Textractor_test/extensions/translatewrapper.cpp

179 lines
6.7 KiB
C++
Raw Normal View History

#include "qtcommon.h"
#include "extension.h"
#include "blockmarkup.h"
2019-06-13 16:01:29 +08:00
#include "network.h"
#include <map>
#include <fstream>
#include <QComboBox>
2019-06-13 16:01:29 +08:00
extern const char* NATIVE_LANGUAGE;
extern const char* TRANSLATE_TO;
extern const char* TRANSLATE_SELECTED_THREAD_ONLY;
extern const char* RATE_LIMIT_ALL_THREADS;
extern const char* RATE_LIMIT_SELECTED_THREAD;
extern const char* USE_TRANS_CACHE;
extern const char* RATE_LIMIT_TOKEN_COUNT;
extern const char* RATE_LIMIT_TOKEN_RESTORE_DELAY;
extern const char* MAX_SENTENCE_SIZE;
extern const char* API_KEY;
2019-06-13 16:01:29 +08:00
extern const wchar_t* TOO_MANY_TRANS_REQUESTS;
extern const char* TRANSLATION_PROVIDER;
extern const char* GET_API_KEY_FROM;
2019-06-13 16:01:29 +08:00
extern QStringList languages;
extern bool translateSelectedOnly, rateLimitAll, rateLimitSelected, useCache;
extern int tokenCount, tokenRestoreDelay, maxSentenceSize;
2020-04-26 10:39:12 +08:00
std::pair<bool, std::wstring> Translate(const std::wstring& text);
2019-06-13 16:01:29 +08:00
2019-08-12 22:44:51 +08:00
const char* LANGUAGE = u8"Language";
2020-12-14 21:26:01 +08:00
const std::string TRANSLATION_CACHE_FILE = FormatString("%s Translation Cache.txt", TRANSLATION_PROVIDER);
2019-08-12 22:44:51 +08:00
QFormLayout* display;
Settings settings;
Synchronized<std::wstring> translateTo = L"en", apiKey;
namespace
{
Synchronized<std::map<std::wstring, std::wstring>> translationCache;
int savedSize;
void SaveCache()
{
std::wstring allTranslations(L"\xfeff");
for (const auto& [sentence, translation] : translationCache.Acquire().contents)
allTranslations.append(L"|SENTENCE|").append(sentence).append(L"|TRANSLATION|").append(translation).append(L"|END|\r\n");
std::ofstream(TRANSLATION_CACHE_FILE, std::ios::binary | std::ios::trunc).write((const char*)allTranslations.c_str(), allTranslations.size() * sizeof(wchar_t));
savedSize = translationCache->size();
}
}
class Window : public QDialog
2019-06-13 16:01:29 +08:00
{
public:
Window() :
QDialog(nullptr, Qt::WindowMinMaxButtonsHint)
2019-06-13 16:01:29 +08:00
{
Localize();
display = new QFormLayout(this);
2019-08-12 22:44:51 +08:00
settings.beginGroup(TRANSLATION_PROVIDER);
auto languageBox = new QComboBox(this);
languageBox->addItems(languages);
int language = -1;
if (settings.contains(LANGUAGE)) language = languageBox->findText(settings.value(LANGUAGE).toString(), Qt::MatchEndsWith);
if (language < 0) language = languageBox->findText(NATIVE_LANGUAGE, Qt::MatchStartsWith);
if (language < 0) language = languageBox->findText("English", Qt::MatchStartsWith);
languageBox->setCurrentIndex(language);
saveLanguage(languageBox->currentText());
display->addRow(TRANSLATE_TO, languageBox);
connect(languageBox, &QComboBox::currentTextChanged, this, &Window::saveLanguage);
for (auto [value, label] : Array<bool&, const char*>{
{ translateSelectedOnly, TRANSLATE_SELECTED_THREAD_ONLY },
{ rateLimitAll, RATE_LIMIT_ALL_THREADS },
{ rateLimitSelected, RATE_LIMIT_SELECTED_THREAD },
{ useCache, USE_TRANS_CACHE },
})
{
value = settings.value(label, value).toBool();
auto checkBox = new QCheckBox(this);
checkBox->setChecked(value);
display->addRow(label, checkBox);
connect(checkBox, &QCheckBox::clicked, [label, &value](bool checked) { settings.setValue(label, value = checked); });
}
for (auto [value, label] : Array<int&, const char*>{
{ tokenCount, RATE_LIMIT_TOKEN_COUNT },
{ tokenRestoreDelay, RATE_LIMIT_TOKEN_RESTORE_DELAY },
{ maxSentenceSize, MAX_SENTENCE_SIZE },
})
{
value = settings.value(label, value).toInt();
auto spinBox = new QSpinBox(this);
spinBox->setRange(0, INT_MAX);
