Java: 如何使用iText从PDF文件中提取选定区域的文本？

Question

我正在开发一个程序，可以从 PDF 文件中 extract texts特定区域，我正在使用 java 和 iText 库。现在，我可以使用以下代码通过点击区域坐标来提取数据：

import java.io.IOException;
import com.itextpdf.text.Rectangle;
import com.itextpdf.text.pdf.PdfReader;
import com.itextpdf.text.pdf.parser.FilteredTextRenderListener;
import com.itextpdf.text.pdf.parser.LocationTextExtractionStrategy;
import com.itextpdf.text.pdf.parser.PdfTextExtractor;
import com.itextpdf.text.pdf.parser.RegionTextRenderFilter;
import com.itextpdf.text.pdf.parser.RenderFilter;
import com.itextpdf.text.pdf.parser.TextExtractionStrategy;

/**
 * Créer par Malek Boubakri le 03/06/2015 à 15:45.
 */

public class ExtractPageContentArea {
 //
    public void parsePdf(float x,float y,float width,float height,String pdf) throws IOException {
        PdfReader reader = new PdfReader(pdf);
        Rectangle rect = new Rectangle(x, y, width, height);
        RenderFilter filter = new RegionTextRenderFilter(rect);
        TextExtractionStrategy strategy;
        for (int i = 1; i <= reader.getNumberOfPages(); i++) {
            strategy = new FilteredTextRenderListener(new LocationTextExtractionStrategy(), filter);
            System.out.println(PdfTextExtractor.getTextFromPage(reader, i, strategy));
        }
        reader.close();
    }
}

并且此代码可以绘制矩形并保存所需的坐标使用：

import java.awt.BorderLayout;
import java.awt.Graphics;
import java.awt.Rectangle;
import java.awt.event.MouseEvent;
import java.awt.event.MouseListener;
import java.awt.event.MouseMotionListener;
import java.util.ArrayList;

import javax.swing.JFrame;
import javax.swing.JLabel;
import javax.swing.SwingConstants;

public class MouseTracker extends JFrame implements MouseListener, MouseMotionListener {

    private static final long serialVersionUID = 1L;
    private final JLabel mousePosition;
    int x1, x2, y1, y2;
    int w, h;
    private final JLabel recStart;
    private final JLabel recStop;
    private final JLabel cords; // set up GUI and register mouse event handlers
    private final ArrayList< Rectangle > rectangles = new ArrayList< Rectangle >();
    private boolean isNewRect = true;

    public MouseTracker() {
        super( "Rectangle Drawer" );

        this.mousePosition = new JLabel();
        this.mousePosition.setHorizontalAlignment( SwingConstants.CENTER );
        getContentPane().add( this.mousePosition, BorderLayout.CENTER );

        JLabel text1 = new JLabel();
        text1.setText( "At the center the mouse pointer's coordinates will be displayed." );
        getContentPane().add( text1, BorderLayout.SOUTH );

        this.recStart = new JLabel();
        getContentPane().add( this.recStart, BorderLayout.WEST );

        this.recStop = new JLabel();
        getContentPane().add( this.recStop, BorderLayout.EAST );

        this.cords = new JLabel();
        getContentPane().add( this.cords, BorderLayout.NORTH );

        addMouseListener( this ); // listens for own mouse and
        addMouseMotionListener( this ); // mouse-motion events

        setSize( 800, 600 );
        setVisible( true );

    }

// MouseListener event handlers // handle event when mouse released immediately after press 
    public void mouseClicked( final MouseEvent event ) {
        this.mousePosition.setText( "Clicked at [" + event.getX() + ", " + event.getY() + "]" );

        repaint();
    }

// handle event when mouse pressed 
    public void mousePressed( final MouseEvent event ) {

        this.mousePosition.setText( "Pressed at [" + ( this.x1 = event.getX() ) + ", " + ( this.y1 = event.getY() ) + "]" );

        this.recStart.setText( "Start:  [" + this.x1 + ", " + this.y1 + "]" );

        repaint();
    }

// handle event when mouse released after dragging 
    public void mouseReleased( final MouseEvent event ) {
        this.mousePosition.setText( "Released at [" + ( this.x2 = event.getX() ) + ", " + ( this.y2 = event.getY() ) + "]" );

        this.recStop.setText( "End:  [" + this.x2 + ", " + this.y2 + "]" );

