- 码表查询从精确匹配改为前缀匹配:精确等长行优先(保住一级/二级简码), 前缀行按权重降序跟随,LIMIT 50 截断(引擎层不读 server 配置的分层边界) - WubiCandidateProvider 注入 RuntimePaths:接通 update_weight / delete_by_key 到 user_dictionary journal;新增 bump_wubi_weight(同码组 MAX+1、主库更新 与 journal upsert 同事务,行缺失不造行) - create_word / cache_dynamic_candidate 补「为何 no-op」注释(后续任务范畴) - 两个 toml 的 wubi_schema 注释与实现对齐:当前仅支持 wubi86 - server 测试:前缀语义、LIMIT 截断、上界构造、调频+journal 回放、删词墓碑
261 lines
10 KiB
C++
261 lines
10 KiB
C++
#include "tests/includes/test_framework.h"
|
|
#include "tests/includes/test_utf8_path.h"
|
|
#include "engine/contracts/assets/assets.h"
|
|
#include "engine/core/runtime_paths.h"
|
|
#include "engine/providers/wubi_candidate_provider.h"
|
|
#include "engine/schemes/wubi_scheme.h"
|
|
#include <filesystem>
|
|
#include <sqlite3.h>
|
|
#include <stdexcept>
|
|
#include <string>
|
|
#include <windows.h>
|
|
|
|
namespace
|
|
{
|
|
void InputLetters(WubiScheme &scheme, const std::string &keys)
|
|
{
|
|
for (const char ch : keys)
|
|
{
|
|
const char upper = ch >= 'a' && ch <= 'z' ? static_cast<char>(ch - ('a' - 'A')) : ch;
|
|
scheme.handle_key(static_cast<UINT>(upper), 0, static_cast<WCHAR>(ch));
|
|
}
|
|
}
|
|
|
|
// 与 test_user_dictionary_journal.cpp 的同名助手同型:打开临时库跑语句、取标量。
|
|
class TestDatabase
|
|
{
|
|
public:
|
|
explicit TestDatabase(const std::filesystem::path &path)
|
|
{
|
|
REQUIRE_EQ(sqlite3_open(test::Utf8(path).c_str(), &db_), SQLITE_OK);
|
|
}
|
|
~TestDatabase()
|
|
{
|
|
if (db_ != nullptr)
|
|
sqlite3_close(db_);
|
|
}
|
|
void exec(const std::string &sql)
|
|
{
|
|
REQUIRE_EQ(sqlite3_exec(db_, sql.c_str(), nullptr, nullptr, nullptr), SQLITE_OK);
|
|
}
|
|
std::int64_t scalar_int64(const std::string &sql)
|
|
{
|
|
sqlite3_stmt *stmt = nullptr;
|
|
REQUIRE_EQ(sqlite3_prepare_v2(db_, sql.c_str(), -1, &stmt, nullptr), SQLITE_OK);
|
|
REQUIRE_EQ(sqlite3_step(stmt), SQLITE_ROW);
|
|
const std::int64_t value = sqlite3_column_int64(stmt, 0);
|
|
sqlite3_finalize(stmt);
|
|
return value;
|
|
}
|
|
std::int64_t scalar_int(const std::string &sql)
|
|
{
|
|
return scalar_int64(sql);
|
|
}
|
|
|
|
private:
|
|
sqlite3 *db_ = nullptr;
|
|
};
|
|
|
|
// journal 路径经 RuntimePaths 注入临时目录,避免测试写真实用户目录;目录须先存在,
|
|
// 与生产环境一致(RuntimePaths 管理的世代目录总是已建好)。
|
|
metasequoia::RuntimePaths PathsIn(const std::filesystem::path &directory)
|
|
{
|
|
metasequoia::RuntimePaths paths;
|
|
paths.resources = directory / "resources";
|
|
paths.user_data = directory / "user";
|
|
paths.cache = directory / "cache";
|
|
paths.dictionaries = directory / "resources";
|
|
std::filesystem::create_directories(paths.user_data);
|
|
return paths;
|
|
}
|
|
|
|
std::filesystem::path CreateWubiDatabase()
|
|
{
|
|
const auto path = std::filesystem::temp_directory_path() / "msime-wubi-provider-test.db";
|
|
std::filesystem::remove(path);
|
|
|
|
sqlite3 *db = nullptr;
|
|
if (sqlite3_open(test::Utf8(path).c_str(), &db) != SQLITE_OK)
|
|
{
|
|
throw std::runtime_error("Failed to create temporary Wubi database.");
|
|
}
|
|
|
|
const char *sql = "CREATE TABLE wubi86 (\"key\" TEXT, \"value\" TEXT, \"weight\" INTEGER);"
|
|
"INSERT INTO wubi86 VALUES ('a', '工', 20);"
|
|
"INSERT INTO wubi86 VALUES ('a', '戈', 10);"
|
|
"INSERT INTO wubi86 VALUES ('aa', '式', 20);"
|
|
// 上界邻居:'y' 前缀查询必须命中 'yh',且排除上界之外的 'z' 行。
|
|
"INSERT INTO wubi86 VALUES ('y', '也', 5);"
|
|
"INSERT INTO wubi86 VALUES ('yh', '又', 6);"
