Jia Suiandfanlusky
93f95d1245
* fix(engine): emoji 查询精确命中优先于前缀命中 (#988) * fix(engine): 颜文字查询同样精确命中优先于前缀命中 (#988) * fix(engine): emoji / 颜文字精确提级只认全拼，并合并为单次排序查询 - 双拼下原始码只按目录顺序参与前缀命中，不再精确提级：小鹤 xl 精确命中简拼键时会压过展开后的 xiang。 - 颜文字只有全拼精确命中提级，简拼精确命中不提级：单音节条目的简拼只有一个字母，打 k 时会整批排到前面。 - 两级查询合并为每个查询码一条 SQL（精确行在 LIMIT 之前），语句只 prepare 一次，循环抽到 local_database 的 query_ranked_entries 共用。 - server 回归测试不再断言 ku 多个精确条目之间的目录顺序。 --------- Co-authored-by: fanlusky <1730976608@qq.com>
94 lines
3.3 KiB
C++
94 lines
3.3 KiB
C++
#include "../contracts/assets/assets.h"
|
|
#include "kaomoji_query.h"
|
|
#include "local_database.h"
|
|
|
|
#include "../core/data_path.h"
|
|
#include "../shuangpin/shuangpin_query.h"
|
|
|
|
#include <algorithm>
|
|
#include <cctype>
|
|
#include <memory>
|
|
#include <vector>
|
|
|
|
namespace metasequoia::local_modes
|
|
{
|
|
namespace
|
|
{
|
|
bool valid_code(const std::string &code)
|
|
{
|
|
return !code.empty() && std::all_of(code.begin(), code.end(), [](unsigned char character) {
|
|
return (character >= 'a' && character <= 'z') || (character >= 'A' && character <= 'Z') || character == '\'';
|
|
});
|
|
}
|
|
|
|
std::string lower_ascii(std::string text)
|
|
{
|
|
std::transform(text.begin(), text.end(), text.begin(),
|
|
[](unsigned char character) { return static_cast<char>(std::tolower(character)); });
|
|
return text;
|
|
}
|
|
|
|
LocalQueryResult query_failure(const char *diagnostic)
|
|
{
|
|
return {{}, std::string(diagnostic)};
|
|
}
|
|
} // namespace
|
|
|
|
LocalQueryResult query_kaomoji(const std::string &code, SchemeType scheme, int limit, const ShuangpinProfile &profile)
|
|
{
|
|
return query_kaomoji(code, scheme, data_file_path(metasequoia::assets::other_dictionary), limit, profile);
|
|
}
|
|
|
|
LocalQueryResult query_kaomoji(const std::string &code, SchemeType scheme, const std::filesystem::path &database_path,
|
|
int limit, const ShuangpinProfile &profile)
|
|
{
|
|
if (!valid_code(code) || limit <= 0)
|
|
{
|
|
return {};
|
|
}
|
|
if (database_path.empty())
|
|
{
|
|
return query_failure("Kaomoji database is unavailable.");
|
|
}
|
|
|
|
const std::string lower = lower_ascii(code);
|
|
std::vector<RankedLookup> lookups{{lower, true}};
|
|
if (scheme == SchemeType::Shuangpin)
|
|
{
|
|
const std::string quanpin = shuangpin::normalize_input(lower, profile);
|
|
if (!quanpin.empty() && quanpin != lower)
|
|
{
|
|
// 同 emoji：双拼下精确命中只认展开后的全拼，原始码只按目录顺序参与前缀命中。
|
|
lookups.front().exact_ranks = false;
|
|
lookups.push_back({quanpin, true});
|
|
}
|
|
}
|
|
|
|
const std::shared_ptr<sqlite3> database = open_local_database(database_path);
|
|
if (!database)
|
|
{
|
|
return query_failure("Kaomoji database is unavailable.");
|
|
}
|
|
|
|
// 同 emoji：全拼精确命中排在更长键的前缀命中之前——bai 的精确命中不能被 baifo 顶掉。
|
|
// 简拼精确命中不提级：单音节条目的简拼就是一个字母，打 k 时所有 ku 的条目都会精确命中 jianpin。
|
|
constexpr const char *kSql = "SELECT kaomoji,sort_order,pinyin IS ?1 FROM kaomoji "
|
|
"WHERE (pinyin>=?1 AND pinyin<?2) OR (jianpin>=?1 AND jianpin<?2) "
|
|
"ORDER BY pinyin IS NOT ?1,sort_order LIMIT ?3";
|
|
const auto entries = query_ranked_entries(database.get(), kSql, lookups, limit);
|
|
if (!entries)
|
|
{
|
|
return query_failure("Kaomoji database could not be queried.");
|
|
}
|
|
const std::size_t count = entries->size();
|
|
LocalQueryResult result;
|
|
result.candidates.reserve(count);
|
|
for (std::size_t index = 0; index < count; ++index)
|
|
{
|
|
result.candidates.emplace_back(lower, (*entries)[index], static_cast<std::int64_t>(count - index),
|
|
CandidateSource::Kaomoji);
|
|
}
|
|
return result;
|
|
}
|
|
} // namespace metasequoia::local_modes
|