NOISSUE hotloading of translations and use of local PO files

The hotloading is still inefficient

5 files changed, 650 insertions, 51 deletions

diff --git a/api/logic/CMakeLists.txt b/api/logic/CMakeLists.txt
index b8e45b30..45b01b9f 100644
--- a/api/logic/CMakeLists.txt
+++ b/api/logic/CMakeLists.txt
@@ -387,6 +387,8 @@ add_unit_test(JavaVersion
 set(TRANSLATIONS_SOURCES
     translations/TranslationsModel.h
     translations/TranslationsModel.cpp
+    translations/POTranslator.h
+    translations/POTranslator.cpp
 )
 
 set(TOOLS_SOURCES
diff --git a/api/logic/translations/POTranslator.cpp b/api/logic/translations/POTranslator.cpp
new file mode 100644
index 00000000..1ffcb9a4
--- /dev/null
+++ b/api/logic/translations/POTranslator.cpp
@@ -0,0 +1,373 @@
+#include "POTranslator.h"
+
+#include <QDebug>
+#include "FileSystem.h"
+
+struct POEntry
+{
+    QString text;
+    bool fuzzy;
+};
+
+struct POTranslatorPrivate
+{
+    QString filename;
+    QHash<QByteArray, POEntry> mapping;
+    QHash<QByteArray, POEntry> mapping_disambiguatrion;
+    bool loaded = false;
+
+    void reload();
+};
+
+class ParserArray : public QByteArray
+{
+public:
+    ParserArray(const QByteArray &in) : QByteArray(in)
+    {
+    }
+    bool chomp(const char * data, int length)
+    {
+        if(startsWith(data))
+        {
+            remove(0, length);
+            return true;
+        }
+        return false;
+    }
+    bool chompString(QByteArray & appendHere)
+    {
+        QByteArray msg;
+        bool escape = false;
+        if(size() < 2)
+        {
+            qDebug() << "String fragment is too short";
+            return false;
+        }
+        if(!startsWith('"'))
+        {
+            qDebug() << "String fragment does not start with \"";
+            return false;
+        }
+        if(!endsWith('"'))
+        {
+            qDebug() << "String fragment does not end with \", instead, there is" << at(size() - 1);
+            return false;
+        }
+        for(int i = 1; i < size() - 1; i++)
+        {
+            char c = operator[](i);
+            if(escape)
+            {
+                switch(c)
+                {
+                    case 'r':
+                        msg += '\r';
+                        break;
+                    case 'n':
+                        msg += '\n';
+                        break;
+                    case 't':
+                        msg += '\t';
+                        break;
+                    case 'v':
+                        msg += '\v';
+                        break;
+                    case 'a':
+                        msg += '\a';
+                        break;
+                    case 'b':
+                        msg += '\b';
+                        break;
+                    case 'f':
+                        msg += '\f';
+                        break;
+                    case '"':
+                        msg += '"';
+                        break;
+                    case '\\':
+                        msg.append('\\');
+                        break;
+                    case '0':
+                    case '1':
+                    case '2':
+                    case '3':
+                    case '4':
+                    case '5':
+                    case '6':
+                    case '7':
+                    {
+                        int octal_start = i;
+                        while ((c = operator[](i)) >= '0' && c <= '7')
+                        {
+                            i++;
+                            if (i == length() - 1)
+                            {
+                                qDebug() << "Something went bad while parsing an octal escape string...";
+                                return false;
+                            }
+                        }
+                        msg += mid(octal_start, i - octal_start).toUInt(0, 8);
+                        break;
+                    }
+                    case 'x':
+                    {
+                        // chomp the 'x'
+                        i++;
+                        int hex_start = i;
+                        while (isxdigit(operator[](i)))
+                        {
+                            i++;
+                            if (i == length() - 1)
+                            {
+                                qDebug() << "Something went bad while parsing a hex escape string...";
+                                return false;
+                            }
+                        }
+                        msg += mid(hex_start, i - hex_start).toUInt(0, 16);
+                        break;
+                    }
+                    default:
+                    {
+                        qDebug() << "Invalid escape sequence character:" << c;
+                        return false;
+                    }
+                }
+                escape = false;
+            }
+            else if(c == '\\')
+            {
+                escape = true;
+            }
+            else
+            {
+                msg += c;
+            }
+        }
+        if(escape)
+        {
+            qDebug() << "Unterminated escape sequence...";
+            return false;
+        }
+        appendHere += msg;
+        return true;
+    }
+};
+
+void POTranslatorPrivate::reload()
+{
+    QFile file(filename);
+    if(!file.open(QFile::OpenMode::enum_type::ReadOnly | QFile::OpenMode::enum_type::Text))
+    {
+        qDebug() << "Failed to open PO file:" << filename;
+        return;
+    }
+
+    QByteArray context;
+    QByteArray disambiguation;
+    QByteArray id;
+    QByteArray str;
+    bool fuzzy = false;
+    bool nextFuzzy = false;
+
+    enum class Mode
+    {
+        First,
+        MessageContext,
+        MessageId,
+        MessageString
+    } mode = Mode::First;
+
+    int lineNumber = 0;
+    QHash<QByteArray, POEntry> newMapping;
+    QHash<QByteArray, POEntry> newMapping_disambiguation;
+    auto endEntry = [&]() {
+        auto strStr = QString::fromUtf8(str);
+        // NOTE: PO header has empty id. We skip it.
+        if(!id.isEmpty())
+        {
+            auto normalKey = context + "|" + id;
+            newMapping.insert(normalKey, {strStr, fuzzy});
+            if(!disambiguation.isEmpty())
+            {
+                auto disambiguationKey = context + "|" + id + "@" + disambiguation;
+                newMapping_disambiguation.insert(disambiguationKey, {strStr, fuzzy});