|
|
"INSERT INTO wubi86 VALUES ('z', '越过上界', 999);";
|
|
const int result = sqlite3_exec(db, sql, nullptr, nullptr, nullptr);
|
|
sqlite3_close(db);
|
|
if (result != SQLITE_OK)
|
|
{
|
|
std::filesystem::remove(path);
|
|
throw std::runtime_error("Failed to initialize temporary Wubi database.");
|
|
}
|
|
return path;
|
|
}
|
|
|
|
QueryRequest WubiRequest(const std::string &code)
|
|
{
|
|
QueryRequest request;
|
|
request.scheme = SchemeType::Wubi;
|
|
request.raw_input = code;
|
|
request.normalized_input = code;
|
|
request.valid = true;
|
|
return request;
|
|
}
|
|
} // namespace
|
|
|
|
TEST_CASE(WubiSchemeBuildsDirectCodeRequest)
|
|
{
|
|
WubiScheme scheme;
|
|
InputLetters(scheme, "aa");
|
|
|
|
const QueryRequest request = scheme.build_request();
|
|
REQUIRE(request.valid);
|
|
REQUIRE_EQ(request.scheme, SchemeType::Wubi);
|
|
REQUIRE_EQ(request.raw_input, std::string("aa"));
|
|
REQUIRE_EQ(request.normalized_input, std::string("aa"));
|
|
REQUIRE_EQ(request.segmentation, std::string("aa"));
|
|
}
|
|
|
|
TEST_CASE(WubiSchemeAcceptsAtMostFourCodesAndDoesNotUseZ)
|
|
{
|
|
WubiScheme scheme;
|
|
InputLetters(scheme, "abcdez");
|
|
REQUIRE_EQ(scheme.build_request().raw_input, std::string("abcd"));
|
|
|
|
scheme.handle_key(VK_BACK, 0, 0);
|
|
REQUIRE_EQ(scheme.build_request().raw_input, std::string("abc"));
|
|
}
|
|
|
|
TEST_CASE(WubiProviderReturnsPrefixMatchesWithExactCodeFirst)
|
|
{
|
|
const auto db_path = CreateWubiDatabase();
|
|
{
|
|
WubiCandidateProvider provider(test::Utf8(db_path));
|
|
|
|
// 1 码:精确等长行在前(按权重),更长前缀行按权重跟随——逐键提示的基础。
|
|
const auto one_code = provider.query(WubiRequest("a"));
|
|
REQUIRE_EQ(one_code.size(), static_cast<size_t>(3));
|
|
REQUIRE_EQ(one_code[0].pinyin, std::string("a"));
|
|
REQUIRE_EQ(one_code[0].word, std::string("工"));
|
|
REQUIRE_EQ(one_code[0].weight, 20);
|
|
REQUIRE_EQ(one_code[1].word, std::string("戈"));
|
|
REQUIRE_EQ(one_code[2].pinyin, std::string("aa"));
|
|
REQUIRE_EQ(one_code[2].word, std::string("式"));
|
|
|
|
// 每一码都是前缀查询:2 码精确命中只剩等长行。
|
|
const auto two_code = provider.query(WubiRequest("aa"));
|
|
REQUIRE_EQ(two_code.size(), static_cast<size_t>(1));
|
|
REQUIRE_EQ(two_code[0].word, std::string("式"));
|
|
|
|
// 精确优先压过权重:'y'(5) 排在前缀行 'yh'(6) 之前;上界构造排除 'z' 行。
|
|
const auto upper_edge = provider.query(WubiRequest("y"));
|
|
REQUIRE_EQ(upper_edge.size(), static_cast<size_t>(2));
|
|
REQUIRE_EQ(upper_edge[0].word, std::string("也"));
|
|
REQUIRE_EQ(upper_edge[1].word, std::string("又"));
|
|
}
|
|
std::filesystem::remove(db_path);
|
|
}
|
|
|
|
TEST_CASE(WubiProviderLimitsPrefixScanToBoundedRows)
|
|
{
|
|
const auto directory =
|
|
std::filesystem::temp_directory_path() / ("msime-wubi-provider-limit-" + std::to_string(GetCurrentProcessId()));
|
|
std::filesystem::create_directories(directory);
|
|
const auto db_path = directory / "limit.db";
|
|
{
|
|
TestDatabase db(db_path);
|
|
std::string sql = "CREATE TABLE wubi86(\"key\" TEXT,\"value\" TEXT,\"weight\" INTEGER);";
|
|
// 55 个精确行 + 5 个前缀行:LIMIT 截断后可见的只能是权重最高的 50 个精确行,
|
|
// 前缀行整体排在精确组之后,不应被返回。
|
|
for (int weight = 1; weight <= 55; ++weight)
|
|
{
|
|
sql += "INSERT INTO wubi86 VALUES('a','字" + std::to_string(weight) + "'," + std::to_string(weight) + ");";
|
|
}
|
|
for (int weight = 1; weight <= 5; ++weight)
|
|
{
|
|