        Rectangle rectangle = getRectangleFromPoints();

        this.rectangles.add( rectangle );

        this.w = this.h = this.x1 = this.y1 = this.x2 = this.y2 = 0;
        this.isNewRect = true;

        repaint();
    }

    private Rectangle getRectangleFromPoints() {
        int width = this.x1 - this.x2;
        int height = this.y1 - this.y2;
        Rectangle rectangle = new Rectangle( width < 0 ? this.x1
            : this.x2, height < 0 ? this.y1
            : this.y2, Math.abs( width ), Math.abs( height ) );

        return rectangle;
    }

// handle event when mouse enters area 
    public void mouseEntered( final MouseEvent event ) {
        this.mousePosition.setText( "Mouse entered at [" + event.getX() + ", " + event.getY() + "]" );
        repaint();
    }

// handle event when mouse exits area 
    public void mouseExited( final MouseEvent event ) {
        this.mousePosition.setText( "Mouse outside window" );
        repaint();
    }

// MouseMotionListener event handlers // handle event when user drags mouse with button pressed 
    public void mouseDragged( final MouseEvent event ) {
        this.mousePosition.setText( "Dragged at [" + ( this.x2 = event.getX() ) + ", " + ( this.y2 = event.getY() ) + "]" ); // call repaint which calls paint repaint();

        this.isNewRect = false;

        repaint();
    }

// handle event when user moves mouse 
    public void mouseMoved( final MouseEvent event ) {
        this.mousePosition.setText( "Moved at [" + event.getX() + ", " + event.getY() + "]" );
        repaint();
    }

    @Override
    public void paint( final Graphics g ) {
        super.paint( g ); // clear the frame surface 
        g.drawString( "Start Rec Here", this.x1, this.y1 );
        g.drawString( "End Rec Here", this.x2, this.y2 );

        Rectangle newRectangle = getRectangleFromPoints();
        if ( !this.isNewRect ) {
            g.drawRect( newRectangle.x, newRectangle.y, newRectangle.width, newRectangle.height );
        }

        for( Rectangle rectangle : this.rectangles ) {
            g.drawRect( rectangle.x, rectangle.y, rectangle.width, rectangle.height );
        }

        this.cords.setText( "w = " + this.w + ", h = " + this.h );

    }

    public static void main( final String args[] ) {
        MouseTracker application = new MouseTracker();
        application.setDefaultCloseOperation( JFrame.EXIT_ON_CLOSE );
    }

}

我想使用这些坐标来指定 PDF 文件中的区域，我真的不知道如何合并这两个功能，如何将 绘图 space 放在文档上方 以及如何使矩形坐标与文本坐标相匹配。

如何在另一个面板上方绘制？
我应该将 PDF 转换为图像并将其放在后面吗？
如果我应该的话，请任何人推荐一个好的免费 OCR 库！

如果有任何模糊之处，请发表评论！谁能把我放在路上！因为我真的迷路了。

等待您的帮助..谢谢（抱歉我的英语不好）

Answer 1

你有一个非常有趣的问题和一个具有挑战性的项目。此 "answer" 可能会提供一些有用的想法，但它不是一个完整的解决方案。

您可以使用所谓的 glass pane 在其他组件之上绘制。

我认为您需要决定的最重要的事情是哪些库最适合您的项目。 iText library 非常好，提供各种 pdf 功能，例如您在问题中显示的文本提取。

但是，据我所知，在 iText 中不支持查看 pdf。您可以使用像 ICEpdf for this (see this example) 这样的库。如果 ICEpdf 也可以支持文本提取，那就太好了，这样您就可以使用一个库，而不是让 ICEpdf 与 iText 或 OCR 一起工作（并处理诸如在 ICEpdf 中缩放 pdf 并在获取文本时进行补偿等问题).

我不确定您是否可以使用 ICEpdf 提取文本，因此在下面的示例代码中目前仍使用 iText：

// File ExtractSelectionFromPdf.java

import com.itextpdf.text.Rectangle;
import com.itextpdf.text.pdf.PdfReader;
import com.itextpdf.text.pdf.parser.*;

import java.awt.Container;
import java.awt.Point;
import java.io.IOException;
import javax.swing.*;

public class ExtractSelectionFromPdf {
    private static String filePath = "[file path to a pdf file]";

    private PdfViewer pdfViewer;

    public static void main(final String[] arguments) {
        SwingUtilities.invokeLater(() -> new ExtractSelectionFromPdf().launchGUI());
    }

    private void launchGUI() {
        final JFrame frame = new JFrame("Extract selected text from a pdf");
        frame.setDefaultCloseOperation(WindowConstants.EXIT_ON_CLOSE);
        final Container contentPane = frame.getContentPane();

