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 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 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 TestPicturesInHeader()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("header_image.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.AreEqual(2, pics.Count);
}
public void TestPicturesWithTable()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("Bug44603.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.AreEqual(2, pics.Count);
}
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);
}
/**
* Pending the missing files being uploaded to
* bug #44937
*/
//[TestMethod]
public void BROKENtestEscherDrawing()
{
HWPFDocument docD = HWPFTestDataSamples.OpenSampleFile(docDFile);
List <Picture> allPictures = docD.GetPicturesTable().GetAllPictures();
Assert.AreEqual(1, allPictures.Count);
Picture pic = allPictures[0];
Assert.IsNotNull(pic);
byte[] picD = ReadFile(imgDFile);
Assert.AreEqual(picD.Length, pic.GetContent().Length);
assertBytesSame(picD, pic.GetContent());
}
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 TestTwoImages()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("two_images.doc");
List <Picture> pics = doc.GetPicturesTable().GetAllPictures();
Assert.IsNotNull(pics);
Assert.AreEqual(pics.Count, 2);
for (int i = 0; i < pics.Count; i++)
{
Picture pic = (Picture)pics[i];
Assert.IsNotNull(pic.SuggestFileExtension());
Assert.IsNotNull(pic.SuggestFullFileName());
}
Picture picA = pics[0];
Picture picB = pics[1];
Assert.AreEqual("jpg", picA.SuggestFileExtension());
Assert.AreEqual("jpg", picA.SuggestFileExtension());
}
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 TestEmfImage()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("vector_image.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());
Assert.IsTrue(pic.Size > 128);
// Check right contents
byte[] emf = POIDataSamples.GetDocumentInstance().ReadFile("vector_image.emf");
byte[] pemf = pic.GetContent();
Assert.AreEqual(emf.Length, pemf.Length);
for (int i = 0; i < emf.Length; i++)
{
Assert.AreEqual(emf[i], pemf[i]);
}
}
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());
}
protected bool ProcessCharacters(HWPFDocumentCore wordDocument, int currentTableLevel, Range range, XmlElement block)
{
if (range == null)
{
return(false);
}
bool 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
*/
IList <Structure> structures = new List <Structure>();
if (wordDocument is HWPFDocument)
{
HWPFDocument doc = (HWPFDocument)wordDocument;
Dictionary <int, List <Bookmark> > rangeBookmarks = doc.GetBookmarks()
.GetBookmarksStartedBetween(range.StartOffset, range.EndOffset);
if (rangeBookmarks != null)
{
foreach (KeyValuePair <int, List <Bookmark> > kv in rangeBookmarks)
{
List <Bookmark> lists = kv.Value;
foreach (Bookmark bookmark in lists)
{
if (!bookmarkStack.Contains(bookmark))
{
AddToStructures(structures, new Structure(bookmark));
}
}
}
}
// TODO: dead fields?
for (int c = 0; c < range.NumCharacterRuns; c++)
{
CharacterRun characterRun = range.GetCharacterRun(c);
if (characterRun == null)
{
throw new NullReferenceException();
}
Field aliveField = ((HWPFDocument)wordDocument).GetFields()
.GetFieldByStartOffset(FieldsDocumentPart.MAIN,
characterRun.StartOffset);
if (aliveField != null)
{
AddToStructures(structures, new Structure(aliveField));
}
}
}
//structures = new ArrayList<Structure>( structures );
//Collections.sort( structures );
SortedList <Structure, Structure> sl = new SortedList <Structure, Structure>();
foreach (Structure s in structures)
{
sl.Add(s, s);
}
structures.Clear();
((List <Structure>)structures).AddRange(sl.Values);
int previous = range.StartOffset;
foreach (Structure structure in structures)
{
if (structure.Start != previous)
{
Range subrange = new Range(previous, structure.Start, range);
//{
// public String toString()
// {
// return "BetweenStructuresSubrange " + super.ToString();
// }
//};
ProcessCharacters(wordDocument, currentTableLevel, subrange, block);
}
if (structure.StructureObject is Bookmark)
{
// other bookmarks with same boundaries
IList <Bookmark> bookmarks = new List <Bookmark>();