+            }
+        }
+        context.clear();
+        disambiguation.clear();
+        id.clear();
+        str.clear();
+        fuzzy = nextFuzzy;
+        nextFuzzy = false;
+    };
+    while (!file.atEnd())
+    {
+        ParserArray line = file.readLine();
+        if(line.endsWith('\n'))
+        {
+            line.resize(line.size() - 1);
+        }
+        if(line.endsWith('\r'))
+        {
+            line.resize(line.size() - 1);
+        }
+
+        if(!line.size())
+        {
+            // NIL
+        }
+        else if(line[0] == '#')
+        {
+            if(line.contains(", fuzzy"))
+            {
+                nextFuzzy = true;
+            }
+        }
+        else if(line.startsWith('"'))
+        {
+            QByteArray temp;
+            QByteArray *out = &temp;
+
+            switch(mode)
+            {
+                case Mode::First:
+                    qDebug() << "Unexpected escaped string during initial state... line:" << lineNumber;
+                    return;
+                case Mode::MessageString:
+                    out = &str;
+                    break;
+                case Mode::MessageContext:
+                    out = &context;
+                    break;
+                case Mode::MessageId:
+                    out = &id;
+                    break;
+            }
+            if(!line.chompString(*out))
+            {
+                qDebug() << "Badly formatted string on line:" << lineNumber;
+                return;
+            }
+        }
+        else if(line.chomp("msgctxt ", 8))
+        {
+            switch(mode)
+            {
+                case Mode::First:
+                    break;
+                case Mode::MessageString:
+                    endEntry();
+                    break;
+                case Mode::MessageContext:
+                case Mode::MessageId:
+                    qDebug() << "Unexpected msgctxt line:" << lineNumber;
+                    return;
+            }
+            if(line.chompString(context))
+            {
+                auto parts = context.split('|');
+                context = parts[0];
+                if(parts.size() > 1 && !parts[1].isEmpty())
+                {
+                    disambiguation = parts[1];
+                }
+                mode = Mode::MessageContext;
+            }
+        }
+        else if (line.chomp("msgid ", 6))
+        {
+            switch(mode)
+            {
+                case Mode::MessageContext:
+                case Mode::First:
+                    break;
+                case Mode::MessageString:
+                    endEntry();
+                    break;
+                case Mode::MessageId:
+                    qDebug() << "Unexpected msgid line:" << lineNumber;
+                    return;
+            }
+            if(line.chompString(id))
+            {
+                mode = Mode::MessageId;
+            }
+        }
+        else if (line.chomp("msgstr ", 7))
+        {
+            switch(mode)
+            {
+                case Mode::First:
+                case Mode::MessageString:
+                case Mode::MessageContext:
+                    qDebug() << "Unexpected msgstr line:" << lineNumber;
+                    return;
+                case Mode::MessageId:
+                    break;
+            }
+            if(line.chompString(str))
+            {
+                mode = Mode::MessageString;
+            }
+        }
+        else
+        {
+            qDebug() << "I did not understand line: " << lineNumber << ":" << QString::fromUtf8(line);
+        }
+        lineNumber++;
+    }
+    endEntry();
+    mapping = std::move(newMapping);
+    mapping_disambiguatrion = std::move(newMapping_disambiguation);
+    loaded = true;
+}
+
+POTranslator::POTranslator(const QString& filename, QObject* parent) : QTranslator(parent)
+{
+    d = new POTranslatorPrivate;
+    d->filename = filename;
+    d->reload();
+}
+
+QString POTranslator::translate(const char* context, const char* sourceText, const char* disambiguation, int n) const
+{
+    if(disambiguation)
+    {
+        auto disambiguationKey = QByteArray(context) + "|" + QByteArray(sourceText) + "@" + QByteArray(disambiguation);
+        auto iter = d->mapping_disambiguatrion.find(disambiguationKey);
+        if(iter != d->mapping_disambiguatrion.end())
+        {
+            auto & entry = *iter;
+            if(entry.text.isEmpty())
+            {
+                qDebug() << "Translation entry has no content:" << disambiguationKey;
+            }
+            if(entry.fuzzy)
+            {
+                qDebug() << "Translation entry is fuzzy:" << disambiguationKey << "->" << entry.text;
+            }
+            return entry.text;
+        }
+    }
+    auto key = QByteArray(context) + "|" + QByteArray(sourceText);
+    auto iter = d->mapping.find(key);
+    if(iter != d->mapping.end())
+    {
+        auto & entry = *iter;
+        if(entry.text.isEmpty())
+        {
+            qDebug() << "Translation entry has no content:" << key;
+        }
+        if(entry.fuzzy)
+        {
+            qDebug() << "Translation entry is fuzzy:" << key << "->" << entry.text;
+        }
+        return entry.text;
+    }
+    return QString();
+}
+
+bool POTranslator::isEmpty() const
+{
+    return !d->loaded;
+}
diff --git a/api/logic/translations/POTranslator.h b/api/logic/translations/POTranslator.h
new file mode 100644
index 00000000..6d518560
--- /dev/null
+++ b/api/logic/translations/POTranslator.h
@@ -0,0 +1,16 @@
+#pragma once
+
+#include <QTranslator>
+
+struct POTranslatorPrivate;
+
+class POTranslator : public QTranslator
+{
+    Q_OBJECT
+public:
+    explicit POTranslator(const QString& filename, QObject * parent = nullptr);
+    QString translate(const char * context, const char * sourceText, const char * disambiguation, int n) const override;
+    bool isEmpty() const override;
+private:
+    POTranslatorPrivate * d;
+};
diff --git a/api/logic/translations/TranslationsModel.cpp b/api/logic/translations/TranslationsModel.cpp
index c32483f0..c9c30212 100644
--- a/api/logic/translations/TranslationsModel.cpp
+++ b/api/logic/translations/TranslationsModel.cpp
@@ -8,16 +8,66 @@
 #include <QDebug>
 #include <FileSystem.h>
 #include <net/NetJob.h>
+#include <net/ChecksumValidator.h>
 #include <Env.h>
 #include <net/URLConstants.h>
+#include "Json.h"
+
+#include "POTranslator.h"
 
