public void TestImageCount()
{
HWPFDocument docA = HWPFTestDataSamples.OpenSampleFile(docAFile);
HWPFDocument docB = HWPFTestDataSamples.OpenSampleFile(docBFile);
Assert.IsNotNull(docA.GetPicturesTable());
Assert.IsNotNull(docB.GetPicturesTable());
PicturesTable picA = docA.GetPicturesTable();
PicturesTable picB = docB.GetPicturesTable();
List <Picture> picturesA = picA.GetAllPictures();
List <Picture> picturesB = picB.GetAllPictures();
Assert.AreEqual(7, picturesA.Count);
Assert.AreEqual(2, picturesB.Count);
}
public void TestEquation()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("equation.doc");
PicturesTable pictures = doc.GetPicturesTable();
List <Picture> allPictures = pictures.GetAllPictures();
Assert.AreEqual(1, allPictures.Count);
Picture picture = allPictures[0];
Assert.IsNotNull(picture);
Assert.AreEqual(PictureType.EMF, picture.SuggestPictureType());
Assert.AreEqual(PictureType.EMF.Extension,
picture.SuggestFileExtension());
Assert.AreEqual(PictureType.EMF.Mime, picture.MimeType);
Assert.AreEqual("0.emf", picture.SuggestFullFileName());
}
public void TestCompressedImageData()
{
HWPFDocument docC = HWPFTestDataSamples.OpenSampleFile(docCFile);
PicturesTable picC = docC.GetPicturesTable();
List <Picture> picturesC = picC.GetAllPictures();
Assert.AreEqual(1, picturesC.Count);
Picture pic = picturesC[0];
Assert.IsNotNull(pic);
// Check the same
byte[] picBytes = ReadFile(imgCFile);
Assert.AreEqual(picBytes.Length, pic.GetContent().Length);
assertBytesSame(picBytes, pic.GetContent());
}
public void TestFloatingPictures()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("FloatingPictures.doc");
PicturesTable pictures = doc.GetPicturesTable();
// There are 19 images in the picture, but some are
// duplicate floating ones
Assert.AreEqual(17, pictures.GetAllPictures().Count);
int plain8s = 0;
int escher8s = 0;
int image1s = 0;
Range r = doc.GetRange();
for (int np = 0; np < r.NumParagraphs; np++)
{
Paragraph p = r.GetParagraph(np);
for (int nc = 0; nc < p.NumCharacterRuns; nc++)
{
CharacterRun cr = p.GetCharacterRun(nc);
if (pictures.HasPicture(cr))
{
image1s++;
}
else if (pictures.HasEscherPicture(cr))
{
escher8s++;
}
else if (cr.Text.StartsWith("\u0008"))
{
plain8s++;
}
}
}
// Total is 20, as the 4 escher 8s all reference
// the same regular image
Assert.AreEqual(16, image1s);
Assert.AreEqual(4, escher8s);
Assert.AreEqual(0, plain8s);
}
public void TestImageData()
{
HWPFDocument docB = HWPFTestDataSamples.OpenSampleFile(docBFile);
PicturesTable picB = docB.GetPicturesTable();
List <Picture> picturesB = picB.GetAllPictures();
Assert.AreEqual(2, picturesB.Count);
Picture pic1 = picturesB[0];
Picture pic2 = picturesB[1];
Assert.IsNotNull(pic1);
Assert.IsNotNull(pic2);
// Check the same
byte[] pic1B = ReadFile(imgAFile);
byte[] pic2B = ReadFile(imgBFile);
Assert.AreEqual(pic1B.Length, pic1.GetContent().Length);
Assert.AreEqual(pic2B.Length, pic2.GetContent().Length);
assertBytesSame(pic1B, pic1.GetContent());
assertBytesSame(pic2B, pic2.GetContent());
}
/// <summary>
/// Initializes a new instance of the <see cref="HWPFDocument"/> class.
/// </summary>
/// <param name="directory">The directory.</param>
public HWPFDocument(DirectoryNode directory)
: base(directory)
{
_endnotes = new NotesImpl(_endnotesTables);
_footnotes = new NotesImpl(_footnotesTables);
// Load the main stream and FIB
// Also handles HPSF bits
// Do the CP Split
_cpSplit = new CPSplitCalculator(_fib);
// Is this document too old for us?
if (_fib.GetNFib() < 106)
{
throw new OldWordFileFormatException("The document is too old - Word 95 or older. Try HWPFOldDocument instead?");
}
// use the fib to determine the name of the table stream.
String name = "0Table";
if (_fib.IsFWhichTblStm())
{
name = "1Table";
}
// Grab the table stream.
