Added ability to automatically fix defined recognition errors.

Recognizer.cpp

@@ -8,10 +8,11 @@
 #include "Settings.h"
 #include "ImageProcessing.h"
 #include "StAssert.h"
+#include "RecognizerHelper.h"
 
 Recognizer::Recognizer (QObject *parent) :
   QObject (parent),
-  engine_ (NULL), imageScale_ (0) {
+  engine_ (NULL), recognizerHelper_ (new RecognizerHelper), imageScale_ (0) {
   applySettings ();
 }
 
@@ -19,6 +20,8 @@ void Recognizer::applySettings () {
   QSettings settings;
   settings.beginGroup (settings_names::recogntionGroup);
 
+  recognizerHelper_->load ();
+
   tessDataDir_ = settings.value (settings_names::tessDataPlace,
                                  settings_values::tessDataPlace).toString ();
   if (tessDataDir_.right (1) != "/") {
@@ -57,8 +60,8 @@ void Recognizer::recognize (ProcessingItem item) {
   bool isCustomLanguage = (!item.ocrLanguage.isEmpty () &&
                            item.ocrLanguage != ocrLanguage_);
   tesseract::TessBaseAPI *engine = (isCustomLanguage) ? NULL : engine_;
+  QString language = (isCustomLanguage) ? item.ocrLanguage : ocrLanguage_;
   if (engine == NULL) {
-    QString language = (isCustomLanguage) ? item.ocrLanguage : ocrLanguage_;
     if (!initEngine (engine, language)) {
       return;
     }
@@ -78,7 +81,7 @@ void Recognizer::recognize (ProcessingItem item) {
   }
 
   if (!result.isEmpty ()) {
-    item.recognized = result;
+    item.recognized = recognizerHelper_->substitute (result, language);
     emit recognized (item);
   }
   else {

Recognizer.h

@@ -9,6 +9,7 @@
 namespace tesseract {
   class TessBaseAPI;
 }
+class RecognizerHelper;
 
 class Recognizer : public QObject {
   Q_OBJECT
@@ -29,6 +30,7 @@ class Recognizer : public QObject {
 
   private:
     tesseract::TessBaseAPI *engine_;
+    RecognizerHelper *recognizerHelper_;
 
     QString tessDataDir_;
     QString ocrLanguage_;

RecognizerHelper.cpp

@@ -0,0 +1,78 @@
+#include <QFile>
+
+#include "RecognizerHelper.h"
+
+RecognizerHelper::RecognizerHelper ()
+  : fileName_ ("subs.csv") {
+}
+
+void RecognizerHelper::load () {
+  subs_.clear ();
+  QFile f (fileName_);
+  if (!f.open (QFile::ReadOnly)) {
+    return;
+  }
+  QByteArray data = f.readAll ();
+  f.close ();
+  QStringList lines = QString::fromUtf8 (data).split ('\n', QString::SkipEmptyParts);
+  foreach (const QString &line, lines) {
+    QStringList parts = line.mid (1, line.size () - 2).split ("\",\""); // remove "
+    if (parts.size () < 3) {
+      continue;
+    }
+    subs_.append (Sub (parts[0], parts[1], parts[2]));
+  }
+}
+
+void RecognizerHelper::save () {
+  QFile f (fileName_);
+  if (!f.open (QFile::WriteOnly)) {
+    return;
+  }
+  foreach (const Sub &sub, subs_) {
+    QStringList parts = QStringList () << sub.language << sub.source << sub.target;
+    QString line = "\"" + parts.join ("\",\"") + "\"\n";
+    f.write (line.toUtf8 ());
+  }
+  f.close ();
+}
+
+QString RecognizerHelper::substitute (const QString &source, const QString &language) const {
+  QString result = source;
+  while (true) {
+    int bestMatchIndex = -1;
+    int bestMatchLen = 0;
+    int index = -1;
+    foreach (const Sub &sub, subs_) {
+      ++index;
+      if (sub.language != language || !result.contains (sub.source)) {
+        continue;
+      }
+      int len = sub.source.length ();
+      if (len > bestMatchLen) {
+        bestMatchLen = len;
+        bestMatchIndex = index;
+      }
+    }
+    if (bestMatchIndex > -1) {
+      const Sub &sub = subs_.at (bestMatchIndex);
+      result.replace (sub.source, sub.target);
+      continue;
+    }
+    break;
+  }
+
+  return result;
+}
+
+const RecognizerHelper::Subs &RecognizerHelper::subs () const {
+  return subs_;
+}
+
+void RecognizerHelper::setSubs (const Subs &subs) {
+  subs_ = subs;
+}
+
+RecognizerHelper::Sub::Sub (const QString &language, const QString &source, const QString &target)
+  : language (language), source (source), target (target) {
+}