 const static QLatin1Literal defaultLangCode("en");
 
+enum class FileType
+{
+    NONE,
+    QM,
+    PO
+};
+
 struct Language
 {
+    Language()
+    {
+        updated = true;
+    }
+    Language(const QString & _key)
+    {
+        key = _key;
+        locale = QLocale(key);
+        updated = (key == defaultLangCode);
+    }
+
+    float percentTranslated() const
+    {
+        if (total == 0)
+        {
+            return 100.0f;
+        }
+        return float(translated) / float(total);
+    }
+
+    void setTranslationStats(unsigned _translated, unsigned _untranslated, unsigned _fuzzy)
+    {
+        translated = _translated;
+        untranslated = _untranslated;
+        fuzzy = _fuzzy;
+        total = translated + untranslated + fuzzy;
+    }
+
     QString key;
     QLocale locale;
     bool updated;
+
+    QString file_name = QString();
+    std::size_t file_size = 0;
+    QString file_sha1 = QString();
+
+    unsigned translated = 0;
+    unsigned untranslated = 0;
+    unsigned fuzzy = 0;
+    unsigned total = 0;
+
+    FileType localFileType = FileType::NONE;
 };
 
 struct TranslationsModel::Private
@@ -25,7 +75,9 @@ struct TranslationsModel::Private
     QDir m_dir;
 
