public void AddDifference(PositionalDifference difference)
{
if (MissingBaselineFile || MissingSecondaryFile)
{
throw new Exception("Cant have differences if a file is missing.");
}
_differences.Add(difference);
}
public static PositionalDifference FromJObject(JObject source)
{
int baselineStart = source.GetValue(nameof(BaselineStartPosition)).ToObject <int>();
string baselineData = source.GetValue(nameof(BaselineData)).ToString();
int secondaryStart = source.GetValue(nameof(SecondaryStartPosition)).ToObject <int>();
string secondaryData = source.GetValue(nameof(SecondaryData)).ToString();
int leeway = source.GetValue(nameof(LocationLeeway)).ToObject <int>();
string classificationString = source.GetValue(nameof(ClassificationString)).ToString();
DifferenceDatatype classification = Enum.Parse <DifferenceDatatype>(classificationString);
PositionalDifference difference = new PositionalDifference(baselineStart, baselineData, secondaryStart, secondaryData, classification, leeway);
return(difference);
}
public static FileDifference FromJObject(JObject source)
{
string filename = source.GetValue(nameof(File)).ToString();
FileDifference fileDifference = new FileDifference(filename);
if (source.TryGetValue(nameof(Differences), out JToken differenceJToken))
{
foreach (JObject diffInfo in (JArray)differenceJToken)
{
PositionalDifference positionalDiff = PositionalDifference.FromJObject(diffInfo);
fileDifference.AddDifference(positionalDiff);
}
}
return(fileDifference);
}
public IReadOnlyList <PositionalDifference> Compare()
{
List <PositionalDifference> differenceList = new List <PositionalDifference>();
PositionalDifference endDifference;
bool unhandledDifference = false;
using (FileStream baselineStream = File.Open(_baselineFilePath, FileMode.Open))
using (FileStream secondaryStream = File.Open(_secondaryFilePath, FileMode.Open))
{
_baselineData = new BufferedReadStream(baselineStream);
_secondaryData = new BufferedReadStream(secondaryStream);
do
{
if (_baselineData.TryReadNext(out byte baselineByte) && _secondaryData.TryReadNext(out byte secondaryByte))
{
if (baselineByte != secondaryByte)
{
if (TryFindDifferenceAtPosition(baselineByte, secondaryByte, out PositionalDifference difference))
{
differenceList.Add(difference);
}
else
{
PositionalDifference tooLongDifference = SetupExplicitlyTypedDifference(DifferenceCheckWindowSize, DifferenceCheckWindowSize, baselineByte, secondaryByte, DifferenceDatatype.TooLong, 0);
differenceList.Add(tooLongDifference);
unhandledDifference = true;
}
}
}
} while (!IsDoneReading(out endDifference) && !unhandledDifference);
}
if (endDifference != null)
{
differenceList.Add(endDifference);
}
return(differenceList);
}
private bool IsDoneReading(out PositionalDifference endDifference)
{
if (_baselineData.DoneReading && _secondaryData.DoneReading)
{
endDifference = null;
return(true);
}
else if (_baselineData.DoneReading)
{
string checkPartialEnd = ReadPartialEndData(_secondaryData);
endDifference = new PositionalDifference(_baselineData.LastReadPosition, string.Empty, _secondaryData.LastReadPosition, checkPartialEnd);
return(true);
}
else if (_secondaryData.DoneReading)
{
string baselinePartialEnd = ReadPartialEndData(_baselineData);
endDifference = new PositionalDifference(_baselineData.LastReadPosition, baselinePartialEnd, _secondaryData.LastReadPosition, string.Empty);
return(true);
}
endDifference = null;
return(false);
}
// naive, greedy determination of where a content difference ends.
private bool TryFindDifferenceAtPosition(byte previousBaselineByte, byte previousCheckByte, out PositionalDifference difference)
{
Dictionary <int, List <int> > checkStartToBaselineStartMatches = FindPotentialRealignments();
if (checkStartToBaselineStartMatches.Count == 0)
{
difference = null;
return(false);
}
// the best difference is the one that has the smallest sum difference of start positions (sum of squares may yield a better result, may have to experiment)
int bestStartDelta = int.MaxValue;
int baselineStartOfBestRealign = int.MaxValue;
int checkStartOfBestRealign = int.MaxValue;
foreach (KeyValuePair <int, List <int> > candidate in checkStartToBaselineStartMatches)
{
int checkStartPosition = candidate.Key;
foreach (int baselineStartPosition in candidate.Value)
{
int startDelta = checkStartPosition + baselineStartPosition;
if (startDelta < bestStartDelta)
{
bestStartDelta = startDelta;
baselineStartOfBestRealign = baselineStartPosition;
checkStartOfBestRealign = checkStartPosition;
}
else if ((startDelta == bestStartDelta) && (baselineStartOfBestRealign < baselineStartPosition))
{
baselineStartOfBestRealign = baselineStartPosition;
checkStartOfBestRealign = checkStartPosition;
}
}
}
if (bestStartDelta == int.MaxValue)
{ // shouldn't be possible, but good to check while developing
difference = null;
return(false);
}
difference = SetupDifference(baselineStartOfBestRealign, checkStartOfBestRealign, previousBaselineByte, previousCheckByte);
return(true);
}
