Dear Apache POI Team,

Please consider a problem: whenever MS-Word document with footnotes/endnotes is being parsed with XWPFWordExtractor, information on the location of footnote/endnote references is lost. This information loss is clearly observed in, for example, Apache Tika output.

To reproduce a problem, please insert the following code to TestXWPFWordExtractor.testFootnotes:

    java.io.FileWriter w = new java.io.FileWriter(new java.io.File(System.getProperty("user.home"), "footnotes.output.txt"));
    try {
      w.write(extractor.getText());
    } finally {
      w.close();
    }

then run tests and inspect the content of "footnotes.output.txt" - it contains "Eto ochen prostoy text so snoskoy", where between "prostoy" and "text" there should be a footnote reference (and it is lost).

SOLUTION:
I suggest to introduce additional markup like [footnoteRef:num], [endnoteRef:num], which will allow applications to correctly render footnote references.

Please, see commit details.