public void ReaderEscapesUnexpectedObject()
{
const string s = @"%PDF-1.7
abcd
1 0 obj
<< /Type /Any >>
endobj
%AZ 0 obj
11 0 obj
769
endobj
%%EOF";
var bytes = new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(s));
var locations = BruteForceSearcher.GetObjectLocations(bytes);
Assert.Equal(2, locations.Count);
var expectedLocations = new long[]
{
s.IndexOf("1 0 obj", StringComparison.OrdinalIgnoreCase),
s.IndexOf("11 0 obj", StringComparison.OrdinalIgnoreCase)
};
Assert.Equal(expectedLocations, locations.Values);
}
public void ReaderEscapesUnexpectedGenerationNumber()
{
const string s = @"%PDF-2.0
abcdefghijklmnop
1 0 obj
256
endobj
16-0 obj
5 0 obj
<< /IsEmpty false >>
endobj";
var bytes = new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(s));
var locations = BruteForceSearcher.GetObjectLocations(bytes);
Assert.Equal(2, locations.Count);
var expectedLocations = new long[]
{
s.IndexOf("1 0 obj", StringComparison.OrdinalIgnoreCase),
s.IndexOf("5 0 obj", StringComparison.OrdinalIgnoreCase)
};
Assert.Equal(expectedLocations, locations.Values);
}
public void BruteForceSearcherFileOffsetsCorrect()
{
using (var fs = File.OpenRead(IntegrationHelpers.GetDocumentPath("Single Page Simple - from inkscape.pdf")))
{
var bytes = new StreamInputBytes(fs);
var searcher = new BruteForceSearcher(bytes);
var locations = searcher.GetObjectLocations();
Assert.Equal(13, locations.Count);
Assert.Equal(6183, locations[new IndirectReference(1, 0)]);
Assert.Equal(244, locations[new IndirectReference(2, 0)]);
Assert.Equal(15, locations[new IndirectReference(3, 0)]);
Assert.Equal(222, locations[new IndirectReference(4, 0)]);
Assert.Equal(5766, locations[new IndirectReference(5, 0)]);
Assert.Equal(353, locations[new IndirectReference(6, 0)]);
Assert.Equal(581, locations[new IndirectReference(7, 0)]);
Assert.Equal(5068, locations[new IndirectReference(8, 0)]);
Assert.Equal(5091, locations[new IndirectReference(9, 0)]);
var s = GetStringAt(bytes, locations[new IndirectReference(3, 0)]);
Assert.StartsWith("3 0 obj", s);
}
}
public void BruteForceSearcherFileOffsetsCorrectOpenOffice()
{
var bytes = new ByteArrayInputBytes(File.ReadAllBytes(IntegrationHelpers.GetDocumentPath("Single Page Simple - from open office.pdf")));
var locations = BruteForceSearcher.GetObjectLocations(bytes);
Assert.Equal(13, locations.Count);
Assert.Equal(17, locations[new IndirectReference(1, 0)]);
Assert.Equal(249, locations[new IndirectReference(2, 0)]);
Assert.Equal(14291, locations[new IndirectReference(3, 0)]);
Assert.Equal(275, locations[new IndirectReference(4, 0)]);
Assert.Equal(382, locations[new IndirectReference(5, 0)]);
Assert.Equal(13283, locations[new IndirectReference(6, 0)]);
Assert.Equal(13309, locations[new IndirectReference(7, 0)]);
Assert.Equal(13556, locations[new IndirectReference(8, 0)]);
Assert.Equal(13926, locations[new IndirectReference(9, 0)]);
Assert.Equal(14183, locations[new IndirectReference(10, 0)]);
Assert.Equal(14224, locations[new IndirectReference(11, 0)]);
Assert.Equal(14428, locations[new IndirectReference(12, 0)]);
Assert.Equal(14488, locations[new IndirectReference(13, 0)]);
var s = GetStringAt(bytes, locations[new IndirectReference(12, 0)]);
Assert.StartsWith("12 0 obj", s);
}
public bool TryGetOffset(IndirectReference reference, out long offset)
{
if (!loadedFromTable)
{
var table = crossReferenceTable.Invoke();
if (table != null)
{
foreach (var objectOffset in table.ObjectOffsets)
{
offsets[objectOffset.Key] = objectOffset.Value;
}
loadedFromTable = true;
}
}
if (offsets.TryGetValue(reference, out offset))
{
return(true);
}
var locations = searcher.GetObjectLocations();
if (locations.TryGetValue(reference, out offset))
{
return(true);
}
return(false);
}
public CosBase Parse(IRandomAccessRead reader, long objectNumber, int objectGeneration,
CosObjectPool pool, CrossReferenceTable crossReferenceTable,
BruteForceSearcher bruteForceSearcher,
bool isLenient,
bool requireExistingObject)
{
if (pool == null)
{
throw new ArgumentNullException(nameof(pool));
}
var key = new CosObjectKey(objectNumber, objectGeneration);
var pdfObject = pool.GetOrCreateDefault(key);
if (pdfObject.GetObject() != null)
{
return(pdfObject.GetObject());
}
if (crossReferenceTable == null)
{
throw new ArgumentNullException(nameof(crossReferenceTable));
}
var offsetOrStreamNumber = TryGet(key, crossReferenceTable.ObjectOffsets);
if (requireExistingObject && (offsetOrStreamNumber == null || offsetOrStreamNumber <= 0))
{
throw new InvalidOperationException("Object must be defined and not compressed: " + key);
}
if (isLenient && offsetOrStreamNumber == null)
{
var locations = bruteForceSearcher.GetObjectLocations();
offsetOrStreamNumber = TryGet(key, locations);
if (offsetOrStreamNumber != null)
{
crossReferenceTable.UpdateOffset(key, offsetOrStreamNumber.Value);