spinBox->setValue(value);
display->addRow(label, spinBox);
connect(spinBox, qOverload<int>(&QSpinBox::valueChanged), [label, &value](int newValue) { settings.setValue(label, value = newValue); });
}
if (GET_API_KEY_FROM)
{
auto keyInput = new QLineEdit(settings.value(API_KEY).toString(), this);
apiKey->assign(S(keyInput->text()));
QObject::connect(keyInput, &QLineEdit::textChanged, [](QString key) { settings.setValue(API_KEY, S(apiKey->assign(S(key)))); });
auto keyLabel = new QLabel(QString("<a href=\"%1\">%2</a>").arg(GET_API_KEY_FROM, API_KEY), this);
keyLabel->setOpenExternalLinks(true);
display->addRow(keyLabel, keyInput);
}
setWindowTitle(TRANSLATION_PROVIDER);
QMetaObject::invokeMethod(this, &QWidget::show, Qt::QueuedConnection);
2019-06-17 07:43:59 +08:00
std::ifstream stream(TRANSLATION_CACHE_FILE, std::ios::binary);
BlockMarkupIterator savedTranslations(stream, Array<std::wstring_view>{ L"|SENTENCE|", L"|TRANSLATION|" });
auto translationCache = ::translationCache.Acquire();
while (auto read = savedTranslations.Next())
{
auto& [sentence, translation] = read.value();
translationCache->try_emplace(std::move(sentence), std::move(translation));
}
savedSize = translationCache->size();
2019-06-13 16:01:29 +08:00
}
~Window()
2019-06-13 16:01:29 +08:00
{
SaveCache();
2019-06-13 16:01:29 +08:00
}
private:
void saveLanguage(QString language)
{
settings.setValue(LANGUAGE, S(translateTo->assign(S(language.split(": ")[1]))));
2019-06-13 16:01:29 +08:00
}
} window;
2019-06-13 16:01:29 +08:00
bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
{
if (sentenceInfo["text number"] == 0 || sentence.size() > maxSentenceSize) return false;
2019-06-13 16:01:29 +08:00
static class
{
public:
bool Request()
{
auto tokens = this->tokens.Acquire();
tokens->push_back(GetTickCount());
if (tokens->size() > tokenCount * 5) tokens->erase(tokens->begin(), tokens->begin() + tokenCount * 3);
tokens->erase(std::remove_if(tokens->begin(), tokens->end(), [](DWORD token) { return GetTickCount() - token > tokenRestoreDelay; }), tokens->end());
2019-06-13 16:01:29 +08:00
return tokens->size() < tokenCount;
}
private:
Synchronized<std::vector<DWORD>> tokens;
} rateLimiter;
auto StripWhitespace = [](std::wstring& text)
{
text.erase(text.begin(), std::find_if_not(text.begin(), text.end(), iswspace));
text.erase(std::find_if_not(text.rbegin(), text.rend(), iswspace).base(), text.end());
};
2019-06-13 16:01:29 +08:00
bool cache = false;
std::wstring translation;
StripWhitespace(sentence);
if (useCache)
{
auto translationCache = ::translationCache.Acquire();
if (auto it = translationCache->find(sentence); it != translationCache->end()) translation = it->second + L"\x200b"; // dumb hack to not try to translate if stored empty translation
}
if (translation.empty() && (!translateSelectedOnly || sentenceInfo["current select"]))
2020-04-26 10:39:12 +08:00
if (rateLimiter.Request() || !rateLimitAll || (!rateLimitSelected && sentenceInfo["current select"])) std::tie(cache, translation) = Translate(sentence);
else translation = TOO_MANY_TRANS_REQUESTS;
StripWhitespace(translation);
if (cache) translationCache->try_emplace(sentence, translation);
if (cache && translationCache->size() > savedSize + 50) SaveCache();
2019-06-13 16:01:29 +08:00
2020-12-14 21:26:01 +08:00
for (int i = 0; i < translation.size(); ++i) if (translation[i] == '\r' && translation[i + 1] == '\n') translation[i] = 0x200b; // for some reason \r appears as newline - no need to double
2020-12-15 10:10:47 +08:00
if (!translation.empty()) (sentence += L"\x200b \n") += translation;
2019-06-13 16:01:29 +08:00
return true;
}
2020-12-14 21:26:01 +08:00
TEST(assert(Translate(L"こんにちは").second.find(L"ello") != std::string::npos));