/**
* 找到关键字的位置
* @author hxw
*
*/
public class PdfKeywordFinder {
public static void main(String[] args) throws IOException {
File pdfFile = new File("E:/temp/off_cnpc_plasthetics.pdf");
byte[] pdfData = new byte[(int) pdfFile.length()];
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) {
}
}
}
//关键字
String keyword = "11.4";
List<float[]> positions = findKeywordPostions(pdfData, keyword);
System.out.println("total:" + positions.size());
if (positions != null && positions.size() > 0) {
for (float[] position : positions) {
System.out.print("pageNum: " + (int) position[0]);
System.out.print("\tx: " + position[1]);
System.out.println("\ty: " + position[2]);
}
}
}
/**
* findKeywordPostions
*
* @param pdfData
* @param keyword
* @return List<float[]> : float[0]:pageNum float[1]:x float[2]:y
* @throws IOException
*/
public static List<float[]> findKeywordPostions(byte[] pdfData, String keyword) throws IOException {
List<float[]> result = new ArrayList<>();
List<PdfPageContentPositions> pdfPageContentPositions = getPdfContentPostionsList(pdfData);
for (PdfPageContentPositions pdfPageContentPosition : pdfPageContentPositions) {
List<float[]> charPositions = findPositions(keyword, pdfPageContentPosition);
if (charPositions == null || charPositions.size() < 1) {
continue;
}
result.addAll(charPositions);
}
return result;
}
private static 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);
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<float[]> positionsList = new ArrayList<>();
for (CharPosition charPosition : charPositions) {
float[] positions = new float[] { 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<float[]> findPositions(String keyword, PdfPageContentPositions pdfPageContentPositions) {
List<float[]> result = new ArrayList<>();
String content = pdfPageContentPositions.getContent();
List<float[]> charPositions = pdfPageContentPositions.getPositions();
for (int pos = 0; pos < content.length();) {
int positionIndex = content.indexOf(keyword, pos);
if (positionIndex == -1) {
break;
}
float[] postions = charPositions.get(positionIndex);
result.add(postions);
pos = positionIndex + 1;
}
return result;
}
private static class PdfPageContentPositions {
private String content;
private List<float[]> positions;
public String getContent() {
return content;
}
public void setContent(String content) {
this.content = content;
}
public List<float[]> getPositions() {
return positions;
}
public void setPostions(List<float[]> positions) {
this.positions = positions;
}
}
private static 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;
}
@Override
public void beginTextBlock() {
}
@Override
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();
double x = rectangle.getMinX();
double y = rectangle.getMaxY();
float xPercent = Math.round(x * 10000) / 10000;
float yPercent = Math.round(y * 10000) / 10000;
CharPosition charPosition = new CharPosition(pageNum, xPercent, yPercent);
charPositions.add(charPosition);
contentBuilder.append(word);
}
}
@Override
public void endTextBlock() {
}
@Override
public void renderImage(ImageRenderInfo renderInfo) {
}
public String getContent() {
return contentBuilder.toString();
}
public List<CharPosition> getcharPositions() {
return charPositions;
}
}
private static class CharPosition {
private int pageNum = 0;
private float x = 0;
private float y = 0;
public CharPosition(int pageNum, float x, float y) {
this.pageNum = pageNum;
this.x = x;
this.y = y;
}
public int getPageNum() {
return pageNum;
}
public float getX() {
return x;
}
public float getY() {
return y;
}
@Override
public String toString() {
return "[pageNum=" + this.pageNum + ",x=" + this.x + ",y=" + this.y + "]";
}
}
}