implemented ocr tool using tesseract

.github/workflows/Linux-pack.yml

@@ -144,6 +144,8 @@ jobs:
             libqt5widgets5 \
             libqt5gui5 \
             libqt5svg5-dev \
+            tesseract-ocr \
+            libtesseract-dev \
             python3 \
             python3-pip
       - name: Prepare cmake(>=3.13.0)

.github/workflows/build_cmake.yml

@@ -50,7 +50,10 @@ jobs:
             libqt5core5a \
             libqt5widgets5 \
             libqt5gui5 \
-            libqt5svg5-dev
+            libqt5svg5-dev \
+            tesseract-ocr \
+            libtesseract-dev
+
 
       - name: Create Build Environment
         # Some projects don't allow in-source building, so create a separate build directory

data/graphics.qrc

@@ -37,6 +37,7 @@
         <file>img/material/black/circle-outline.svg</file>
         <file>img/material/black/pixelate.svg</file>
         <file>img/material/black/arrow-bottom-left.svg</file>
+        <file>img/material/black/ocr.svg</file>
         <file>img/material/white/undo-variant.svg</file>
         <file>img/material/white/text.svg</file>
         <file>img/material/white/square.svg</file>
@@ -74,5 +75,6 @@
         <file>img/material/white/shortcut.svg</file>
         <file>img/material/black/filepath.svg</file>
         <file>img/material/white/filepath.svg</file>
+        <file>img/material/white/ocr.svg</file>
     </qresource>
 </RCC>

src/CMakeLists.txt

@@ -159,6 +159,8 @@ target_link_libraries(
         Qt5::Widgets
         SingleApplication::SingleApplication
         spdlog::spdlog
+        tesseract
+        leptonica
 )
 
 if (APPLE)

src/tools/CMakeLists.txt

@@ -35,6 +35,8 @@ target_sources(
 target_sources(flameshot PRIVATE rectangle/rectangletool.h rectangle/rectangletool.cpp)
 target_sources(flameshot PRIVATE redo/redotool.h redo/redotool.cpp)
 target_sources(flameshot PRIVATE save/savetool.h save/savetool.cpp)
+target_sources(flameshot PRIVATE ocr/ocrtool.h ocr/ocrtool.cpp)
+target_sources(flameshot PRIVATE ocr/tesseract_tool.h)
 target_sources(flameshot PRIVATE selection/selectiontool.h selection/selectiontool.cpp)
 target_sources(flameshot PRIVATE sizeindicator/sizeindicatortool.h sizeindicator/sizeindicatortool.cpp)
 target_sources(

src/tools/capturetool.h

@@ -45,7 +45,8 @@ enum class ToolType
     SIZEINDICATOR,
     TEXT,
     UNDO,
-    UPLOAD
+    UPLOAD,
+    OCR
 };
 
 class CaptureTool : public QObject

src/tools/ocr/ocrtool.cpp

@@ -0,0 +1,84 @@
+// Copyright(c) 2017-2019 Alejandro Sirgo Rica & Contributors
+//
+// This file is part of Flameshot.
+//
+//     Flameshot is free software: you can redistribute it and/or modify
+//     it under the terms of the GNU General Public License as published by
+//     the Free Software Foundation, either version 3 of the License, or
+//     (at your option) any later version.
+//
+//     Flameshot is distributed in the hope that it will be useful,
+//     but WITHOUT ANY WARRANTY; without even the implied warranty of
+//     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+//     GNU General Public License for more details.
+//
+//     You should have received a copy of the GNU General Public License
+//     along with Flameshot.  If not, see <http://www.gnu.org/licenses/>.
+
+#include "ocrtool.h"
+#include "tesseract_tool.h"
+#include "src/utils/screenshotsaver.h"
+#include <QPainter>
+#include <tesseract/basetesseractApi.h>
+#include <leptonica/allheaders.h>
+
+OcrTool::OcrTool(QObject* parent)
+  : AbstractActionTool(parent)
+{}
+
+bool OcrTool::closeOnButtonPressed() const
+{
+    return true;
+}
+
+QIcon OcrTool::icon(const QColor& background, bool inEditor) const
+{
+    Q_UNUSED(inEditor);
+    return QIcon(iconPath(background) + "ocr.svg");
+}
+QString OcrTool::name() const
+{
+    return "OCR";
+}
+
+ToolType OcrTool::nameID() const
+{
+    return ToolType::OCR;
+}
+
+QString OcrTool::description() const
+{
+    return "Copy text in Capture to Clipboard";
+}
+
+CaptureTool* OcrTool::copy(QObject* parent)
+{
+    return new OcrTool(parent);
+}
+
+void OcrTool::pressed(const CaptureContext& context)
+{
+    char *outText;
+
+    tesseract::TessBaseAPI *tesseractApi = new tesseract::TessBaseAPI();
+
+    // TODO: tesseract language configs?
+    if (tesseractApi->Init(NULL, "eng")) {
+        // TODO: error system notification?
+        return;
+    }
+
+    Pix *image = TesseractTool::qImage2PIX(context.selectedScreenshotArea().toImage());
+    tesseractApi->SetImage(image);
+
+    outText = tesseractApi->GetUTF8Text();
+    printf("OCR output:\n%s", outText);
+
+    const QString qString = outText;
+
+    ScreenshotSaver().saveToClipboard(qString);
+
+    tesseractApi->End();
+    delete tesseractApi;
+    delete [] outText;
+}