RecognizerHelper.h

@@ -0,0 +1,33 @@
+#ifndef RECOGNIZERHELPER_H
+#define RECOGNIZERHELPER_H
+
+#include <QString>
+
+class RecognizerHelper {
+  public:
+    struct Sub {
+      Sub (const QString &language = QString (), const QString &source = QString (),
+           const QString &target = QString ());
+      QString language;
+      QString source;
+      QString target;
+    };
+    typedef QList<Sub> Subs;
+
+  public:
+    RecognizerHelper ();
+
+    void load ();
+    void save ();
+
+    QString substitute (const QString &source, const QString& language) const;
+
+    const Subs &subs () const;
+    void setSubs (const Subs &subs);
+
+  private:
+    QString fileName_;
+    Subs subs_;
+};
+
+#endif // RECOGNIZERHELPER_H

ScreenTranslator.pro

@@ -39,7 +39,8 @@ SOURCES += main.cpp\
     LanguageHelper.cpp \
     WebTranslator.cpp \
     WebTranslatorProxy.cpp \
-    TranslatorHelper.cpp
+    TranslatorHelper.cpp \
+    RecognizerHelper.cpp
 
 HEADERS  += \
     Manager.h \
@@ -55,7 +56,8 @@ HEADERS  += \
     WebTranslator.h \
     WebTranslatorProxy.h \
     StAssert.h \
-    TranslatorHelper.h
+    TranslatorHelper.h \
+    RecognizerHelper.h
 
 FORMS    += \
     SettingsEditor.ui \

SettingsEditor.cpp

@@ -2,6 +2,8 @@
 #include "ui_SettingsEditor.h"
 #include "LanguageHelper.h"
 #include "TranslatorHelper.h"
+#include "RecognizerHelper.h"
+#include "StAssert.h"
 
 #include <QSettings>
 #include <QFileDialog>
@@ -11,7 +13,8 @@
 
 SettingsEditor::SettingsEditor (const LanguageHelper &dictionary, QWidget *parent) :
   QDialog (parent),
-  ui (new Ui::SettingsEditor), translatorHelper_ (new TranslatorHelper), dictionary_ (dictionary),
+  ui (new Ui::SettingsEditor), translatorHelper_ (new TranslatorHelper),
+  recognizerHelper_ (new RecognizerHelper), dictionary_ (dictionary),
   buttonGroup_ (new QButtonGroup (this)) {
   ui->setupUi (this);
 
@@ -22,6 +25,9 @@ SettingsEditor::SettingsEditor (const LanguageHelper &dictionary, QWidget *paren
   connect (ui->tessdataEdit, SIGNAL (textChanged (const QString &)),
            SLOT (initOcrLangCombo (const QString &)));
 
+  connect (ui->recognizerFixTable, SIGNAL (itemChanged (QTableWidgetItem *)),
+           SLOT (recognizerFixTableItemChanged (QTableWidgetItem *)));
+
   ui->translateLangCombo->addItems (dictionary_.translateLanguagesUi ());
   loadSettings ();
   loadState ();
@@ -29,6 +35,7 @@ SettingsEditor::SettingsEditor (const LanguageHelper &dictionary, QWidget *paren
 
 SettingsEditor::~SettingsEditor () {
   saveState ();
+  delete recognizerHelper_;
   delete translatorHelper_;
   delete ui;
 }
@@ -58,8 +65,29 @@ void SettingsEditor::saveSettings () const {
   QString ocrLanguageVal = dictionary_.ocrUiToCode (ui->ocrLangCombo->currentText ());
   settings.setValue (ocrLanguage, ocrLanguageVal);
   settings.setValue (imageScale, ui->imageScaleSpin->value ());
-  settings.endGroup ();
 