     // initial state is just english
-    QVector<Language> m_languages = {{defaultLangCode, QLocale(defaultLangCode), false}};
+    QVector<Language> m_languages = {Language (defaultLangCode)};
+    QMap<QString, int> m_languageLookup = {{defaultLangCode, 0}};
+
     QString m_selectedLanguage = defaultLangCode;
     std::unique_ptr<QTranslator> m_qt_translator;
     std::unique_ptr<QTranslator> m_app_translator;
@@ -35,19 +87,168 @@ struct TranslationsModel::Private
     NetJobPtr m_dl_job;
     NetJobPtr m_index_job;
     QString m_nextDownload;
+
+    std::unique_ptr<POTranslator> m_po_translator;
+    QFileSystemWatcher *watcher;
 };
 
 TranslationsModel::TranslationsModel(QString path, QObject* parent): QAbstractListModel(parent)
 {
     d.reset(new Private);
     d->m_dir.setPath(path);
-    loadLocalIndex();
+    reloadLocalFiles();
+
+    d->watcher = new QFileSystemWatcher(this);
+    connect(d->watcher, &QFileSystemWatcher::directoryChanged, this, &TranslationsModel::translationDirChanged);
+    d->watcher->addPath(d->m_dir.canonicalPath());
 }
 
 TranslationsModel::~TranslationsModel()
 {
 }
 
+void TranslationsModel::translationDirChanged(const QString& path)
+{
+    qDebug() << "Dir changed:" << path;
+    reloadLocalFiles();
+    selectLanguage(selectedLanguage());
+}
+
+void TranslationsModel::indexRecieved()
+{
+    qDebug() << "Got translations index!";
+    d->m_index_job.reset();
+    if(d->m_selectedLanguage != defaultLangCode)
+    {
+        downloadTranslation(d->m_selectedLanguage);
+    }
+}
+
+namespace {
+void readIndex(const QString & path, QVector<Language>& languages, QMap<QString, int>& languagesLookup)
+{
+    QByteArray data;
+    try
+    {
+        data = FS::read(path);
+    }
+    catch (const Exception &e)
+    {
+        qCritical() << "Translations Download Failed: index file not readable";
+        return;
+    }
+
+    int index = 1;
+    try
+    {
+        auto doc = Json::requireObject(Json::requireDocument(data));
+        auto file_type = Json::requireString(doc, "file_type");
+        if(file_type != "MMC-TRANSLATION-INDEX")
+        {
+            qCritical() << "Translations Download Failed: index file is of unknown file type" << file_type;
+            return;
+        }
+        auto version = Json::requireInteger(doc, "version");
+        if(version > 2)
+        {
+            qCritical() << "Translations Download Failed: index file is of unknown format version" << file_type;
+            return;
+        }
+        auto langObjs = Json::requireObject(doc, "languages");
+        for(auto iter = langObjs.begin(); iter != langObjs.end(); iter++)
+        {
+            Language lang(iter.key());
+
+            auto langObj =  Json::requireObject(iter.value());
+            lang.setTranslationStats(
+                Json::ensureInteger(langObj, "translated", 0),
+                Json::ensureInteger(langObj, "untranslated", 0),
+                Json::ensureInteger(langObj, "fuzzy", 0)
+            );
+            lang.file_name = Json::requireString(langObj, "file");
+            lang.file_sha1 = Json::requireString(langObj, "sha1");
+            lang.file_size = Json::requireInteger(langObj, "size");
+
+            languages.append(std::move(lang));
+            languagesLookup[iter.key()] = index;
+            index++;
+        }
+    }
+    catch (Json::JsonException & e)
+    {
+        qCritical() << "Translations Download Failed: index file could not be parsed as json";
+    }
+}
+}
+
+void TranslationsModel::reloadLocalFiles()
+{
+    QVector<Language> languages = {Language (defaultLangCode)};
+    QMap<QString, int> languageLookup = {{defaultLangCode, 0}};
+
+    readIndex(d->m_dir.absoluteFilePath("index_v2.json"), languages, languageLookup);
+    auto fileTypeToString = [](FileType ft) -> QString
+    {
+        switch(ft)
+        {
+            case FileType::NONE:
+                return QString();
+            case FileType::QM:
+                return "QM";
+            case FileType::PO:
+                return "PO";
+        }
+        return QString();
+    };
+    auto entries = d->m_dir.entryInfoList({"mmc_*.qm", "*.po"}, QDir::Files | QDir::NoDotAndDotDot);
+    for(auto & entry: entries)
+    {
