VAR/VAR.PdfTools
1
1
Fork 0
Code Issues 2 Pull Requests Packages Projects Releases Wiki Activity

Robust stream parsing

Browse Source
This commit is contained in:
Valeriano A.R.
2016-06-21 16:33:23 +02:00
parent b1eecbf902
commit 5e96ee22d8
4 changed files with 130 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
VAR.PdfTools/PdfDocument.cs
View File

@@ -210,7 +210,7 @@ namespace VAR.PdfTools
var parser = new PdfParser(data); var parser = new PdfParser(data);
do do
{ {
PdfObject obj = parser.ParseObject(); PdfObject obj = parser.ParseObject(doc.Objects);
if (obj != null) if (obj != null)
{ {
if (obj.Data is PdfStream) if (obj.Data is PdfStream)

20
VAR.PdfTools/PdfElements.cs
View File

@@ -168,6 +168,10 @@ namespace VAR.PdfTools
{ {
public static double GetReal(IPdfElement elem, double defaultValue) public static double GetReal(IPdfElement elem, double defaultValue)
{ {
if(elem == null)
{
return defaultValue;
}
if (elem is PdfInteger) if (elem is PdfInteger)
{ {
return ((PdfInteger)elem).Value; return ((PdfInteger)elem).Value;
@@ -178,5 +182,21 @@ namespace VAR.PdfTools
} }
return defaultValue; return defaultValue;
} }
public static long GetInt(IPdfElement elem, long defaultValue)
{
if (elem == null)
{
return defaultValue;
}
if (elem is PdfInteger)
{
return ((PdfInteger)elem).Value;
}
if (elem is PdfReal)
{
return (long)((PdfReal)elem).Value;
}
return defaultValue;
}
} }
} }

7
VAR.PdfTools/PdfFont.cs
View File

@@ -15,6 +15,8 @@ namespace VAR.PdfTools
private double _height = 1.0; private double _height = 1.0;
private bool _tainted = false;
#endregion #endregion
#region Properties #region Properties
@@ -23,6 +25,8 @@ namespace VAR.PdfTools
public double Height { get { return _height; } } public double Height { get { return _height; } }
public bool Tainted { get { return _tainted; } }
#endregion #endregion
#region Life cycle #region Life cycle
@@ -33,7 +37,8 @@ namespace VAR.PdfTools
string type = baseData.GetParamAsString("Type"); string type = baseData.GetParamAsString("Type");
if (type != "Font") if (type != "Font")
{ {
throw new Exception(string.Format("PdfFont: Expected dictionary of type:\"Font\". Found: {0}", type)); // NOTE: Type="Font" is Required by the standard, continuing anyway
_tainted = true;
} }
if (baseData.Values.ContainsKey("ToUnicode")) if (baseData.Values.ContainsKey("ToUnicode"))