}
}
if (offsetOrStreamNumber == null)
{
return(CosNull.Null);
}
var isCompressedStreamObject = offsetOrStreamNumber <= 0;
if (!isCompressedStreamObject)
{
return(ParseObjectFromFile(offsetOrStreamNumber.Value, reader, key, pool, isLenient));
}
return(ParseCompressedStreamObject(reader, -offsetOrStreamNumber.Value, objectNumber, pool, crossReferenceTable, bruteForceSearcher, isLenient));
}
public void BruteForceSearcherCorrectlyFindsAllObjectsWhenOffset()
{
var input = new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(TestData));
input.Seek(593);
var locations = BruteForceSearcher.GetObjectLocations(input);
Assert.Equal(TestDataOffsets, locations.Values);
}
public void ReaderOnlyCallsOnce()
{
var reader = StringBytesTestConverter.Convert(TestData, false);
var searcher = new BruteForceSearcher(reader.Bytes);
var locations = searcher.GetObjectLocations();
Assert.Equal(4, locations.Count);
var newLocations = searcher.GetObjectLocations();
Assert.Equal(4, locations.Count);
foreach (var keyValuePair in locations)
{
Assert.Contains(newLocations.Keys, x => x.Equals(keyValuePair.Key));
}
}
public void SearcherFindsCorrectObjects()
{
var input = new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(TestData));
var locations = BruteForceSearcher.GetObjectLocations(input);
Assert.Equal(4, locations.Count);
Assert.Equal(TestDataOffsets, locations.Values);
}
private IReadOnlyDictionary <IndirectReference, long> getBFCosObjectOffsets()
{
if (objectKeyOffsets == null)
{
var offsets = bruteForceSearcher.GetObjectLocations();
objectKeyOffsets = offsets;
}
return(objectKeyOffsets);
}
public void ReaderOnlyCallsOnce()
{
var bytes = OtherEncodings.StringAsLatin1Bytes(TestData);
var reader = new ThrowingReader(new RandomAccessBuffer(bytes));
var searcher = new BruteForceSearcher(reader);
var locations = searcher.GetObjectLocations();
Assert.Equal(4, locations.Count);
reader.Throw = true;
var newLocations = searcher.GetObjectLocations();
Assert.Equal(4, locations.Count);
foreach (var keyValuePair in locations)
{
Assert.Contains(newLocations.Keys, x => ReferenceEquals(x, keyValuePair.Key));
}
}
public CosBase Parse(IndirectReference indirectReference, IRandomAccessRead reader, bool isLenientParsing = true, bool requireExistingObject = false)
{
var key = new CosObjectKey(indirectReference.ObjectNumber, indirectReference.Generation);
var pdfObject = objectPool.GetOrCreateDefault(key);
if (pdfObject.GetObject() != null)
{
return(pdfObject.GetObject());
}
var offsetOrStreamNumber = TryGet(key, crossReferenceTable.ObjectOffsets);
if (requireExistingObject && (offsetOrStreamNumber == null || offsetOrStreamNumber <= 0))
{
throw new InvalidOperationException("Object must be defined and not compressed: " + key);
}
if (isLenientParsing && offsetOrStreamNumber == null)
{
var locations = bruteForceSearcher.GetObjectLocations();
offsetOrStreamNumber = TryGet(key, locations);
if (offsetOrStreamNumber != null)
{
crossReferenceTable.UpdateOffset(key, offsetOrStreamNumber.Value);
}
}
if (offsetOrStreamNumber == null)
{
if (isLenientParsing)
{
return(CosNull.Null);
}
throw new InvalidOperationException($"Could not locate the object {key.Number} which was not found in the cross reference table.");
}
var isCompressedStreamObject = offsetOrStreamNumber <= 0;
if (!isCompressedStreamObject)
{
return(ParseObjectFromFile(offsetOrStreamNumber.Value, reader, key, objectPool, isLenientParsing));
}
return(ParseCompressedStreamObject(reader, -offsetOrStreamNumber.Value, indirectReference.ObjectNumber, isLenientParsing));
}
public void SearcherFindsCorrectObjects()
{
var input = new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(TestData));
var searcher = new BruteForceSearcher(input);
var locations = searcher.GetObjectLocations();
Assert.Equal(4, locations.Count);
Assert.Equal(locations.Values, new long[]
{
TestData.IndexOf("2 17 obj", StringComparison.OrdinalIgnoreCase),
TestData.IndexOf("3 0 obj", StringComparison.OrdinalIgnoreCase),
TestData.IndexOf("4 0 obj", StringComparison.OrdinalIgnoreCase),
TestData.IndexOf("5 0 obj", StringComparison.OrdinalIgnoreCase)
});
}
public void SearcherFindsCorrectObjects()
{
var bytes = OtherEncodings.StringAsLatin1Bytes(TestData);
var reader = new RandomAccessBuffer(bytes);
var searcher = new BruteForceSearcher(reader);
var locations = searcher.GetObjectLocations();
Assert.Equal(4, locations.Count);
Assert.Equal(locations.Values, new long[]
{
TestData.IndexOf("2 0 obj", StringComparison.OrdinalIgnoreCase),
TestData.IndexOf("3 0 obj", StringComparison.OrdinalIgnoreCase),
TestData.IndexOf("4 0 obj", StringComparison.OrdinalIgnoreCase),
TestData.IndexOf("5 0 obj", StringComparison.OrdinalIgnoreCase)
});
}
/// <summary>
/// Check that the offsets in the cross reference are correct.
