/// <summary>Parses an inline image from the provided content parser.</summary>
/// <remarks>
/// Parses an inline image from the provided content parser. The parser must be positioned immediately following the BI operator in the content stream.
/// The parser will be left with current position immediately following the EI operator that terminates the inline image
/// </remarks>
/// <param name="ps">the content parser to use for reading the image.</param>
/// <param name="colorSpaceDic">a color space dictionary</param>
/// <returns>the parsed image</returns>
/// <exception cref="System.IO.IOException">if anything goes wring with the parsing</exception>
/// <exception cref="InlineImageParseException">if parsing of the inline image failed due to issues specific to inline image processing
/// </exception>
public static PdfStream Parse(PdfCanvasParser ps, PdfDictionary colorSpaceDic)
{
PdfDictionary inlineImageDict = ParseDictionary(ps);
byte[] samples = ParseSamples(inlineImageDict, colorSpaceDic, ps);
PdfStream inlineImageAsStreamObject = new PdfStream(samples);
inlineImageAsStreamObject.PutAll(inlineImageDict);
return(inlineImageAsStreamObject);
}
/// <summary>Parses the next inline image dictionary from the parser.</summary>
/// <remarks>
/// Parses the next inline image dictionary from the parser. The parser must be positioned immediately following the BI operator.
/// The parser will be left with position immediately following the whitespace character that follows the ID operator that ends the inline image dictionary.
/// </remarks>
/// <param name="ps">the parser to extract the embedded image information from</param>
/// <returns>the dictionary for the inline image, with any abbreviations converted to regular image dictionary keys and values
/// </returns>
/// <exception cref="System.IO.IOException">if the parse fails</exception>
private static PdfDictionary ParseDictionary(PdfCanvasParser ps)
{
// by the time we get to here, we have already parsed the BI operator
PdfDictionary dict = new PdfDictionary();
for (PdfObject key = ps.ReadObject(); key != null && !"ID".Equals(key.ToString()); key = ps.ReadObject())
{
PdfObject value = ps.ReadObject();
PdfName resolvedKey = inlineImageEntryAbbreviationMap.Get((PdfName)key);
if (resolvedKey == null)
{
resolvedKey = (PdfName)key;
}
dict.Put(resolvedKey, GetAlternateValue(resolvedKey, value));
}
int ch = ps.GetTokeniser().Read();
if (!PdfTokenizer.IsWhitespace(ch))
{
throw new InlineImageParsingUtils.InlineImageParseException(PdfException.UnexpectedCharacter1FoundAfterIDInInlineImage
).SetMessageParams(ch);
}
return(dict);
}
public virtual void InnerArraysInContentStreamTest()
{
String inputFileName = sourceFolder + "innerArraysInContentStream.pdf";
PdfDocument pdfDocument = new PdfDocument(new PdfReader(inputFileName));
byte[] docInBytes = pdfDocument.GetFirstPage().GetContentBytes();
RandomAccessSourceFactory factory = new RandomAccessSourceFactory();
PdfTokenizer tokeniser = new PdfTokenizer(new RandomAccessFileOrArray(factory.CreateSource(docInBytes)));
PdfResources resources = pdfDocument.GetPage(1).GetResources();
PdfCanvasParser ps = new PdfCanvasParser(tokeniser, resources);
IList <PdfObject> actual = ps.Parse(null);
IList <PdfObject> expected = new List <PdfObject>();
expected.Add(new PdfString("Cyan"));
expected.Add(new PdfArray(new int[] { 1, 0, 0, 0 }));
expected.Add(new PdfString("Magenta"));
expected.Add(new PdfArray(new int[] { 0, 1, 0, 0 }));
expected.Add(new PdfString("Yellow"));
expected.Add(new PdfArray(new int[] { 0, 0, 1, 0 }));
PdfArray cmpArray = new PdfArray(expected);
NUnit.Framework.Assert.IsTrue(new CompareTool().CompareArrays(cmpArray, (((PdfDictionary)actual[1]).GetAsArray
(new PdfName("ColorantsDef")))));
}
/// <summary>
/// Parses the samples of the image from the underlying content parser, accounting for filters
/// The parser must be positioned immediately after the ID operator that ends the inline image's dictionary.
