Java PDDocument.getNumberOfPages方法代码示例

import org.apache.pdfbox.pdmodel.PDDocument; //导入方法依赖的package包/类
public void process(File pdf, File output){
        PDDocument pdDoc;
        try {//Kudos for closing: http://stackoverflow.com/questions/156508/closing-a-java-fileinputstream
            File tmpfile = File.createTempFile(String.format("txttmp-%s", UUID.randomUUID().toString()), null);
            RandomAccessFile raf = new RandomAccessFile(tmpfile, "rw");
            pdDoc = PDDocument.loadNonSeq(pdf, raf);
            FileWriter writer = new FileWriter(output);
            try {
                PDFTextStripper stripper = new PDFTextStripper();
                int numberOfPages = pdDoc.getNumberOfPages();

                for (int j = 1; j < numberOfPages+1; j++) {
                    stripper.setStartPage(j);
                    stripper.setEndPage(j);
                    writer.write(stripper.getText(pdDoc));
                    writer.flush();
                }
            } finally {
                pdDoc.close();
                raf.close();
                tmpfile.delete();
                writer.close();
            }
        } catch (IOException ioe) {
//            log.warn(String.format("Failed to create txt for file: %s", pdf.getName()), ioe);
        }
    }
 

import org.apache.pdfbox.pdmodel.PDDocument; //导入方法依赖的package包/类
/**
 * Given a pdf document, creates a printable booklet.
 * @param doc original document to be converted
 * @return booklet form of the original document.
 */
public static PDDocument make(PDDocument doc) {
	PDDocument booklet = new PDDocument();
	
	// "effective pages" defines the effective number of pages in the booklet (including blanks)
	int effectivePages = doc.getNumberOfPages() + doc.getNumberOfPages() % 4;
	// "pages" defines the actual number of sides of paper the booklet will use
	int pages = effectivePages / 2;
	
	
	boolean left = false; // smaller page number goes on the left?
	
	for(int i=0; i<pages; i++) {
		int page1 = i;
		int page2 = effectivePages - i - 1;
		
		PDPage page;
		if(left) page = createBookPage(doc, page1, page2);
		else page = createBookPage(doc, page2, page1);
		
		booklet.addPage(page);
		
		left = !left;
	}
	
	return booklet;
}
 

import org.apache.pdfbox.pdmodel.PDDocument; //导入方法依赖的package包/类
/**
 * Selects every 'period-th' page starting with 'start'
 * @param doc document from which pages are selected
 * @param start starting page which is selected
 * @param period periodic interval at which pages a selected
 * @return
 */
public static PageSelector pagesSequence(PDDocument doc, int start, int period) {
	PageSelector br = new PageSelector();
	
	for(int i=start; i<doc.getNumberOfPages(); i+=period)
		br.addPage(doc, i);
	
	return br;
}
 

import org.apache.pdfbox.pdmodel.PDDocument; //导入方法依赖的package包/类
public static PDImageXObject pageToImage(PDDocument doc, int page, int scale) throws IOException {
	// return empty pages if page number excedes number of pages
	if(page >= doc.getNumberOfPages()) 
		return null;
	
	PDFRenderer pdfRenderer = new PDFRenderer(doc);
	BufferedImage bim = pdfRenderer.renderImage(page, scale);
	PDImageXObject img = JPEGFactory.createFromImage(doc, bim);
	
	return img;
}