Line | Branch | Exec | Source |
---|---|---|---|
1 | #include "test.h" | ||
2 | #include "encoding.h" | ||
3 | |||
4 | 1 | static void test_detect_encoding_from_bom(TestContext *ctx) | |
5 | { | ||
6 | 1 | static const struct bom_test { | |
7 | EncodingType encoding; | ||
8 | const unsigned char *text; | ||
9 | size_t size; | ||
10 | } tests[] = { | ||
11 | {UTF8, STRN("\xef\xbb\xbfHello")}, | ||
12 | {UTF32BE, STRN("\x00\x00\xfe\xffHello")}, | ||
13 | {UTF32LE, STRN("\xff\xfe\x00\x00Hello")}, | ||
14 | {UTF16BE, STRN("\xfe\xffHello")}, | ||
15 | {UTF16LE, STRN("\xff\xfeHello")}, | ||
16 | {UTF16LE, STRN("\xff\xfe")}, | ||
17 | {UNKNOWN_ENCODING, STRN("\xff\xff\x00\x00Hello")}, | ||
18 | {UNKNOWN_ENCODING, STRN("\x00\xef\xbb\xbfHello")}, | ||
19 | {UNKNOWN_ENCODING, STRN("\xef\xbb")}, | ||
20 | {UNKNOWN_ENCODING, STRN("\xef\xbb#")}, | ||
21 | {UNKNOWN_ENCODING, STRN("\xee\xbb\xbf")}, | ||
22 | {UNKNOWN_ENCODING, STRN("\xff\xbb\xbf")}, | ||
23 | {UNKNOWN_ENCODING, STRN("\xbf\xbb\xef")}, | ||
24 | {UNKNOWN_ENCODING, STRN("\x00\x00\xfe")}, | ||
25 | {UNKNOWN_ENCODING, STRN("\x00")}, | ||
26 | {UNKNOWN_ENCODING, "", 0}, | ||
27 | {UNKNOWN_ENCODING, NULL, 0}, | ||
28 | }; | ||
29 |
2/2✓ Branch 0 taken 17 times.
✓ Branch 1 taken 1 times.
|
18 | FOR_EACH_I(i, tests) { |
30 | 17 | const struct bom_test *t = &tests[i]; | |
31 | 17 | EncodingType result = detect_encoding_from_bom(t->text, t->size); | |
32 | 17 | IEXPECT_EQ(result, t->encoding); | |
33 | } | ||
34 | 1 | } | |
35 | |||
36 | 1 | static void test_lookup_encoding(TestContext *ctx) | |
37 | { | ||
38 | 1 | static const struct { | |
39 | EncodingType encoding; | ||
40 | const char *name; | ||
41 | } tests[] = { | ||
42 | {UTF8, "UTF-8"}, | ||
43 | {UTF8, "UTF8"}, | ||
44 | {UTF8, "utf-8"}, | ||
45 | {UTF8, "utf8"}, | ||
46 | {UTF8, "Utf8"}, | ||
47 | {UTF16BE, "UTF16"}, | ||
48 | {UTF16BE, "UTF-16"}, | ||
49 | {UTF32BE, "utf32"}, | ||
50 | {UTF32BE, "utf-32"}, | ||
51 | {UTF32LE, "utf-32le"}, | ||
52 | {UTF32LE, "ucs-4le"}, | ||
53 | {UTF32BE, "ucs-4BE"}, | ||
54 | {UTF32BE, "ucs-4"}, | ||
55 | {UTF32BE, "ucs4"}, | ||
56 | {UNKNOWN_ENCODING, "UTF8_"}, | ||
57 | {UNKNOWN_ENCODING, "UTF"}, | ||
58 | }; | ||
59 |
2/2✓ Branch 0 taken 16 times.
✓ Branch 1 taken 1 times.
|
17 | FOR_EACH_I(i, tests) { |
60 | 16 | EncodingType result = lookup_encoding(tests[i].name); | |
61 | 16 | IEXPECT_EQ(result, tests[i].encoding); | |
62 | } | ||
63 | 1 | } | |
64 | |||
65 | 1 | static void test_encoding_from_type(TestContext *ctx) | |
66 | { | ||
67 | 1 | const char *a = encoding_from_type(UTF8); | |
68 | 1 | EXPECT_STREQ(a, "UTF-8"); | |
69 | 1 | EXPECT_TRUE(encoding_is_utf8(a)); | |
70 | // Ensure returned value is an "interned" string | ||
71 | 1 | EXPECT_PTREQ(a, encoding_normalize("utf8")); | |
72 | 1 | EXPECT_PTREQ(a, encoding_from_type(UTF8)); | |
73 | 1 | } | |
74 | |||
75 | 1 | static void test_get_bom_for_encoding(TestContext *ctx) | |
76 | { | ||
77 | 1 | const ByteOrderMark *bom = get_bom_for_encoding(UTF8); | |
78 | 1 | EXPECT_MEMEQ(bom->bytes, bom->len, "\xef\xbb\xbf", 3); | |
79 | |||
80 | 1 | bom = get_bom_for_encoding(UTF32LE); | |
81 | 1 | EXPECT_MEMEQ(bom->bytes, bom->len, "\xff\xfe\0\0", 4); | |
82 | |||
83 | 1 | bom = get_bom_for_encoding(UTF16BE); | |
84 | 1 | EXPECT_MEMEQ(bom->bytes, bom->len, "\xfe\xff", 2); | |
85 | |||
86 | 1 | EXPECT_NULL(get_bom_for_encoding(UNKNOWN_ENCODING)); | |
87 | 1 | EXPECT_FALSE(encoding_type_has_bom(UNKNOWN_ENCODING)); | |
88 | 1 | EXPECT_TRUE(encoding_type_has_bom(UTF8)); | |
89 | 1 | EXPECT_TRUE(encoding_type_has_bom(UTF32LE)); | |
90 | 1 | } | |
91 | |||
92 | static const TestEntry tests[] = { | ||
93 | TEST(test_detect_encoding_from_bom), | ||
94 | TEST(test_lookup_encoding), | ||
95 | TEST(test_encoding_from_type), | ||
96 | TEST(test_get_bom_for_encoding), | ||
97 | }; | ||
98 | |||
99 | const TestGroup encoding_tests = TEST_GROUP(tests); | ||
100 |