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