use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.
the class TestTableRow method testInnerTableCellsDetection.
@Test
public void testInnerTableCellsDetection() throws IOException {
InputStream is = SAMPLES.openResourceAsStream("innertable.doc");
HWPFDocument hwpfDocument = new HWPFDocument(is);
is.close();
hwpfDocument.getRange();
Range documentRange = hwpfDocument.getRange();
Paragraph startOfInnerTable = documentRange.getParagraph(6);
Table innerTable = documentRange.getTable(startOfInnerTable);
assertEquals(2, innerTable.numRows());
TableRow tableRow = innerTable.getRow(0);
assertEquals(2, tableRow.numCells());
hwpfDocument.close();
}
use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.
the class EmbeddedObjects method main.
@SuppressWarnings("unused")
public static void main(String[] args) throws Exception {
POIFSFileSystem fs = new POIFSFileSystem(new FileInputStream(args[0]));
HSSFWorkbook workbook = new HSSFWorkbook(fs);
for (HSSFObjectData obj : workbook.getAllEmbeddedObjects()) {
//the OLE2 Class Name of the object
String oleName = obj.getOLE2ClassName();
DirectoryNode dn = (obj.hasDirectoryEntry()) ? (DirectoryNode) obj.getDirectory() : null;
Closeable document = null;
if (oleName.equals("Worksheet")) {
document = new HSSFWorkbook(dn, fs, false);
} else if (oleName.equals("Document")) {
document = new HWPFDocument(dn);
} else if (oleName.equals("Presentation")) {
document = new HSLFSlideShow(dn);
} else {
if (dn != null) {
// The DirectoryEntry is a DocumentNode. Examine its entries to find out what it is
for (Entry entry : dn) {
String name = entry.getName();
}
} else {
// There is no DirectoryEntry
// Recover the object's data from the HSSFObjectData instance.
byte[] objectData = obj.getObjectData();
}
}
if (document != null) {
document.close();
}
}
workbook.close();
}
use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.
the class LoadEmbedded method loadEmbedded.
public static void loadEmbedded(HSSFWorkbook workbook) throws IOException {
for (HSSFObjectData obj : workbook.getAllEmbeddedObjects()) {
//the OLE2 Class Name of the object
String oleName = obj.getOLE2ClassName();
if (oleName.equals("Worksheet")) {
DirectoryNode dn = (DirectoryNode) obj.getDirectory();
HSSFWorkbook embeddedWorkbook = new HSSFWorkbook(dn, false);
embeddedWorkbook.close();
} else if (oleName.equals("Document")) {
DirectoryNode dn = (DirectoryNode) obj.getDirectory();
HWPFDocument embeddedWordDocument = new HWPFDocument(dn);
embeddedWordDocument.close();
} else if (oleName.equals("Presentation")) {
DirectoryNode dn = (DirectoryNode) obj.getDirectory();
SlideShow<?, ?> embeddedSlieShow = new HSLFSlideShow(dn);
embeddedSlieShow.close();
} else {
if (obj.hasDirectoryEntry()) {
// The DirectoryEntry is a DocumentNode. Examine its entries to find out what it is
DirectoryNode dn = (DirectoryNode) obj.getDirectory();
for (Entry entry : dn) {
//System.out.println(oleName + "." + entry.getName());
}
} else {
// There is no DirectoryEntry
// Recover the object's data from the HSSFObjectData instance.
byte[] objectData = obj.getObjectData();
}
}
}
}
use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.
the class AbstractWordConverter method processCharacters.
protected boolean processCharacters(final HWPFDocumentCore wordDocument, final int currentTableLevel, final Range range, final Element block) {
if (range == null)
return false;
boolean haveAnyText = false;
/*
* In text there can be fields, bookmarks, may be other structures (code
* below allows extension). Those structures can overlaps, so either we
* should process char-by-char (slow) or find a correct way to
* reconstruct the structure of range -- sergey
*/
List<Structure> structures = new LinkedList<Structure>();
if (wordDocument instanceof HWPFDocument) {
final HWPFDocument doc = (HWPFDocument) wordDocument;
Map<Integer, List<Bookmark>> rangeBookmarks = doc.getBookmarks().getBookmarksStartedBetween(range.getStartOffset(), range.getEndOffset());
if (rangeBookmarks != null) {
for (List<Bookmark> lists : rangeBookmarks.values()) {
for (Bookmark bookmark : lists) {
if (!bookmarkStack.contains(bookmark))
addToStructures(structures, new Structure(bookmark));
}
}
}
// TODO: dead fields?