+  {  //Recognizer substitutions
+    RecognizerHelper::Subs subs;
+    QTableWidget *t = ui->recognizerFixTable; // Shortcut
+    for (int i = 0, end = t->rowCount () - 1; i < end; ++i) {
+      QComboBox *combo = static_cast<QComboBox *>(t->cellWidget (i, SubsColLanguage));
+      QString langUi = combo->currentText ();
+      RecognizerHelper::Sub sub;
+      sub.language = dictionary_.ocrUiToCode (langUi);
+#define GET(COL) (t->item (i, COL) ? t->item (i, COL)->text () : QString ())
+      sub.source = GET (SubsColSource);
+      sub.target = GET (SubsColTarget);
+#undef GET
+      if (langUi.isEmpty () || sub.language == langUi || sub.source.isEmpty ()) {
+        continue;
+      }
+      subs.append (sub);
+    }
+    recognizerHelper_->setSubs (subs);
+    recognizerHelper_->save ();
+  }
+
+  settings.endGroup ();
 
   settings.beginGroup (translationGroup);
   settings.setValue (doTranslation, ui->doTranslationCheck->isChecked ());
@@ -115,8 +143,28 @@ void SettingsEditor::loadSettings () {
   QString ocrLanguage = dictionary_.ocrCodeToUi (GET (ocrLanguage).toString ());
   ui->ocrLangCombo->setCurrentText (ocrLanguage);
   ui->imageScaleSpin->setValue (GET (imageScale).toInt ());
+
+  {//Recognizer substitutions
+    recognizerHelper_->load ();
+    RecognizerHelper::Subs subs = recognizerHelper_->subs ();
+    ui->recognizerFixTable->setRowCount (subs.size ());
+    int row = 0;
+    foreach (const RecognizerHelper::Sub & sub, subs) {
+      if (!initSubsTableRow (row, sub.language)) {
+        continue;
+      }
+      ui->recognizerFixTable->setItem (row, SubsColSource, new QTableWidgetItem (sub.source));
+      ui->recognizerFixTable->setItem (row, SubsColTarget, new QTableWidgetItem (sub.target));
+      ++row;
+    }
+    ui->recognizerFixTable->setRowCount (row + 1);
+    initSubsTableRow (row);
+    ui->recognizerFixTable->resizeColumnsToContents ();
+  }
+
   settings.endGroup ();
 
+
   settings.beginGroup (settings_names::translationGroup);
   ui->doTranslationCheck->setChecked (GET (doTranslation).toBool ());
   ui->translatorDebugCheck->setChecked (GET (translationDebugMode).toBool ());
@@ -138,6 +186,23 @@ void SettingsEditor::loadSettings () {
 #undef GET
 }
 
+bool SettingsEditor::initSubsTableRow (int row, const QString &languageCode) {
+  QString lang = dictionary_.ocrCodeToUi (languageCode);
+  if (!languageCode.isEmpty () && lang == languageCode) {
+    return false;
+  }
+  QComboBox *langCombo = new QComboBox (ui->recognizerFixTable);
+  langCombo->setModel (ui->ocrLangCombo->model ());
+  if (!languageCode.isEmpty ()) {
+    langCombo->setCurrentText (lang);
+  }
+  else {
+    langCombo->setCurrentIndex (ui->ocrLangCombo->currentIndex ());
+  }
+  ui->recognizerFixTable->setCellWidget (row, SubsColLanguage, langCombo);
+  return true;
+}
+
 void SettingsEditor::saveState () const {
   QSettings settings;
   settings.beginGroup (settings_names::guiGroup);
@@ -154,3 +219,21 @@ void SettingsEditor::initOcrLangCombo (const QString &path) {
   ui->ocrLangCombo->clear ();
   ui->ocrLangCombo->addItems (dictionary_.availableOcrLanguagesUi (path));
 }
+
+void SettingsEditor::recognizerFixTableItemChanged (QTableWidgetItem *item) {
+  ST_ASSERT (item->column () < 3);
+  int row = item->row ();
+  QTableWidget *t = ui->recognizerFixTable;
+#define CHECK(COL) (!t->item (row, COL) || t->item (row, COL)->text ().isEmpty ())
+  bool isRowEmpty = CHECK (SubsColSource) && CHECK (SubsColTarget);
+#undef CHECK
+  int lastRow = ui->recognizerFixTable->rowCount () - 1;
+  if (isRowEmpty && row != lastRow) {
+    ui->recognizerFixTable->removeRow (row);
+  }
+  else if (!isRowEmpty && row == lastRow) {
+    int newRow = lastRow + 1;
+    ui->recognizerFixTable->insertRow (newRow);
+    initSubsTableRow (newRow);
+  }
+}

