获取 pdf 关键字坐标

标签:lin   bytes   stringbu   位置   getwidth   误差   puts   方式   idt   

package Demo.qd;

import com.itextpdf.awt.geom.Rectangle2D.Float;
import com.itextpdf.text.pdf.PdfDictionary;
import com.itextpdf.text.pdf.PdfName;
import com.itextpdf.text.pdf.PdfReader;
import com.itextpdf.text.pdf.parser.*;

import java.io.File;
import java.io.FileInputStream;
import java.io.IOException;
import java.math.BigDecimal;
import java.util.ArrayList;
import java.util.List;

public class PdfPositionTool {

    public static void main(String[] args) throws Exception {
        PdfPositionTool pdfPositionTool = new PdfPositionTool();
        List<double[]> positions = pdfPositionTool.getPositions("/Users/yourouniu/Desktop/111.pdf", "%盖章处%");
        if (positions != null && positions.size() > 0) {
            for (double[] position : positions) {
                System.out.println("pageNum: " + (int) position[0]);
                System.out.println("x: " + position[1]);
                System.out.println("y: " + position[2]);
            }
        }
    }

    /**
     * @return List<float [ ]> 坐标数组:float[0]:页码,float[1]:x ,float[2]:y
     * @Description 获取关键字坐标
     * @Param filePath:pdf 路径
     * @Param keyword:关键字
     */
    public List<double[]> getPositions(String filePath, String keyword) throws IOException {
        PdfPositionTool pdfPositionTool = new PdfPositionTool();
        //1.给定文件
        File pdfFile = new File(filePath);
        //2.定义一个byte数组,长度为文件的长度
        byte[] pdfData = new byte[(int) pdfFile.length()];
        //3.IO流读取文件内容到byte数组
        FileInputStream inputStream = null;
        try {
            inputStream = new FileInputStream(pdfFile);
            inputStream.read(pdfData);
        } catch (IOException e) {
            throw e;
        } finally {
            if (inputStream != null) {
                try {
                    inputStream.close();
                } catch (IOException e) {
                }
            }
        }
        //5.调用方法,给定关键字和文件
        List<double[]> positions = pdfPositionTool.findKeywordPostions(pdfData, keyword);
        return positions;
    }

    /**
     * @Description pdf 坐标转换为 ofd 坐标,比值为 25.4/72 ,该转换存在误差
     * 最好的转换方式为按距离原点的百分比计算
     */
    private double transForPosition(double pdfPosition) {
        double ofdPosition = pdfPosition * 25.4 / 72;
        return ofdPosition;
    }


    /**
     * @param pdfData 通过IO流 PDF文件转化的byte数组
     * @param keyword 关键字
     * @return List<float [ ]> : float[0]:pageNum float[1]:x float[2]:y
     * @throws IOException
     */
    public List<double[]> findKeywordPostions(byte[] pdfData, String keyword) throws IOException {
        List<double[]> result = new ArrayList<>();
        List<PdfPageContentPositions> pdfPageContentPositions = getPdfContentPostionsList(pdfData);
        for (PdfPageContentPositions pdfPageContentPosition : pdfPageContentPositions) {
            List<double[]> charPositions = findPositions(keyword, pdfPageContentPosition);
            if (charPositions == null || charPositions.size() < 1) {
                continue;
            }
            result.addAll(charPositions);
        }
        return result;
    }


    private List<PdfPageContentPositions> getPdfContentPostionsList(byte[] pdfData) throws IOException {
        PdfReader reader = new PdfReader(pdfData);
        List<PdfPageContentPositions> result = new ArrayList<>();
        int pages = reader.getNumberOfPages();
        for (int pageNum = 1; pageNum <= pages; pageNum++) {
            float width = reader.getPageSize(pageNum).getWidth();
            float height = reader.getPageSize(pageNum).getHeight();
            PdfRenderListener pdfRenderListener = new PdfRenderListener(pageNum, width, height);
            //解析pdf,定位位置
            PdfContentStreamProcessor processor = new PdfContentStreamProcessor(pdfRenderListener);
            PdfDictionary pageDic = reader.getPageN(pageNum);
            PdfDictionary resourcesDic = pageDic.getAsDict(PdfName.RESOURCES);
            try {
                processor.processContent(ContentByteUtils.getContentBytesForPage(reader, pageNum), resourcesDic);
            } catch (IOException e) {
                reader.close();
                throw e;
            }
            String content = pdfRenderListener.getContent();
            List<CharPosition> charPositions = pdfRenderListener.getcharPositions();
            List<double[]> positionsList = new ArrayList<>();
            for (CharPosition charPosition : charPositions) {
                double[] positions = new double[]{charPosition.getPageNum(), charPosition.getX(), charPosition.getY()};
                positionsList.add(positions);
            }
            PdfPageContentPositions pdfPageContentPositions = new PdfPageContentPositions();
            pdfPageContentPositions.setContent(content);
            pdfPageContentPositions.setPostions(positionsList);
            result.add(pdfPageContentPositions);
        }
        reader.close();
        return result;
    }

