xref: /openbmc/linux/fs/unicode/utf8-selftest.c (revision 2cf1c348)
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 unsigned int failed_tests;
18 unsigned int total_tests;
19 
20 /* Tests will be based on this version. */
21 #define UTF8_LATEST	UNICODE_AGE(12, 1, 0)
22 
23 #define _test(cond, func, line, fmt, ...) do {				\
24 		total_tests++;						\
25 		if (!cond) {						\
26 			failed_tests++;					\
27 			pr_err("test %s:%d Failed: %s%s",		\
28 			       func, line, #cond, (fmt?":":"."));	\
29 			if (fmt)					\
30 				pr_err(fmt, ##__VA_ARGS__);		\
31 		}							\
32 	} while (0)
33 #define test_f(cond, fmt, ...) _test(cond, __func__, __LINE__, fmt, ##__VA_ARGS__)
34 #define test(cond) _test(cond, __func__, __LINE__, "")
35 
36 static const struct {
37 	/* UTF-8 strings in this vector _must_ be NULL-terminated. */
38 	unsigned char str[10];
39 	unsigned char dec[10];
40 } nfdi_test_data[] = {
41 	/* Trivial sequence */
42 	{
43 		/* "ABba" decomposes to itself */
44 		.str = "aBba",
45 		.dec = "aBba",
46 	},
47 	/* Simple equivalent sequences */
48 	{
49                /* 'VULGAR FRACTION ONE QUARTER' cannot decompose to
50                   'NUMBER 1' + 'FRACTION SLASH' + 'NUMBER 4' on
51                   canonical decomposition */
52                .str = {0xc2, 0xbc, 0x00},
53 	       .dec = {0xc2, 0xbc, 0x00},
54 	},
55 	{
56 		/* 'LATIN SMALL LETTER A WITH DIAERESIS' decomposes to
57 		   'LETTER A' + 'COMBINING DIAERESIS' */
58 		.str = {0xc3, 0xa4, 0x00},
59 		.dec = {0x61, 0xcc, 0x88, 0x00},
60 	},
61 	{
62 		/* 'LATIN SMALL LETTER LJ' can't decompose to
63 		   'LETTER L' + 'LETTER J' on canonical decomposition */
64 		.str = {0xC7, 0x89, 0x00},
65 		.dec = {0xC7, 0x89, 0x00},
66 	},
67 	{
68 		/* GREEK ANO TELEIA decomposes to MIDDLE DOT */
69 		.str = {0xCE, 0x87, 0x00},
70 		.dec = {0xC2, 0xB7, 0x00}
71 	},
72 	/* Canonical ordering */
73 	{
74 		/* A + 'COMBINING ACUTE ACCENT' + 'COMBINING OGONEK' decomposes
75 		   to A + 'COMBINING OGONEK' + 'COMBINING ACUTE ACCENT' */
76 		.str = {0x41, 0xcc, 0x81, 0xcc, 0xa8, 0x0},
77 		.dec = {0x41, 0xcc, 0xa8, 0xcc, 0x81, 0x0},
78 	},
79 	{
80 		/* 'LATIN SMALL LETTER A WITH DIAERESIS' + 'COMBINING OGONEK'
81 		   decomposes to
82 		   'LETTER A' + 'COMBINING OGONEK' + 'COMBINING DIAERESIS' */
83 		.str = {0xc3, 0xa4, 0xCC, 0xA8, 0x00},
84 
85 		.dec = {0x61, 0xCC, 0xA8, 0xcc, 0x88, 0x00},
86 	},
87 
88 };
89 
90 static const struct {
91 	/* UTF-8 strings in this vector _must_ be NULL-terminated. */
92 	unsigned char str[30];
93 	unsigned char ncf[30];
94 } nfdicf_test_data[] = {
95 	/* Trivial sequences */
96 	{
97 		/* "ABba" folds to lowercase */
98 		.str = {0x41, 0x42, 0x62, 0x61, 0x00},
99 		.ncf = {0x61, 0x62, 0x62, 0x61, 0x00},
100 	},
101 	{
102 		/* All ASCII folds to lower-case */
103 		.str = "ABCDEFGHIJKLMNOPQRSTUVWXYZ0.1",
104 		.ncf = "abcdefghijklmnopqrstuvwxyz0.1",
105 	},
106 	{
107 		/* LATIN SMALL LETTER SHARP S folds to
108 		   LATIN SMALL LETTER S + LATIN SMALL LETTER S */
109 		.str = {0xc3, 0x9f, 0x00},
110 		.ncf = {0x73, 0x73, 0x00},