/// </summary>
public static bool ValidateCrossReferenceOffsets(IInputBytes bytes, CrossReferenceTable crossReferenceTable, ILog log,
out IReadOnlyDictionary <IndirectReference, long> actualOffsets)
{
actualOffsets = crossReferenceTable.ObjectOffsets;
if (ValidateXrefOffsets(bytes, crossReferenceTable.ObjectOffsets, log))
{
return(true);
}
var builderOffsets = new Dictionary <IndirectReference, long>();
var bruteForceOffsets = BruteForceSearcher.GetObjectLocations(bytes);
if (bruteForceOffsets.Count > 0)
{
// find all object streams
foreach (var entry in crossReferenceTable.ObjectOffsets)
{
var offset = entry.Value;
if (offset < 0)
{
// Trust stream offsets for now.
// TODO: more validation of streams.
builderOffsets[entry.Key] = entry.Value;
}
foreach (var item in bruteForceOffsets)
{
builderOffsets[item.Key] = item.Value;
}
}
actualOffsets = builderOffsets;
}
return(false);
}
public void ReaderNull_Throws()
{
Action action = () => BruteForceSearcher.GetObjectLocations(null);
Assert.Throws <ArgumentNullException>(action);
}
/// <summary>
/// Check that the offsets in the cross reference are correct.
/// </summary>
public static bool ValidateCrossReferenceOffsets(IInputBytes bytes, CrossReferenceTable crossReferenceTable, ILog log,
out IReadOnlyDictionary <IndirectReference, long> actualOffsets)
{
actualOffsets = crossReferenceTable.ObjectOffsets;
if (ValidateXrefOffsets(bytes, crossReferenceTable.ObjectOffsets, log))
{
return(true);
}
var builderOffsets = new Dictionary <IndirectReference, long>();
var bruteForceOffsets = BruteForceSearcher.GetObjectLocations(bytes);
if (bruteForceOffsets.Count > 0)
{
var objStreams = new List <IndirectReference>();
// find all object streams
foreach (var entry in crossReferenceTable.ObjectOffsets)
{
var offset = entry.Value;
if (offset < 0)
{
var objStream = new IndirectReference(-offset, 0);
if (!objStreams.Contains(objStream))
{
objStreams.Add(new IndirectReference(-offset, 0));
}
}
// remove all found object streams
if (objStreams.Count > 0)
{
foreach (var key in objStreams)
{
if (bruteForceOffsets.ContainsKey(key))
{
// remove all parsed objects which are part of an object stream
//ISet<long> objects = xrefTrailerResolver
// .getContainedObjectNumbers((int)(key.Number));
//foreach (long objNr in objects)
//{
// CosObjectKey streamObjectKey = new CosObjectKey(objNr, 0);
// if (bfCOSObjectKeyOffsets.TryGetValue(streamObjectKey, out long streamObjectOffset) && streamObjectOffset > 0)
// {
// bfCOSObjectKeyOffsets.Remove(streamObjectKey);
// }
//}
}
else
{
// remove all objects which are part of an object stream which wasn't found
//ISet<long> objects = xrefTrailerResolver
// .getContainedObjectNumbers((int)(key.Number));
//foreach (long objNr in objects)
//{
// xrefOffset.Remove(new CosObjectKey(objNr, 0));
//}
}
}
}
foreach (var item in bruteForceOffsets)
{
builderOffsets[item.Key] = item.Value;
}
}
actualOffsets = builderOffsets;
}
return(false);
}