        pdfViewer = new PdfViewer();
        contentPane.add(pdfViewer);

        pdfViewer.openDocument(filePath);

        final CustomGlassPane customGlassPane = new CustomGlassPane(this, contentPane);
        frame.setGlassPane(customGlassPane);
        customGlassPane.setVisible(true);

        frame.setBounds(60, 10, 1800, 1000);
        frame.setVisible(true);
    }

    public void handleSelection(final Point topLeft, final Point bottomRight) {
        final int width = bottomRight.x - topLeft.x;
        final int height = bottomRight.y - topLeft.x;
        final String text = parsePdf(topLeft.x, topLeft.y, width, height, filePath);
        System.out.println("text: " + text);
    }

    public String parsePdf(final int x, final int y, final int width, final int height, 
                           final String pdfFilePath) {
        String text = null;

        try {
            final PdfReader pdfReader = new PdfReader(pdfFilePath);
            final int pageNumber = pdfViewer.getCurrentPageNumber() + 1;
            System.out.println("Page number: " + pageNumber);
            final Rectangle selection = new Rectangle(x, y, width, height);
            final RenderFilter renderFilter = new RegionTextRenderFilter(selection);
            final LocationTextExtractionStrategy delegate 
                    = new LocationTextExtractionStrategy();
            final TextExtractionStrategy extractionStrategy 
                    = new FilteredTextRenderListener(delegate, renderFilter);
            text = PdfTextExtractor.getTextFromPage(pdfReader, pageNumber, 
                                                    extractionStrategy);
            pdfReader.close();
        } catch (final IOException e) {
            e.printStackTrace();
        }

        return text;
    }
}


// File PdfViewer.java

import java.util.ResourceBundle;
import javax.swing.*;
import org.icepdf.ri.common.*;
import org.icepdf.ri.common.views.DocumentViewController;
import org.icepdf.ri.util.PropertiesManager;

public class PdfViewer extends JPanel {
    private final SwingController controller;

    public PdfViewer() {
        controller = new SwingController();
        controller.setIsEmbeddedComponent(true);

        final String bundleName = PropertiesManager.DEFAULT_MESSAGE_BUNDLE;
        final ResourceBundle messageBundle = ResourceBundle.getBundle(bundleName);
        final Properties systemProperties = System.getProperties();
        final PropertiesManager properties = new PropertiesManager(systemProperties,
                                                                   messageBundle);

        properties.set(PropertiesManager.PROPERTY_DEFAULT_ZOOM_LEVEL, "1");

        final SwingViewBuilder factory = new SwingViewBuilder(controller, properties);

        final DocumentViewController viewController 
                = controller.getDocumentViewController();
        viewController.setAnnotationCallback(new MyAnnotationCallback(viewController));

        final JScrollPane scrollPane = new JScrollPane(factory.buildViewerPanel());
        final int horizontalPolicy = ScrollPaneConstants.HORIZONTAL_SCROLLBAR_ALWAYS;
        final int verticalPolicy = ScrollPaneConstants.VERTICAL_SCROLLBAR_ALWAYS;
        scrollPane.setHorizontalScrollBarPolicy(horizontalPolicy);
        scrollPane.setVerticalScrollBarPolicy(verticalPolicy);
        add(scrollPane);
    }

    public void openDocument(final String filePath) {
        controller.openDocument(filePath);
    }

    public int getCurrentPageNumber() {
        return controller.getCurrentPageNumber();
    }
}


// File CustomGlassPane.java

import java.awt.*;
import javax.swing.JComponent;

public class CustomGlassPane extends JComponent {
    private Point topLeftPoint;
    private Point bottomRightPoint;

    public CustomGlassPane(final ExtractSelectionFromPdf extractSelectionFromPdf, 
                           final Container contentPane) {
        final MouseEventsListener listener 
                = new MouseEventsListener(extractSelectionFromPdf, this, contentPane);
        addMouseListener(listener);
        addMouseMotionListener(listener);
    }

    public void setSelection(final Point topLeftPoint, final Point bottomRightPoint) {
        this.topLeftPoint = topLeftPoint;
        this.bottomRightPoint = bottomRightPoint;
    }

    protected void paintComponent(final Graphics graphics) {
        if (topLeftPoint != null && bottomRightPoint != null) {
            graphics.setColor(Color.BLACK);
            graphics.drawRect(topLeftPoint.x, 
                              topLeftPoint.y,
                              bottomRightPoint.x - topLeftPoint.x, 
                              bottomRightPoint.y - topLeftPoint.y);
        }
    }
}


