xref: /linux/fs/unicode/utf8-selftest.c (revision b77e0ce62d63a761ffb7f7245a215a49f5921c2f)
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 latest_maj 12
22 #define latest_min 1
23 #define latest_rev 0
24 
25 #define _test(cond, func, line, fmt, ...) do {				\
26 		total_tests++;						\
27 		if (!cond) {						\
28 			failed_tests++;					\
29 			pr_err("test %s:%d Failed: %s%s",		\
30 			       func, line, #cond, (fmt?":":"."));	\
31 			if (fmt)					\
32 				pr_err(fmt, ##__VA_ARGS__);		\
33 		}							\
34 	} while (0)
35 #define test_f(cond, fmt, ...) _test(cond, __func__, __LINE__, fmt, ##__VA_ARGS__)
36 #define test(cond) _test(cond, __func__, __LINE__, "")
37 
38 static const struct {
39 	/* UTF-8 strings in this vector _must_ be NULL-terminated. */
40 	unsigned char str[10];
41 	unsigned char dec[10];
42 } nfdi_test_data[] = {
43 	/* Trivial sequence */
44 	{
45 		/* "ABba" decomposes to itself */
46 		.str = "aBba",
47 		.dec = "aBba",
48 	},
49 	/* Simple equivalent sequences */
50 	{
51                /* 'VULGAR FRACTION ONE QUARTER' cannot decompose to
52                   'NUMBER 1' + 'FRACTION SLASH' + 'NUMBER 4' on
53                   canonical decomposition */
54                .str = {0xc2, 0xbc, 0x00},
55 	       .dec = {0xc2, 0xbc, 0x00},
56 	},
57 	{
58 		/* 'LATIN SMALL LETTER A WITH DIAERESIS' decomposes to
59 		   'LETTER A' + 'COMBINING DIAERESIS' */
60 		.str = {0xc3, 0xa4, 0x00},
61 		.dec = {0x61, 0xcc, 0x88, 0x00},
62 	},
63 	{
64 		/* 'LATIN SMALL LETTER LJ' can't decompose to
65 		   'LETTER L' + 'LETTER J' on canonical decomposition */
66 		.str = {0xC7, 0x89, 0x00},
67 		.dec = {0xC7, 0x89, 0x00},
68 	},
69 	{
70 		/* GREEK ANO TELEIA decomposes to MIDDLE DOT */
71 		.str = {0xCE, 0x87, 0x00},
72 		.dec = {0xC2, 0xB7, 0x00}
73 	},
74 	/* Canonical ordering */
75 	{
76 		/* A + 'COMBINING ACUTE ACCENT' + 'COMBINING OGONEK' decomposes
77 		   to A + 'COMBINING OGONEK' + 'COMBINING ACUTE ACCENT' */
78 		.str = {0x41, 0xcc, 0x81, 0xcc, 0xa8, 0x0},
79 		.dec = {0x41, 0xcc, 0xa8, 0xcc, 0x81, 0x0},
80 	},
81 	{
82 		/* 'LATIN SMALL LETTER A WITH DIAERESIS' + 'COMBINING OGONEK'
83 		   decomposes to
84 		   'LETTER A' + 'COMBINING OGONEK' + 'COMBINING DIAERESIS' */
85 		.str = {0xc3, 0xa4, 0xCC, 0xA8, 0x00},
86 
87 		.dec = {0x61, 0xCC, 0xA8, 0xcc, 0x88, 0x00},
88 	},
89 
90 };
91 
92 static const struct {
93 	/* UTF-8 strings in this vector _must_ be NULL-terminated. */
94 	unsigned char str[30];
95 	unsigned char ncf[30];
96 } nfdicf_test_data[] = {
97 	/* Trivial sequences */
98 	{
99 		/* "ABba" folds to lowercase */
100 		.str = {0x41, 0x42, 0x62, 0x61, 0x00},
101 		.ncf = {0x61, 0x62, 0x62, 0x61, 0x00},
102 	},
103 	{
104 		/* All ASCII folds to lower-case */
105 		.str = "ABCDEFGHIJKLMNOPQRSTUVWXYZ0.1",
106 		.ncf = "abcdefghijklmnopqrstuvwxyz0.1",
107 	},
108 	{
109 		/* LATIN SMALL LETTER SHARP S folds to
110 		   LATIN SMALL LETTER S + LATIN SMALL LETTER S */
