public void TestOrderedLists()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Lists.doc");
Range r = doc.GetRange();
Assert.AreEqual(40, r.NumParagraphs);
Assert.AreEqual("Next up is an ordered list:\r", r.GetParagraph(5).Text);
Assert.AreEqual("Ordered list 1\r", r.GetParagraph(6).Text);
Assert.AreEqual("OL 2\r", r.GetParagraph(7).Text);
Assert.AreEqual("OL 3\r", r.GetParagraph(8).Text);
Assert.AreEqual("Now for an un-ordered list with a different bullet style:\r", r.GetParagraph(9).Text);
Assert.AreEqual(9, r.GetParagraph(5).GetLvl());
Assert.AreEqual(9, r.GetParagraph(6).GetLvl());
Assert.AreEqual(9, r.GetParagraph(7).GetLvl());
Assert.AreEqual(9, r.GetParagraph(8).GetLvl());
Assert.AreEqual(9, r.GetParagraph(9).GetLvl());
Assert.AreEqual(0, r.GetParagraph(5).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(6).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(7).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(8).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(9).GetIlvl());
}
public void Test49820()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Bug49820.doc");
Range documentRange = doc.GetRange();
StyleSheet styleSheet = doc.GetStyleSheet();
// JUnit asserts
assertLevels(documentRange, styleSheet, 0, 0, 0);
assertLevels(documentRange, styleSheet, 1, 1, 1);
assertLevels(documentRange, styleSheet, 2, 2, 2);
assertLevels(documentRange, styleSheet, 3, 3, 3);
assertLevels(documentRange, styleSheet, 4, 4, 4);
assertLevels(documentRange, styleSheet, 5, 5, 5);
assertLevels(documentRange, styleSheet, 6, 6, 6);
assertLevels(documentRange, styleSheet, 7, 7, 7);
assertLevels(documentRange, styleSheet, 8, 8, 8);
assertLevels(documentRange, styleSheet, 9, 9, 9);
assertLevels(documentRange, styleSheet, 10, 9, 0);
assertLevels(documentRange, styleSheet, 11, 9, 4);
// output to console
for (int i = 0; i < documentRange.NumParagraphs; i++)
{
Paragraph par = documentRange.GetParagraph(i);
int styleLvl = styleSheet.GetParagraphStyle(par.GetStyleIndex()).GetLvl();
int parLvl = par.GetLvl();
Console.WriteLine("Style level: " + styleLvl + ", paragraph level: " + parLvl + ", text: " + par.Text);
}
}
private static String GetHtmlText(String sampleFileName,
bool emulatePictureStorage)
{
HWPFDocument hwpfDocument = new HWPFDocument(POIDataSamples
.GetDocumentInstance().OpenResourceAsStream(sampleFileName));
XmlDocument newDocument = new XmlDocument();
WordToHtmlConverter wordToHtmlConverter = new WordToHtmlConverter(
newDocument);
if (emulatePictureStorage)
{
//wordToHtmlConverter.SetPicturesManager( new PicturesManager()
//{
// public String SavePicture( byte[] content,
// PictureType pictureType, String suggestedName )
// {
// return suggestedName;
// }
//} );
}
wordToHtmlConverter.ProcessDocument(hwpfDocument);
;
return(wordToHtmlConverter.Document.InnerXml);
}
public void TestAsciiParts()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("ThreeColHeadFoot.doc");
TextPieceTable tbl = doc.TextTable;
// All ascii, so stored in one big lump
Assert.AreEqual(1, tbl.TextPieces.Count);
TextPiece tp = (TextPiece)tbl.TextPieces[0];
Assert.AreEqual(0, tp.Start);
Assert.AreEqual(339, tp.End);
Assert.AreEqual(339, tp.CharacterLength);
Assert.AreEqual(339, tp.BytesLength);
Assert.IsTrue(tp.GetStringBuilder().ToString().StartsWith("This is a sample word document"));
// Save and re-load
HWPFDocument docB = SaveAndReload(doc);
tbl = docB.TextTable;
Assert.AreEqual(1, tbl.TextPieces.Count);
tp = (TextPiece)tbl.TextPieces[0];
Assert.AreEqual(0, tp.Start);
Assert.AreEqual(339, tp.End);
Assert.AreEqual(339, tp.CharacterLength);
Assert.AreEqual(339, tp.BytesLength);
Assert.IsTrue(tp.GetStringBuilder().ToString().StartsWith("This is a sample word document"));
}