DocumentEntry tableProps;
try
{
tableProps =
(DocumentEntry)directory.GetEntry(name);
}
catch (FileNotFoundException)
{
throw new InvalidOperationException("Table Stream '" + name + "' wasn't found - Either the document is corrupt, or is Word95 (or earlier)");
}
// read in the table stream.
_tableStream = new byte[tableProps.Size];
directory.CreatePOIFSDocumentReader(name).Read(_tableStream);
_fib.FillVariableFields(_mainStream, _tableStream);
// read in the data stream.
try
{
DocumentEntry dataProps =
(DocumentEntry)directory.GetEntry("Data");
_dataStream = new byte[dataProps.Size];
directory.CreatePOIFSDocumentReader("Data").Read(_dataStream);
}
catch (FileNotFoundException)
{
_dataStream = new byte[0];
}
// Get the cp of the start of text in the main stream
// The latest spec doc says this is always zero!
int fcMin = 0;
//fcMin = _fib.GetFcMin()
// Start to load up our standard structures.
_dop = new DocumentProperties(_tableStream, _fib.GetFcDop());
_cft = new ComplexFileTable(_mainStream, _tableStream, _fib.GetFcClx(), fcMin);
TextPieceTable _tpt = _cft.GetTextPieceTable();
// Now load the rest of the properties, which need to be adjusted
// for where text really begin
_cbt = new CHPBinTable(_mainStream, _tableStream, _fib.GetFcPlcfbteChpx(), _fib.GetLcbPlcfbteChpx(), _tpt);
_pbt = new PAPBinTable(_mainStream, _tableStream, _dataStream, _fib.GetFcPlcfbtePapx(), _fib.GetLcbPlcfbtePapx(), _tpt);
_text = _tpt.Text;
/*
* in this mode we preserving PAPX/CHPX structure from file, so text may
* miss from output, and text order may be corrupted
*/
bool preserveBinTables = false;
try
{
preserveBinTables = Boolean.Parse(
ConfigurationManager.AppSettings[PROPERTY_PRESERVE_BIN_TABLES]);
}
catch (Exception)
{
// ignore;
}
if (!preserveBinTables)
{
_cbt.Rebuild(_cft);
_pbt.Rebuild(_text, _cft);
}
/*
* Property to disable text rebuilding. In this mode changing the text
* will lead to unpredictable behavior
*/
bool preserveTextTable = false;
try
{
preserveTextTable = Boolean.Parse(
ConfigurationManager.AppSettings[PROPERTY_PRESERVE_TEXT_TABLE]);
}
catch (Exception)
{
// ignore;
}
if (!preserveTextTable)
{
_cft = new ComplexFileTable();
_tpt = _cft.GetTextPieceTable();
TextPiece textPiece = new SinglentonTextPiece(_text);
_tpt.Add(textPiece);
_text = textPiece.GetStringBuilder();
}
// Read FSPA and Escher information
// _fspa = new FSPATable(_tableStream, _fib.getFcPlcspaMom(),
// _fib.getLcbPlcspaMom(), getTextTable().getTextPieces());
_fspaHeaders = new FSPATable(_tableStream, _fib,
FSPADocumentPart.HEADER);
_fspaMain = new FSPATable(_tableStream, _fib, FSPADocumentPart.MAIN);
if (_fib.GetFcDggInfo() != 0)
{
_dgg = new EscherRecordHolder(_tableStream, _fib.GetFcDggInfo(), _fib.GetLcbDggInfo());
}
else
{
_dgg = new EscherRecordHolder();
}
// read in the pictures stream
_pictures = new PicturesTable(this, _dataStream, _mainStream, _fspa, _dgg);
// And the art shapes stream
_officeArts = new ShapesTable(_tableStream, _fib);
// And escher pictures
_officeDrawingsHeaders = new OfficeDrawingsImpl(_fspaHeaders, _dgg, _mainStream);
_officeDrawingsMain = new OfficeDrawingsImpl(_fspaMain, _dgg, _mainStream);
_st = new SectionTable(_mainStream, _tableStream, _fib.GetFcPlcfsed(), _fib.GetLcbPlcfsed(), fcMin, _tpt, _cpSplit);
_ss = new StyleSheet(_tableStream, _fib.GetFcStshf());
_ft = new FontTable(_tableStream, _fib.GetFcSttbfffn(), _fib.GetLcbSttbfffn());
int listOffset = _fib.GetFcPlcfLst();
int lfoOffset = _fib.GetFcPlfLfo();
if (listOffset != 0 && _fib.GetLcbPlcfLst() != 0)
{
_lt = new ListTables(_tableStream, _fib.GetFcPlcfLst(), _fib.GetFcPlfLfo());
}
int sbtOffset = _fib.GetFcSttbSavedBy();
int sbtLength = _fib.GetLcbSttbSavedBy();
if (sbtOffset != 0 && sbtLength != 0)
{
_sbt = new SavedByTable(_tableStream, sbtOffset, sbtLength);
}
int rmarkOffset = _fib.GetFcSttbfRMark();
int rmarkLength = _fib.GetLcbSttbfRMark();
if (rmarkOffset != 0 && rmarkLength != 0)
{
_rmat = new RevisionMarkAuthorTable(_tableStream, rmarkOffset, rmarkLength);
}