src/tools/ocr/ocrtool.h

@@ -0,0 +1,41 @@
+// Copyright(c) 2017-2019 Alejandro Sirgo Rica & Contributors
+//
+// This file is part of Flameshot.
+//
+//     Flameshot is free software: you can redistribute it and/or modify
+//     it under the terms of the GNU General Public License as published by
+//     the Free Software Foundation, either version 3 of the License, or
+//     (at your option) any later version.
+//
+//     Flameshot is distributed in the hope that it will be useful,
+//     but WITHOUT ANY WARRANTY; without even the implied warranty of
+//     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+//     GNU General Public License for more details.
+//
+//     You should have received a copy of the GNU General Public License
+//     along with Flameshot.  If not, see <http://www.gnu.org/licenses/>.
+
+#pragma once
+
+#include "src/tools/abstractactiontool.h"
+
+class OcrTool : public AbstractActionTool
+{
+    Q_OBJECT
+public:
+    explicit OcrTool(QObject* parent = nullptr);
+
+    bool closeOnButtonPressed() const;
+
+    QIcon icon(const QColor& background, bool inEditor) const override;
+    QString name() const override;
+    QString description() const override;
+
+    CaptureTool* copy(QObject* parent = nullptr) override;
+
+protected:
+    ToolType nameID() const override;
+
+public slots:
+    void pressed(const CaptureContext& context) override;
+};

src/tools/ocr/tesseract_tool.h

@@ -0,0 +1,43 @@
+#pragma once
+
+#include <QPainter>
+#include <leptonica/allheaders.h>
+#include <tesseract/baseapi.h>
+
+// Thanks to Stackoverflow user "user898678": https://stackoverflow.com/a/10019508
+// https://github.com/zdenop/qt-box-editor/blob/master/src/TessTools.cpp
+// LICENSE: https://github.com/zdenop/qt-box-editor/blob/master/LICENSE
+// modified version
+
+class TesseractTool
+{
+public:
+    static PIX* qImageToPIX(const QImage& qImage)
+    {
+        PIX* pixs;
+        l_uint32* lines;
+
+        QImage qImageCopy = qImage.copy();
+
+        qImageCopy = qImageCopy.rgbSwapped();
+        int width = qImageCopy.width();
+        int height = qImageCopy.height();
+        int depth = qImageCopy.depth();
+        int wpl = qImageCopy.bytesPerLine() / 4;
+
+        pixs = pixCreate(width, height, depth);
+        pixSetWpl(pixs, wpl);
+        pixSetColormap(pixs, NULL);
+        l_uint32* datas = pixs->data;
+
+        for (int y = 0; y < height; y++) {
+            lines = datas + y * wpl;
+            QByteArray a((const char*)qImageCopy.scanLine(y),
+                         qImageCopy.bytesPerLine());
+            for (int j = 0; j < a.size(); j++) {
+                *((l_uint8*)lines + j) = a[j];
+            }
+        }
+        return pixEndianByteSwapNew(pixs);
+    }
+};

src/tools/toolfactory.cpp

@@ -32,6 +32,7 @@
 #include "rectangle/rectangletool.h"
 #include "redo/redotool.h"
 #include "save/savetool.h"
+#include "ocr/ocrtool.h"
 #include "selection/selectiontool.h"
 #include "sizeindicator/sizeindicatortool.h"
 #include "src/utils/confighandler.h"
@@ -110,6 +111,9 @@ CaptureTool* ToolFactory::CreateTool(CaptureToolButton::ButtonType t,
         case CaptureToolButton::TYPE_CIRCLECOUNT:
             tool = new CircleCountTool(parent);
             break;
+        case CaptureToolButton::TYPE_OCR:
+            tool = new OcrTool(parent);
+            break;
         default:
             tool = nullptr;
             break;

src/utils/confighandler.cpp

@@ -65,7 +65,8 @@ QVector<CaptureToolButton::ButtonType> ConfigHandler::getButtons()
                 << CaptureToolButton::TYPE_OPEN_APP
 #endif
                 << CaptureToolButton::TYPE_PIN << CaptureToolButton::TYPE_TEXT
-                << CaptureToolButton::TYPE_CIRCLECOUNT;
+                << CaptureToolButton::TYPE_CIRCLECOUNT
+                << CaptureToolButton::TYPE_OCR;
     }
 