/// </summary>
/// <remarks>
/// Parses the samples of the image from the underlying content parser, accounting for filters
/// The parser must be positioned immediately after the ID operator that ends the inline image's dictionary.
/// The parser will be left positioned immediately following the EI operator.
/// <b>Note:</b>This implementation does not actually apply the filters at this time
/// </remarks>
/// <param name="imageDictionary">the dictionary of the inline image</param>
/// <param name="ps">the content parser</param>
/// <returns>the samples of the image</returns>
/// <exception cref="System.IO.IOException">if anything bad happens during parsing</exception>
private static byte[] ParseSamples(PdfDictionary imageDictionary, PdfDictionary colorSpaceDic, PdfCanvasParser
ps)
{
// by the time we get to here, we have already parsed the ID operator
if (!imageDictionary.ContainsKey(PdfName.Filter) && ImageColorSpaceIsKnown(imageDictionary, colorSpaceDic)
)
{
return(ParseUnfilteredSamples(imageDictionary, colorSpaceDic, ps));
}
// read all content until we reach an EI operator surrounded by whitespace.
// The following algorithm has two potential issues: what if the image stream
// contains <ws>EI<ws> ?
// Plus, there are some streams that don't have the <ws> before the EI operator
// it sounds like we would have to actually decode the content stream, which
// I'd rather avoid right now.
MemoryStream baos = new MemoryStream();
MemoryStream accumulated = new MemoryStream();
int ch;
int found = 0;
PdfTokenizer tokeniser = ps.GetTokeniser();
while ((ch = tokeniser.Read()) != -1)
{
if (found == 0 && PdfTokenizer.IsWhitespace(ch))
{
found++;
accumulated.Write(ch);
}
else
{
if (found == 1 && ch == 'E')
{
found++;
accumulated.Write(ch);
}
else
{
if (found == 1 && PdfTokenizer.IsWhitespace(ch))
{
// this clause is needed if we have a white space character that is part of the image data
// followed by a whitespace character that precedes the EI operator. In this case, we need
// to flush the first whitespace, then treat the current whitespace as the first potential
// character for the end of stream check. Note that we don't increment 'found' here.
baos.Write(accumulated.ToArray());
accumulated.JReset();
accumulated.Write(ch);
}
else
{
if (found == 2 && ch == 'I')
{
found++;
accumulated.Write(ch);
}
else
{
if (found == 3 && PdfTokenizer.IsWhitespace(ch))
{
byte[] tmp = baos.ToArray();
if (InlineImageStreamBytesAreComplete(tmp, imageDictionary))
{
return(tmp);
}
baos.Write(accumulated.ToArray());
accumulated.JReset();
baos.Write(ch);
found = 0;
}
else
{
baos.Write(accumulated.ToArray());
accumulated.JReset();
baos.Write(ch);
found = 0;
}
}
}
}
}
}
throw new InlineImageParsingUtils.InlineImageParseException(PdfException.CannotFindImageDataOrEI);
}
/// <summary>Parses the samples of the image from the underlying content parser, ignoring all filters.</summary>
/// <remarks>
/// Parses the samples of the image from the underlying content parser, ignoring all filters.
/// The parser must be positioned immediately after the ID operator that ends the inline image's dictionary.
/// The parser will be left positioned immediately following the EI operator.
/// This is primarily useful if no filters have been applied.
/// </remarks>
/// <param name="imageDictionary">the dictionary of the inline image</param>
/// <param name="ps">the content parser</param>
/// <returns>the samples of the image</returns>
/// <exception cref="System.IO.IOException">if anything bad happens during parsing</exception>
private static byte[] ParseUnfilteredSamples(PdfDictionary imageDictionary, PdfDictionary colorSpaceDic, PdfCanvasParser
ps)
{
// special case: when no filter is specified, we just read the number of bits
// per component, multiplied by the width and height.