public void TestDocStructure()
{
HWPFDocument daDoc = HWPFTestDataSamples.OpenSampleFile(illustrativeDocFile);
Range range = daDoc.GetRange();
Assert.AreEqual(1, range.NumSections);
Section section = range.GetSection(0);
Assert.AreEqual(3, section.NumParagraphs);
Paragraph para = section.GetParagraph(2);
Assert.AreEqual(originalText, para.Text);
Assert.AreEqual(3, para.NumCharacterRuns);
String text =
para.GetCharacterRun(0).Text +
para.GetCharacterRun(1).Text +
para.GetCharacterRun(2).Text
;
Assert.AreEqual(originalText, text);
Assert.AreEqual(insertionPoint, para.StartOffset);
}
public void Init(string path)
{
if (path.EndsWith(".doc"))
{
HWPFDocument hwpf;
using (FileStream file = new FileStream(path, FileMode.Open, FileAccess.Read))
{
hwpf = new HWPFDocument(file);
}
this.textBody = hwpf.Text.ToString();
}
else if (path.EndsWith(".docx"))
{
XWPFDocument xwpf;
using (FileStream file = new FileStream(path, FileMode.Open, FileAccess.Read))
{
xwpf = new XWPFDocument(file);
}
XWPFWordExtractor ex = new XWPFWordExtractor(xwpf);
this.textBody = ex.Text;
}
textBody = textBody.Replace("（", "(").Replace("）", ")");
}
public void TestIndentedText()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Lists.doc");
Range r = doc.GetRange();
Assert.AreEqual(40, r.NumParagraphs);
Assert.AreEqual("Finally we want some indents, to tell the difference\r", r.GetParagraph(34).Text);
Assert.AreEqual("Indented once\r", r.GetParagraph(35).Text);
Assert.AreEqual("Indented twice\r", r.GetParagraph(36).Text);
Assert.AreEqual("Indented three times\r", r.GetParagraph(37).Text);
Assert.AreEqual("The end!\r", r.GetParagraph(38).Text);
Assert.AreEqual(9, r.GetParagraph(34).GetLvl());
Assert.AreEqual(9, r.GetParagraph(35).GetLvl());
Assert.AreEqual(9, r.GetParagraph(36).GetLvl());
Assert.AreEqual(9, r.GetParagraph(37).GetLvl());
Assert.AreEqual(9, r.GetParagraph(38).GetLvl());
Assert.AreEqual(9, r.GetParagraph(39).GetLvl());
Assert.AreEqual(0, r.GetParagraph(34).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(35).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(36).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(37).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(38).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(39).GetIlvl());
// TODO Test the indent
}
protected override void ProcessEndnoteAutonumbered(HWPFDocument wordDocument,
int noteIndex, XmlElement block, Range endnoteTextRange)
{
String textIndex;// = (internalLinkCounter.incrementAndGet()).ToString();
lock (objLinkCounter)
{
internalLinkCounter++;
textIndex = internalLinkCounter.ToString();
}
String forwardLinkName = "endnote_" + textIndex;
String backwardLinkName = "endnote_back_" + textIndex;
XmlElement forwardLink = foDocumentFacade
.CreateBasicLinkInternal(forwardLinkName);
forwardLink.AppendChild(CreateNoteInline(textIndex));
SetId(forwardLink, backwardLinkName);
block.AppendChild(forwardLink);
XmlElement endnote = foDocumentFacade.CreateBlock();
XmlElement backwardLink = foDocumentFacade
.CreateBasicLinkInternal(backwardLinkName);
backwardLink.AppendChild(CreateNoteInline(textIndex + " "));
SetId(backwardLink, forwardLinkName);
endnote.AppendChild(backwardLink);
ProcessCharacters(wordDocument, int.MinValue, endnoteTextRange, endnote);
WordToFoUtils.CompactInlines(endnote);
this.endnotes.Add(endnote);
}
public void TestDifferentImages()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("testPictures.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.IsNotNull(pics);
Assert.AreEqual(7, pics.Count);
for (int i = 0; i < pics.Count; i++)
{
Picture pic = (Picture)pics[i];
Assert.IsNotNull(pic.SuggestFileExtension());
Assert.IsNotNull(pic.SuggestFullFileName());
}
Assert.AreEqual("jpg", pics[0].SuggestFileExtension());
