add unit tests for Encoding module

This commit is contained in:
Josh Holtrop 2016-10-06 22:45:38 -04:00
parent ac439d30bc
commit 24e6eb7f04

69
test/src/test_Encoding.cc Normal file
View File

@ -0,0 +1,69 @@
#include "gtest/gtest.h"
#include "Encoding.h"
#include <string>
TEST(Encoding_detect_encoding, returns_UTF_8_for_all_ascii)
{
uint8_t buffer[] = "ABCDEFG 12345";
EXPECT_EQ(Encoding::UTF_8, Encoding::detect_encoding(buffer, sizeof(buffer) - 1));
}
TEST(Encoding_detect_encoding, returns_UTF_8_for_valid_encodings)
{
uint8_t buffer[] = "%\xC1\x82\xE1\x82\x83\xF1\x82\x83\x84\xF9\x82\x83\x84\x85\xFD\x82\x83\x84\x85\x86yay";
EXPECT_EQ(Encoding::UTF_8, Encoding::detect_encoding(buffer, sizeof(buffer) - 1));
}
TEST(Encoding_detect_encoding, returns_CP_1252_if_not_valid_utf_8)
{
uint8_t buffer[] = "ABC\xEE\xCC";
EXPECT_EQ(Encoding::CP_1252, Encoding::detect_encoding(buffer, sizeof(buffer) - 1));
}
TEST(Encoding_num_bytes_in_code_point, returns_1_for_UTF_8_ascii_character)
{
uint8_t buffer[] = "Z";
EXPECT_EQ(1, Encoding::num_bytes_in_code_point(Encoding::UTF_8, buffer));
}
TEST(Encoding_num_bytes_in_code_point, returns_1_for_CP_1252_character)
{
uint8_t buffer[] = "\x96";
EXPECT_EQ(1, Encoding::num_bytes_in_code_point(Encoding::CP_1252, buffer));
}
TEST(Encoding_num_bytes_in_code_point, returns_3_for_UTF_8_multi_byte)
{
uint8_t buffer[] = "\xE1\x82\x83";
EXPECT_EQ(3, Encoding::num_bytes_in_code_point(Encoding::UTF_8, buffer));
}
TEST(Encoding_beginning_of_code_point, returns_pointer_to_beginning_of_code_point_for_UTF_8)
{
uint8_t buffer[] = "12\xE1\x82\x83";
EXPECT_EQ(&buffer[2], Encoding::beginning_of_code_point(Encoding::UTF_8, &buffer[4]));
}
TEST(Encoding_beginning_of_code_point, returns_pointer_to_beginning_of_code_point_for_CP_1252)
{
uint8_t buffer[] = "12\xE1\x82\x83";
EXPECT_EQ(&buffer[4], Encoding::beginning_of_code_point(Encoding::CP_1252, &buffer[4]));
}
TEST(Encoding_decode, decodes_UTF_8_correctly)
{
EXPECT_EQ((uint32_t)'%', Encoding::decode(Encoding::UTF_8, (const uint8_t *)"%"));
EXPECT_EQ(0x42u, Encoding::decode(Encoding::UTF_8, (const uint8_t *)"\xC1\x82"));
EXPECT_EQ(0x1083u, Encoding::decode(Encoding::UTF_8, (const uint8_t *)"\xE1\x82\x83"));
EXPECT_EQ(0x420C4u, Encoding::decode(Encoding::UTF_8, (const uint8_t *)"\xF1\x82\x83\x84"));
EXPECT_EQ(0x1083105u, Encoding::decode(Encoding::UTF_8, (const uint8_t *)"\xF9\x82\x83\x84\x85"));
EXPECT_EQ(0x420C4146u, Encoding::decode(Encoding::UTF_8, (const uint8_t *)"\xFD\x82\x83\x84\x85\x86"));
}
TEST(Encoding_decode, decodes_CP_1252_correctly)
{
EXPECT_EQ(0x99u, Encoding::decode(Encoding::CP_1252, (const uint8_t *)"\x99"));
}