_bookmarksTables = new BookmarksTables(_tableStream, _fib);
_bookmarks = new BookmarksImpl(_bookmarksTables);
_endnotesTables = new NotesTables(NoteType.ENDNOTE, _tableStream, _fib);
_endnotes = new NotesImpl(_endnotesTables);
_footnotesTables = new NotesTables(NoteType.FOOTNOTE, _tableStream, _fib);
_footnotes = new NotesImpl(_footnotesTables);
_fieldsTables = new FieldsTables(_tableStream, _fib);
_fields = new FieldsImpl(_fieldsTables);
}
public void TestEmbededDocumentIcon()
{
// This file has two embeded excel files, an embeded powerpoint
// file and an embeded word file, in that order
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("word_with_embeded.doc");
// Check we don't break loading the pictures
doc.GetPicturesTable().GetAllPictures();
PicturesTable pictureTable = doc.GetPicturesTable();
// Check the Text, and its embeded images
Paragraph p;
Range r = doc.GetRange();
Assert.AreEqual(1, r.NumSections);
Assert.AreEqual(5, r.NumParagraphs);
p = r.GetParagraph(0);
Assert.AreEqual(2, p.NumCharacterRuns);
Assert.AreEqual("I have lots of embedded files in me\r", p.Text);
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(0)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(1)));
p = r.GetParagraph(1);
Assert.AreEqual(5, p.NumCharacterRuns);
Assert.AreEqual("\u0013 EMBED Excel.Sheet.8 \u0014\u0001\u0015\r", p.Text);
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(0)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(1)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(2)));
Assert.AreEqual(true, pictureTable.HasPicture(p.GetCharacterRun(3)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(4)));
p = r.GetParagraph(2);
Assert.AreEqual(6, p.NumCharacterRuns);
Assert.AreEqual("\u0013 EMBED Excel.Sheet.8 \u0014\u0001\u0015\r", p.Text);
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(0)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(1)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(2)));
Assert.AreEqual(true, pictureTable.HasPicture(p.GetCharacterRun(3)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(4)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(5)));
p = r.GetParagraph(3);
Assert.AreEqual(6, p.NumCharacterRuns);
Assert.AreEqual("\u0013 EMBED PowerPoint.Show.8 \u0014\u0001\u0015\r", p.Text);
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(0)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(1)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(2)));
Assert.AreEqual(true, pictureTable.HasPicture(p.GetCharacterRun(3)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(4)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(5)));
p = r.GetParagraph(4);
Assert.AreEqual(6, p.NumCharacterRuns);
Assert.AreEqual("\u0013 EMBED Word.Document.8 \\s \u0014\u0001\u0015\r", p.Text);
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(0)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(1)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(2)));
Assert.AreEqual(true, pictureTable.HasPicture(p.GetCharacterRun(3)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(4)));
Assert.AreEqual(false, pictureTable.HasPicture(p.GetCharacterRun(5)));
// Look at the pictures table
List <Picture> pictures = pictureTable.GetAllPictures();
Assert.AreEqual(4, pictures.Count);
Picture picture = pictures[0];
Assert.AreEqual("", picture.SuggestFileExtension());
Assert.AreEqual("0", picture.SuggestFullFileName());
Assert.AreEqual("image/unknown", picture.MimeType);
picture = pictures[1];
Assert.AreEqual("", picture.SuggestFileExtension());
Assert.AreEqual("469", picture.SuggestFullFileName());
Assert.AreEqual("image/unknown", picture.MimeType);
picture = pictures[2];
Assert.AreEqual("", picture.SuggestFileExtension());
Assert.AreEqual("8c7", picture.SuggestFullFileName());
Assert.AreEqual("image/unknown", picture.MimeType);
picture = pictures[3];
Assert.AreEqual("", picture.SuggestFileExtension());
Assert.AreEqual("10a8", picture.SuggestFullFileName());
Assert.AreEqual("image/unknown", picture.MimeType);
}