int skipUntil = -1;
for (int c = 0; c < range.numCharacterRuns(); c++) {
CharacterRun characterRun = range.getCharacterRun(c);
if (characterRun == null)
throw new AssertionError();
if (characterRun.getStartOffset() < skipUntil)
continue;
String text = characterRun.text();
if (text == null || text.length() == 0 || text.charAt(0) != FIELD_BEGIN_MARK)
continue;
Field aliveField = ((HWPFDocument) wordDocument).getFields().getFieldByStartOffset(FieldsDocumentPart.MAIN, characterRun.getStartOffset());
if (aliveField != null) {
addToStructures(structures, new Structure(aliveField));
} else {
int[] separatorEnd = tryDeadField_lookupFieldSeparatorEnd(wordDocument, range, c);
if (separatorEnd != null) {
addToStructures(structures, new Structure(new DeadFieldBoundaries(c, separatorEnd[0], separatorEnd[1]), characterRun.getStartOffset(), range.getCharacterRun(separatorEnd[1]).getEndOffset()));
c = separatorEnd[1];
}
}
}
}
structures = new ArrayList<Structure>(structures);
Collections.sort(structures);
int previous = range.getStartOffset();
for (Structure structure : structures) {
if (structure.start != previous) {
Range subrange = new Range(previous, structure.start, range) {
@Override
public String toString() {
return "BetweenStructuresSubrange " + super.toString();
}
};
processCharacters(wordDocument, currentTableLevel, subrange, block);
}
if (structure.structure instanceof Bookmark) {
// other bookmarks with same boundaries
List<Bookmark> bookmarks = new LinkedList<Bookmark>();
for (Bookmark bookmark : ((HWPFDocument) wordDocument).getBookmarks().getBookmarksStartedBetween(structure.start, structure.start + 1).values().iterator().next()) {
if (bookmark.getStart() == structure.start && bookmark.getEnd() == structure.end) {
bookmarks.add(bookmark);
}
}
bookmarkStack.addAll(bookmarks);
try {
int end = Math.min(range.getEndOffset(), structure.end);
Range subrange = new Range(structure.start, end, range) {
@Override
public String toString() {
return "BookmarksSubrange " + super.toString();
}
};
processBookmarks(wordDocument, block, subrange, currentTableLevel, bookmarks);
} finally {
bookmarkStack.removeAll(bookmarks);
}
} else if (structure.structure instanceof Field) {
Field field = (Field) structure.structure;
processField((HWPFDocument) wordDocument, range, currentTableLevel, field, block);
} else if (structure.structure instanceof DeadFieldBoundaries) {
DeadFieldBoundaries boundaries = (DeadFieldBoundaries) structure.structure;
processDeadField(wordDocument, block, range, currentTableLevel, boundaries.beginMark, boundaries.separatorMark, boundaries.endMark);
} else {
throw new UnsupportedOperationException("NYI: " + structure.structure.getClass());
}
previous = Math.min(range.getEndOffset(), structure.end);
}
if (previous != range.getStartOffset()) {
if (previous > range.getEndOffset()) {
logger.log(POILogger.WARN, "Latest structure in ", range, " ended at #" + previous, " after range boundaries [", range.getStartOffset() + "; " + range.getEndOffset(), ")");
return true;
}
if (previous < range.getEndOffset()) {
Range subrange = new Range(previous, range.getEndOffset(), range) {
@Override
public String toString() {
return "AfterStructureSubrange " + super.toString();
}
};
processCharacters(wordDocument, currentTableLevel, subrange, block);
}
return true;
}
for (int c = 0; c < range.numCharacterRuns(); c++) {
CharacterRun characterRun = range.getCharacterRun(c);
if (characterRun == null)
throw new AssertionError();
if (wordDocument instanceof HWPFDocument && ((HWPFDocument) wordDocument).getPicturesTable().hasPicture(characterRun)) {
HWPFDocument newFormat = (HWPFDocument) wordDocument;