Assert.AreEqual("image/jpeg", pics[0].MimeType);
Assert.AreEqual("jpg", pics[1].SuggestFileExtension());
Assert.AreEqual("image/jpeg", pics[1].MimeType);
Assert.AreEqual("png", pics[3].SuggestFileExtension());
Assert.AreEqual("image/png", pics[3].MimeType);
Assert.AreEqual("png", pics[4].SuggestFileExtension());
Assert.AreEqual("image/png", pics[4].MimeType);
Assert.AreEqual("wmf", pics[5].SuggestFileExtension());
Assert.AreEqual("image/x-wmf", pics[5].MimeType);
Assert.AreEqual("jpg", pics[6].SuggestFileExtension());
Assert.AreEqual("image/jpeg", pics[6].MimeType);
}
public void TestEmfComplexImage()
{
// Commenting out this Test case temporarily. The file emf_2003_image does not contain any
// pictures. Instead it has an office drawing object. Need to rewrite this Test after
// revisiting the implementation of office drawing objects.
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Bug41898.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.IsNotNull(pics);
Assert.AreEqual(1, pics.Count);
Picture pic = pics[0];
Assert.IsNotNull(pic.SuggestFileExtension());
Assert.IsNotNull(pic.SuggestFullFileName());
// This one's tricky
// TODO: Fix once we've sorted bug #41898
Assert.IsNotNull(pic.GetContent());
Assert.IsNotNull(pic.GetRawContent());
// These are probably some sort of offSet, need to figure them out
Assert.AreEqual(4, pic.Size);
Assert.AreEqual((uint)0x80000000, LittleEndian.GetUInt(pic.GetContent()));
Assert.AreEqual((uint)0x80000000, LittleEndian.GetUInt(pic.GetRawContent()));
}
public void TestReadWrite()
{
// This document is widely available on the internet as "blair.doc".
// I tried stripping the content and saving the document but my version
// of Word (from Office XP) strips this table out.
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("saved-by-table.doc");
// Check what we just Read.
for(int i=0;i<expected.Count;i++)
{
Assert.AreEqual(expected[i],doc.GetSavedByTable().GetEntries()[i], "List of saved-by entries was not as expected");
}
// Now write the entire document out, and read it back in...
MemoryStream byteStream = new MemoryStream();
doc.Write(byteStream);
Stream copyStream = new MemoryStream(byteStream.ToArray());
HWPFDocument copy = new HWPFDocument(copyStream);
// And check again.
for (int i = 0; i < expected.Count; i++)
{
Assert.AreEqual(
expected[i], copy.GetSavedByTable().GetEntries()[i], "List of saved-by entries was incorrect after writing");
}
}
/**
* Test that we can replace text in our Range with Unicode text.
*/
public void TestRangeReplacementAll()
{
HWPFDocument daDoc = HWPFTestDataSamples.OpenSampleFile(illustrativeDocFile);
Range range = daDoc.GetRange();
Assert.AreEqual(1, range.NumSections);
Section section = range.GetSection(0);
Assert.AreEqual(5, section.NumParagraphs);
Paragraph para = section.GetParagraph(2);
String text = para.Text;
Assert.AreEqual(originalText, text);
range.ReplaceText(searchText, ReplacementText);
Assert.AreEqual(1, range.NumSections);
section = range.GetSection(0);
Assert.AreEqual(5, section.NumParagraphs);
para = section.GetParagraph(2);
text = para.Text;
Assert.AreEqual(expectedText2, text);
para = section.GetParagraph(3);
text = para.Text;
Assert.AreEqual(expectedText3, text);
}
protected void ProcessDrawnObject(HWPFDocument doc,
CharacterRun characterRun, XmlElement block)
{
if (GetPicturesManager() == null)
{
return;
}
// TODO: support headers
OfficeDrawing officeDrawing = doc.GetOfficeDrawingsMain().GetOfficeDrawingAt(characterRun.StartOffset);
if (officeDrawing == null)
{
logger.Log(POILogger.WARN, "Characters #" + characterRun
+ " references missing drawn object");
return;
}
byte[] pictureData = officeDrawing.GetPictureData();
if (pictureData == null)
{
// usual shape?