     using bt = CaptureToolButton::ButtonType;

src/utils/configshortcuts.cpp

@@ -140,6 +140,9 @@ const QKeySequence& ConfigShortcuts::captureShortcutDefault(
         case CaptureToolButton::ButtonType::TYPE_TEXT:
             m_ks = QKeySequence(Qt::Key_T);
             break;
+        case CaptureToolButton::ButtonType::TYPE_OCR:
+            m_ks = QKeySequence(Qt::CTRL + Qt::Key_T);
+            break;
         default:
             break;
     }

src/utils/screenshotsaver.cpp

@@ -56,6 +56,13 @@ void ScreenshotSaver::saveToClipboard(const QPixmap& capture)
     }
 }
 
+void ScreenshotSaver::saveToClipboard(const QString& text)
+{
+    SystemNotification().sendMessage(
+          QObject::tr("Text saved to clipboard"));
+    QApplication::clipboard()->setText(text);
+}
+
 bool ScreenshotSaver::saveToFilesystem(const QPixmap& capture,
                                        const QString& path,
                                        const QString& messagePrefix)

src/utils/screenshotsaver.h

@@ -27,6 +27,7 @@ class ScreenshotSaver
     ScreenshotSaver(const unsigned id);
 
     void saveToClipboard(const QPixmap& capture);
+    void saveToClipboard(const QString& text);
     bool saveToFilesystem(const QPixmap& capture,
                           const QString& path,
                           const QString& messagePrefix);

src/widgets/capture/capturetoolbutton.cpp

@@ -138,8 +138,10 @@ static std::map<CaptureToolButton::ButtonType, int> buttonTypeOrder
         defined(Q_OS_MACX))
       { CaptureToolButton::TYPE_OPEN_APP, 17 },
       { CaptureToolButton::TYPE_EXIT, 18 }, { CaptureToolButton::TYPE_PIN, 19 },
+      { CaptureToolButton::TYPE_OCR, 20 },
 #else
       { CaptureToolButton::TYPE_EXIT, 17 }, { CaptureToolButton::TYPE_PIN, 18 },
+      { CaptureToolButton::TYPE_OCR, 19 },
 #endif
 };
 
@@ -175,4 +177,5 @@ QVector<CaptureToolButton::ButtonType>
 #endif
       CaptureToolButton::TYPE_PIN,
       CaptureToolButton::TYPE_CIRCLECOUNT,
+      CaptureToolButton::TYPE_OCR
   };

src/widgets/capture/capturetoolbutton.h

@@ -53,7 +53,8 @@ class CaptureToolButton : public CaptureButton
         TYPE_REDO = 16,
         TYPE_PIN = 17,
         TYPE_TEXT = 18,
-        TYPE_CIRCLECOUNT = 19
+        TYPE_CIRCLECOUNT = 19,
+        TYPE_OCR = 20
     };
     Q_ENUM(ButtonType)
 

src/widgets/capture/capturewidget.cpp

@@ -217,6 +217,7 @@ void CaptureWidget::updateButtons()
             case CaptureToolButton::ButtonType::TYPE_UNDO:
             case CaptureToolButton::ButtonType::TYPE_REDO:
             case CaptureToolButton::ButtonType::TYPE_IMAGEUPLOADER:
+            case CaptureToolButton::ButtonType::TYPE_OCR:
                 // nothing to do, just skip non-dynamic buttons with existing
                 // hard coded slots
                 break;
@@ -1045,6 +1046,10 @@ void CaptureWidget::initShortcuts()
       QVariant::fromValue(CaptureToolButton::ButtonType::TYPE_IMAGEUPLOADER)
         .toString());
 
+    shortcut = ConfigHandler().shortcut(
+      QVariant::fromValue(CaptureToolButton::ButtonType::TYPE_OCR).toString());
+    new QShortcut(QKeySequence(shortcut), this, SLOT(ocr()));
+
     new QShortcut(QKeySequence(ConfigHandler().shortcut("TYPE_TOGGLE_PANEL")),
                   this,
                   SLOT(togglePanel()));
@@ -1208,6 +1213,18 @@ void CaptureWidget::saveScreenshot()
     close();
 }
 
+void CaptureWidget::ocr()
+{
+    m_captureDone = true;
+    if (m_activeTool != nullptr) {
+        QPainter painter(&m_context.screenshot);
+        m_activeTool->process(painter, m_context.screenshot, true);
+    }
+    hide();
+    // Process here!
+    close();
+}
+
 void CaptureWidget::undo()
 {
     m_undoStack.undo();

src/widgets/capture/capturewidget.h

@@ -77,6 +77,7 @@ private slots:
     // TODO replace with tools
     void copyScreenshot();
     void saveScreenshot();
+    void ocr();
     void undo();
     void redo();
     void togglePanel();