111 	},
112 	{
113 		/* LATIN CAPITAL LETTER A WITH RING ABOVE folds to
114 		   LATIN SMALL LETTER A + COMBINING RING ABOVE */
115 		.str = {0xC3, 0x85, 0x00},
116 		.ncf = {0x61, 0xcc, 0x8a, 0x00},
117 	},
118 	/* Introduced by UTF-8.0.0. */
119 	/* Cherokee letters are interesting test-cases because they fold
120 	   to upper-case.  Before 8.0.0, Cherokee lowercase were
121 	   undefined, thus, the folding from LC is not stable between
122 	   7.0.0 -> 8.0.0, but it is from UC. */
123 	{
124 		/* CHEROKEE SMALL LETTER A folds to CHEROKEE LETTER A */
125 		.str = {0xea, 0xad, 0xb0, 0x00},
126 		.ncf = {0xe1, 0x8e, 0xa0, 0x00},
127 	},
128 	{
129 		/* CHEROKEE SMALL LETTER YE folds to CHEROKEE LETTER YE */
130 		.str = {0xe1, 0x8f, 0xb8, 0x00},
131 		.ncf = {0xe1, 0x8f, 0xb0, 0x00},
132 	},
133 	{
134 		/* OLD HUNGARIAN CAPITAL LETTER AMB folds to
135 		   OLD HUNGARIAN SMALL LETTER AMB */
136 		.str = {0xf0, 0x90, 0xb2, 0x83, 0x00},
137 		.ncf = {0xf0, 0x90, 0xb3, 0x83, 0x00},
138 	},
139 	/* Introduced by UTF-9.0.0. */
140 	{
141 		/* OSAGE CAPITAL LETTER CHA folds to
142 		   OSAGE SMALL LETTER CHA */
143 		.str = {0xf0, 0x90, 0x92, 0xb5, 0x00},
144 		.ncf = {0xf0, 0x90, 0x93, 0x9d, 0x00},
145 	},
146 	{
147 		/* LATIN CAPITAL LETTER SMALL CAPITAL I folds to
148 		   LATIN LETTER SMALL CAPITAL I */
149 		.str = {0xea, 0x9e, 0xae, 0x00},
150 		.ncf = {0xc9, 0xaa, 0x00},
151 	},
152 	/* Introduced by UTF-11.0.0. */
153 	{
154 		/* GEORGIAN SMALL LETTER AN folds to GEORGIAN MTAVRULI
155 		   CAPITAL LETTER AN */
156 		.str = {0xe1, 0xb2, 0x90, 0x00},
157 		.ncf = {0xe1, 0x83, 0x90, 0x00},
158 	}
159 };
160 
161 static ssize_t utf8len(const struct unicode_map *um, enum utf8_normalization n,
162 		const char *s)
163 {
164 	return utf8nlen(um, n, s, (size_t)-1);
165 }
166 
167 static int utf8cursor(struct utf8cursor *u8c, const struct unicode_map *um,
168 		enum utf8_normalization n, const char *s)
169 {
170 	return utf8ncursor(u8c, um, n, s, (unsigned int)-1);
171 }
172 
173 static void check_utf8_nfdi(struct unicode_map *um)
174 {
175 	int i;
176 	struct utf8cursor u8c;
177 
178 	for (i = 0; i < ARRAY_SIZE(nfdi_test_data); i++) {
179 		int len = strlen(nfdi_test_data[i].str);
180 		int nlen = strlen(nfdi_test_data[i].dec);
181 		int j = 0;
182 		unsigned char c;
183 
184 		test((utf8len(um, UTF8_NFDI, nfdi_test_data[i].str) == nlen));
185 		test((utf8nlen(um, UTF8_NFDI, nfdi_test_data[i].str, len) ==
186 			nlen));
187 
188 		if (utf8cursor(&u8c, um, UTF8_NFDI, nfdi_test_data[i].str) < 0)
189 			pr_err("can't create cursor\n");
190 
191 		while ((c = utf8byte(&u8c)) > 0) {
192 			test_f((c == nfdi_test_data[i].dec[j]),
193 			       "Unexpected byte 0x%x should be 0x%x\n",
194 			       c, nfdi_test_data[i].dec[j]);
195 			j++;
196 		}
197 
198 		test((j == nlen));
199 	}
200 }
201 
202 static void check_utf8_nfdicf(struct unicode_map *um)
203 {
204 	int i;
205 	struct utf8cursor u8c;
206 
207 	for (i = 0; i < ARRAY_SIZE(nfdicf_test_data); i++) {
208 		int len = strlen(nfdicf_test_data[i].str);
