如何使用 PDFBox 在 pdf 中查找表格边框线？

Question

我正在尝试在 pdf 中查找表格边框线。我用PrintTextLocations pdfBox 类来造词。现在我正在寻找形成表格的不同线条的坐标。我尝试使用org.apache.pdfbox.pdfviewer.PageDrawer，但找不到包含这些行的任何字符/图形。我尝试了两种方法：

第一：

Graphics g = null;
Dimension d = new Dimension();
d.setSize(700, 700);
PageDrawer pageDrawer = new PageDrawer();
pageDrawer.drawPage(g, myPage, d);

它给了我空指针异常。所以其次，我试图覆盖processStream 函数，但我无法获得任何中风。请帮帮我。我愿意使用任何其他提供表格中线条坐标的库。另一个快速的问题是，pdfbox 中的那些表格边框线是什么类型的对象？这些是图形还是这些字符？

这是我要解析的示例 pdf 的链接： http://stats.bls.gov/news.release/pdf/empsit.pdf 并尝试获取第 8 页上的表格行。

编辑：我遇到了另一个问题，在解析这个 pdf 的第 1 页时，我无法得到任何行，因为 printPath() 函数中的 pathIterator 是空的，尽管 strokePath() 函数为每一行调用。如何处理这个 pdf？

Answer 1

在 1.8.* 版本中，PDFBox 解析功能以一种不太通用的方式实现，特别是 OperatorProcessor 实现与特定解析器类紧密相关，例如处理路径绘制操作的实现假定与PageDrawer 实例交互。

因此，除非想要复制和粘贴所有 OperatorProcessor 类并进行细微更改，否则必须从这样一个特定的解析器类派生。

因此，在您的情况下，我们也将从PageDrawer 派生我们的解析器，毕竟我们对路径绘制操作感兴趣：

public class PrintPaths extends PageDrawer
{
    //
    // constructor
    //
    public PrintPaths() throws IOException
    {
        super();
    }

    //
    // method overrides for mere path observation
    //
    // ignore text
    @Override
    protected void processTextPosition(TextPosition text) { }

    // ignore bitmaps
    @Override
    public void drawImage(Image awtImage, AffineTransform at) { }

    // ignore shadings
    @Override
    public void shFill(COSName shadingName) throws IOException { }

    @Override
    public void processStream(PDPage aPage, PDResources resources, COSStream cosStream) throws IOException
    {
        PDRectangle cropBox = aPage.findCropBox();
        this.pageSize = cropBox.createDimension();
        super.processStream(aPage, resources, cosStream);
    }

    @Override
    public void fillPath(int windingRule) throws IOException
    {
        printPath();
        System.out.printf("Fill; windingrule: %s\n\n", windingRule);
        getLinePath().reset();
    }

    @Override
    public void strokePath() throws IOException
    {
        printPath();
        System.out.printf("Stroke; unscaled width: %s\n\n", getGraphicsState().getLineWidth());
        getLinePath().reset();
    }

    void printPath()
    {
        GeneralPath path = getLinePath();
        PathIterator pathIterator = path.getPathIterator(null);

        double x = 0, y = 0;
        double coords[] = new double[6];
        while (!pathIterator.isDone()) {
            switch (pathIterator.currentSegment(coords)) {
            case PathIterator.SEG_MOVETO:
                System.out.printf("Move to (%s %s)\n", coords[0], fixY(coords[1]));
                x = coords[0];
                y = coords[1];
                break;
            case PathIterator.SEG_LINETO:
                double width = getEffectiveWidth(coords[0] - x, coords[1] - y);
                System.out.printf("Line to (%s %s), scaled width %s\n", coords[0], fixY(coords[1]), width);
                x = coords[0];
                y = coords[1];
                break;
            case PathIterator.SEG_QUADTO:
                System.out.printf("Quad along (%s %s) and (%s %s)\n", coords[0], fixY(coords[1]), coords[2], fixY(coords[3]));
                x = coords[2];
                y = coords[3];
                break;
            case PathIterator.SEG_CUBICTO:
                System.out.printf("Cubic along (%s %s), (%s %s), and (%s %s)\n", coords[0], fixY(coords[1]), coords[2], fixY(coords[3]), coords[4], fixY(coords[5]));
                x = coords[4];
                y = coords[5];
                break;
            case PathIterator.SEG_CLOSE:
                System.out.println("Close path");
            }
            pathIterator.next();
        }
    }

    double getEffectiveWidth(double dirX, double dirY)
    {
        if (dirX == 0 && dirY == 0)
            return 0;
        Matrix ctm = getGraphicsState().getCurrentTransformationMatrix();
        double widthX = dirY;
        double widthY = -dirX;
        double widthXTransformed = widthX * ctm.getValue(0, 0) + widthY * ctm.getValue(1, 0);
        double widthYTransformed = widthX * ctm.getValue(0, 1) + widthY * ctm.getValue(1, 1);
        double factor = Math.sqrt((widthXTransformed*widthXTransformed + widthYTransformed*widthYTransformed) / (widthX*widthX + widthY*widthY));
        return getGraphicsState().getLineWidth() * factor;
    }
}

(PrintPaths.java)

由于我们不想真正绘制页面，而只是提取将要绘制的路径，因此我们必须像这样剥离PageDrawer。

此示例解析器输出路径绘制操作以显示如何执行此操作。显然，您可以将它们收集起来进行自动化处理...

你可以像这样使用解析器：

PDDocument document = PDDocument.load(resource);
List<?> allPages = document.getDocumentCatalog().getAllPages();
int i = 7; // page 8

System.out.println("\n\nPage " + (i+1));
PrintPaths printPaths = new PrintPaths();

PDPage page = (PDPage) allPages.get(i);
PDStream contents = page.getContents();
if (contents != null)
{
    printPaths.processStream(page, page.findResources(), page.getContents().getStream());
}

(ExtractPaths.java)

输出是：

Page 8
Move to (35.92070007324219 724.6490478515625)
Line to (574.72998046875 724.6490478515625), scaled width 0.5981000089123845
Stroke; unscaled width: 5.981

Move to (35.92070007324219 694.4660034179688)
Line to (574.72998046875 694.4660034179688), scaled width 0.5981000089123845
Stroke; unscaled width: 5.981

Move to (292.2610168457031 468.677001953125)
Line to (292.8590087890625 468.677001953125), scaled width 512.9430076434463
Stroke; unscaled width: 5129.43

Move to (348.9360046386719 468.677001953125)
Line to (349.53399658203125 468.677001953125), scaled width 512.9430076434463
Stroke; unscaled width: 5129.43

Move to (405.6090087890625 468.677001953125)
Line to (406.2070007324219 468.677001953125), scaled width 512.9430076434463
Stroke; unscaled width: 5129.43

Move to (462.281982421875 468.677001953125)
Line to (462.8799743652344 468.677001953125), scaled width 512.9430076434463
Stroke; unscaled width: 5129.43

Move to (518.9549560546875 468.677001953125)
Line to (519.553955078125 468.677001953125), scaled width 512.9430076434463
Stroke; unscaled width: 5129.43

Move to (35.92070007324219 725.447998046875)
Line to (574.72998046875 725.447998046875), scaled width 0.5981000089123845
Stroke; unscaled width: 5.981

Move to (35.92070007324219 212.5050048828125)
Line to (574.72998046875 212.5050048828125), scaled width 0.5981000089123845
Stroke; unscaled width: 5.981

非常奇特：垂直线实际上画得很短（约 0.6 单位）很粗（约 513 单位）水平线...