SettingsEditor.h

@@ -5,15 +5,21 @@
 #include <QButtonGroup>
 #include <QMap>
 
+class QTableWidgetItem;
 namespace Ui {
   class SettingsEditor;
 }
 class LanguageHelper;
 class TranslatorHelper;
+class RecognizerHelper;
 
 class SettingsEditor : public QDialog {
   Q_OBJECT
 
+  enum SubsCol {
+    SubsColLanguage = 0, SubsColSource, SubsColTarget
+  };
+
   public:
     explicit SettingsEditor (const LanguageHelper &dictionary, QWidget *parent = 0);
     ~SettingsEditor ();
@@ -28,15 +34,18 @@ class SettingsEditor : public QDialog {
     void saveSettings () const;
     void openTessdataDialog ();
     void initOcrLangCombo (const QString &path);
+    void recognizerFixTableItemChanged (QTableWidgetItem *item);
 
   private:
     void loadSettings ();
     void saveState () const;
     void loadState ();
+    bool initSubsTableRow (int row, const QString &languageCode = QString ());
 
   private:
     Ui::SettingsEditor *ui;
     TranslatorHelper *translatorHelper_;
+    RecognizerHelper *recognizerHelper_;
     const LanguageHelper &dictionary_;
     QButtonGroup *buttonGroup_;
 };

SettingsEditor.ui

@@ -140,7 +140,7 @@
       <attribute name="title">
        <string>Распознавание</string>
       </attribute>
-      <layout class="QGridLayout" name="gridLayout_3">
+      <layout class="QGridLayout" name="gridLayout_2">
        <item row="0" column="0">
         <widget class="QLabel" name="label_2">
          <property name="toolTip">
@@ -196,18 +196,43 @@
        <item row="2" column="1" colspan="2">
         <widget class="QSpinBox" name="imageScaleSpin"/>
        </item>
-       <item row="3" column="0">
+       <item row="3" column="0" colspan="3">
-        <spacer name="verticalSpacer_3">
+        <widget class="QLabel" name="label_11">
-         <property name="orientation">
+         <property name="toolTip">
-          <enum>Qt::Vertical</enum>
+          <string>&lt;html&gt;&lt;head/&gt;&lt;body&gt;&lt;p&gt;Символы, регулярно распознаваемые с ошибками. При обнаружении будут заменены на указанные.&lt;/p&gt;&lt;/body&gt;&lt;/html&gt;</string>
          </property>
-         <property name="sizeHint" stdset="0">
+         <property name="text">
-          <size>
+          <string>Исправления:</string>
-           <width>20</width>
-           <height>132</height>
-          </size>
          </property>
-        </spacer>
+         <property name="alignment">
+          <set>Qt::AlignCenter</set>
+         </property>
+        </widget>
+       </item>
+       <item row="4" column="0" colspan="3">
+        <widget class="QTableWidget" name="recognizerFixTable">
+         <property name="selectionBehavior">
+          <enum>QAbstractItemView::SelectRows</enum>
+         </property>
+         <property name="sortingEnabled">
+          <bool>true</bool>
+         </property>
+         <column>
+          <property name="text">
+           <string>Язык</string>
+          </property>
+         </column>
+         <column>
+          <property name="text">
+           <string>Исходный текст</string>
+          </property>
+         </column>
+         <column>
+          <property name="text">
+           <string>Исправление</string>
+          </property>
+         </column>
+        </widget>
        </item>
       </layout>
      </widget>