Search in sources :

Example 1 with Bookmark

use of org.apache.poi.hwpf.usermodel.Bookmark 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;
}
Also used : CharacterRun(org.apache.poi.hwpf.usermodel.CharacterRun) Range(org.apache.poi.hwpf.usermodel.Range) LinkedList(java.util.LinkedList) HWPFDocument(org.apache.poi.hwpf.HWPFDocument) Field(org.apache.poi.hwpf.usermodel.Field) Bookmark(org.apache.poi.hwpf.usermodel.Bookmark) Picture(org.apache.poi.hwpf.usermodel.Picture) ArrayList(java.util.ArrayList) LinkedList(java.util.LinkedList) List(java.util.List) HWPFList(org.apache.poi.hwpf.usermodel.HWPFList)

Example 2 with Bookmark

use of org.apache.poi.hwpf.usermodel.Bookmark in project poi by apache.

the class WordToHtmlConverter method processBookmarks.

@Override
protected void processBookmarks(HWPFDocumentCore wordDocument, Element currentBlock, Range range, int currentTableLevel, List<Bookmark> rangeBookmarks) {
    Element parent = currentBlock;
    for (Bookmark bookmark : rangeBookmarks) {
        Element bookmarkElement = htmlDocumentFacade.createBookmark(bookmark.getName());
        parent.appendChild(bookmarkElement);
        parent = bookmarkElement;
    }
    if (range != null) {
        processCharacters(wordDocument, currentTableLevel, range, parent);
    }
}
Also used : Bookmark(org.apache.poi.hwpf.usermodel.Bookmark) Element(org.w3c.dom.Element)

Example 3 with Bookmark

use of org.apache.poi.hwpf.usermodel.Bookmark 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());
    }
}
Also used : HWPFDocument(org.apache.poi.hwpf.HWPFDocument) Bookmarks(org.apache.poi.hwpf.usermodel.Bookmarks) Bookmark(org.apache.poi.hwpf.usermodel.Bookmark)

Example 4 with Bookmark

use of org.apache.poi.hwpf.usermodel.Bookmark in project poi by apache.

the class WordToFoConverter method processBookmarks.

@Override
protected void processBookmarks(HWPFDocumentCore wordDocument, Element currentBlock, Range range, int currentTableLevel, List<Bookmark> rangeBookmarks) {
    Element parent = currentBlock;
    for (Bookmark bookmark : rangeBookmarks) {
        Element bookmarkElement = foDocumentFacade.createInline();
        final String idName = "bookmark_" + bookmark.getName();
        // make sure ID used once
        if (setId(bookmarkElement, idName)) {
            /*
                 * if it just empty fo:inline without "id" attribute doesn't
                 * making sense to add it to DOM
                 */
            parent.appendChild(bookmarkElement);
            parent = bookmarkElement;
        }
    }
    if (range != null)
        processCharacters(wordDocument, currentTableLevel, range, parent);
}
Also used : Bookmark(org.apache.poi.hwpf.usermodel.Bookmark) Element(org.w3c.dom.Element)

Example 5 with Bookmark

use of org.apache.poi.hwpf.usermodel.Bookmark in project poi by apache.

the class TestBookmarksTables method testReplaceTextAfter.

public void testReplaceTextAfter() {
    HWPFDocument doc = HWPFTestDataSamples.openSampleFile("pageref.doc");
    Bookmark bookmark = doc.getBookmarks().getBookmark(0);
    Range range = new Range(bookmark.getStart(), bookmark.getEnd(), doc);
    range.replaceText("1destin2ation3", true);
    bookmark = doc.getBookmarks().getBookmark(0);
    assertEquals("userref", bookmark.getName());
    assertEquals(27, bookmark.getStart());
    assertEquals(41, bookmark.getEnd());
}
Also used : HWPFDocument(org.apache.poi.hwpf.HWPFDocument) Bookmark(org.apache.poi.hwpf.usermodel.Bookmark) Range(org.apache.poi.hwpf.usermodel.Range)

Aggregations

Bookmark (org.apache.poi.hwpf.usermodel.Bookmark)8 HWPFDocument (org.apache.poi.hwpf.HWPFDocument)6 Range (org.apache.poi.hwpf.usermodel.Range)4 Bookmarks (org.apache.poi.hwpf.usermodel.Bookmarks)2 Element (org.w3c.dom.Element)2 ArrayList (java.util.ArrayList)1 LinkedList (java.util.LinkedList)1 List (java.util.List)1 CharacterRun (org.apache.poi.hwpf.usermodel.CharacterRun)1 Field (org.apache.poi.hwpf.usermodel.Field)1 HWPFList (org.apache.poi.hwpf.usermodel.HWPFList)1 Picture (org.apache.poi.hwpf.usermodel.Picture)1