return;
}
PictureType type = PictureType.FindMatchingType(pictureData);
String path = GetPicturesManager().SavePicture(pictureData, type,
"s" + characterRun.StartOffset + "." + type);
ProcessDrawnObject(doc, characterRun, officeDrawing, path, block);
}
private static HWPFDocument Reload(HWPFDocument hwpfDocument)
{
MemoryStream baos = new MemoryStream();
hwpfDocument.Write(baos);
return(new HWPFDocument(new MemoryStream(baos.ToArray())));
}
public void TestUnicodeParts()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("HeaderFooterUnicode.doc");
TextPieceTable tbl = doc.TextTable;
// In three bits, split every 512 bytes
Assert.AreEqual(3, tbl.TextPieces.Count);
TextPiece tpA = (TextPiece)tbl.TextPieces[0];
TextPiece tpB = (TextPiece)tbl.TextPieces[1];
TextPiece tpC = (TextPiece)tbl.TextPieces[2];
Assert.IsTrue(tpA.IsUnicode);
Assert.IsTrue(tpB.IsUnicode);
Assert.IsTrue(tpC.IsUnicode);
Assert.AreEqual(256, tpA.CharacterLength);
Assert.AreEqual(256, tpB.CharacterLength);
Assert.AreEqual(19, tpC.CharacterLength);
Assert.AreEqual(512, tpA.BytesLength);
Assert.AreEqual(512, tpB.BytesLength);
Assert.AreEqual(38, tpC.BytesLength);
Assert.AreEqual(0, tpA.Start);
Assert.AreEqual(256, tpA.End);
Assert.AreEqual(256, tpB.Start);
Assert.AreEqual(512, tpB.End);
Assert.AreEqual(512, tpC.Start);
Assert.AreEqual(531, tpC.End);
// Save and re-load
HWPFDocument docB = SaveAndReload(doc);
tbl = docB.TextTable;
Assert.AreEqual(3, tbl.TextPieces.Count);
tpA = (TextPiece)tbl.TextPieces[0];
tpB = (TextPiece)tbl.TextPieces[1];
tpC = (TextPiece)tbl.TextPieces[2];
Assert.IsTrue(tpA.IsUnicode);
Assert.IsTrue(tpB.IsUnicode);
Assert.IsTrue(tpC.IsUnicode);
Assert.AreEqual(256, tpA.CharacterLength);
Assert.AreEqual(256, tpB.CharacterLength);
Assert.AreEqual(19, tpC.CharacterLength);
Assert.AreEqual(512, tpA.BytesLength);
Assert.AreEqual(512, tpB.BytesLength);
Assert.AreEqual(38, tpC.BytesLength);
Assert.AreEqual(0, tpA.Start);
Assert.AreEqual(256, tpA.End);
Assert.AreEqual(256, tpB.Start);
Assert.AreEqual(512, tpB.End);
Assert.AreEqual(512, tpC.Start);
Assert.AreEqual(531, tpC.End);
}
protected void ProcessNote(HWPFDocument wordDocument, XmlElement block,
Range noteTextRange)
{
int noteIndex;
lock (objCounters)
{
noteIndex = noteCounters++;
}
block.AppendChild(textDocumentFacade
.CreateText(UNICODECHAR_ZERO_WIDTH_SPACE + "[" + noteIndex
+ "]" + UNICODECHAR_ZERO_WIDTH_SPACE));
if (notes == null)
{
notes = textDocumentFacade.CreateBlock();
}
XmlElement note = textDocumentFacade.CreateBlock();
notes.AppendChild(note);
note.AppendChild(textDocumentFacade.CreateText("^" + noteIndex
+ "\t "));
ProcessCharacters(wordDocument, int.MinValue, noteTextRange, note);
note.AppendChild(textDocumentFacade.CreateText("\n"));
}
public void TestCroppedPictures()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("testCroppedPictures.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.IsNotNull(pics);
Assert.AreEqual(2, pics.Count);
Picture pic1 = pics[0];
Assert.AreEqual(27, pic1.AspectRatioX);
Assert.AreEqual(270, pic1.HorizontalScalingFactor);
Assert.AreEqual(27, pic1.AspectRatioY);
Assert.AreEqual(271, pic1.VerticalScalingFactor);
Assert.AreEqual(12000, pic1.DxaGoal); // 21.17 cm / 2.54 cm/inch * 72dpi * 20 = 12000
Assert.AreEqual(9000, pic1.DyaGoal); // 15.88 cm / 2.54 cm/inch * 72dpi * 20 = 9000
Assert.AreEqual(0, pic1.DxaCropLeft);
