private void analyzeWordFile(File fileName) throws IOException {
XWPFDocument doc = new XWPFDocument(new FileInputStream(fileName));
XWPFWordExtractor extractor = new XWPFWordExtractor(doc);
String[] docLines = extractor.getText().split("\n");
for (int i = 0; i < docLines.length; i++) {
checkForKeywords(fileName.getName(), docLines[i], i);
}
}
public void ReadAll(String path, String filename) {
try {
FileInputStream fis = new FileInputStream(path + filename + ".doc");
XWPFDocument xdoc = new XWPFDocument(OPCPackage.open(fis));
XWPFWordExtractor extractor = new XWPFWordExtractor(xdoc);
System.out.println(extractor.getText());
} catch (Exception ex) {
ex.printStackTrace();
}
}
private String sampleDocx(String resourceName) {
String content = null;
try {
InputStream input = new BOMInputStream(TestUtil.class.getResourceAsStream(resourceName));
XWPFDocument document = new XWPFDocument(OPCPackage.open(input));
@SuppressWarnings("resource")
XWPFWordExtractor wordExtractor = new XWPFWordExtractor(document);
content = wordExtractor.getText();
} catch (Exception exep) {
exep.printStackTrace();
}
return content;
}
public ArrayList<String> getText(File file) {
String text = "";
try {
XWPFDocument docx = new XWPFDocument(new FileInputStream(file));
XWPFWordExtractor we = new XWPFWordExtractor(docx);
text = we.getText();
we.close();
} catch (IOException e) {
e.printStackTrace();
}
return new ArrayList<String>(Arrays.asList(text.split("\n")));
}
/**
* save the converted text (without any processing) to the given file.
*
* @param filename
* @return
*/
public void DOCX2Text(String filename) {
try {
File output =
new File(filename); // The text file where you are going to store the extracted data
XWPFWordExtractor ex = new XWPFWordExtractor(dx);
String fileData = ex.getText();
wr = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(output)));
wr.write(fileData);
wr.close();
} catch (Exception e) {
e.printStackTrace();
}
}
/**
* Extracts text from a Word document and stores it in the document.
*
* @param inputStream An input stream pointing to the Word document to be read.
* @throws IOException
*/
private static char[] loadMSWordDocx(InputStream inputStream) throws IOException {
XWPFDocument docx = new XWPFDocument(inputStream);
XWPFWordExtractor extractor = new XWPFWordExtractor(docx);
return extractor.getText().toCharArray();
}
