10 files changed, 119 insertions, 8 deletions

diff --git a/rbutil/rbutilqt/base/rbsettings.cpp b/rbutil/rbutilqt/base/rbsettings.cpp
index 592e5c1148..29c893c66c 100644
--- a/rbutil/rbutilqt/base/rbsettings.cpp
+++ b/rbutil/rbutilqt/base/rbsettings.cpp
@@ -48,6 +48,7 @@ const static struct {
 #else
     { RbSettings::Tts,                  "tts",                  "espeak" },
 #endif
+    { RbSettings::UseTtsCorrections,    "use_tts_corrections",  "true" },
     { RbSettings::LastTalkedFolder,     "last_talked_folder",   "" },
     { RbSettings::VoiceLanguage,        "voicelanguage",        "" },
     { RbSettings::TtsLanguage,          ":tts:/language",       "" },
diff --git a/rbutil/rbutilqt/base/rbsettings.h b/rbutil/rbutilqt/base/rbsettings.h
index 277b2197d1..ab001c4916 100644
--- a/rbutil/rbutilqt/base/rbsettings.h
+++ b/rbutil/rbutilqt/base/rbsettings.h
@@ -42,6 +42,7 @@ class RbSettings : public QObject
             Platform,
             Language,
             Tts,
+            UseTtsCorrections,
             LastTalkedFolder,
             VoiceLanguage,
             TtsLanguage,
diff --git a/rbutil/rbutilqt/base/talkfile.cpp b/rbutil/rbutilqt/base/talkfile.cpp
index 3813912347..badb77f42b 100644
--- a/rbutil/rbutilqt/base/talkfile.cpp
+++ b/rbutil/rbutilqt/base/talkfile.cpp
@@ -50,6 +50,7 @@ bool TalkFileCreator::createTalkFiles()
     // generate entries
     {
         TalkGenerator generator(this);
+        // no string corrections yet: do not set language for TalkGenerator.
         connect(&generator,SIGNAL(done(bool)),this,SIGNAL(done(bool)));
         connect(&generator,SIGNAL(logItem(QString,int)),this,SIGNAL(logItem(QString,int)));
         connect(&generator,SIGNAL(logProgress(int,int)),this,SIGNAL(logProgress(int,int)));
diff --git a/rbutil/rbutilqt/base/talkgenerator.cpp b/rbutil/rbutilqt/base/talkgenerator.cpp
index 9b0cbf4066..4dffe69a42 100644
--- a/rbutil/rbutilqt/base/talkgenerator.cpp
+++ b/rbutil/rbutilqt/base/talkgenerator.cpp
@@ -25,6 +25,7 @@
 TalkGenerator::TalkGenerator(QObject* parent): QObject(parent), encFutureWatcher(this), ttsFutureWatcher(this)
 {
     m_userAborted = false;
+    m_lang = "";
 }
 
 //! \brief Creates Talkfiles.
@@ -113,6 +114,11 @@ TalkGenerator::Status TalkGenerator::voiceList(QList<TalkEntry>* list,int wavtri
         (*list)[i].refs.tts = m_tts;
         (*list)[i].refs.wavtrim = wavtrimth;
         (*list)[i].refs.generator = this;
+        // enable voice corrections only if a language is set.
+        if(!m_lang.isEmpty()) {
+            QString s = (*list)[i].toSpeak;
+            (*list)[i].toSpeak = correctString(s);
+        }
 
         // skip duplicated wav entries
         if(!duplicates.contains(list->at(i).wavfilename))
@@ -247,7 +253,7 @@ TalkGenerator::Status TalkGenerator::encodeList(QList<TalkEntry>* list)
                                                TalkGenerators.*/
     }
 
-    connect(&encFutureWatcher, SIGNAL(progressValueChanged(int)), 
+    connect(&encFutureWatcher, SIGNAL(progressValueChanged(int)),
             this, SLOT(encProgress(int)));
     encFutureWatcher.setFuture(QtConcurrent::map(*list, &TalkGenerator::encEntryPoint));
 
