Crude Java class (depends on pdfbox-1.8.5 and common-logging) to recursively scan directories of PDF files to determine if they contain text or not (or are non-PDF files)

PDFScanner.java

package prpsutil;

import java.io.*;
import java.util.*;
import org.apache.pdfbox.pdmodel.*;
import org.apache.pdfbox.pdmodel.font.*;

public class PDFScanner {

    private static final byte[] expectedMagicNumber = "%PDF".getBytes();

    private static final FileFilter FILES = new FileFilter() {
        @Override 
        public boolean accept(File file) {
            return file.isFile();
        }
    };

    private static final FileFilter DIRECTORIES = new FileFilter() {
        @Override 
        public boolean accept(File file) {
            return file.isDirectory();
        }
    };

    private static class PdfMeta {
        public int pages;
        public boolean textual;
        public long fileSize;
    }

    public static void main(String[] args) {
        if (args.length < 1) {
            throw new IllegalArgumentException("Usage: [input dir]");
        }
        String inDir = args[0];

        File dir = new File(inDir);
        if (dir.isFile()) {
            analyzeFile(dir);
        } else {
            analyzeDirectory(dir);
        }
    }

    static void analyzeDirectory(File dir) {
        File[] files = dir.listFiles(FILES);
        if (files != null) {
            for (File file : files) {
                analyzeFile(file);
            }
        }

        File[] dirs = dir.listFiles(DIRECTORIES);
        if (dirs != null) {
            for (File dirFile : dirs) {
                analyzeDirectory(dirFile);
            }
        }
    } 

    static void analyzeFile(File file) {
        try {
            if (isPdf(file)) {
                PdfMeta meta = analyzePdf(file);
                if (!meta.textual) {
                    System.out.print(file.getAbsolutePath() + ",IMAGEPDF,");
                } else {
                    System.out.print(file.getAbsolutePath() + ",TEXTPDF,");
                }
                System.out.println(meta.pages + "," + meta.fileSize);
            } else {
                System.out.println(file.getAbsolutePath() + ",NONPDF,,");
            }
        } catch (Exception e) {
            System.out.print(file.getAbsolutePath() + ",ERROR,,");
        }
    }

    static boolean isPdf(File file) throws IOException {
        byte[] actualMagicNumber = new byte[expectedMagicNumber.length];
        InputStream in = null;
        try {
            in = new FileInputStream(file);
            in.read(actualMagicNumber);
            for (int i = 0; i < expectedMagicNumber.length; ++i) {
                if (actualMagicNumber[i] != expectedMagicNumber[i]) {
                    return false;
                }
            }
            return true;
        } finally {
            try {
                in.close();
            } catch (IOException e) {
                // ignore
            }
        }
    }

    static PdfMeta analyzePdf(File pdfFile) throws Exception {
        PdfMeta meta = new PdfMeta();
        PDDocument doc = PDDocument.load(pdfFile);
        try {
            List<PDPage> pages = doc.getDocumentCatalog().getAllPages();
            meta.textual = isTextual(pages);
            meta.pages = pages.size();
            meta.fileSize = pdfFile.length();
            return meta;
        } finally {
            doc.close();
        }
    }

    static boolean isTextual(List<PDPage> pages) {
        if (pages != null) {
            for (PDPage page : pages) {
                Map<String, PDFont> pageFonts = page.getResources().getFonts();
                if (!pageFonts.isEmpty()) {
                    return true;
                }
            }
        }
        return false;
    }
}