+        auto completeSuffix = entry.completeSuffix();
+        QString langCode;
+        FileType fileType = FileType::NONE;
+        if(completeSuffix == "qm")
+        {
+            langCode = entry.baseName().remove(0,4);
+            fileType = FileType::QM;
+        }
+        else if(completeSuffix == "po")
+        {
+            langCode = entry.baseName();
+            fileType = FileType::PO;
+        }
+        else
+        {
+            continue;
+        }
+
+        auto langIter = languageLookup.find(langCode);
+        if(langIter != languageLookup.end())
+        {
+            auto & language = languages[*langIter];
+            if(int(fileType) > int(language.localFileType))
+            {
+                qDebug() << "Found" << fileTypeToString(fileType) << "local file for language" << langCode;
+                language.localFileType = fileType;
+            }
+        }
+        else
+        {
+            if(fileType == FileType::PO)
+            {
+                Language localFound(langCode);
+                localFound.localFileType = FileType::PO;
+                languages.append(localFound);
+                qDebug() << "Found standalone translation PO file: " << langCode;
+            }
+        }
+
+    }
+
+    beginResetModel();
+    d->m_languages.swap(languages);
+    endResetModel();
+}
+
 QVariant TranslationsModel::data(const QModelIndex& index, int role) const
 {
     if (!index.isValid())
@@ -153,18 +354,48 @@ bool TranslationsModel::selectLanguage(QString key)
         d->m_qt_translator.reset();
     }
 
-    d->m_app_translator.reset(new QTranslator());
-    if (d->m_app_translator->load("mmc_" + langCode, d->m_dir.path()))
+    if(langPtr->localFileType == FileType::PO)
     {
         qDebug() << "Loading Application Language File for" << langCode.toLocal8Bit().constData() << "...";
-        if (!QCoreApplication::installTranslator(d->m_app_translator.get()))
+        auto poTranslator = new POTranslator(FS::PathCombine(d->m_dir.path(), langCode + ".po"));
+        if(!poTranslator->isEmpty())
+        {
+            if (!QCoreApplication::installTranslator(poTranslator))
+            {
+                delete poTranslator;
+                qCritical() << "Installing Application Language File failed.";
+            }
+            else
+            {
+                d->m_app_translator.reset(poTranslator);
+                successful = true;
+            }
+        }
+        else
         {
             qCritical() << "Loading Application Language File failed.";
             d->m_app_translator.reset();
         }
+    }
+    else if(langPtr->localFileType == FileType::QM)
+    {
+        d->m_app_translator.reset(new QTranslator());
+        if (d->m_app_translator->load("mmc_" + langCode, d->m_dir.path()))
+        {
+            qDebug() << "Loading Application Language File for" << langCode.toLocal8Bit().constData() << "...";
+            if (!QCoreApplication::installTranslator(d->m_app_translator.get()))
+            {
+                qCritical() << "Installing Application Language File failed.";
+                d->m_app_translator.reset();
+            }
+            else
+            {
+                successful = true;
+            }
+        }
         else
         {
-            successful = true;
+            d->m_app_translator.reset();
         }
     }
     else
@@ -199,55 +430,15 @@ void TranslationsModel::downloadIndex()
     }
     qDebug() << "Downloading Translations Index...";
     d->m_index_job.reset(new NetJob("Translations Index"));
-    MetaEntryPtr entry = ENV.metacache()->resolveEntry("translations", "index");
-    d->m_index_task = Net::Download::makeCached(QUrl("https://files.multimc.org/translations/index"), entry);
+    MetaEntryPtr entry = ENV.metacache()->resolveEntry("translations", "index_v2.json");
+    entry->setStale(true);
+    d->m_index_task = Net::Download::makeCached(QUrl("https://files.multimc.org/translations/index_v2.json"), entry);
     d->m_index_job->addNetAction(d->m_index_task);
     connect(d->m_index_job.get(), &NetJob::failed, this, &TranslationsModel::indexFailed);
     connect(d->m_index_job.get(), &NetJob::succeeded, this, &TranslationsModel::indexRecieved);
     d->m_index_job->start();
 }
 
