Examples with HWPFDocument org.apache.poi.hwpf.HWPFDocument used on opensource projects

Example 31 with HWPFDocument

use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.

the class HWPFLister method dumpPapx.

public void dumpPapx(boolean withProperties, boolean withSprms) throws Exception {
    if (_doc instanceof HWPFDocument) {
        System.out.println("binary PAP pages ");
        HWPFDocument doc = (HWPFDocument) _doc;
        byte[] mainStream = _doc.getMainStream();
        PlexOfCps binTable = new PlexOfCps(doc.getTableStream(), doc.getFileInformationBlock().getFcPlcfbtePapx(), doc.getFileInformationBlock().getLcbPlcfbtePapx(), 4);
        List<PAPX> papxs = new ArrayList<PAPX>();
        int length = binTable.length();
        for (int x = 0; x < length; x++) {
            GenericPropertyNode node = binTable.getProperty(x);
            int pageNum = LittleEndian.getInt(node.getBytes());
            int pageOffset = POIFSConstants.SMALLER_BIG_BLOCK_SIZE * pageNum;
            PAPFormattedDiskPage pfkp = new PAPFormattedDiskPage(mainStream, doc.getDataStream(), pageOffset, doc.getTextTable());
            System.out.println("* PFKP: " + pfkp);
            for (PAPX papx : pfkp.getPAPXs()) {
                System.out.println("** " + papx);
                papxs.add(papx);
                if (papx != null && withSprms) {
                    SprmIterator sprmIt = new SprmIterator(papx.getGrpprl(), 2);
                    dumpSprms(sprmIt, "*** ");
                }
            }
        }
        Collections.sort(papxs);
        System.out.println("* Sorted by END");
        for (PAPX papx : papxs) {
            System.out.println("** " + papx);
            if (papx != null && withSprms) {
                SprmIterator sprmIt = new SprmIterator(papx.getGrpprl(), 2);
                dumpSprms(sprmIt, "*** ");
            }
        }
    }
    for (PAPX papx : _doc.getParagraphTable().getParagraphs()) {
        System.out.println(papx);
        if (withProperties) {
            Paragraph paragraph = Paragraph.newParagraph(_doc.getOverallRange(), papx);
            System.out.println(paragraph.getProps());
        }
        SprmIterator sprmIt = new SprmIterator(papx.getGrpprl(), 2);
        dumpSprms(sprmIt, "\t");
    }
}

Example 32 with HWPFDocument

use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.

the class TestTextPieceTable method testUnicodeParts.

/**
     * Check that we do the positions correctly when working with a mix ascii,
     * unicode file
     */
@Test
public void testUnicodeParts() throws Exception {
    HWPFDocument doc = HWPFTestDataSamples.openSampleFile("HeaderFooterUnicode.doc");
    TextPieceTable tbl = doc.getTextTable();
    // In three bits, split every 512 bytes
    assertEquals(3, tbl.getTextPieces().size());
    TextPiece tpA = tbl.getTextPieces().get(0);
    TextPiece tpB = tbl.getTextPieces().get(1);
    TextPiece tpC = tbl.getTextPieces().get(2);
    assertTrue(tpA.isUnicode());
    assertTrue(tpB.isUnicode());
    assertTrue(tpC.isUnicode());
    assertEquals(256, tpA.characterLength());
    assertEquals(256, tpB.characterLength());
    assertEquals(19, tpC.characterLength());
    assertEquals(512, tpA.bytesLength());
    assertEquals(512, tpB.bytesLength());
    assertEquals(38, tpC.bytesLength());
    assertEquals(0, tpA.getStart());
    assertEquals(256, tpA.getEnd());
    assertEquals(256, tpB.getStart());
    assertEquals(512, tpB.getEnd());
    assertEquals(512, tpC.getStart());
    assertEquals(531, tpC.getEnd());
    // Save and re-load
    HWPFDocument docB = saveAndReload(doc);
    tbl = docB.getTextTable();
    assertEquals(3, tbl.getTextPieces().size());
    tpA = tbl.getTextPieces().get(0);
    tpB = tbl.getTextPieces().get(1);
    tpC = tbl.getTextPieces().get(2);
    assertTrue(tpA.isUnicode());
    assertTrue(tpB.isUnicode());
    assertTrue(tpC.isUnicode());
    assertEquals(256, tpA.characterLength());
    assertEquals(256, tpB.characterLength());
    assertEquals(19, tpC.characterLength());
    assertEquals(512, tpA.bytesLength());
    assertEquals(512, tpB.bytesLength());
    assertEquals(38, tpC.bytesLength());
    assertEquals(0, tpA.getStart());
    assertEquals(256, tpA.getEnd());
    assertEquals(256, tpB.getStart());
    assertEquals(512, tpB.getEnd());
    assertEquals(512, tpC.getStart());
    assertEquals(531, tpC.getEnd());
}

Example 33 with HWPFDocument

use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.

the class TestTextPieceTable method saveAndReload.

protected HWPFDocument saveAndReload(HWPFDocument doc) throws Exception {
    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    doc.write(baos);
    return new HWPFDocument(new ByteArrayInputStream(baos.toByteArray()));
}

Example 34 with HWPFDocument

use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.

the class TestTextPieceTable method test56549_CharIndexRange.

@Test
public void test56549_CharIndexRange() {
    HWPFDocument doc = HWPFTestDataSamples.openSampleFile("ThreeColHeadFoot.doc");
    // there is one range from 2048 - 2387
    TextPieceTable tbl = doc.getTextTable();
    int[][] range = tbl.getCharIndexRanges(0, 0);
    assertEquals(0, range.length);
    range = tbl.getCharIndexRanges(0, 1);
    assertEquals(0, range.length);
    range = tbl.getCharIndexRanges(0, 338);
    assertEquals(0, range.length);
    range = tbl.getCharIndexRanges(0, 339);
    assertEquals(0, range.length);
    range = tbl.getCharIndexRanges(0, 340);
    assertEquals(0, range.length);
    range = tbl.getCharIndexRanges(2030, 2048);
    assertEquals(0, range.length);
    range = tbl.getCharIndexRanges(2030, 2049);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 0, 1 }, range[0]);
    range = tbl.getCharIndexRanges(2048, 2049);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 0, 1 }, range[0]);
    range = tbl.getCharIndexRanges(2048, 2300);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 0, 252 }, range[0]);
    range = tbl.getCharIndexRanges(2049, 2300);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 1, 252 }, range[0]);
    range = tbl.getCharIndexRanges(2049, 2300);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 1, 252 }, range[0]);
    range = tbl.getCharIndexRanges(2049, 2387);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 1, 339 }, range[0]);
    range = tbl.getCharIndexRanges(2049, 2388);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 1, 339 }, range[0]);
    range = tbl.getCharIndexRanges(2387, 2388);
    assertEquals(1, range.length);
    assertArrayEquals(new int[] { 339, 339 }, range[0]);
}

Example 35 with HWPFDocument

use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.

the class TestSprms method testInnerTable.

/**
     * Test correct processing of "sprmPItap" (0x6649) and "sprmPFInTable"
     * (0x2416)
     */
public void testInnerTable() throws Exception {
    InputStream resourceAsStream = POIDataSamples.getDocumentInstance().openResourceAsStream("innertable.doc");
    HWPFDocument hwpfDocument = new HWPFDocument(resourceAsStream);
    resourceAsStream.close();
    testInnerTable(hwpfDocument);
    hwpfDocument = reload(hwpfDocument);
    testInnerTable(hwpfDocument);
}