@@ -302,3 +308,76 @@ void TalkGenerator::abort()
     m_userAborted = true;
 }
 
+QString TalkGenerator::correctString(QString s)
+{
+    QString corrected = s;
+    int i = 0;
+    int max = m_corrections.size();
+    while(i < max) {
+        corrected = corrected.replace(QRegExp(m_corrections.at(i).search,
+                m_corrections.at(i).modifier.contains("i")
+                    ? Qt::CaseInsensitive : Qt::CaseSensitive),
+                m_corrections.at(i).replace);
+        i++;
+    }
+
+    if(corrected != s)
+        qDebug() << "[VoiceFileCreator] corrected string" << s << "to" << corrected;
+
+    return corrected;
+}
+
+
+void TalkGenerator::setLang(QString name)
+{
+    m_lang = name;
+
+    // re-initialize corrections list
+    m_corrections.clear();
+    QFile correctionsFile(":/builtin/voice-corrections.txt");
+    correctionsFile.open(QIODevice::ReadOnly);
+
+    QString engine = RbSettings::value(RbSettings::Tts).toString();
+    TTSBase* tts = TTSBase::getTTS(this,RbSettings::value(RbSettings::Tts).toString());
+    QString vendor = tts->voiceVendor();
+    delete tts;
+
+    if(m_lang.isEmpty())
+        m_lang = "english";
+    qDebug() << "[TalkGenerator] building string corrections list for"
+             << m_lang << engine << vendor;
+    QTextStream stream(&correctionsFile);
+    while(!stream.atEnd()) {
+        QString line = stream.readLine();
+        if(line.startsWith(" ") || line.length() < 10)
+            continue;
+        // separator is first character
+        QString separator = line.at(0);
+        line.remove(0, 1);
+        QStringList items = line.split(separator);
+        // we need to have at least 6 separate entries.
+        if(items.size() < 6)
+            continue;
+
+        QRegExp re_lang(items.at(0));
+        QRegExp re_engine(items.at(1));
+        QRegExp re_vendor(items.at(2));
+        if(!re_lang.exactMatch(m_lang)) {
+            continue;
+        }
+        if(!re_vendor.exactMatch(vendor)) {
+            continue;
+        }
+        if(!re_engine.exactMatch(engine)) {
+            continue;
+        }
+        struct CorrectionItems co;
+        co.search = items.at(3);
+        co.replace = items.at(4);
+        // Qt uses backslash for back references, Perl uses dollar sign.
+        co.replace.replace(QRegExp("\\$(\\d+)"), "\\\\1");
+        co.modifier = items.at(5);
+        m_corrections.append(co);
+    }
+    correctionsFile.close();
+}
diff --git a/rbutil/rbutilqt/base/talkgenerator.h b/rbutil/rbutilqt/base/talkgenerator.h
index d0cbcd0af3..8617b27238 100644
--- a/rbutil/rbutilqt/base/talkgenerator.h
+++ b/rbutil/rbutilqt/base/talkgenerator.h
@@ -67,11 +67,13 @@ public:
 
     TalkGenerator(QObject* parent);
     Status process(QList<TalkEntry>* list,int wavtrimth = -1);
+    QString correctString(QString s);
 
 public slots:
     void abort();
     void encProgress(int value);
     void ttsProgress(int value);
+    void setLang(QString name);
 
 signals:
     void done(bool);
@@ -95,6 +97,15 @@ private:
 
     bool m_ttsWarnings;
     bool m_userAborted;
+    QString m_lang;
+
+    struct CorrectionItems
+    {
+        QString search;
+        QString replace;
+        QString modifier;
+    };
+    QList<struct CorrectionItems> m_corrections;
 };
 
 
diff --git a/rbutil/rbutilqt/base/voicefile.cpp b/rbutil/rbutilqt/base/voicefile.cpp
index ce4596828f..04825f6d91 100644
--- a/rbutil/rbutilqt/base/voicefile.cpp
+++ b/rbutil/rbutilqt/base/voicefile.cpp
@@ -132,6 +132,7 @@ void VoiceFileCreator::downloadDone(bool error)
     QString id, voice;
     bool idfound = false;
     bool voicefound=false;
+    bool useCorrection = RbSettings::value(RbSettings::UseTtsCorrections).toBool();
     while (!in.atEnd())
     {
         QString line = in.readLine();
@@ -151,7 +152,8 @@ void VoiceFileCreator::downloadDone(bool error)
             TalkGenerator::TalkEntry entry;
             entry.toSpeak = voice;
             entry.wavfilename = m_path + "/" + id + ".wav";
-            entry.talkfilename = m_path + "/" + id + ".mp3";   //voicefont wants them with .mp3 extension
+            //voicefont wants them with .mp3 extension
+            entry.talkfilename = m_path + "/" + id + ".mp3";
             entry.voiced = false;
             entry.encoded = false;
             if(id == "VOICE_PAUSE")
@@ -178,6 +180,9 @@ void VoiceFileCreator::downloadDone(bool error)
     // generate files
     {
         TalkGenerator generator(this);
+        // set language for string correction. If not set no correction will be made.
+        if(useCorrection)
+            generator.setLang(m_lang);
         connect(&generator,SIGNAL(done(bool)),this,SIGNAL(done(bool)));
         connect(&generator,SIGNAL(logItem(QString,int)),this,SIGNAL(logItem(QString,int)));
         connect(&generator,SIGNAL(logProgress(int,int)),this,SIGNAL(logProgress(int,int)));
diff --git a/rbutil/rbutilqt/base/voicefile.h b/rbutil/rbutilqt/base/voicefile.h
index 72f905e4f2..94aea6c643 100644
--- a/rbutil/rbutilqt/base/voicefile.h
+++ b/rbutil/rbutilqt/base/voicefile.h
@@ -40,7 +40,7 @@ public:
     bool createVoiceFile();
 
