#include <cstdint>
#include <cstring>
#include <vector>
#include "DartsDict.hpp"
#include "TestUtilsUTF8.hpp"
#include "TextDictTestBase.hpp"
namespace opencc {
class DartsDictTest : public TextDictTestBase {
protected:
DartsDictTest()
: dartsDict(DartsDict::NewFromDict(*textDict.get())),
fileName("dict.ocd"){};
static std::string WriteMalformedDartsFile(uint32_t dartsSize) {
const std::string path = "malformed_darts.ocd";
FILE* fp = fopen(path.c_str(), "wb");
const char* header = "OPENCCDARTS1";
fwrite(header, sizeof(char), strlen(header), fp);
fwrite(&dartsSize, sizeof(uint32_t), 1, fp);
uint32_t fakeFirstUnit = 0x00000400U;
fwrite(&fakeFirstUnit, sizeof(uint32_t), 1, fp);
fclose(fp);
return path;
}
static void SerializeAsLegacy64(const DartsDictPtr& dict,
const std::string& path) {
const std::string tmp = path + ".tmp";
dict->opencc::SerializableDict::SerializeToFile(tmp);
FILE* src = fopen(tmp.c_str(), "rb");
fseek(src, 12, SEEK_SET);
uint32_t dartsSize32 = 0;
fread(&dartsSize32, sizeof(uint32_t), 1, src);
size_t numUnits = dartsSize32 / sizeof(uint32_t);
std::vector<uint32_t> units32(numUnits);
fread(units32.data(), sizeof(uint32_t), numUnits, src);
long binaryStart = ftell(src);
FILE* dst = fopen(path.c_str(), "wb");
const char* header = "OPENCCDARTS1";
fwrite(header, sizeof(char), strlen(header), dst);
uint64_t dartsSize64 =
static_cast<uint64_t>(numUnits) * sizeof(uint64_t);
fwrite(&dartsSize64, sizeof(uint64_t), 1, dst);
for (size_t i = 0; i < numUnits; ++i) {
uint64_t unit64 = units32[i]; fwrite(&unit64, sizeof(uint64_t), 1, dst);
}
fseek(src, binaryStart, SEEK_SET);
char buf[4096];
size_t n;
while ((n = fread(buf, 1, sizeof(buf), src)) > 0) {
fwrite(buf, 1, n, dst);
}
fclose(src);
fclose(dst);
std::remove(tmp.c_str());
}
static void CorruptDartsRootUnit(const std::string& path) {
FILE* fp = fopen(path.c_str(), "r+b");
const char* header = "OPENCCDARTS1";
fseek(fp, static_cast<long>(strlen(header) + sizeof(uint32_t)), SEEK_SET);
uint32_t invalidRoot = 0;
fwrite(&invalidRoot, sizeof(uint32_t), 1, fp);
fclose(fp);
}
static void CorruptFirstDartsValue(const std::string& path) {
FILE* fp = fopen(path.c_str(), "r+b");
const char* header = "OPENCCDARTS1";
fseek(fp, static_cast<long>(strlen(header)), SEEK_SET);
uint32_t dartsSize = 0;
fread(&dartsSize, sizeof(uint32_t), 1, fp);
const long dartsOffset =
static_cast<long>(strlen(header) + sizeof(uint32_t));
const size_t numUnits = static_cast<size_t>(dartsSize) / sizeof(uint32_t);
for (size_t i = 0; i < numUnits; i++) {
uint32_t unit = 0;
fseek(fp, dartsOffset + static_cast<long>(i * sizeof(uint32_t)),
SEEK_SET);
fread(&unit, sizeof(uint32_t), 1, fp);
const uint32_t label = unit & ((1U << 31) | 0xFF);
if (label > 0xFF) {
const uint32_t invalidValue = (1U << 31) | ((1U << 31) - 1);
fseek(fp, dartsOffset + static_cast<long>(i * sizeof(uint32_t)),
SEEK_SET);
fwrite(&invalidValue, sizeof(uint32_t), 1, fp);
break;
}
}
fclose(fp);
}
const DartsDictPtr dartsDict;
const std::string fileName;
};
TEST_F(DartsDictTest, DictTest) { TestDict(dartsDict); }
TEST_F(DartsDictTest, Serialization) {
dartsDict->opencc::SerializableDict::SerializeToFile(fileName);
}
TEST_F(DartsDictTest, Deserialization) {
const DartsDictPtr& deserialized =
SerializableDict::NewFromFile<DartsDict>(fileName);
TestDict(deserialized);
const LexiconPtr& lex1 = dartsDict->GetLexicon();
const LexiconPtr& lex2 = deserialized->GetLexicon();
EXPECT_EQ(lex1->Length(), lex2->Length());
for (size_t i = 0; i < lex1->Length(); i++) {