IEnumerator <List <Bookmark> > iterator = ((HWPFDocument)wordDocument).GetBookmarks().GetBookmarksStartedBetween(structure.Start, structure.Start + 1).Values.GetEnumerator();
iterator.MoveNext();
foreach (Bookmark bookmark in iterator.Current)
{
if (bookmark.Start == structure.Start &&
bookmark.End == structure.End)
{
bookmarks.Add(bookmark);
}
}
bookmarkStack.AddRange(bookmarks);
try
{
int end = Math.Min(range.EndOffset, structure.End);
Range subrange = new Range(structure.Start, end, range);
/*{
* public String toString()
* {
* return "BookmarksSubrange " + super.ToString();
* }
* };*/
ProcessBookmarks(wordDocument, block, subrange,
currentTableLevel, bookmarks);
}
finally
{
bookmarkStack.RemoveAll((e) => { return(bookmarks.Contains(e)); });
}
}
else if (structure.StructureObject is Field)
{
Field field = (Field)structure.StructureObject;
ProcessField((HWPFDocument)wordDocument, range, currentTableLevel, field, block);
}
else
{
throw new NotSupportedException("NYI: " + structure.StructureObject.GetType().ToString());
}
previous = Math.Min(range.EndOffset, structure.End);
}
if (previous != range.StartOffset)
{
if (previous > range.EndOffset)
{
logger.Log(POILogger.WARN, "Latest structure in ", range,
" ended at #" + previous, " after range boundaries [",
range.StartOffset + "; " + range.EndOffset,
")");
return(true);
}
if (previous < range.EndOffset)
{
Range subrange = new Range(previous, range.EndOffset, 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 NullReferenceException();
}
if (wordDocument is HWPFDocument && ((HWPFDocument)wordDocument).GetPicturesTable().HasPicture(characterRun))
{
HWPFDocument newFormat = (HWPFDocument)wordDocument;
Picture picture = newFormat.GetPicturesTable().ExtractPicture(characterRun, true);
ProcessImage(block, characterRun.Text[0] == 0x01, picture);
continue;
}
string text = characterRun.Text;
byte[] textByte = System.Text.Encoding.GetEncoding("iso-8859-1").GetBytes(text);
//if ( text.getBytes().length == 0 )
if (textByte.Length == 0)
{
continue;
}
if (characterRun.IsSpecialCharacter())
{
if (text[0] == SPECCHAR_AUTONUMBERED_FOOTNOTE_REFERENCE && (wordDocument is HWPFDocument))
{
HWPFDocument doc = (HWPFDocument)wordDocument;
ProcessNoteAnchor(doc, characterRun, block);
continue;
}
if (text[0] == SPECCHAR_DRAWN_OBJECT &&
(wordDocument is HWPFDocument))
{
HWPFDocument doc = (HWPFDocument)wordDocument;
ProcessDrawnObject(doc, characterRun, block);
continue;
}
if (characterRun.IsOle2() && (wordDocument is HWPFDocument))
{
HWPFDocument doc = (HWPFDocument)wordDocument;
ProcessOle2(doc, characterRun, block);
continue;
}
}
if (textByte[0] == FIELD_BEGIN_MARK)
//if ( text.getBytes()[0] == FIELD_BEGIN_MARK )
{
if (wordDocument is HWPFDocument)
{
Field aliveField = ((HWPFDocument)wordDocument).GetFields().GetFieldByStartOffset(
FieldsDocumentPart.MAIN, characterRun.StartOffset);
if (aliveField != null)
{
ProcessField(((HWPFDocument)wordDocument), range,
currentTableLevel, aliveField, block);
int continueAfter = aliveField.GetFieldEndOffset();
while (c < range.NumCharacterRuns &&
range.GetCharacterRun(c).EndOffset <= continueAfter)
{
c++;
}
if (c < range.NumCharacterRuns)
{
c--;
}
continue;
}
}
int skipTo = TryDeadField(wordDocument, range,
currentTableLevel, c, block);
if (skipTo != c)
{
c = skipTo;
continue;
}
continue;
}
if (textByte[0] == FIELD_SEPARATOR_MARK)
{
// shall not appear without FIELD_BEGIN_MARK
continue;
}
if (textByte[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[text.Length - 1] == BEL_MARK && currentTableLevel != int.MinValue))
{
text = text.Substring(0, text.Length - 1);
}
{
// line breaks
StringBuilder stringBuilder = new StringBuilder();
foreach (char charChar in text.ToCharArray())
{
if (charChar == 11)
{
if (stringBuilder.Length > 0)
{
OutputCharacters(block, characterRun,
stringBuilder.ToString());
stringBuilder.Length = 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.Length = 0;
}
}
haveAnyText |= text.Trim().Length != 0;
}
return(haveAnyText);
}
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);
}
public void TestFastSaved3()
{
HWPFDocument doc = HWPFTestDataSamples.OpenSampleFile("ob_is.doc");
doc.GetPicturesTable().GetAllPictures(); // just check that we do not throw Exception
}