summaryrefslogtreecommitdiff
path: root/utf/tst/test_utf8.cc
diff options
context:
space:
mode:
Diffstat (limited to 'utf/tst/test_utf8.cc')
-rw-r--r--utf/tst/test_utf8.cc204
1 files changed, 131 insertions, 73 deletions
diff --git a/utf/tst/test_utf8.cc b/utf/tst/test_utf8.cc
index 10df969..8bdeba4 100644
--- a/utf/tst/test_utf8.cc
+++ b/utf/tst/test_utf8.cc
@@ -2,187 +2,245 @@
#include "utf_error.hh"
+#include <array>
+#include <gmock/gmock.h>
#include <gtest/gtest.h>
+#include <span>
-TEST(utf8, sanity) {
- std::string_view str("$");
+TEST(utf8, read_sanity) {
size_t offset = 0;
- auto ret = utf::read8(str, offset);
+ auto ret = utf::read8(std::array<uint8_t, 1>({'$'}), offset);
EXPECT_EQ('$', ret);
EXPECT_EQ(1, offset);
- str = "\xC2\xA3";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 2>({0xC2, 0xA3}), offset);
EXPECT_EQ(0xa3, ret);
EXPECT_EQ(2, offset);
- str = "\xD0\x98";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 2>({0xD0, 0x98}), offset);
EXPECT_EQ(0x418, ret);
EXPECT_EQ(2, offset);
- str = "\xE0\xA4\xB9";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 3>({0xE0, 0xA4, 0xB9}), offset);
EXPECT_EQ(0x939, ret);
EXPECT_EQ(3, offset);
- str = "\xE2\x82\xAC";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 3>({0xE2, 0x82, 0xAC}), offset);
EXPECT_EQ(0x20AC, ret);
EXPECT_EQ(3, offset);
- str = "\xED\x95\x9C";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 3>({0xED, 0x95, 0x9C}), offset);
EXPECT_EQ(0xD55C, ret);
EXPECT_EQ(3, offset);
- str = "\xF0\x90\x8D\x88";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 4>({0xF0, 0x90, 0x8D, 0x88}), offset);
EXPECT_EQ(0x10348, ret);
EXPECT_EQ(4, offset);
}
-TEST(utf8, overlong) {
- std::string_view str("\xF0\x82\x82\xAC");
+TEST(utf8, write_sanity) {
+ std::array<uint8_t, 10> out;
size_t offset = 0;
- auto ret = utf::read8(str, offset);
+ EXPECT_TRUE(utf::write8('$', out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre('$'));
+ EXPECT_EQ(1, offset);
+
+ offset = 0;
+ EXPECT_TRUE(utf::write8(0xa3, out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre(0xC2, 0xA3));
+ EXPECT_EQ(2, offset);
+
+ offset = 0;
+ EXPECT_TRUE(utf::write8(0x418, out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre(0xD0, 0x98));
+ EXPECT_EQ(2, offset);
+
+ offset = 0;
+ EXPECT_TRUE(utf::write8(0x939, out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre(0xE0, 0xA4, 0xB9));
+ EXPECT_EQ(3, offset);
+
+ offset = 0;
+ EXPECT_TRUE(utf::write8(0x20AC, out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre(0xE2, 0x82, 0xAC));
+ EXPECT_EQ(3, offset);
+
+ offset = 0;
+ EXPECT_TRUE(utf::write8(0xD55C, out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre(0xED, 0x95, 0x9C));
+ EXPECT_EQ(3, offset);
+
+ offset = 0;
+ EXPECT_TRUE(utf::write8(0x10348, out, offset));
+ EXPECT_THAT(std::span(out).subspan(0, offset),
+ testing::ElementsAre(0xF0, 0x90, 0x8D, 0x88));
+ EXPECT_EQ(4, offset);
+}
+
+TEST(utf8, read_overlong) {
+ size_t offset = 0;
+ auto ret = utf::read8(
+ std::array<uint8_t, 4>({0xF0, 0x82, 0x82, 0xAC}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "\xE0\x81\x81";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 3>({0xE0, 0x81, 0x81}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "\xC0\x80";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 2>({0xC0, 0x80}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
}
-TEST(utf8, invalid) {
- std::string_view str("\xED\xB0\x80");
+TEST(utf8, read_invalid) {
size_t offset = 0;
- auto ret = utf::read8(str, offset);
+ auto ret = utf::read8(std::array<uint8_t, 3>({0xED, 0xB0, 0x80}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "\xFB\xFF\xFF";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 3>({0xFB, 0xFF, 0xFF}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "\xFF\xFF\xFF\xFF\xFF";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(
+ std::array<uint8_t, 5>({0xFF, 0xFF, 0xFF, 0xFF, 0xFF}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 0>(), offset);