111 		.str = {0xc3, 0x9f, 0x00},
112 		.ncf = {0x73, 0x73, 0x00},
113 	},
114 	{
115 		/* LATIN CAPITAL LETTER A WITH RING ABOVE folds to
116 		   LATIN SMALL LETTER A + COMBINING RING ABOVE */
117 		.str = {0xC3, 0x85, 0x00},
118 		.ncf = {0x61, 0xcc, 0x8a, 0x00},
119 	},
120 	/* Introduced by UTF-8.0.0. */
121 	/* Cherokee letters are interesting test-cases because they fold
122 	   to upper-case.  Before 8.0.0, Cherokee lowercase were
123 	   undefined, thus, the folding from LC is not stable between
124 	   7.0.0 -> 8.0.0, but it is from UC. */
125 	{
126 		/* CHEROKEE SMALL LETTER A folds to CHEROKEE LETTER A */
127 		.str = {0xea, 0xad, 0xb0, 0x00},
128 		.ncf = {0xe1, 0x8e, 0xa0, 0x00},
129 	},
130 	{
131 		/* CHEROKEE SMALL LETTER YE folds to CHEROKEE LETTER YE */
132 		.str = {0xe1, 0x8f, 0xb8, 0x00},
133 		.ncf = {0xe1, 0x8f, 0xb0, 0x00},
134 	},
135 	{
136 		/* OLD HUNGARIAN CAPITAL LETTER AMB folds to
137 		   OLD HUNGARIAN SMALL LETTER AMB */
138 		.str = {0xf0, 0x90, 0xb2, 0x83, 0x00},
139 		.ncf = {0xf0, 0x90, 0xb3, 0x83, 0x00},
140 	},
141 	/* Introduced by UTF-9.0.0. */
142 	{
143 		/* OSAGE CAPITAL LETTER CHA folds to
144 		   OSAGE SMALL LETTER CHA */
145 		.str = {0xf0, 0x90, 0x92, 0xb5, 0x00},
146 		.ncf = {0xf0, 0x90, 0x93, 0x9d, 0x00},
147 	},
148 	{
149 		/* LATIN CAPITAL LETTER SMALL CAPITAL I folds to
150 		   LATIN LETTER SMALL CAPITAL I */
151 		.str = {0xea, 0x9e, 0xae, 0x00},
152 		.ncf = {0xc9, 0xaa, 0x00},
153 	},
154 	/* Introduced by UTF-11.0.0. */
155 	{
156 		/* GEORGIAN SMALL LETTER AN folds to GEORGIAN MTAVRULI
157 		   CAPITAL LETTER AN */
158 		.str = {0xe1, 0xb2, 0x90, 0x00},
159 		.ncf = {0xe1, 0x83, 0x90, 0x00},
160 	}
161 };
162 
163 static void check_utf8_nfdi(void)
164 {
165 	int i;
166 	struct utf8cursor u8c;
167 	const struct utf8data *data;
168 
169 	data = utf8nfdi(UNICODE_AGE(latest_maj, latest_min, latest_rev));
170 	if (!data) {
171 		pr_err("%s: Unable to load utf8-%d.%d.%d. Skipping.\n",
172 		       __func__, latest_maj, latest_min, latest_rev);
173 		return;
174 	}
175 
176 	for (i = 0; i < ARRAY_SIZE(nfdi_test_data); i++) {
177 		int len = strlen(nfdi_test_data[i].str);
178 		int nlen = strlen(nfdi_test_data[i].dec);
179 		int j = 0;
180 		unsigned char c;
181 
182 		test((utf8len(data, nfdi_test_data[i].str) == nlen));
183 		test((utf8nlen(data, nfdi_test_data[i].str, len) == nlen));
184 
185 		if (utf8cursor(&u8c, data, 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(void)
200 {
201 	int i;
202 	struct utf8cursor u8c;
203 	const struct utf8data *data;
204 
205 	data = utf8nfdicf(UNICODE_AGE(latest_maj, latest_min, latest_rev));
206 	if (!data) {
207 		pr_err("%s: Unable to load utf8-%d.%d.%d. Skipping.\n",
208 		       __func__, latest_maj, latest_min, latest_rev);
209 		return;
210 	}
211 
212 	for (i = 0; i < ARRAY_SIZE(nfdicf_test_data); i++) {
213 		int len = strlen(nfdicf_test_data[i].str);
214 		int nlen = strlen(nfdicf_test_data[i].ncf);