Assert.AreEqual(0, pic1.DxaCropRight);
Assert.AreEqual(0, pic1.DyaCropTop);
Assert.AreEqual(0, pic1.DyaCropBottom);
Picture pic2 = pics[1];
Assert.AreEqual(76, pic2.AspectRatioX);
Assert.AreEqual(764, pic2.HorizontalScalingFactor);
Assert.AreEqual(68, pic2.AspectRatioY);
Assert.AreEqual(685, pic2.VerticalScalingFactor);
Assert.AreEqual(12000, pic2.DxaGoal); // 21.17 cm / 2.54 cm/inch * 72dpi * 20 = 12000
Assert.AreEqual(9000, pic2.DyaGoal); // 15.88 cm / 2.54 cm/inch * 72dpi * 20 = 9000
Assert.AreEqual(0, pic2.DxaCropLeft); // TODO YK: The Picture is cropped but HWPF reads the crop parameters all zeros
Assert.AreEqual(0, pic2.DxaCropRight);
Assert.AreEqual(0, pic2.DyaCropTop);
Assert.AreEqual(0, pic2.DyaCropBottom);
}
public void TestReadWrite()
{
// This document is widely available on the internet as "blair.doc".
// I tried stripping the content and saving the document but my version
// of Word (from Office XP) strips this table out.
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("saved-by-table.doc");
// Check what we just Read.
for (int i = 0; i < expected.Count; i++)
{
Assert.AreEqual(expected[i], doc.GetSavedByTable().GetEntries()[i], "List of saved-by entries was not as expected");
}
// Now write the entire document out, and read it back in...
MemoryStream byteStream = new MemoryStream();
doc.Write(byteStream);
Stream copyStream = new MemoryStream(byteStream.ToArray());
HWPFDocument copy = new HWPFDocument(copyStream);
// And check again.
for (int i = 0; i < expected.Count; i++)
{
Assert.AreEqual(
expected[i], copy.GetSavedByTable().GetEntries()[i], "List of saved-by entries was incorrect after writing");
}
}
public void TestMultiLevelLists()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Lists.doc");
Range r = doc.GetRange();
Assert.AreEqual(40, r.NumParagraphs);
Assert.AreEqual("Multi-level un-ordered list:\r", r.GetParagraph(12).Text);
Assert.AreEqual("ML 1:1\r", r.GetParagraph(13).Text);
Assert.AreEqual("ML 1:2\r", r.GetParagraph(14).Text);
Assert.AreEqual("ML 2:1\r", r.GetParagraph(15).Text);
Assert.AreEqual("ML 2:2\r", r.GetParagraph(16).Text);
Assert.AreEqual("ML 2:3\r", r.GetParagraph(17).Text);
Assert.AreEqual("ML 3:1\r", r.GetParagraph(18).Text);
Assert.AreEqual("ML 4:1\r", r.GetParagraph(19).Text);
Assert.AreEqual("ML 5:1\r", r.GetParagraph(20).Text);
Assert.AreEqual("ML 5:2\r", r.GetParagraph(21).Text);
Assert.AreEqual("ML 2:4\r", r.GetParagraph(22).Text);
Assert.AreEqual("ML 1:3\r", r.GetParagraph(23).Text);
Assert.AreEqual("Multi-level ordered list:\r", r.GetParagraph(24).Text);
Assert.AreEqual("OL 1\r", r.GetParagraph(25).Text);
Assert.AreEqual("OL 2\r", r.GetParagraph(26).Text);
Assert.AreEqual("OL 2.1\r", r.GetParagraph(27).Text);
Assert.AreEqual("OL 2.2\r", r.GetParagraph(28).Text);
Assert.AreEqual("OL 2.2.1\r", r.GetParagraph(29).Text);
Assert.AreEqual("OL 2.2.2\r", r.GetParagraph(30).Text);
Assert.AreEqual("OL 2.2.2.1\r", r.GetParagraph(31).Text);
Assert.AreEqual("OL 2.2.3\r", r.GetParagraph(32).Text);
Assert.AreEqual("OL 3\r", r.GetParagraph(33).Text);
Assert.AreEqual("Finally we want some indents, to tell the difference\r", r.GetParagraph(34).Text);
for (int i = 12; i <= 34; i++)
{
Assert.AreEqual(9, r.GetParagraph(12).GetLvl());
}
Assert.AreEqual(0, r.GetParagraph(12).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(13).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(14).GetIlvl());