    private static List<double[]> findPositions(String keyword, PdfPageContentPositions pdfPageContentPositions) {
        List<double[]> result = new ArrayList<>();
        String content = pdfPageContentPositions.getContent();
        List<double[]> charPositions = pdfPageContentPositions.getPositions();
        for (int pos = 0; pos < content.length(); ) {
            int positionIndex = content.indexOf(keyword, pos);
            if (positionIndex == -1) {
                break;
            }
            double[] postions = charPositions.get(positionIndex);
            result.add(postions);
            pos = positionIndex + 1;
        }
        return result;
    }

    private class PdfPageContentPositions {
        private String content;
        private List<double[]> positions;

        public String getContent() {
            return content;
        }

        public void setContent(String content) {
            this.content = content;
        }

        public List<double[]> getPositions() {
            return positions;
        }

        public void setPostions(List<double[]> positions) {
            this.positions = positions;
        }
    }

    private class PdfRenderListener implements RenderListener {
        private int pageNum;
        private float pageWidth;
        private float pageHeight;
        private StringBuilder contentBuilder = new StringBuilder();
        private List<CharPosition> charPositions = new ArrayList<>();

        public PdfRenderListener(int pageNum, float pageWidth, float pageHeight) {
            this.pageNum = pageNum;
            this.pageWidth = pageWidth;
            this.pageHeight = pageHeight;
        }

        public void beginTextBlock() {
        }

        /**
         * @Description 计算转换后的 ofd 坐标值
         * 如有需要,可转为计算距离原点的百分比值。在知道 ofd 长宽的情况下,用百分比重新计算坐标更精确
         */
        public void renderText(TextRenderInfo renderInfo) {
            List<TextRenderInfo> characterRenderInfos = renderInfo.getCharacterRenderInfos();
            for (TextRenderInfo textRenderInfo : characterRenderInfos) {
                String word = textRenderInfo.getText();
                if (word.length() > 1) {
                    word = word.substring(word.length() - 1, word.length());
                }
                Float rectangle = textRenderInfo.getAscentLine().getBoundingRectange();
                float x = (float) rectangle.getX();
                float y = (float) rectangle.getY();
                //这两个是关键字在所在页面的XY轴的百分比
                float xPercent = Math.round(x / pageWidth * 10000) / 10000f;
                // pdf 原点在左下,ofd 原点在左上
                float yPercent = Math.round((1 - y / pageHeight) * 10000) / 10000f;
                CharPosition charPosition = new CharPosition(pageNum, transForPosition(x),
                        transForPosition((yPercent) * pageHeight));
                charPositions.add(charPosition);
                contentBuilder.append(word);
            }
        }

        public void endTextBlock() {
        }

        public void renderImage(ImageRenderInfo renderInfo) {
        }

        public String getContent() {
            return contentBuilder.toString();
        }

        public List<CharPosition> getcharPositions() {
            return charPositions;
        }
    }

    private class CharPosition {
        private int pageNum = 0;
        private double x = 0;
        private double y = 0;

        public CharPosition(int pageNum, double x, double y) {
            this.pageNum = pageNum;
            this.x = x;
            this.y = y;
        }

        public int getPageNum() {
            return pageNum;
        }

        public double getX() {
            return x;
        }

        public double getY() {
            return y;
        }

        @Override
        public String toString() {
            return "[pageNum=" + this.pageNum + ",x=" + this.x + ",y=" + this.y + "]";
        }
    }
}

 

获取 pdf 关键字坐标

标签:lin   bytes   stringbu   位置   getwidth   误差   puts   方式   idt   

原文地址:https://www.cnblogs.com/niuyourou/p/14235096.html

版权声明:完美者 发表于 2021-01-07 12:22:14。
转载请注明:获取 pdf 关键字坐标 | 完美导航

暂无评论

暂无评论...