| 1 | #include "catch.hpp" |
| 2 | #include "duckdb/common/types/vector.hpp" |
| 3 | #include "duckdb/main/appender.hpp" |
| 4 | #include "test_helpers.hpp" |
| 5 | |
| 6 | using namespace duckdb; |
| 7 | using namespace std; |
| 8 | |
| 9 | static void test_valid_str(Vector &a, const char *str) { |
| 10 | Value s(str); |
| 11 | REQUIRE_NOTHROW(a.SetValue(0, s)); |
| 12 | REQUIRE(a.GetValue(0) == s); |
| 13 | } |
| 14 | |
| 15 | TEST_CASE("UTF8 error checking" , "[utf8]" ) { |
| 16 | Vector a(TypeId::VARCHAR); |
| 17 | |
| 18 | test_valid_str(a, "a" ); |
| 19 | test_valid_str(a, "\xc3\xb1" ); |
| 20 | test_valid_str(a, "\xE2\x82\xA1" ); |
| 21 | test_valid_str(a, "\xF0\x9F\xA6\x86" ); // a duck! |
| 22 | test_valid_str(a, "\xf0\x90\x8c\xbc" ); |
| 23 | |
| 24 | REQUIRE_THROWS(a.SetValue(0, Value("\xc3\x28" ))); |
| 25 | REQUIRE_THROWS(a.SetValue(0, Value("\xa0\xa1" ))); |
| 26 | REQUIRE_THROWS(a.SetValue(0, Value("\xe2\x28\xa1" ))); |
| 27 | REQUIRE_THROWS(a.SetValue(0, Value("\xe2\x82\x28" ))); |
| 28 | REQUIRE_THROWS(a.SetValue(0, Value("\xf0\x28\x8c\xbc" ))); |
| 29 | REQUIRE_THROWS(a.SetValue(0, Value("\xf0\x90\x28\xbc" ))); |
| 30 | REQUIRE_THROWS(a.SetValue(0, Value("\xf0\x28\x8c\x28" ))); |
| 31 | REQUIRE_THROWS(a.SetValue(0, Value("\xf8\xa1\xa1\xa1\xa1" ))); |
| 32 | REQUIRE_THROWS(a.SetValue(0, Value("\xfc\xa1\xa1\xa1\xa1\xa1" ))); |
| 33 | } |
| 34 | |
| 35 | TEST_CASE("UTF8 NFC tests" , "[utf8]" ) { |
| 36 | // check NFC equivalence in Value API |
| 37 | REQUIRE(Value("a" ) == Value("a" )); |
| 38 | REQUIRE(Value("a" ) != Value("b" )); |
| 39 | REQUIRE(Value("\xc3\xbc" ) == Value("\xc3\xbc" )); |
| 40 | REQUIRE(Value("\xc3\xbc" ) == Value("\x75\xcc\x88" )); |
| 41 | |
| 42 | // also in SQL |
| 43 | DuckDB db(nullptr); |
| 44 | Connection con(db); |
| 45 | con.EnableQueryVerification(); |
| 46 | unique_ptr<QueryResult> result; |
| 47 | |
| 48 | result = con.Query("SELECT 'a'='a'" ); |
| 49 | REQUIRE(CHECK_COLUMN(result, 0, {Value::BOOLEAN(true)})); |
| 50 | |
| 51 | result = con.Query("SELECT '\xc3\xbc'='\xc3\xbc'" ); |
| 52 | REQUIRE(CHECK_COLUMN(result, 0, {Value::BOOLEAN(true)})); |
| 53 | |
| 54 | result = con.Query("SELECT '\xc3\xbc'='\x75\xcc\x88'" ); |
| 55 | REQUIRE(CHECK_COLUMN(result, 0, {Value::BOOLEAN(true)})); |
| 56 | |
| 57 | // also through appenders |
| 58 | REQUIRE_NO_FAIL(con.Query("CREATE TABLE strings (s STRING)" )); |
| 59 | Appender appender(con, DEFAULT_SCHEMA, "strings" ); |
| 60 | appender.BeginRow(); |
| 61 | appender.Append("\x75\xcc\x88" ); |
| 62 | appender.EndRow(); |
| 63 | appender.Close(); |
| 64 | |
| 65 | result = con.Query("SELECT s = '\xc3\xbc' FROM strings" ); |
| 66 | REQUIRE(CHECK_COLUMN(result, 0, {Value::BOOLEAN(true)})); |
| 67 | |
| 68 | result = con.Query("SELECT s = '\x75\xcc\x88' FROM strings" ); |
| 69 | REQUIRE(CHECK_COLUMN(result, 0, {Value::BOOLEAN(true)})); |
| 70 | |
| 71 | // related to bug 539 |
| 72 | REQUIRE_NO_FAIL(con.Query("\x49\x4E\x53\x45\x52\x54\x20\x49\x4E\x54\x4F\x20\x73\x74\x72\x69\x6E\x67\x73\x20\x56\x41" |
| 73 | "\x4C\x55\x45\x53\x28\x27\x61\x27\x29" )); |
| 74 | REQUIRE_NO_FAIL(con.Query("\x49\x4E\x53\x45\x52\x54\x20\x49\x4E\x54\x4F\x20\x73\x74\x72\x69\x6E\x67\x73\x20\x56\x41" |
| 75 | "\x4C\x55\x45\x53\x28\x27\x3F\x27\x29" )); |
| 76 | |
| 77 | // also through CSV reader |
| 78 | } |
| 79 | |