Picture picture = newFormat.getPicturesTable().extractPicture(characterRun, true);
processImage(block, characterRun.text().charAt(0) == 0x01, picture);
continue;
}
String text = characterRun.text();
if (text.isEmpty())
continue;
if (characterRun.isSpecialCharacter()) {
if (text.charAt(0) == SPECCHAR_AUTONUMBERED_FOOTNOTE_REFERENCE && (wordDocument instanceof HWPFDocument)) {
HWPFDocument doc = (HWPFDocument) wordDocument;
processNoteAnchor(doc, characterRun, block);
continue;
}
if (text.charAt(0) == SPECCHAR_DRAWN_OBJECT && (wordDocument instanceof HWPFDocument)) {
HWPFDocument doc = (HWPFDocument) wordDocument;
processDrawnObject(doc, characterRun, block);
continue;
}
if (characterRun.isOle2() && (wordDocument instanceof HWPFDocument)) {
HWPFDocument doc = (HWPFDocument) wordDocument;
processOle2(doc, characterRun, block);
continue;
}
if (characterRun.isSymbol() && (wordDocument instanceof HWPFDocument)) {
HWPFDocument doc = (HWPFDocument) wordDocument;
processSymbol(doc, characterRun, block);
continue;
}
}
if (text.charAt(0) == FIELD_BEGIN_MARK) {
if (wordDocument instanceof HWPFDocument) {
Field aliveField = ((HWPFDocument) wordDocument).getFields().getFieldByStartOffset(FieldsDocumentPart.MAIN, characterRun.getStartOffset());
if (aliveField != null) {
processField(((HWPFDocument) wordDocument), range, currentTableLevel, aliveField, block);
int continueAfter = aliveField.getFieldEndOffset();
while (c < range.numCharacterRuns() && range.getCharacterRun(c).getEndOffset() <= continueAfter) c++;
if (c < range.numCharacterRuns())
c--;
continue;
}
}
int skipTo = tryDeadField(wordDocument, range, currentTableLevel, c, block);
if (skipTo != c) {
c = skipTo;
continue;
}
continue;
}
if (text.charAt(0) == FIELD_SEPARATOR_MARK) {
// shall not appear without FIELD_BEGIN_MARK
continue;
}
if (text.charAt(0) == FIELD_END_MARK) {
// shall not appear without FIELD_BEGIN_MARK
continue;
}
if (characterRun.isSpecialCharacter() || characterRun.isObj() || characterRun.isOle2()) {
continue;
}
if (text.endsWith("\r") || (text.charAt(text.length() - 1) == BEL_MARK && currentTableLevel != Integer.MIN_VALUE))
text = text.substring(0, text.length() - 1);
{
// line breaks
StringBuilder stringBuilder = new StringBuilder();
for (char charChar : text.toCharArray()) {
if (charChar == 11) {
if (stringBuilder.length() > 0) {
outputCharacters(block, characterRun, stringBuilder.toString());
stringBuilder.setLength(0);
}
processLineBreak(block, characterRun);
} else if (charChar == 30) {
// Non-breaking hyphens are stored as ASCII 30
stringBuilder.append(UNICODECHAR_NONBREAKING_HYPHEN);
} else if (charChar == 31) {
// Non-required hyphens to zero-width space
stringBuilder.append(UNICODECHAR_ZERO_WIDTH_SPACE);
} else if (charChar >= 0x20 || charChar == 0x09 || charChar == 0x0A || charChar == 0x0D) {
stringBuilder.append(charChar);
}
}
if (stringBuilder.length() > 0) {
outputCharacters(block, characterRun, stringBuilder.toString());
stringBuilder.setLength(0);
}
}
haveAnyText |= text.trim().length() != 0;
}
return haveAnyText;
}
use of org.apache.poi.hwpf.HWPFDocument in project poi by apache.
the class HWPFLister method dumpBookmarks.
private void dumpBookmarks() {
if (!(_doc instanceof HWPFDocument)) {
System.out.println("Word 95 not supported so far");
return;
}
HWPFDocument document = (HWPFDocument) _doc;
Bookmarks bookmarks = document.getBookmarks();
for (int b = 0; b < bookmarks.getBookmarksCount(); b++) {
Bookmark bookmark = bookmarks.getBookmark(b);
System.out.println("[" + bookmark.getStart() + "; " + bookmark.getEnd() + "): " + bookmark.getName());
}
}
Aggregations