if (imageDictionary.ContainsKey(PdfName.Filter))
{
throw new ArgumentException("Dictionary contains filters");
}
PdfNumber h = imageDictionary.GetAsNumber(PdfName.Height);
int bytesToRead = ComputeBytesPerRow(imageDictionary, colorSpaceDic) * h.IntValue();
byte[] bytes = new byte[bytesToRead];
PdfTokenizer tokeniser = ps.GetTokeniser();
int shouldBeWhiteSpace = tokeniser.Read();
// skip next character (which better be a whitespace character - I suppose we could check for this)
// from the PDF spec: Unless the image uses ASCIIHexDecode or ASCII85Decode as one of its filters, the ID operator shall be followed by a single white-space character, and the next character shall be interpreted as the first byte of image data.
// unfortunately, we've seen some PDFs where there is no space following the ID, so we have to capture this case and handle it
int startIndex = 0;
if (!PdfTokenizer.IsWhitespace(shouldBeWhiteSpace) || shouldBeWhiteSpace == 0)
{
// tokeniser treats 0 as whitespace, but for our purposes, we shouldn't
bytes[0] = (byte)shouldBeWhiteSpace;
startIndex++;
}
for (int i = startIndex; i < bytesToRead; i++)
{
int ch = tokeniser.Read();
if (ch == -1)
{
throw new InlineImageParsingUtils.InlineImageParseException(PdfException.EndOfContentStreamReachedBeforeEndOfImageData
);
}
bytes[i] = (byte)ch;
}
PdfObject ei = ps.ReadObject();
if (!ei.ToString().Equals("EI"))
{
// Some PDF producers seem to add another non-whitespace character after the image data.
// Let's try to handle that case here.
PdfObject ei2 = ps.ReadObject();
if (!ei2.ToString().Equals("EI"))
{
throw new InlineImageParsingUtils.InlineImageParseException(PdfException.OperatorEINotFoundAfterEndOfImageData
);
}
}
return(bytes);
}
/// <summary>
/// Parses the samples of the image from the underlying content parser, accounting for filters
/// The parser must be positioned immediately after the ID operator that ends the inline image's dictionary.
/// </summary>
/// <remarks>
/// Parses the samples of the image from the underlying content parser, accounting for filters
/// The parser must be positioned immediately after the ID operator that ends the inline image's dictionary.
/// The parser will be left positioned immediately following the EI operator.
/// <b>Note:</b>This implementation does not actually apply the filters at this time
/// </remarks>
/// <param name="imageDictionary">the dictionary of the inline image</param>
/// <param name="ps">the content parser</param>
/// <returns>the samples of the image</returns>
private static byte[] ParseSamples(PdfDictionary imageDictionary, PdfDictionary colorSpaceDic, PdfCanvasParser
ps)
{
// by the time we get to here, we have already parsed the ID operator
if (!imageDictionary.ContainsKey(PdfName.Filter) && ImageColorSpaceIsKnown(imageDictionary, colorSpaceDic)
)
{
return(ParseUnfilteredSamples(imageDictionary, colorSpaceDic, ps));
}
// read all content until we reach an EI operator followed by whitespace.
// then decode the content stream to check that bytes that were parsed are really all image bytes
MemoryStream baos = new MemoryStream();
int ch;
int found = 0;
PdfTokenizer tokeniser = ps.GetTokeniser();
while ((ch = tokeniser.Read()) != -1)
{
if (ch == 'E')
{
// probably some bytes were preserved so write them
baos.Write(EI, 0, found);
// just preserve 'E' and do not write it immediately
found = 1;
}
else
{
if (found == 1 && ch == 'I')
{
// just preserve 'EI' and do not write it immediately
found = 2;
}
else
{
if (found == 2 && PdfTokenizer.IsWhitespace(ch))
{
byte[] tmp = baos.ToArray();
if (InlineImageStreamBytesAreComplete(tmp, imageDictionary))
{
return(tmp);
}
}
// probably some bytes were preserved so write them
baos.Write(EI, 0, found);
baos.Write(ch);
found = 0;
}
}
}
throw new InlineImageParsingUtils.InlineImageParseException(PdfException.CannotFindImageDataOrEI);
}