Assert.AreEqual(1, r.GetParagraph(15).GetIlvl());
Assert.AreEqual(1, r.GetParagraph(16).GetIlvl());
Assert.AreEqual(1, r.GetParagraph(17).GetIlvl());
Assert.AreEqual(2, r.GetParagraph(18).GetIlvl());
Assert.AreEqual(3, r.GetParagraph(19).GetIlvl());
Assert.AreEqual(4, r.GetParagraph(20).GetIlvl());
Assert.AreEqual(4, r.GetParagraph(21).GetIlvl());
Assert.AreEqual(1, r.GetParagraph(22).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(23).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(24).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(25).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(26).GetIlvl());
Assert.AreEqual(1, r.GetParagraph(27).GetIlvl());
Assert.AreEqual(1, r.GetParagraph(28).GetIlvl());
Assert.AreEqual(2, r.GetParagraph(29).GetIlvl());
Assert.AreEqual(2, r.GetParagraph(30).GetIlvl());
Assert.AreEqual(3, r.GetParagraph(31).GetIlvl());
Assert.AreEqual(2, r.GetParagraph(32).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(33).GetIlvl());
Assert.AreEqual(0, r.GetParagraph(34).GetIlvl());
}
protected override void ProcessDrawnObject(HWPFDocument doc,
CharacterRun characterRun, OfficeDrawing officeDrawing,
String path, XmlElement block)
{
XmlElement externalGraphic = foDocumentFacade.CreateExternalGraphic(path);
block.AppendChild(externalGraphic);
}
public void TestPicturesInHeader()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("header_image.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.AreEqual(2, pics.Count);
}
/** @link dependency
* @stereotype instantiate*/
/*# Picture lnkPicture; */
/**
*
* @param _document
* @param _dataStream
*/
public PicturesTable(HWPFDocument _document, byte[] _dataStream, byte[] _mainStream, FSPATable fspa, EscherRecordHolder dgg)
{
this._document = _document;
this._dataStream = _dataStream;
this._mainStream = _mainStream;
this._fspa = fspa;
this._dgg = dgg;
}
public void TestEmptyDocument()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("empty.doc");
RevisionMarkAuthorTable rmt = doc.GetRevisionMarkAuthorTable();
Assert.IsNull(rmt);
}
public void TestBug47731()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Bug47731.doc");
String foundText = WordToTextConverter.GetText(doc);
Assert.IsTrue(foundText
.Contains("Soak the rice in water for three to four hours"));
}
public void TestPicturesWithTable()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Bug44603.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.AreEqual(2, pics.Count);
}
/** @link dependency
* @stereotype instantiate*/
/*# Picture lnkPicture; */
/**
*
* @param _document
* @param _dataStream
*/
public PicturesTable(HWPFDocument _document, byte[] _dataStream, byte[] _mainStream, FSPATable fspa, EscherRecordHolder dgg)
{
this._document = _document;
this._dataStream = _dataStream;
this._mainStream = _mainStream;
this._fspa = fspa;
this._dgg = dgg;
}
public void TestParagraphPAPXNoParent45877()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Bug45877.doc");
Assert.AreEqual(17, doc.GetRange().NumParagraphs);
Assert.AreEqual("First paragraph\r", doc.GetRange().GetParagraph(0).Text);
Assert.AreEqual("After Crashing Part\r", doc.GetRange().GetParagraph(13).Text);
}
/**
* Writes a spreadsheet to a <tt>MemoryStream</tt> and Reads it back
* from a <tt>MemoryStream</tt>.<p/>
* Useful for verifying that the serialisation round trip
*/