111
VAR.PdfTools/PdfParser.cs
View File

@@ -90,7 +90,7 @@ namespace VAR.PdfTools
} }
position++; position++;
} while (position < _stream.Length); } while (position < _stream.Length);
return 0; return -1;
} }
private byte PeekChar() private byte PeekChar()
@@ -197,6 +197,30 @@ namespace VAR.PdfTools
} }
} }
private void SkipWhitespaceBack()
{
while (IsWhitespace(PeekChar()))
{
if (_streamPosition == 0)
{
break;
}
_streamPosition--;
}
}
private void SkipDigitsBack()
{
while (IsDigit(PeekChar()))
{
if (_streamPosition == 0)
{
break;
}
_streamPosition--;
}
}
private void SkipEndOfLine() private void SkipEndOfLine()
{ {
byte lineFeed = 0x0A; byte lineFeed = 0x0A;
@@ -718,11 +742,23 @@ namespace VAR.PdfTools
return Encoding.BigEndianUnicode.GetString(byteArray); return Encoding.BigEndianUnicode.GetString(byteArray);
} }
public IPdfElement SearchObjectID(List<PdfObject> knownObjects, long objectID)
{
foreach (PdfObject obj in knownObjects)
{
if (obj.ObjectID == objectID)
{
return obj.Data;
}
}
return null;
}
#endregion #endregion
#region Public methods #region Public methods
public PdfObject ParseObject() public PdfObject ParseObject(List<PdfObject> knownObjects)
{ {
PdfObject obj = null; PdfObject obj = null;
long startPosition = _streamPosition; long startPosition = _streamPosition;
@@ -757,12 +793,19 @@ namespace VAR.PdfTools
throw new Exception(string.Format("Stream after a not dictionary element at: {0}", _streamPosition)); throw new Exception(string.Format("Stream after a not dictionary element at: {0}", _streamPosition));
} }
SkipEndOfLine(); SkipEndOfLine();
long length;
if (streamDict.Values.ContainsKey("Length") && streamDict.Values["Length"] is PdfInteger) // Find the length of the stream
long length = -1;
if (streamDict.Values.ContainsKey("Length") )
{ {
length = ((PdfInteger)streamDict.Values["Length"]).Value; length = PdfElementUtils.GetInt(streamDict.Values["Length"], -1);
if (length == -1 && streamDict.Values["Length"] is PdfObjectReference)
{
IPdfElement lenghtObj = SearchObjectID(knownObjects, ((PdfObjectReference) streamDict.Values["Length"]).ObjectID);
length = PdfElementUtils.GetInt(lenghtObj, -1);
}
} }
else if(length == -1)
{ {
byte lineFeed = 0x0A; byte lineFeed = 0x0A;
byte carriageReturn = 0x0D; byte carriageReturn = 0x0D;
@@ -773,6 +816,8 @@ namespace VAR.PdfTools
new char[] {'e', 'n', 'd', 's', 't', 'r', 'e', 'a', 'm', (char)carriageReturn, (char)lineFeed}, new char[] {'e', 'n', 'd', 's', 't', 'r', 'e', 'a', 'm', (char)carriageReturn, (char)lineFeed},
}); });
} }
// Get the stream
byte[] streamBody = GetRawData(length); byte[] streamBody = GetRawData(length);
SkipEndOfLine(); SkipEndOfLine();
endToken = ParseToken(); endToken = ParseToken();
@@ -857,7 +902,58 @@ namespace VAR.PdfTools
SkipWhitespace(); SkipWhitespace();
continue; continue;
} }
throw new Exception(string.Format("Expected objectID at {0}", startPosition));
// Try to find an object marker
byte lineFeed = 0x0A;
byte carriageReturn = 0x0D;
long distToObject = MeasureToMarkers(new char[][] {
new char[] {' ', 'o', 'b', 'j', (char)lineFeed},
new char[] {' ', 'o', 'b', 'j', (char)carriageReturn, (char)lineFeed},
});
if (distToObject > 0)
{
// Object marker found, backtrack and retry
long originalPosition = _streamPosition;
_streamPosition += distToObject;
long marker = _streamPosition;
SkipWhitespaceBack();
if (_streamPosition == marker)
{
// Abort backtrack, skip garbage
_streamPosition = originalPosition + distToObject + 4;
continue;
}
marker = _streamPosition;
SkipDigitsBack();
if (_streamPosition == marker)
{
// Abort backtrack, skip garbage
_streamPosition = originalPosition + distToObject + 4;
continue;
}
marker = _streamPosition;
SkipWhitespaceBack();
if (_streamPosition == marker)
{
// Abort backtrack, skip garbage
_streamPosition = originalPosition + distToObject + 4;
continue;
}
marker = _streamPosition;
SkipDigitsBack();
if (_streamPosition == marker)
{
// Abort backtrack, skip garbage
_streamPosition = originalPosition + distToObject + 4;
continue;
}
NextChar();
}
else
{
// No more obj markers found, abort all.
_streamPosition = _stream.Length;
}
} }
} while (IsEndOfStream() == false); } while (IsEndOfStream() == false);
return obj; return obj;
@@ -919,7 +1015,6 @@ namespace VAR.PdfTools
if (string.IsNullOrEmpty(token)) if (string.IsNullOrEmpty(token))
{ {
break; break;
//throw new Exception(string.Format("ParseContet: Expected token found nothing, at: {0}", _streamPosition));
} }
PdfContentAction action = new PdfContentAction(token, elems); PdfContentAction action = new PdfContentAction(token, elems);
elems = new List<IPdfElement>(); elems = new List<IPdfElement>();