209 		int nlen = strlen(nfdicf_test_data[i].ncf);
210 		int j = 0;
211 		unsigned char c;
212 
213 		test((utf8len(um, UTF8_NFDICF, nfdicf_test_data[i].str) ==
214 				nlen));
215 		test((utf8nlen(um, UTF8_NFDICF, nfdicf_test_data[i].str, len) ==
216 				nlen));
217 
218 		if (utf8cursor(&u8c, um, UTF8_NFDICF,
219 				nfdicf_test_data[i].str) < 0)
220 			pr_err("can't create cursor\n");
221 
222 		while ((c = utf8byte(&u8c)) > 0) {
223 			test_f((c == nfdicf_test_data[i].ncf[j]),
224 			       "Unexpected byte 0x%x should be 0x%x\n",
225 			       c, nfdicf_test_data[i].ncf[j]);
226 			j++;
227 		}
228 
229 		test((j == nlen));
230 	}
231 }
232 
233 static void check_utf8_comparisons(struct unicode_map *table)
234 {
235 	int i;
236 
237 	for (i = 0; i < ARRAY_SIZE(nfdi_test_data); i++) {
238 		const struct qstr s1 = {.name = nfdi_test_data[i].str,
239 					.len = sizeof(nfdi_test_data[i].str)};
240 		const struct qstr s2 = {.name = nfdi_test_data[i].dec,
241 					.len = sizeof(nfdi_test_data[i].dec)};
242 
243 		test_f(!utf8_strncmp(table, &s1, &s2),
244 		       "%s %s comparison mismatch\n", s1.name, s2.name);
245 	}
246 
247 	for (i = 0; i < ARRAY_SIZE(nfdicf_test_data); i++) {
248 		const struct qstr s1 = {.name = nfdicf_test_data[i].str,
249 					.len = sizeof(nfdicf_test_data[i].str)};
250 		const struct qstr s2 = {.name = nfdicf_test_data[i].ncf,
251 					.len = sizeof(nfdicf_test_data[i].ncf)};
252 
253 		test_f(!utf8_strncasecmp(table, &s1, &s2),
254 		       "%s %s comparison mismatch\n", s1.name, s2.name);
255 	}
256 }
257 
258 static void check_supported_versions(struct unicode_map *um)
259 {
260 	/* Unicode 7.0.0 should be supported. */
261 	test(utf8version_is_supported(um, UNICODE_AGE(7, 0, 0)));
262 
263 	/* Unicode 9.0.0 should be supported. */
264 	test(utf8version_is_supported(um, UNICODE_AGE(9, 0, 0)));
265 
266 	/* Unicode 1x.0.0 (the latest version) should be supported. */
267 	test(utf8version_is_supported(um, UTF8_LATEST));
268 
269 	/* Next versions don't exist. */
270 	test(!utf8version_is_supported(um, UNICODE_AGE(13, 0, 0)));
271 	test(!utf8version_is_supported(um, UNICODE_AGE(0, 0, 0)));
272 	test(!utf8version_is_supported(um, UNICODE_AGE(-1, -1, -1)));
273 }
274 
275 static int __init init_test_ucd(void)
276 {
277 	struct unicode_map *um;
278 
279 	failed_tests = 0;
280 	total_tests = 0;
281 
282 	um = utf8_load(UTF8_LATEST);
283 	if (IS_ERR(um)) {
284 		pr_err("%s: Unable to load utf8 table.\n", __func__);
285 		return PTR_ERR(um);
286 	}
287 
288 	check_supported_versions(um);
289 	check_utf8_nfdi(um);
290 	check_utf8_nfdicf(um);
291 	check_utf8_comparisons(um);
292 
293 	if (!failed_tests)
294 		pr_info("All %u tests passed\n", total_tests);
295 	else
296 		pr_err("%u out of %u tests failed\n", failed_tests,
297 		       total_tests);
298 	utf8_unload(um);
299 	return 0;
300 }
301 
302 static void __exit exit_test_ucd(void)
303 {
304 }
305 
306 module_init(init_test_ucd);
307 module_exit(exit_test_ucd);
308 
309 MODULE_AUTHOR("Gabriel Krisman Bertazi <krisman@collabora.co.uk>");
310 MODULE_LICENSE("GPL");
311