public static HWPFDocument WriteOutAndReadBack(HWPFDocument original)
{
MemoryStream baos = new MemoryStream(4096);
original.Write(baos);
MemoryStream bais = new MemoryStream(baos.ToArray());
return(new HWPFDocument(bais));
}
private static String getFoText(String sampleFileName)
{
HWPFDocument hwpfDocument = new HWPFDocument(POIDataSamples.GetDocumentInstance().OpenResourceAsStream(sampleFileName));
WordToFoConverter wordToFoConverter = new WordToFoConverter(new XmlDocument());
wordToFoConverter.ProcessDocument(hwpfDocument);
return(wordToFoConverter.Document.InnerXml);
}
protected HWPFDocument SaveAndReload(HWPFDocument doc)
{
MemoryStream baos = new MemoryStream();
doc.Write(baos);
return(new HWPFDocument(
new MemoryStream(baos.ToArray())
));
}
public ToxyDocument Parse()
{
if (!File.Exists(Context.Path))
{
throw new FileNotFoundException("File " + Context.Path + " is not found");
}
bool extractHeader = false;
if (Context.Properties.ContainsKey("ExtractHeader"))
{
extractHeader = Utility.IsTrue(Context.Properties["ExtractHeader"]);
}
bool extractFooter = false;
if (Context.Properties.ContainsKey("ExtractFooter"))
{
extractFooter = Utility.IsTrue(Context.Properties["ExtractFooter"]);
}
ToxyDocument rdoc = new ToxyDocument();
using (FileStream stream = File.OpenRead(Context.Path))
{
HWPFDocument worddoc = new HWPFDocument(stream);
if (extractHeader && worddoc.GetHeaderStoryRange() != null)
{
StringBuilder sb = new StringBuilder();
rdoc.Header = worddoc.GetHeaderStoryRange().Text;
}
if (extractFooter && worddoc.GetFootnoteRange() != null)
{
StringBuilder sb = new StringBuilder();
rdoc.Footer = worddoc.GetFootnoteRange().Text;
}
for (int i = 0; i < worddoc.GetRange().NumParagraphs; i++)
{
Paragraph para = worddoc.GetRange().GetParagraph(i);
string text = para.Text;
ToxyParagraph p = new ToxyParagraph();
p.Text = text;
//var runs = para.Runs;
p.StyleID = para.GetStyleIndex().ToString();
//for (int i = 0; i < runs.Count; i++)
//{
// var run = runs[i];
//}
rdoc.Paragraphs.Add(p);
}
}
return(rdoc);
}
public void TestInnerTable()
{
Stream resourceAsStream = POIDataSamples.GetDocumentInstance()
.OpenResourceAsStream("innertable.doc");
HWPFDocument hwpfDocument = new HWPFDocument(resourceAsStream);
resourceAsStream.Close();
TestInnerTable(hwpfDocument);
hwpfDocument = Reload(hwpfDocument);
TestInnerTable(hwpfDocument);
}
public void TestWriteProperties()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("SampleDoc.doc");
Assert.AreEqual("Nick Burch", doc.SummaryInformation.Author);
// Write and read
HWPFDocument doc2 = WriteOutAndRead(doc);
Assert.AreEqual("Nick Burch", doc2.SummaryInformation.Author);
}
public void TestSprmPJc()
{
Stream resourceAsStream = POIDataSamples.GetDocumentInstance()
.OpenResourceAsStream("Bug49820.doc");
HWPFDocument hwpfDocument = new HWPFDocument(resourceAsStream);
resourceAsStream.Close();
Assert.AreEqual(1, hwpfDocument.GetStyleSheet().GetParagraphStyle(8)
.GetJustification());
hwpfDocument = Reload(hwpfDocument);
Assert.AreEqual(1, hwpfDocument.GetStyleSheet().GetParagraphStyle(8)
.GetJustification());
}
private void TestInnerTable(HWPFDocument hwpfDocument)
{
Range range = hwpfDocument.GetRange();
for (int p = 0; p < range.NumParagraphs; p++)
{
Paragraph paragraph = range.GetParagraph(p);
char first = paragraph.Text.ToLower()[0];
if ('1' <= first && first < '4')
{
Assert.IsTrue(paragraph.IsInTable());