EXPECT_EQ(utf::NEED_MORE, ret);
EXPECT_EQ(0, offset);
- str = "\x80";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 1>({0x80}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "\xC2";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 1>({0xC2}), offset);
EXPECT_EQ(utf::NEED_MORE, ret);
EXPECT_EQ(0, offset);
- str = "\xC2\x03";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 2>({0xC2, 0x03}), offset);
EXPECT_EQ(utf::INVALID, ret);
EXPECT_EQ(0, offset);
- str = "\xE0\xA4";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 2>({0xE0, 0xA4}), offset);
EXPECT_EQ(utf::NEED_MORE, ret);
EXPECT_EQ(0, offset);
- str = "\xF0\x90\x8D";
offset = 0;
- ret = utf::read8(str, offset);
+ ret = utf::read8(std::array<uint8_t, 3>({0xF0, 0x90, 0x8D}), offset);
EXPECT_EQ(utf::NEED_MORE, ret);
EXPECT_EQ(0, offset);
}
-TEST(utf8, multiple1) {
- std::string_view str("\x4D\xC3\xAC\x6E\x68\x20\x6E\xC3\xB3\x69\x20\x74\x69"
- "\xE1\xBA\xBF\x6E\x67\x20\x56\x69\xE1\xBB\x87\x74");
+TEST(utf8, read_multiple1) {
+ std::array<uint8_t, 25> data({
+ 0x4D, 0xC3, 0xAC, 0x6E, 0x68, 0x20, 0x6E, 0xC3, 0xB3, 0x69,
+ 0x20, 0x74, 0x69, 0xE1, 0xBA, 0xBF, 0x6E, 0x67, 0x20, 0x56,
+ 0x69, 0xE1, 0xBB, 0x87, 0x74
+ });
size_t offset = 0;
- auto ret = utf::read8(str, offset);
+ auto ret = utf::read8(data, offset);
EXPECT_EQ('M', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0xEC, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('n', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('h', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(' ', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('n', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0xF3, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('i', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(' ', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('t', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('i', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0x1EBF, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('n', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('g', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(' ', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('V', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('i', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0x1EC7, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ('t', ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(utf::NEED_MORE, ret);
- EXPECT_EQ(str.size(), offset);
+ EXPECT_EQ(data.size(), offset);
}
-TEST(utf8, multiple2) {
- std::string_view str("\xF0\xA8\x89\x9F\xE5\x91\x90\xE3\x97\x82\xE8\xB6\x8A");
+TEST(utf8, read_multiple2) {
+ std::array<uint8_t, 13> data({
+ 0xF0, 0xA8, 0x89, 0x9F, 0xE5, 0x91, 0x90, 0xE3, 0x97, 0x82,
+ 0xE8, 0xB6, 0x8A,
+ });
size_t offset = 0;
- auto ret = utf::read8(str, offset);
+ auto ret = utf::read8(data, offset);
EXPECT_EQ(0x2825F, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0x5450, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0x35C2, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(0x8D8A, ret);
- ret = utf::read8(str, offset);
+ ret = utf::read8(data, offset);
EXPECT_EQ(utf::NEED_MORE, ret);
- EXPECT_EQ(str.size(), offset);
+ EXPECT_EQ(data.size(), offset);
+}
+
+TEST(utf8, write_no_space) {
+ std::array<uint8_t, 10> data;
+ std::span<uint8_t> out(data);
+ size_t offset = 0;
+ EXPECT_FALSE(utf::write8('$', out.subspan(0, 0), offset));
+ EXPECT_EQ(0u, offset);
+
+ EXPECT_FALSE(utf::write8(0xa3, out.subspan(0, 1), offset));
+ EXPECT_EQ(0u, offset);
+ EXPECT_FALSE(utf::write8(0x418, out.subspan(0, 0), offset));
+ EXPECT_EQ(0u, offset);
+
+ EXPECT_FALSE(utf::write8(0x939, out.subspan(0, 2), offset));
+ EXPECT_EQ(0u, offset);
+ EXPECT_FALSE(utf::write8(0x20AC, out.subspan(0, 0), offset));
+ EXPECT_EQ(0u, offset);
+
+ EXPECT_FALSE(utf::write8(0x10348, out.subspan(0, 3), offset));
+ EXPECT_EQ(0u, offset);
}
generated by cgit v1.2.3-70-g09d2 (git 2.52.0) at 2025-12-06 23:51:27 +0000