/**
* Constructor sets the input and output file and convert the pdf, docx and doc files to text .
*
* @param infile,outfile
* @return
*/
public TextConvertor(String infile, String outfile) {
try {
File input = new File(infile); // The file from where you would like to extract
FileInputStream fis = new FileInputStream(input.getAbsolutePath());
int x = fis.read();
int y = fis.read();
fis = new FileInputStream(input.getAbsolutePath());
if (x == 37 && y == 80) {
filetype = "pdf";
pd = PDDocument.load(input);
PDF2Text(outfile);
} else if (x == 80 && y == 75) {
filetype = "docx";
dx = new XWPFDocument(fis);
DOCX2Text(outfile);
} else if (x == 208 && y == 207) {
filetype = "doc";
dc = new HWPFDocument(fis);
DOC2Text(outfile);
}
} catch (Exception e) {
e.printStackTrace();
}
}
public static String[] getFiles(String dirname, FilenameFilter filter) {
File dir = new File(dirname);
String[] files = dir.list(filter);
return files;
}