// File MouseEventsListener.java

import java.awt.*;
import java.awt.event.MouseEvent;
import javax.swing.SwingUtilities;
import javax.swing.event.MouseInputAdapter;

public class MouseEventsListener extends MouseInputAdapter {
    private ExtractSelectionFromPdf extractSelectionFromPdf;
    private CustomGlassPane customGlassPane;
    private Container contentPane;
    private Point topLeftPoint;
    private Point bottomRightPoint;

    public MouseEventsListener(final ExtractSelectionFromPdf extractSelectionFromPdf,
                               final CustomGlassPane customGlassPane,
                               final Container contentPane) {
        this.extractSelectionFromPdf = extractSelectionFromPdf;
        this.customGlassPane = customGlassPane;
        this.contentPane = contentPane;
    }

    public void mousePressed(final MouseEvent mouseEvent) {
        topLeftPoint = mouseEvent.getPoint();

        redispatchMouseEvent(mouseEvent);
    }

    public void mouseDragged(final MouseEvent mouseEvent) {
        bottomRightPoint = mouseEvent.getPoint();

        redispatchMouseEvent(mouseEvent, topLeftPoint != null, false);
    }

    public void mouseReleased(final MouseEvent mouseEvent) {
        bottomRightPoint = mouseEvent.getPoint();

        redispatchMouseEvent(mouseEvent, true, true);
    }

    public void mouseMoved(final MouseEvent mouseEvent) {
        redispatchMouseEvent(mouseEvent);
    }

    public void mouseClicked(final MouseEvent mouseEvent) {
        redispatchMouseEvent(mouseEvent);
    }

    public void mouseEntered(final MouseEvent mouseEvent) {
        redispatchMouseEvent(mouseEvent);
    }

    public void mouseExited(final MouseEvent mouseEvent) {
        redispatchMouseEvent(mouseEvent);
    }

    private void redispatchMouseEvent(final MouseEvent mouseEvent) {
        redispatchMouseEvent(mouseEvent, false, false);
    }

    private void redispatchMouseEvent(final MouseEvent mouseEvent, 
                                      final boolean repaint,
                                      final boolean extract) {
        final Point glassPanePoint = mouseEvent.getPoint();
        final Point containerPoint = SwingUtilities.convertPoint(customGlassPane, 
                                                                 glassPanePoint, 
                                                                 contentPane);

        if (containerPoint.y >= 0) {
            final Component component
                    = SwingUtilities.getDeepestComponentAt(contentPane,
                                                           containerPoint.x,
                                                           containerPoint.y);

            if (component != null) {
                final Point componentPoint 
                        = SwingUtilities.convertPoint(customGlassPane, 
                                                      glassPanePoint,
                                                      component);

                // Forward events to the component under the glass pane.
                component.dispatchEvent(new MouseEvent(component,
                                                       mouseEvent.getID(),
                                                       mouseEvent.getWhen(),
                                                       mouseEvent.getModifiers(),
                                                       componentPoint.x,
                                                       componentPoint.y,
                                                       mouseEvent.getClickCount(),
                                                       mouseEvent.isPopupTrigger()));
            }
        }

        // Update the glass pane if requested.
        if (repaint) {
            if (extract) {
                extractSelectionFromPdf.handleSelection(topLeftPoint, bottomRightPoint);

                topLeftPoint = null;
                bottomRightPoint = null;
            }

            customGlassPane.setSelection(topLeftPoint, bottomRightPoint);
            customGlassPane.repaint();
        }
    }
}

上面代码的玻璃面板部分的灵感来自 GlassPaneDemo example。

以上代码中已知的遗留问题：

出于某种原因，在页面 Up/Down 和箭头 Up/Down 键起作用之前，必须单击一次 pdf 查看器的向下滚动按钮。
当前实际提取的文本似乎在所选矩形下方。

Answer 2

经过多次尝试，我确信 "GlassPane" 不是我的应用程序或任何类似应用程序的正确解决方案。因为：

它不能在特定区域或组件之上。
它可用仅用于根窗格..
阅读 PDF 文件的最佳方法是将其转换为图片，然后在平板电脑上阅读ImagePane.
在这种情况下不需要OCR[=33=]..

我现在正在研究另一个解决方案，一切顺利。如果有人在做这样的项目，请评论、标记或升级问题，我准备好了解任何细节。

Java: 如何使用iText从PDF文件中提取选定区域的文本？

Java: How to extract text by a selected area from a PDF file using iText?

java

pdf

swing

awt

itext