import org.openxmlformats.schemas.wordprocessingml.x2006.main.*;

sclass WordDocumentTextReplacer {

  replace Run with XWPFRun.

  replace Paragraph with XWPFParagraph.

  File inFile, outFile;

  S regularExpression;

  swappable S getReplacement(S text, LS groups) { ret "[REPLACED]"; }

  run {

    assertNempty(+regularExpression);

    assertNotNull(+inFile);

    assertNotNull(+outFile);

    XWPFDocument doc = loadDocx(print("Loading", inFile));

    print("Document loaded");

    new XWPFDocument docOut;

    CTBody body = doc.getDocument().getBody();

    CTSectPr sectPr = body.getSectPr();

    CTBody bodyOut = docOut.getDocument().getBody();

    bodyOut.setSectPr(sectPr);

    L<Paragraph> paragraphs = doc.getParagraphs();

    for (Paragraph para : paragraphs) {

      Paragraph paraOut = docOut.createParagraph();

      L<Run> runs = para.getRuns();

      //print(n2(runs, "run"));

      new LPair<Run, S> runs2;

      for (Run r : runs)

        addPair(runs2, r, unnull(r.getText(0)));

      S fullText = join(pairsB(runs2));

      //print(quote(fullText));

      //printIfNempty(regexpExtractAll(regexp, fullText));

      for (int i = 0; i < l(runs2); i++) {

        for (int j = i+1; j <= l(runs2); j++) {

          S text = join(pairsB(subList(runs2, i, j)));

          print(+text);

          IntRange range = regexpFindRange(regularExpression, text);

          if (range == null) continue; // no match

          //print("Match: " + substring(text, range));

          // we have a match, find out run indices

          // skip runs left of match

          while (i < l(runs2) && range.start >= l(runs2.get(i).b)) {

            range = shiftIntRange(range, -l(runs2.get(i).b));

            i++;

          }

          text = join(pairsB(subList(runs2, i, j)));

          print("Found match: " + substring(text, range));

          // replace all matched runs with one or two runs at i

          removeSubList(runs2, i+1, j);

          Run run = runs2.get(i).a;

          S found = substring(text, range);

          LS groups = regexpFirstGroups(regularExpression, found);

          S replacement = getReplacement(found, groups);

          print("Replacing with: " + replacement);

          S text1 = takeFirst(text, range.start) + replacement;

          if (nempty(text1)) {

            runs2.add(i, pair(run, text1));

            ++i;

          }

          S rest = substring(text, range.end);

          if (empty(rest))

            runs2.remove(i--);

          else

            runs2.get(i).b = rest;

        }

      }

      //for (int k = l(runs)-1; k >= 0; k--) para.removeRun(k);

      //print("Adding " + nRuns(runs2));

      for (Pair<Run, S> p : runs2) {

        //paraOut.addRun(run);

        Run run = p.a;

        Run runOut = paraOut.createRun();

        runOut.setText(p.b);

        // copy run attributes

        runOut.setColor(run.getColor());

        runOut.setFontFamily(run.getFontFamily());

        runOut.setFontSize(run.getFontSize());

        runOut.setBold(run.isBold());

        runOut.setItalic(run.isItalic());

        runOut.setUnderline(run.getUnderline());

        paraOut.addRun(runOut);

      }

    }

    saveDocx(docOut, outFile);

    printFileInfo(outFile);

  }

}