, s, (unsigned int)-1);
168 }
169 
170 static void check_utf8_nfdi(struct unicode_map *um)
171 {
172 	int i;
173 	struct utf8cursor u8c;
174 
175 	for (i = 0; i < ARRAY_SIZE(nfdi_test_data); i++) {
176 		int len = strlen(nfdi_test_data[i].str);
177 		int nlen = strlen(nfdi_test_data[i].dec);
178 		int j = 0;
179 		unsigned char c;
180 
181 		test((utf8len(um, UTF8_NFDI, nfdi_test_data[i].str) == nlen));
182 		test((utf8nlen(um, UTF8_NFDI, nfdi_test_data[i].str, len) ==
183 			nlen));
184 
185 		if (utf8cursor(&u8c, um, UTF8_NFDI, nfdi_test_data[i].str) < 0)
186 			pr_err("can't create cursor\n");
187 
188 		while ((c = utf8byte(&u8c)) > 0) {
189 			test_f((c == nfdi_test_data[i].dec[j]),
190 			       "Unexpected byte 0x%x should be 0x%x\n",
191 			       c, nfdi_test_data[i].dec[j]);
192 			j++;
193 		}
194 
195 		test((j == nlen));
196 	}
197 }
198 
199 static void check_utf8_nfdicf(struct unicode_map *um)
200 {
201 	int i;
202 	struct utf8cursor u8c;
203 
204 	for (i = 0; i < ARRAY_SIZE(nfdicf_test_data); i++) {
205 		int len = strlen(nfdicf_test_data[i].str);
206 		int nlen = strlen(nfdicf_test_data[i].ncf);
207 		int j = 0;
208 		unsigned char c;
209 
210 		test((utf8len(um, UTF8_NFDICF, nfdicf_test_data[i].str) ==
211 				nlen));
212 		test((utf8nlen(um, UTF8_NFDICF, nfdicf_test_data[i].str, len) ==
213 				nlen));
214 
215 		if (utf8cursor(&u8c, um, UTF8_NFDICF,
216 				nfdicf_test_data[i].str) < 0)
217 			pr_err("can't create cursor\n");
218 
219 		while ((c = utf8byte(&u8c)) > 0) {
220 			test_f((c == nfdicf_test_data[i].ncf[j]),
221 			       "Unexpected byte 0x%x should be 0x%x\n",
222 			       c, nfdicf_test_data[i].ncf[j]);
223 			j++;
224 		}
225 
226 		test((j == nlen));
227 	}
228 }
229 
230 static void check_utf8_comparisons(struct unicode_map *table)
231 {
232 	int i;
233 
234 	for (i = 0; i < ARRAY_SIZE(nfdi_test_data); i++) {
235 		const struct qstr s1 = {.name = nfdi_test_data[i].str,
236 					.len = sizeof(nfdi_test_data[i].str)};
237 		const struct qstr s2 = {.name = nfdi_test_data[i].dec,
238 					.len = sizeof(nfdi_test_data[i].dec)};
239 
240 		test_f(!utf8_strncmp(table, &s1, &s2),
241 		       "%s %s comparison mismatch\n", s1.name, s2.name);
242 	}
243 
244 	for (i = 0; i < ARRAY_SIZE(nfdicf_test_data); i++) {
245 		const struct qstr s1 = {.name = nfdicf_test_data[i].str,
246 					.len = sizeof(nfdicf_test_data[i].str)};
247 		const struct qstr s2 = {.name = nfdicf_test_data[i].ncf,
248 					.len = sizeof(nfdicf_test_data[i].ncf)};
249 
250 		test_f(!utf8_strncasecmp(table, &s1, &s2),
251 		       "%s %s comparison mismatch\n", s1.name, s2.name);
252 	}
253 }
254 
255 static void check_supported_versions(struct unicode_map *um)
256 {
257 	/* Unicode 7.0.0 should be supported. */
258 	test(utf8version_is_supported(um, UNICODE_AGE(7, 0, 0)));
259 
260 	/* Unicode 9.0.0 should be supported. */
261 	test(utf8version_is_supported(um, UNICODE_AGE(9, 0, 0)));
262 
263 	/* Unicode 1x.0.0 (the latest version) should be supported. */
264 	test(utf8version_is_supported(um, UTF8_LATEST));
265 
266 	/* Next versions don't exist. */
267 	test(!utf8version_is_supported(um, UNICODE_AGE(13, 0, 0)));
268 	test(!utf8version_is_supported(um, UNICODE_AGE(0, 0, 0)));
269 	test(!utf8version_is_supported(um, UNICODE_AGE(-1, -1, -1)));
270 }
271 
272 static int __init init_test_ucd(void)
273 {
274 	struct unicode_map *um;
275 
276 	failed_tests = 0;
277 	total_tests = 0;
278 
279 	um = utf8_load(UTF8_LATEST);
280 	if (IS_ERR(um)) {
281 		pr_err("%s: Unable to load utf8 table.\n", __func__);
282 		return PTR_ERR(um);
283 	}
284 
285 	check_supported_versions(um);
286 	check_utf8_nfdi(um);
287 	check_utf8_nfdicf(um);
288 	check_utf8_comparisons(um);
289 
290 	if (!failed_tests)
291 		pr_info("All %u tests passed\n", total_tests);
292 	else
293 		pr_err("%u out of %u tests failed\n", failed_tests,
294 		       total_tests);
295 	utf8_unload(um);
296 	return 0;
297 }
298 
299 static void __exit exit_test_ucd(void)
300 {
301 }
302 
303 module_init(init_test_ucd);
304 module_exit(exit_test_ucd);
305 
306 MODULE_AUTHOR("Gabriel Krisman Bertazi <krisman@collabora.co.uk>");
307 MODULE_DESCRIPTION("Kernel module for testing utf-8 support");
308 MODULE_LICENSE("GPL");
309