sql += "INSERT INTO wubi86 VALUES('ab','前" + std::to_string(weight) + "'," + std::to_string(weight) + ");";
|
|
}
|
|
db.exec(sql);
|
|
}
|
|
{
|
|
WubiCandidateProvider provider(test::Utf8(db_path));
|
|
const auto candidates = provider.query(WubiRequest("a"));
|
|
REQUIRE_EQ(candidates.size(), static_cast<size_t>(50));
|
|
REQUIRE_EQ(candidates[0].weight, 55);
|
|
REQUIRE_EQ(candidates[49].weight, 6);
|
|
}
|
|
std::filesystem::remove_all(directory);
|
|
}
|
|
|
|
TEST_CASE(WubiProviderLearnsSelectionIntoWeightAndJournal)
|
|
{
|
|
const auto directory =
|
|
std::filesystem::temp_directory_path() / ("msime-wubi-provider-learn-" + std::to_string(GetCurrentProcessId()));
|
|
std::filesystem::create_directories(directory);
|
|
const auto db_path = directory / "main.db";
|
|
const auto paths = PathsIn(directory);
|
|
{
|
|
TestDatabase db(db_path);
|
|
db.exec("CREATE TABLE wubi86(\"key\" TEXT,\"value\" TEXT,\"weight\" INTEGER);"
|
|
"INSERT INTO wubi86 VALUES('aa','式',20);"
|
|
"INSERT INTO wubi86 VALUES('aa','工',10);");
|
|
}
|
|
{
|
|
WubiCandidateProvider provider(test::Utf8(db_path), paths);
|
|
// 选中次位候选:权重升到同码组最高(MAX+1),与全拼 build_sql_for_updating_word 语义一致。
|
|
REQUIRE_EQ(provider.update_weight_by_pinyin_and_word(SchemeType::Wubi, "aa", "工"), 0);
|
|
const auto candidates = provider.query(WubiRequest("aa"));
|
|
REQUIRE_EQ(candidates.size(), static_cast<size_t>(2));
|
|
REQUIRE_EQ(candidates[0].word, std::string("工"));
|
|
REQUIRE_EQ(candidates[0].weight, 21);
|
|
|
|
// 行不存在不造行。
|
|
REQUIRE_EQ(provider.update_weight_by_pinyin_and_word(SchemeType::Wubi, "ab", "不存在"), -1);
|
|
}
|
|
{
|
|
TestDatabase journal(paths.user(metasequoia::assets::user_journal));
|
|
REQUIRE_EQ(journal.scalar_int("SELECT COUNT(*) FROM user_dictionary_operations WHERE dictionary='wubi' "
|
|
"AND key='aa' AND value='工' AND operation='upsert' AND weight=21"),
|
|
1);
|
|
REQUIRE_EQ(journal.scalar_int("SELECT COUNT(*) FROM user_dictionary_operations WHERE value='不存在'"), 0);
|
|
}
|
|
std::filesystem::remove_all(directory);
|
|
}
|
|
|
|
TEST_CASE(WubiProviderDeletesWubiCandidateThroughJournal)
|
|
{
|
|
const auto directory = std::filesystem::temp_directory_path() /
|
|
("msime-wubi-provider-delete-" + std::to_string(GetCurrentProcessId()));
|
|
std::filesystem::create_directories(directory);
|
|
const auto db_path = directory / "main.db";
|
|
const auto paths = PathsIn(directory);
|
|
{
|
|
TestDatabase db(db_path);
|
|
db.exec("CREATE TABLE wubi86(\"key\" TEXT,\"value\" TEXT,\"weight\" INTEGER);"
|
|
"INSERT INTO wubi86 VALUES('aa','式',20);"
|
|
"INSERT INTO wubi86 VALUES('aa','工',10);");
|
|
}
|
|
{
|
|
WubiCandidateProvider provider(test::Utf8(db_path), paths);
|
|
REQUIRE_EQ(provider.delete_by_pinyin_and_word(SchemeType::Wubi, "aa", "式"), 0);
|
|
// 再删已不存在的行:changes()==0,失败且不写墓碑。
|
|
REQUIRE_EQ(provider.delete_by_pinyin_and_word(SchemeType::Wubi, "aa", "式"), -1);
|
|
const auto remaining = provider.query(WubiRequest("aa"));
|
|
REQUIRE_EQ(remaining.size(), static_cast<size_t>(1));
|
|
REQUIRE_EQ(remaining[0].word, std::string("工"));
|
|
}
|
|
{
|
|
TestDatabase journal(paths.user(metasequoia::assets::user_journal));
|
|
REQUIRE_EQ(journal.scalar_int("SELECT COUNT(*) FROM user_dictionary_operations WHERE dictionary='wubi' "
|
|
"AND key='aa' AND value='式' AND operation='delete'"),
|
|
1);
|
|
}
|
|
std::filesystem::remove_all(directory);
|
|
}
|