-void TranslationsModel::indexRecieved()
-{
-    qDebug() << "Got translations index!";
-    d->m_index_job.reset();
-    loadLocalIndex();
-    if(d->m_selectedLanguage != defaultLangCode)
-    {
-        downloadTranslation(d->m_selectedLanguage);
-    }
-}
-
-void TranslationsModel::loadLocalIndex()
-{
-    QByteArray data;
-    try
-    {
-        data = FS::read(d->m_dir.absoluteFilePath("index"));
-    }
-    catch (const Exception &e)
-    {
-        qCritical() << "Translations Download Failed: index file not readable";
-        return;
-    }
-    QVector<Language> languages;
-    QList<QByteArray> lines = data.split('\n');
-    // add the default english.
-    languages.append({defaultLangCode, QLocale(defaultLangCode), true});
-    for (const auto line : lines)
-    {
-        if(!line.isEmpty())
-        {
-            auto str = QString::fromLatin1(line);
-            str.remove(".qm");
-            languages.append({str, QLocale(str), false});
-        }
-    }
-    beginResetModel();
-    d->m_languages.swap(languages);
-    endResetModel();
-}
-
 void TranslationsModel::updateLanguage(QString key)
 {
     if(key == defaultLangCode)
@@ -274,13 +465,28 @@ void TranslationsModel::downloadTranslation(QString key)
         d->m_nextDownload = key;
         return;
     }
+    auto lang = findLanguage(key);
+    if(!lang)
+    {
+        qWarning() << "Will not download an unknown translation" << key;
+        return;
+    }
+
     d->m_downloadingTranslation = key;
     MetaEntryPtr entry = ENV.metacache()->resolveEntry("translations", "mmc_" + key + ".qm");
     entry->setStale(true);
+
+    auto dl = Net::Download::makeCached(QUrl(URLConstants::TRANSLATIONS_BASE_URL + lang->file_name), entry);
+    auto rawHash = QByteArray::fromHex(lang->file_sha1.toLatin1());
+    dl->addValidator(new Net::ChecksumValidator(QCryptographicHash::Sha1, rawHash));
+    dl->m_total_progress = lang->file_size;
+
     d->m_dl_job.reset(new NetJob("Translation for " + key));
-    d->m_dl_job->addNetAction(Net::Download::makeCached(QUrl(URLConstants::TRANSLATIONS_BASE_URL + key + ".qm"), entry));
+    d->m_dl_job->addNetAction(dl);
+
     connect(d->m_dl_job.get(), &NetJob::succeeded, this, &TranslationsModel::dlGood);
     connect(d->m_dl_job.get(), &NetJob::failed, this, &TranslationsModel::dlFailed);
+
     d->m_dl_job->start();
 }
 
diff --git a/api/logic/translations/TranslationsModel.h b/api/logic/translations/TranslationsModel.h
index 8a9298d4..5bf769c3 100644
--- a/api/logic/translations/TranslationsModel.h
+++ b/api/logic/translations/TranslationsModel.h
@@ -40,7 +40,7 @@ public:
 
 private:
     Language *findLanguage(const QString & key);
-    void loadLocalIndex();
+    void reloadLocalFiles();
     void downloadTranslation(QString key);
     void downloadNext();
 
@@ -54,6 +54,8 @@ private slots:
     void indexFailed(QString reason);
     void dlFailed(QString reason);
     void dlGood();
+    void translationDirChanged(const QString &path);
+
 
 private: /* data */
     struct Private;