     void setMountPoint(QString mountpoint) {m_mountpoint =mountpoint; }
-    void setLang(QString name){m_lang =name;}
+    void setLang(QString name) { m_lang = name; }
     void setWavtrimThreshold(int th){m_wavtrimThreshold = th;}
     
 public slots:
@@ -56,8 +56,9 @@ private slots:
     void downloadDone(bool error);
 
 private:
+
     void cleanup();
-    
+
     HttpGet *getter;
     QString filename;  //the temporary file
     QString m_mountpoint;  //mountpoint of the device
diff --git a/rbutil/rbutilqt/configure.cpp b/rbutil/rbutilqt/configure.cpp
index 9e1978974f..4a6fb67a43 100644
--- a/rbutil/rbutilqt/configure.cpp
+++ b/rbutil/rbutilqt/configure.cpp
@@ -205,6 +205,7 @@ void Config::accept()
     RbSettings::setValue(RbSettings::CacheOffline, ui.cacheOfflineMode->isChecked());
 
     // tts settings
+    RbSettings::setValue(RbSettings::UseTtsCorrections, ui.ttsCorrections->isChecked());
     int i = ui.comboTts->currentIndex();
     RbSettings::setValue(RbSettings::Tts, ui.comboTts->itemData(i).toString());
 
@@ -288,6 +289,9 @@ void Config::setUserSettings()
     ui.cacheDisable->setChecked(RbSettings::value(RbSettings::CacheDisabled).toBool());
     ui.cacheOfflineMode->setChecked(RbSettings::value(RbSettings::CacheOffline).toBool());
     updateCacheInfo(RbSettings::value(RbSettings::CachePath).toString());
+
+    // TTS tab
+    ui.ttsCorrections->setChecked(RbSettings::value(RbSettings::UseTtsCorrections).toBool());
 }
 
 
diff --git a/rbutil/rbutilqt/configurefrm.ui b/rbutil/rbutilqt/configurefrm.ui
index 5d2de0647b..3c0afdc15b 100644
--- a/rbutil/rbutilqt/configurefrm.ui
+++ b/rbutil/rbutilqt/configurefrm.ui
@@ -449,6 +449,13 @@
             </property>
            </widget>
           </item>
+          <item row="2" column="0" colspan="3">
+           <widget class="QCheckBox" name="ttsCorrections">
+            <property name="text">
+             <string>&amp;Use string corrections for TTS</string>
+            </property>
+           </widget>
+          </item>
          </layout>
         </widget>
        </item>
diff --git a/rbutil/rbutilqt/rbutilqt.qrc b/rbutil/rbutilqt/rbutilqt.qrc
index 4138db78bd..28ab8fa370 100644
--- a/rbutil/rbutilqt/rbutilqt.qrc
+++ b/rbutil/rbutilqt/rbutilqt.qrc
@@ -1,12 +1,13 @@
 <RCC>
-    <qresource prefix="/" >
+    <qresource prefix="/">
         <file>../../docs/CREDITS</file>
         <file>../../docs/gpl-2.0.html</file>
     </qresource>
-     <qresource>
+    <qresource>
         <file alias="builtin/VOICE_PAUSE.wav">../../tools/VOICE_PAUSE.wav</file>
+        <file alias="builtin/voice-corrections.txt">../../tools/voice-corrections.txt</file>
     </qresource>
-    <qresource prefix="/" >
+    <qresource prefix="/">
         <file>icons/audio-input-microphone.png</file>
         <file>icons/bootloader_btn.png</file>
         <file>icons/dialog-error.png</file>
@@ -36,7 +37,7 @@
         <file>icons/wizard.jpg</file>
         <file alias="icons/rockbox-clef.svg">../../docs/logo/rockbox-clef.svg</file>
     </qresource>
-    <qresource prefix="/ini" >
+    <qresource prefix="/ini">
         <file>rbutil.ini</file>
     </qresource>
 </RCC>