215 		int j = 0;
216 		unsigned char c;
217 
218 		test((utf8len(data, nfdicf_test_data[i].str) == nlen));
219 		test((utf8nlen(data, nfdicf_test_data[i].str, len) == nlen));
220 
221 		if (utf8cursor(&u8c, data, nfdicf_test_data[i].str) < 0)
222 			pr_err("can't create cursor\n");
223 
224 		while ((c = utf8byte(&u8c)) > 0) {
225 			test_f((c == nfdicf_test_data[i].ncf[j]),
226 			       "Unexpected byte 0x%x should be 0x%x\n",
227 			       c, nfdicf_test_data[i].ncf[j]);
228 			j++;
229 		}
230 
231 		test((j == nlen));
232 	}
233 }
234 
235 static void check_utf8_comparisons(void)
236 {
237 	int i;
238 	struct unicode_map *table = utf8_load("12.1.0");
239 
240 	if (IS_ERR(table)) {
241 		pr_err("%s: Unable to load utf8 %d.%d.%d. Skipping.\n",
242 		       __func__, latest_maj, latest_min, latest_rev);
243 		return;
244 	}
245 
246 	for (i = 0; i < ARRAY_SIZE(nfdi_test_data); i++) {
247 		const struct qstr s1 = {.name = nfdi_test_data[i].str,
248 					.len = sizeof(nfdi_test_data[i].str)};
249 		const struct qstr s2 = {.name = nfdi_test_data[i].dec,
250 					.len = sizeof(nfdi_test_data[i].dec)};
251 
252 		test_f(!utf8_strncmp(table, &s1, &s2),
253 		       "%s %s comparison mismatch\n", s1.name, s2.name);
254 	}
255 
256 	for (i = 0; i < ARRAY_SIZE(nfdicf_test_data); i++) {
257 		const struct qstr s1 = {.name = nfdicf_test_data[i].str,
258 					.len = sizeof(nfdicf_test_data[i].str)};
259 		const struct qstr s2 = {.name = nfdicf_test_data[i].ncf,
260 					.len = sizeof(nfdicf_test_data[i].ncf)};
261 
262 		test_f(!utf8_strncasecmp(table, &s1, &s2),
263 		       "%s %s comparison mismatch\n", s1.name, s2.name);
264 	}
265 
266 	utf8_unload(table);
267 }
268 
269 static void check_supported_versions(void)
270 {
271 	/* Unicode 7.0.0 should be supported. */
272 	test(utf8version_is_supported(7, 0, 0));
273 
274 	/* Unicode 9.0.0 should be supported. */
275 	test(utf8version_is_supported(9, 0, 0));
276 
277 	/* Unicode 1x.0.0 (the latest version) should be supported. */
278 	test(utf8version_is_supported(latest_maj, latest_min, latest_rev));
279 
280 	/* Next versions don't exist. */
281 	test(!utf8version_is_supported(13, 0, 0));
282 	test(!utf8version_is_supported(0, 0, 0));
283 	test(!utf8version_is_supported(-1, -1, -1));
284 }
285 
286 static int __init init_test_ucd(void)
287 {
288 	failed_tests = 0;
289 	total_tests = 0;
290 
291 	check_supported_versions();
292 	check_utf8_nfdi();
293 	check_utf8_nfdicf();
294 	check_utf8_comparisons();
295 
296 	if (!failed_tests)
297 		pr_info("All %u tests passed\n", total_tests);
298 	else
299 		pr_err("%u out of %u tests failed\n", failed_tests,
300 		       total_tests);
301 	return 0;
302 }
303 
304 static void __exit exit_test_ucd(void)
305 {
306 }
307 
308 module_init(init_test_ucd);
309 module_exit(exit_test_ucd);
310 
311 MODULE_AUTHOR("Gabriel Krisman Bertazi <krisman@collabora.co.uk>");
312 MODULE_LICENSE("GPL");
313