Assert.AreEqual(2, paragraph.GetTableLevel());
}
if ('a' <= first && first < 'z')
{
Assert.IsTrue(paragraph.IsInTable());
Assert.AreEqual(1, paragraph.GetTableLevel());
}
}
}
public HeaderStories(HWPFDocument doc)
{
this.headerStories = doc.GetHeaderStoryRange();
FileInformationBlock fib = doc.GetFileInformationBlock();
// If there's no PlcfHdd, nothing to do
if (fib.GetCcpHdd() == 0)
{
return;
}
if (fib.GetPlcfHddSize() == 0)
{
return;
}
// Handle the PlcfHdd
plcfHdd = new PlexOfCps(
doc.GetTableStream(), fib.GetPlcfHddOffset(),
fib.GetPlcfHddSize(), 0
);
}
protected HWPFDocument SaveAndReload(HWPFDocument doc)
{
MemoryStream baos = new MemoryStream();
doc.Write(baos);
return new HWPFDocument(
new MemoryStream(baos.ToArray())
);
}
protected override void ProcessDrawnObject(HWPFDocument doc, CharacterRun characterRun, OfficeDrawing officeDrawing, string path, XmlElement block)
{
XmlElement img = htmlDocumentFacade.CreateImage(path);
block.AppendChild(img);
}
protected override void ProcessEndnoteAutonumbered(HWPFDocument wordDocument, int noteIndex, XmlElement block, Range endnoteTextRange)
{
ProcessNoteAutonumbered(wordDocument, "end", noteIndex, block, endnoteTextRange);
}
private void ProcessNoteAutonumbered(HWPFDocument wordDocument, string type, int noteIndex, XmlElement block, Range noteTextRange)
{
String textIndex = (noteIndex + 1).ToString();
String textIndexClass = htmlDocumentFacade.GetOrCreateCssClass("a", "a", "vertical-align:super;font-size:smaller;");
String forwardNoteLink = type + "note_" + textIndex;
String backwardNoteLink = type + "note_back_" + textIndex;
XmlElement anchor = htmlDocumentFacade.CreateHyperlink("#" + forwardNoteLink);
anchor.SetAttribute("name", backwardNoteLink);
anchor.SetAttribute("class", textIndexClass + " " + type + "noteanchor");
anchor.InnerText = textIndex;
block.AppendChild(anchor);
if (notes == null)
{
notes = htmlDocumentFacade.CreateBlock();
notes.SetAttribute("class", "notes");
}
XmlElement note = htmlDocumentFacade.CreateBlock();
note.SetAttribute("class", type + "note");
notes.AppendChild(note);
XmlElement bookmark = htmlDocumentFacade.CreateBookmark(forwardNoteLink);
bookmark.SetAttribute("href", "#" + backwardNoteLink);
bookmark.InnerText = (textIndex);
bookmark.SetAttribute("class", textIndexClass + " " + type + "noteindex");
note.AppendChild(bookmark);
note.AppendChild(htmlDocumentFacade.CreateText(" "));
XmlElement span = htmlDocumentFacade.Document.CreateElement("span");
span.SetAttribute("class", type + "notetext");
note.AppendChild(span);
this.blocksProperies.Push(new BlockProperies("", -1));
try
{
ProcessCharacters(wordDocument, int.MinValue, noteTextRange, span);
}
finally
{
this.blocksProperies.Pop();
}
}
/**
* Create a new Word Extractor
* @param doc The HWPFDocument to extract from
*/
public WordExtractor(HWPFDocument doc)
: base(doc)
{
this.doc = doc;
}
protected override void ProcessFootnoteAutonumbered(HWPFDocument wordDocument, int noteIndex, XmlElement block, Range footnoteTextRange)
{
ProcessNoteAutonumbered(wordDocument, "foot", noteIndex, block, footnoteTextRange);
}
private static HWPFDocument Reload(HWPFDocument hwpfDocument)
{
MemoryStream baos = new MemoryStream();
hwpfDocument.Write(baos);
return new HWPFDocument(new MemoryStream(baos.ToArray()));
}
public DocumentPosition(HWPFDocument doc, int pos):base(pos, pos, doc)
{
}