EXPECT_EQ(lex1->At(i)->Key(), lex2->At(i)->Key());
EXPECT_EQ(lex1->At(i)->NumValues(), lex2->At(i)->NumValues());
}
const TextDictPtr deserializedTextDict(new TextDict(lex2));
TestDict(deserializedTextDict);
}
TEST_F(DartsDictTest, ExactMatch) {
auto there = dartsDict->Match("積羽沉舟", 12);
EXPECT_FALSE(there.IsNull());
auto dictEntry = there.Get();
EXPECT_EQ(1, dictEntry->NumValues());
EXPECT_EQ(utf8("羣輕折軸"), dictEntry->GetDefault());
auto nowhere = dartsDict->Match("積羽沉舟衆口鑠金", 24);
EXPECT_TRUE(nowhere.IsNull());
}
TEST_F(DartsDictTest, RejectsHugeDartsSize) {
std::string path = WriteMalformedDartsFile(0x13000000U);
EXPECT_THROW(SerializableDict::NewFromFile<DartsDict>(path), InvalidFormat);
std::remove(path.c_str());
}
TEST_F(DartsDictTest, RejectsMisalignedDartsSize) {
std::string path = WriteMalformedDartsFile(1);
FILE* fp = fopen(path.c_str(), "ab");
char padding = '\0';
fwrite(&padding, sizeof(char), 1, fp);
fclose(fp);
EXPECT_THROW(SerializableDict::NewFromFile<DartsDict>(path), InvalidFormat);
std::remove(path.c_str());
}
TEST_F(DartsDictTest, RejectsInvalidDartsRoot) {
const std::string path = "invalid_darts_root.ocd";
dartsDict->opencc::SerializableDict::SerializeToFile(path);
CorruptDartsRootUnit(path);
EXPECT_THROW(SerializableDict::NewFromFile<DartsDict>(path), InvalidFormat);
std::remove(path.c_str());
}
TEST_F(DartsDictTest, RejectsInvalidDartsValue) {
const std::string path = "invalid_darts_value.ocd";
dartsDict->opencc::SerializableDict::SerializeToFile(path);
CorruptFirstDartsValue(path);
EXPECT_THROW(SerializableDict::NewFromFile<DartsDict>(path), InvalidFormat);
std::remove(path.c_str());
}
TEST_F(DartsDictTest, LoadsLegacy64bitFormat) {
const std::string path = "legacy64_dict.ocd";
SerializeAsLegacy64(dartsDict, path);
const DartsDictPtr& loaded = SerializableDict::NewFromFile<DartsDict>(path);
TestDict(loaded);
auto there = loaded->Match("積羽沉舟", 12);
EXPECT_FALSE(there.IsNull());
EXPECT_EQ(utf8("羣輕折軸"), there.Get()->GetDefault());
std::remove(path.c_str());
}
TEST_F(DartsDictTest, RejectsInvalidLegacy64Root) {
const std::string path = "invalid_legacy64_root.ocd";
SerializeAsLegacy64(dartsDict, path);
FILE* fp = fopen(path.c_str(), "r+b");
fseek(fp, 12 + 8, SEEK_SET);
uint64_t zeroUnit = 0;
fwrite(&zeroUnit, sizeof(uint64_t), 1, fp);
fclose(fp);
EXPECT_THROW(SerializableDict::NewFromFile<DartsDict>(path), InvalidFormat);
std::remove(path.c_str());
}
TEST_F(DartsDictTest, Legacy64PrefixSearchBeyond64Matches) {
const size_t kEntries = 65;
LexiconPtr lex(new Lexicon);
for (size_t i = 1; i <= kEntries; ++i) {
lex->Add(DictEntryFactory::New(std::string(i, 'a'), "v" + std::to_string(i)));
}
lex->Sort();
TextDictPtr deepText(new TextDict(lex));
DartsDictPtr deepDict = DartsDict::NewFromDict(*deepText);
const std::string path = "legacy64_deep.ocd";
SerializeAsLegacy64(deepDict, path);
const DartsDictPtr& loaded = SerializableDict::NewFromFile<DartsDict>(path);
auto result = loaded->MatchPrefix(std::string(kEntries, 'a').c_str(), kEntries);
ASSERT_FALSE(result.IsNull());
EXPECT_EQ("v65", result.Get()->GetDefault());
std::remove(path.c_str());
}
TEST_F(DartsDictTest, Legacy64SerializeToFileThrows) {
const std::string path = "legacy64_for_serialize.ocd";
SerializeAsLegacy64(dartsDict, path);
const DartsDictPtr& loaded = SerializableDict::NewFromFile<DartsDict>(path);
const std::string outPath = "legacy64_out.ocd";
EXPECT_THROW(loaded->opencc::SerializableDict::SerializeToFile(outPath),
InvalidFormat);
std::remove(path.c_str());
std::remove(outPath.c_str());
}
}